Why IE is no paragon (WAS: Penutian (and Sino-Tibetan)

[bwald]

----------------------------Original message----------------------------
I found much of value in AMR's answer to my question:
 
>> Why?  What is wrong with the PRINCIPLES / METHODS upon which IE is based?
 
I hope it is further useful to separate the points in AMR's answer that are
most valuable, from those that I think are less valuable.
 
>...the first point is that IE is not really all that widely
>known: I am above all trying to combat the tendency of too many people
>to cite little undigested bits of IE they picked up from textbooks of
>Hist Lx w/o actually anything about IE and IE lx.  If someone is really
>speaking with knowledge about IE, that's fine, of course,except that--
.... the supposed methodological
>standards of IE only exist in some virtual sense, that is, in the
>sense that every IEnist learns to ignore most of the published stuff
>and somehow decides for him/herself what to pay attention to.  But
>this is very tricky, because it is not a matter of public record.
 
OK.  I think that everyone would agree that there are a lot of
controversial aspects to IE reconstruction, and that some familiar
proposals are quite weak.  Since I am more concerned with principles and
methods in general, my point is that those principles and methods which are
part of the consensus are more familiar to most readers in the case of IE
than other language families.  Therefore, that prior familiarity requires
less to be said in preparation to exemplifying a principle (or problem),
if appeal can be made to an example from IE rather than some less widely
known language family of comparable complexity / variability.  Of course,
if some other language family, but not IE, instantiates a general
principle, then that is extremely interesting, and it is worth going into
more detail in preparation to discuss or defend it.  That's what I'm
looking for.  So, let's continue with AMR's reply, since he does try to
address my concerns.
 
>(c) IE is atypical of the problems we face in comparrative and esp.
>classificatory lx because Proto-IE is reckoned by those who like
>to play with numbers (e.g., Watkins) to be less than 2000 years, maybe
>only 1000 years, older than the oldest attested languages (OLd Latin,
>Mycenean Greek, Hittite, Vedic, etc.), so it is a very YOUNG
>family.  Almost every other nontrivial linguistic grouping involves
>much greater time depth.
 
This goes back to my comments about the potential for "time-depth" to be
used as an "excuse" to not even approach the (consensus) standards set by
IE.  The final passage is also contentious, since it implies that some
linguistic groupings are "trivial", which seems to me an expression of
taste, not something worth arguing about.  It also assumes that the
relative time-depth of different groupings can be estimated, an assumption
that, first of all depends on the validity of these groupings to begin
with.  This could lead to circularity, if the implication is that we have
to "lower" our consensus methological standards in order to accomodate such
claims.  I will stick to focussing on the issue of what those consensus
standards might be, and how the problems of other groupings either alter or
amplify them.
 
>(d) IE is atypical also in the sense that it is one of the few
>lg families which was originally established largely if not wholly on
>the basis of MORHOLOGICAL parallels, rather than LEXICAL ones.
>Uto-Aztecan, Finno-Ugric, Samoyedic, Turkic, Mongolic, Tungusic,
>are all examples of the reverse (although Afro-Asiatic is like
>IE in this regard, I think).
 
I am not sure of what line of development of IE reconstruction AMR has in
mind.  The "originally established" suggests the typological approach of
Bopp at the beginning of the 19th c.  However, that is irrelevant to the
current (and long-time) consensus.  The unity (such as it is) of IE reached
consensus on the basis of the reconstruction of a large amount of LEXICAL
material.  The reconstruction of IE morphology has always been more
controversial because sound correspondences have not been sufficient for
reconstructing most of its morphological forms.  At the same time,
the sound correspondences in grammatical morphemes were suggestive of other
principles of linguistic change, paradigm rearrangements and whatever, to
which some attention was focussed and remains in a state of development.
There is still disagreement about whether certain cognate inflections in IE
languages reflect regular sound changes or some other kind of change.
 
 I don't believe the situation is different for IE than for any of the
other language families AMR mentioned above.  Consensus and controversy in
reconstruction reside in the same areas.  Certainly, grammatical morphemes
as well as lexical morphemes have been reconstructed for these groups, and
the problems in reconstructing the grammatical inventory and grammar of one
group are basically comparable to the problems in reconstructing any other
group -- to a degree worthy of further discussion.  (If I remember
correctly almost every consonant that has been reconstructed for
Finno-Ugric, or maybe it was Uralic, has also been reconstructed as a
suffix of one sort or another.)
 
AMR goes on in a longish passage to illustrate various other problems in
the reconstruction of IE.
One, which he exemplifies, but does not identify as a general problem is
the problem of sub-grouping, such that many characteristics attributed to
IE may turn out to be only characteristics of a sub-group of IE.  Again,
this is a common problem for reconstruction of all language families
everywhere, as far as I can tell.  It does not make IE "atypical".  In
short, this is a general problem in reconstruction, and points out a
weakness in the assumptions underlying the application of the classical
comparative method which remains a general problem (generally handled by
reconstructing older areal groupings and applying diffusion / convergence
theories to them).
 
AMR concludes:
> In short, IE lx is not a model for comparative
>lx necessarily, and the IE family is much too shallow (because its
>oldest representatives are so anciently attested) to be a good model
>for other work in the field anyway.  Of course, the BEST work on
>IE is awe-inspiring, but then so is the best work in any other
>language family--and work on much deeper families is I think
>more awe-inspiring still.
 
There are qualifications in this conclusion, but again IE's implied
"trivial" time-depth is appealed to (to be measured against the problems of
accurately sub-grouping its members and revising its earliest
reconstructable state), once again raising my suspicions that an argument
involving lowering standards is lurking beyond the surface arguments.  In
any case, it is indeed the BEST work in IE and in other families that I
would like to see identified for the principles they are based on, and I
would like to know which of those principles apply to the reconstruction
/demonstration of relationship of one family but not another, and why.
 
I appreciate AMR's effort to explain his earlier contention that IE is not
a good reference point for illustrating general linguistic principles /
methods.  However, it remains unclear to me why it is worse as a reference
point than any other family of its own or greater variety (whether that is
due to time-depth or whatever).  All I see so far is that Alexis thinks
using examples from IE to illustrate general points *prejudices* judgments
of what might be possible in reconstruction.  However, I do not see why
such prejudices cannot be overcome with sound arguments, if they exist,
which demonstrate principles which do not apply to IE.  It might be clearer
if AMR responds more directly to my second question, which he quoted at the
outset of the reply:
 
> What is DIFFERENT about the PRINCIPLES / METHODS upon which Uto-Aztecan is
> based?
 
All I understand from his present message is that Uto-Aztecan is based on
LEXICAL comparisons, not ("originally") morphological ones (prefixes and
suffixes, I assume).  However, I questioned the relevance of this claim to
the consensus methods by which IE has been established, and how the methods
used to reconstruct Altaic have failed to reach as general a consensus --
if that's the case.
 
Contrary to what AMR's reply might imply, I am not aware that Altaic is
disputed because its morphological reconstruction is disputed *and its
lexical reconstruction is not*.  And it is certainly not the case that IE
is accepted as a family because it was *originally* recognized on the basis
of a similar scheme of declensions and conjugations among its classical
members.  That turned out to be gravy, and lumpy gravy at that (though
maybe not as lumpy gravy as reconstructing Altaic morphology, not to
mention "Ural-Altaic", or should it be "Japanese-Korean-Altaic"?)
 
Finally, if Uto-Aztecan indeed is a better model for the principles of
reconstruction, does that mean that somehow it informs the reconstruction
of Altaic in a way that IE cannot?
 
P.S.  It's not about ethnocentrically making IE a standard referent point
for methodology, but the convenience of using examples from it to
illustrate a point which the largest number of readers can understand from
their training (esp. if not a matter of great detail).  Then it is very
interesting and COMPREHENSIBLE to criticise a supposed general principle
based on IE as, in fact, not a general principle at all, but one which is
specific to certain (types of) languages.  Again, as far as I can tell, all
controversial language families that have maintained the interest of what
AMR calls "competent linguists" are basically controversial for the same
reasons.