phonetic resemblances

[Sally Thomason]

----------------------------Original message----------------------------


re Alexis Manaster Ramer's comments:

  I wouldn't deny that Joseph Greenberg is a historical linguist
at all -- he did some historical linguistics in the past
(for instance on aspects of Semitic, if I'm remembering right,
and on the history of Bantu tones).  But I would certainly deny
that he has been doing historical linguistics recently; all he's
been doing, at least as reflected by his publications, is inspecting
wordlists and a few affixes for phonetic resemblances and grouping
languages on that basis.  That isn't what historical linguistics is
about.  (Merritt Ruhlen, at least, explicitly distinguishes what he
calls "classification" from historical linguistics, and says that he's
doing classification.  I don't believe Greenberg himself has explicitly
said that he's doing classification rather than historical linguistics,
but he's doing the same thing that Ruhlen, following him, is doing.)

   The problem with Comecrudan is that there's so little data
available for *any* comparison.  Goddard's proposal is actually
quite cautious: "There is a basis, then, for postulating a Comecrudan
family consisting of Comecrudo, Garza, and Mamulique..." (1979:380,
in Campbell & Mithun, eds., The Languages of Native America).  He
clearly isn't claiming that the family is fully established.  But
what's most relevant for Manaster Ramer's comment is that Goddard
does not, repeat not, postulate the family on the basis of
resemblances alone; in fact, in the same passage he argues strongly
for *rejecting* certain groupings based on similarities (on the grounds
that the words are too similar and therefore most likely due to borrowing).
Here's the crucial sentence about Comecrudan: "The case for relating
Garza and Comecrudo seems strong; the agreement in the words for `sun'
and `road' is particularly striking and shows a consistent phonological
pattern suggestive of an historical sound law" (Goddard 1979:380).
 -- that is: a recurring sound correspondence.  He then goes on to say
that "The sparse Mamulique data compare well with Garza and Comecrudo
as far as they go."  And that's the postulated family.  The reason
he only discusses lexical items is that that's all the data there is
-- wordlists.  Campbell, in his 1997 book, accepts Goddard's
classification; he doesn't say anything to suggest that he has carried
out an independent analysis.

   And finally, as Larry Trask and others have pointed out, spotting
phonetic resemblances in lexical items is a common stimulus to a
systematic investigation of the possibility that two or more languages
are related.  So of course I wouldn't quarrel with Hamp's statement that
comparison typically begins with items that are phonetically similar.
The whole point is that no one who is doing historical linguistics would
claim that that's enough to establish genetic relationship.  Greenberg &
Ruhlen do claim that; but their view is, to put it mildly, not popular
among historical linguists (including the Russian Nostraticists, who, as
Manaster Ramer and others have insisted, should not be lumped...no pun
intended...with Greenberg & Ruhlen in their methodology).

  -- Sally Thomason
     sally at thomason.org