Penutian (and Sino-Tibetan)

[bwald]

----------------------------Original message----------------------------
AMR writes:
 
>Note 4: IE is not a very good standard by which to judge
>language families; nor has it been unchallenged in modern
>times (see Trubetzkoy and his followers again). Uto-aztecan
>is a far better standard and the one I always use.
 
inviting the following questions:
 
Why?  What is wrong with the PRINCIPLES / METHODS upon which IE is based?
 
What is DIFFERENT about the PRINCIPLES / METHODS upon which Uto-Aztecan is
based?
 
In conjunction with the last question, why do you use a standard that is
much less widely known (conceding that your answer to the first and second
questions may be sufficient to answer the third question).
 
A general point on Afro-Asiatic.  First, a disclaimer, which is, I don't
really know much about it. But... from what I have seen and know about
individual languages in various branches (Hausa in Chadic, Somali and Galla
in Cushitic, Arabic, etc. in Semitic, Egyptian and Coptic in Egyptian
...nothing I really know well in Berber just the general outline) AA seems
like a reasonable guess, and a promising HYPOTHESIS that needs a whole lot
more work (and it's getting it).  Just as I read (most of ) Miller's old
book on Altaic and Japanese (without judgment), I felt that a lot could be
learned about these languages and their histories even if the hypothesis
turns out to be wrong, and that that was sufficient reason to ATTEMPT to
TEST the *hypothesis* that they're related, i.e., that the relationship is
DEMONSTRABLE (not necessarily ALREADY demonstrated).  Such larger
considerations as more distant affinities really magnifies the attention
that clearly demonstrable groups get, with respect to their details and
earliest reconstructable characteristics.
 
NB.  My point of view is that we're not "proving" or "disproving"
something.  We're "testing" it.  This emphasizes that we have "tools" for
testing, and they need to be inspected, improved, made as explicit as
possible, and, esp, their weaknesses have to be recognized. "proving" (<
"probing") may not really be different, but it is vaguer, suggesting that
slight of hand,  slight of mind, mental gymnastics, debating skills, etc
can also be involved, and distracts from what really needs to be taken into
account.  So with respect to "tests", Basque-anything doesn't work (yet),
so the conclusion is either "the tools are no good", which is absurd in
view of what else they have accomplished, or forget about Basque-anything
until you have some *tested* tools that do work for it.
 
Returning to AA, I have been to conferences where Chris Ehret has tried to
reconstruct the phoneme inventory of AA.  The typical criticism -- without
necessarily denying the validity or USEFULNESS of the hypothesis -- is that
there are TOO MANY consonants reconstructed (cf. Brugmann's IE), which
weakens the credibility of valid correspondences (i.e., proposed cognates).
The natural problem here would be to try to determine whether certain
conditioning factors have been obscured.  That will naturally take a lot of
work, and if successful, strengthen the hypothesis of cognate-dom.
Meanwhile, stemming from Greenberg's original hypothesis about AA was the
later recognition that OMOTIC, which he assigned to a sub-branch of
Cushitic, may not be Cushitic at all -- and maybe not even AA (I haven't
followed that lately).  Similarly, the earlier notion that one of the
Kordofanian groups might indicate a genetic link between Niger-Congo and
Nilo-Saharan later gave way to general consensus that the key group was
Nilo-Saharan, and that its similarity with a nearby Niger-Congo Kordofanian
group was due to convergence.
 
Such things did not lead to violent emotional outbursts, as far as I know,
but only to appreciation of the original AA scheme and how it has been
helpful to subsequent research in recognizing what is more likely and what
is less likely on the basis of current evidence -- and to Omotic as a
special problem of *special* interest in its own right as well as with
respect to AA.   All this is productive, amd seems to me quite different
from hysteria about whether certain families are or are not related or
relatable to other families.  Hypotheses are hypotheses, not divinities to
be adored, worshipped and celebrated.  Our business is to test them, and in
so doing keep building better tools to perform those tests.  Maybe physics
has reached a point where philosophical considerations have become more
important than more experimental data and methods of eliciting them (many
think so), but I have yet to hear such an argument in linguistics, and am
not inviting it.  (Incidentally, this business about a "floor" -- or is it
"ceiling" -- to time-depth might implicate such a claim, i.e., we can't go
any further, but the qualification could be "with our present agreed-upon
methods".  Enough on that, since it is not clear to me that there  is even
a demonstration that makes such a qualification advisable.)