Retrodictions

From: Gregory L. Eyink
Message: 2368
Date: 2000-05-06

Hello, List!

Some time ago, in the midst of one of his interesting recurrent
debates with Glen, Piotr wrote the following:

>>

From: Piotr Gasiorowski <gpiotr@...>
Date: Thu Mar 30, 2000 5:21am
Subject: Re: IE athematics

As in science in general, there is no single criterion of validity for
a new theory. Cases regarded as convincing typically involve a variety
of arguments converging on a central conclusion. Linguistic
relationship becomes obvious when:

1) A consistent reconstruction of the phonological system and
structure is offered, with correspondence rules which not only relate
the initially compared pairs of alleged cognates but allow you to
detect further such pairs, previously unsuspected of being related.
This is the closest that we can get to empirical testing.

2) The morphological system of the protolanguage is reconstructed,
with the same logical feedback as above. With (1) and (2) done, it
becomes possible to apply internal reconstruction to the protolanguage
itself just to make sure that "tout se tient" (with a bit of luck you
may even discover new patterns in this way, like the laryngeal theory,
make new predictions and verify them again to strengthen your case).

>>

I meant to comment at the time, but got too busy to do so. Better late
than never, I would like now to pick up again that thread.

As a physicist, coming out of a traditional "hard science", I have
always been fascinated with the difficulties in validating the
conclusions of historical linguistics. In physics, we can generally
perform experiments to check the predictions of our theories. Many of
us, at least, look askance at theories (e.g. string theory) which deal
with length or energy scales remotely inaccessible to experimental
test. One may, as many of string theorists insist, use internal
consistency and elegance or economy, in judging theories, but this
only takes one so far. One wants to know the truth, not just
speculations!

In historical linguistics, the best one can hope for, it seems, is to
make successful "retrodictions" rather than predictions. Thus, as new
linguistic data becomes available from archaeological discovery---as
in the cases of Hittite and Tocharian---one may verify the conclusions
of reconstruction. Perhaps the most dramatic example of this I know,
to which Piotr alluded, is Saussure's proposal in his Memoire of 1878
that PIE long vowels had developed from a short vowel plus "sonant
coefficients". It seems to be now generally agreed that his hypothesis
was confirmed after Hittite was discovered. J. Kurylowicz in 1927
pointed out that the Hittite consonants transcribed with "hh"
corresponded in some cognates to the "laryngeals" (as they are now
termed) which Saussure had suggested purely on the basis of
phonological analysis. There seem to be still some dissenters,
but the laryngeal hypothesis is now pretty widely accepted
on the basis of this empirical evidence and internal reconstruction.

Here is a question for the list: are there other striking examples of
successful "retrodictions" in PIE linguistics? I guess that Michael
Ventris' decipherment of Linear B in 1954 provides such an example,
in terms of the reconstructed labiovelars. The syllabary that was used
in Crete after 1450 B.C. for the archaic Mycenaean Greek had special
symbols for syllables "qa","qe","qi","qo" that began with a
reconstructed labiovelar, such as (I believe) "qe-to-ro" for "four".
Possibly Ogham script could provide another example, since it had also
a special symbol for the reconstructed labiovelar "kw" or "q", as in
the common word "maqi", genitive case of "son". However, I do not know
the history well enough to say whether Ogham was an element leading to
the original proposal of labiovelars, or whether it was understood
after the fact to give corroboration. Can anyone say? And, most
importantly, are there other striking examples of such successful
"retrodictions"?

All of this leads to a more weighty question: how can one really hope
to substantiate theories about language before the appearance of ANY
scripts (alphabets, syllabaries, hieroglyphics, etc.)? Such questions
bear directly on the efforts of the Nostraticists. Internal
consistency or logical criteria can certainly help. These days it
seems that cultural archaeology is also being more thoroughly
employed. However, in the end, it must be the case that MANY
possible past histories could account for the linguistic facts that
are directly evidenced, in terms of living languages or ancient
inscriptions. It seems that the best one may be able to do is to make
a catalogue of all possible historical reconstructions with, perhaps,
a weighting as to their relative likelihoods.

Whither then historical linguistics?

Best,

Gregory

P.S. I am leaving tomorrow for a trip of a few days. I'll look forward
to reading responses when I return next week.