🚀 go-pugleaf

I believe that a kind of prediction *is* possible in synchronic
linguistics.  If I make a generalization, it lends itself to a kind of
prediction:  it predicts that no one in the field will find any
counterexamples.  Usually one thinks of negative predictions based on
typological generalization - "no language can exist that contains [y] but
not [u]".  The record has not been generally good for negative predictions
made from the armchair (as it happens, Yeongdinq Koongfu contains [y] but
not [u]), but there have been some triumphs for positive predictions.  Li
Fang Kuei is said to have predicted the existence of a post-velar
fricative initial that was eventually discovered in some Northern Chinese
dialects, based on his knowledge of medieval Chinese phonological
categories (whether he had not actually heard such dialects before is not
honestly known, however).  And Li is also said to have "predicted" the
existence of certain consonant + r clusters in Tai, based on strict
comparative principles, which were later found by William Gedney in Srek
(to be fair, not all of Li's clusters show up in Srek, and Srek has other
clusters that are not accounted for in Li).  

Li's Tai "predictions" were actually made as "reconstructions", and later
interpreted as predictions.  I think it would be valuable to discuss the
meaning of reconstruction.  Does it have actual historical meaning?  Or is
it a mere tool of comparison - a diasystem, a metasystem, a supersystem. 
I lean toward the latter view.  

I am leaving the country - kindly respond to my email account if possible.  


David Prager Branner, Yuen Ren Society
Asian L&L, DO-21, University of Washington
Seattle, WA  98195				<charmii@u.washington.edu>