Darwin-L Message Log 5:99 (January 1994)

Academic Discussion on the History and Theory of the Historical Sciences

This is one message from the Archives of Darwin-L (1993–1997), a professional discussion group on the history and theory of the historical sciences.

Note: Additional publications on evolution and the historical sciences by the Darwin-L list owner are available on SSRN.

---

<5:99>From sally@pogo.isp.pitt.edu  Tue Jan 18 14:23:25 1994

To: darwin-l@ukanaix.cc.ukans.edu
Subject: Re: Systematics and linguistics
Date: Tue, 18 Jan 94 15:25:52 -0500
From: Sally Thomason <sally@pogo.isp.pitt.edu>

   I'm not sure just why we are confused, Kent Holsinger and I,
about the notions of convergent evolution vs. hybridization --
but it must have to do, at some level, with different ways of
talking about these things in our different disciplines.  I also
don't remember exactly how I phrased my earlier posting about
language mixture (hybridization), but it's quite possible that I
was unclear or even (in my effort to avoid technical linguistic
terms that only linguists could love) misleading.  It's like
this: borrowing between separate languages (as opposed to borrowing
between dialects) usually takes place without causing any
disruption in the family tree -- the main lines of descent are
normally quite clear, provided (and this was what Scott was talking
about) that the relationships are shallow enough chronologically
that you can find systematic correspondences in all grammatical
subsystems, including, crucially, the basic vocabulary.  But
borrowing can and sometimes does produce more extreme effects,
under the right kinds of social circumstances (fairly unusual
social circumstances, as far as historical linguists can tell
-- but stories I've heard lately about language contacts in
East Africa make me a bit dubious about that conventional
wisdom).  There are (in my view) no limits whatsoever on what
CAN be transferred from one lg. to another; I have counterexamples
to all the proposals that have been made along those lines, at
least all the proposals I've seen.  But that's not the usual sort
of situation; and in more ordinary social circumstances, normal
transmission of a whole lg. from one generation to the next gives
you, eventually, a fairly tidy family tree, at time depths up to
-- roughly -- somewhere between 5000 and 10,000 years.

   All that is quite different from convergent evolution.  Basically,
what I think Scott is saying, and what I know I'm saying, is this:
there isn't any convergent evolution in the sense that unrelated lgs.
get more similar, *without* borrowing, to the point where they look
as if they're related.  The reason is that, for such a situation
to exist, you'd have to get convergence in sound/meaning chunks
-- i.e. in vocabulary -- and that just doesn't happen to any
significant degree.  (I haven't seen the dissertation Victor
Golla mentioned, about sound symbolism, but I think there really
is sufficient evidence to rule out sound symbolism as the source
of widespread systematic correspondences in whole words.  I wouldn't
expect the picture to change even iss of sound symbolism has
has been greatly underestimated.)

   It *is* true that typological characteristics -- the kinds of
sound features that I talked about in my previous posting, and also
word-structure and sentence-structure features (e.g. word order,
presence of suffixes rather than prefixes, ....) -- appear widely
in unrelated languages.  It's also true that such similarities
have been taken as evidence for relatedness, for instance in the
Uralic & Altaic languages.  That's more like convergent evolution,
but there's a big difference: as soon as you start looking at
sound/meaning chunks, especially basic vocabulary, the picture
changes dramatically.  (Actually, people still argue about whether
Uralic and Altaic languages are related -- these families include
Finnish and Hungarian among the Uralic lgs., and Turkish and
Mongolian among the Altaic lgs. -- but no one nowadays would take
their word-order patterns, vowel-harmony rules, and other
structural features to be primary evidence for such a relationship.
The vocabularies do not match anywhere near as closely as some
of the structure does.  It's possible that borrowing is the source
of some of those structural similarities, and there are quite a
few people who no longer even believe that Altaic is a valid
family, so there are lots of complications.)

   I'm sorry to have been unclear before, and even sorrier if
(as I suspect) I'm not much clearer this time.  Probably someone
else should take over.  My remaining question would be about what
Kent has in mind in saying "Well, if reticulation isn't the
answer, then convergence is the only alternative" -- my question
is, answer to what?

    Sally Thomason
    sally@pogo.isp.pitt.edu

---

Your Amazon purchases help support this website. Thank you!