isolates

[Sarah G. Thomason]

  In discussing the hypothesis of Japanese-Korean relationship,
Alexander Vovin says, rightly, that I shouldn't have referred
to a review without checking the exact reference.  So I have now
checked: the reviewer was David Solnit, the review was of Paul
Benedict's JAPANESE/AUSTRO-TAI (Ann Arbor: Karoma), and the review
appeared in LANGUAGE 68:188-96 (1992).  Solnit concludes his review
as follows: "Finally, the correspondences with Austronesian and with
Altaic, to the extent that both are valid, need to be evaluated and
placed in relation to each other, whether that entails choosing one as
inherited and the other as borrowed, or whether Japanese is one of
those rare cases having in its past a break in normal genetic
transmission."
 
   Larry Trask has already made it clear in his own response to A.V.'s
posting that  not all specialists agree that a relationship between Japanese
and Korean has been established.
 
   A.V. goes on to say that I have grossly misrepresented his own work,
in my reference to his reconstruction of *hd- for a correspondence set
consisting mainly of w's.  My methodological point was that a claim of
genetic relationship that rests in part on such reconstructions is not
one that I, at least, would place very much confidence in.  He asks if
I can offer a better reconstruction for the correspondence set in
question.  I do believe that I can: I propose *w.  The fact that *w would,
on the evidence of the dialects, be a rare Proto-Ainu phoneme doesn't bother
me nearly as much as A.V.'s own proposal.  Here is the relevant material
from his book A RECONSTRUCTION OF PROTO-AINU (Leiden: E.J. Brill, 1993):
 
1.  The correspondence set (which I call large because it ranges over a
    large number of dialects):
 
    w : w : w : w : w : w : w : w : w : w : u-/w- : -G- : v : v : -gu-
 
 
2.  A.V.'s rationale for reconstructing *hd-: He does not reconstruct *w
because "[w] is extremely rare in common Ainu.  It occurs initially in
five words, one of which is doubtful, and in medial position in three
words....the distribution of [w] is quite peculiar -- it occurs mostly
before [a], once before [e], and once before [o]...."  He also considers,
and rejects, reconstructing *hw- or *gw-, saying that "the existence of
such a cluster seems quite unnatural in a system which lacks [w] itself."
Therefore, he continues, "we have to look for some sound which existed in
PA and could produce the [w] sound in the process of development."  He
likes *h for the first segment because it "can be easily reconstructed on
the basis of [two particular dialects, the ones with -G- and -gu-]"; and
he likes *d for the second segment in the cluster because "it could undergo
spirantisation *d > <dh> [eth, sorry no voiced interdental fricative on
my computer], and the shift *<dh> > w is rather probable."
 
   The consonant system that A.V. reconstructs for Proto-Ainu is
/p t k q d g m n s y h r/ (where q = glottal stop).  I no longer have the
book at hand, but I don't recall any elaborate C clusters at all, much
less something as weird as hd-, in any of the dialects.  The sound changes
A.V. posits from this very strange C cluster are themselves not compelling.
Since we are talking about dialects of the same language, there also isn't
much time for the weird cluster to undergo all these INDEPENDENT  changes
in every single dialect....changes that just happen to lead to the very
same rather surprising result in ten dialects, for instance.
 
   I leave it to other HISTLINGers to judge the plausibility of this
reconstruction (and others in the book).
 
  -- Sally Thomason
     sally at isp.pitt.edu
 
P.S. The point I made in the same message  about isolates in remote
mountain valleys was a methodological one, not a claim either about
polygenesis of human language or about the existence of  such cases
...though Basque comes close.  I made no reference to complete
homogeneity of my hypothetical speech community, as A.V. suggests in
his reply.