Conservative dilemma

[Larry Trask]

On Thu, 9 Sep 1999, Herb Stahlke wrote:

> Unfortunately nearly all of the reference to Greenberg's work in
> this discussion has been to his Language in the Americas.  His
> earlier work in Africa not only established at least two previously
> unknown genetic gouping, Niger-Congo and Afro-Asiatic, but has stood
> the test of four decades of research.

Yes, of course.  Curiously, I have on occasion found myself defending
Greenberg's African work against what I saw as unfair criticism of it.

However, I would maintain that the linguistic position in Africa is (and
was) really rather different from the position obtaining in the other
parts of the world to which Greenberg has directed his attention, in
that major groupings were already well established before Greenberg came
along.

Maybe it's a little generous to give Greenberg credit for establishing
Afro-Asiatic, whose unity was already widely recognized.  True, G
clarified its membership and its internal structure, but he didn't
invent the grouping.

> The Afro-Asiatic classification has been modified largely by
> Bender's work on Omotic, and Niger-Congo has seen some interesting
> internal restructuring, most of it reported in Bendor-Samuel's
> (1989) The Niger-Congo Languages.  In the Niger-Congo case, the
> restructuring has confirmed doubts that Greenberg expressed about
> some of the subgrouping, e.g., the status of Benue-Congo and Kwa.
> However, before Greenberg, Mande and West Atlantic were not commonly
> thought to be part of the same group that contained B-C and Kwa, and
> the Adamawa-Eastern languages and the Kordofanian languages had
> simply been dropped in the geographical/ethnic catch-all "Sudanic".

Perhaps G's major achievement in Africa was in clarifying the positions
of the assorted `Sudanic' languages, at least to the extent of providing
better working hypotheses.  But, even today, it remains debatable
whether all the languages assigned by G to his Kordofanian group really
belong to Niger-Congo, or even whether they constitute a valid taxon at
all.

> The full Nilo-Saharan, especially the inclusion of Songhay, remains
> uncertain,

Yes, agreed.  I think Nilo-Saharan is best regarded as a geographical
`residue' grouping, containing all the languages in the area that don't
appear to belong to one of the neighboring large families.  Some of the
Nilo-Saharan languages are certainly related, but it has not been shown
that they all are, and I suggest that Nilo-Saharan should be regarded
as, at best, a working hypothesis, and not as an established family.

I have been told that G came very close to positing no fewer than ten
distinct families before finally tossing all ten into his `Nilo-Saharan'
pot.

> and the inclusion of Hatsa and Sandawe into Khoisan is at about the
> same level.

Or perhaps even more doubtful.  `Khoisan' is pretty clearly another
residue group, containing four units no two of which have yet been shown
to be related.

> However, the Hamitic of Nilo-Hamitic is something Greenberg
> effectively debunked.

Agreed.

> Perhaps this is why many Africanists are bemused at the intensity of
> the Americanist reaction to Greenberg's work.

G gave the Africanists something that was better than what they had
before, as perhaps all Africanists would agree.  But it is hard to see
that G has done as much for the Americanists.  If we put aside
Eskimo-Aleut and Na-Dene, both of which were already widely accepted
anyway, then G found the Americanists with 140-odd secure families and a
number of ambitious but unsubstantiated proposals of larger groupings,
and he gave them a single monster family instead -- much as had earlier
done in the Papuan case.  It is perhaps not so surprising that the
Americanists have failed to see this as a step forward.

> As to Greenberg's alleged absolutism in his claims of relationship,
> what is relevant is what the field does with his work, not what he
> thinks it means.  As Bill Welmers used to say of G's Niger-Congo, "G
> hasn't proved that the languages are genetically related; he's made
> it inconceivable that they aren't."

A nice comment.  But who would want to make the equivalent remark for
Amerind, or for Indo-Pacific?

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk