Sino-Tibetan

[Ralf-Stefan Georg]

----------------------------Original message----------------------------
At 8:41 Uhr -0500 05.02.1999, Scott DeLancey wrote:

>But if you're looking for a clear example of an indisputable
>family established without any morphological basis, Tai will
>do fine.  Actually, I mentioned this to Goddard once when we
>were disagreeing about exactly this point, and what I remember
>him saying (again, he's not responsible for my memory) was,
>essentially, well then, we can never really be sure that the
>relationship among these languages is genetic.  Since Goddard
>hasn't ever looked at Tai, I suppose I can imagine how he might
>say such a thing, but really--this is a relationship that is
>inspectionally evident.

I think this is important, and the matter of the necessity to have surefire
morphological comparisons to secure a relationship is as well, but I think
some clarifications are possible.
First of all, I may recall (one of) Meillet's famous formulation(s) on the
issue, from his 1914  "Le probleme de la parente des langues", Scientia XV
(accents omitted):

La demonstration de parente est parfaite si l'on peut expliquer par la
transformation des memes elements anciens l'ensemble du systeme grammatical
de deux langues distinctes" Meillet 1914, 93

The crucial word here is, I think, "parfaite", since it seems to imply
that, even for Meillet, "less-than-perfect", but still viable and to a
degree acceptable demonstrations of relationship might exist.
Meillet was of course the leading I.E.'ist of his time, and his experiences
with I.E. are palpably reflected in these words.
I think now, the Tai example doesn't really make a point against morphology
being important, nor should it be interpreted as - since morphology is so
indespensable - that Tai is only a weakly demonstrable language family. It
isn't, and I know since I have spent that half an hour with Li's textbook.
Scott DeLancey says that the relationship is inspectionally obvious, and
this also my impression. Since I unfortunately did no more study of Tai
than that half of an hour, I may be allowed to speculate that this being
inspectionally obvious of the relationship is comparable to that we find
in, say, Slavic or Romance. Both language families actually never did have
to be proposed for the first time - there being no Slavic or Romance Jones,
Bopp, or Sajnovics - there closeness is so great that even prescientific
inspection will reveal that some notion of "relatedness" - without
necessarily a clear idea about how this may have come about, of course - to
even the superficial observer. Semitic may be added to this. And in fact
this notion of relatedness is generally part of the linguistically
untrained speaker of those languages - if at all s/he has access to other
members of these families. I repeat, I don't know whether this is
comparable to the situation in Tai, but I suspect it to be (please, correct
me). In such a case, i.e. with such a close degree of relatedness, the
question of "proof" simply is next to irrelevant, I'd say, i.e. (again I'm
suspecting only that this is the case with Tai) if the amount of sared
lexical items is so overwhelmingly great, and, possibly, long pieces of
texts from lg. A may be made comprehensible for speakers of related lg. B
by pointing out a limited set of sound-laws and some explanations of, say,
divergent syntax.
The more so, if the languages in question do not possess anything in the
way of a large quantity of bound affixes, organized in intricate paradigms.
Again, I take it Tai is more like Chinese in this respect than like
Kiranti, Chukchee or Abkhaz.
The point where morphology does gain some importance, and considerable
importance, is where a) the relationship is not close enough to be
inspectionally obvious, and b) where the languages in question *do* possess
such a morphological system.
All I want to say at this stage is that, with languages of this kind, I'd
expect any claim of relatedness to tell me at least something (the more the
better) about the coming-about of these system, since *explaining* things
we see is what making hypotheses is all about. I short: we don't need much
morphological arguments for any claim of relatedness if

a) the languages are as closely related as, say, Slavic, Romance or maybe
Tai (for argument's sake, please overlook here that, in the case of Slavic
and -less so - Romace, common morphology is of course present and
definitely part of the impression of relatedness it makes for the untrained
eye - but I suspect it could do without); just as a whim of the moment
(please, don't press me anyone on it) : with Semitic, morphology is so
close that we perhaps could do without vocabulary ????

b): a) holds and the languages don't have any morphology worth speaking of
anyway

BUT: if a) doesn't hold and b) doesn't hold either, morphology should play
a more important role, at least in the sense of Meillet's formulation.
Otherwise, an inevitable consequence would be to say that, while the
relationship seems to be sure, all the bound morphology of the languages
has been developed after the split-up of the parent language, which then
inevitably has to be dated far back in time. Such a scenario is certainly
not impossible, one may at times be forced to say this, and believe it,
too. But such a demonstration of relationship would be - no, not wrong,
misguided, or nonsensical - it would only not so perfect as in other cases.
This may even mean that proper reconstruction of the parent language may be
so difficult as to border on being impossible (though I don't want to
verdict this); S-T may, just may, be an example for this. While languages
may be related *closely* or *distantly*  ( as Slavic lgs. are surely closer
related that I.E. lgs.), a notion of being *more* or *less* rlated seems to
make no sense. What does seem to make sense, though, is a notion of *more*
or *le* *transparently* related, and for the determination of the latter,
morphology - if present in the languages - does play an important role.

St.G.

Stefan Georg
Heerstrasse 7
D-53111 Bonn
FRG
+49-228-69-13-32