LINGUIST List 2.115

Thursday, 4 Apr 1991

Misc: Lost correspondence, Structuralists, Language Families

Editor for this issue: <>

Directory

1. , Functionalism: Lost Correspondence
2. "Michael Kac", More on what American Structuralists REALLY believed
3. Herb Stahlke, Language families: African language classification

Our host machine went down during a collating process, and managed
in the process to trash our entire file of messages collected on the
subject of functionalism. If you submitted a message on this topic
and have not yet seen it, then the odds are that it was part of the
holocaust which has just taken place. You will have to resubmit 
your posting.

Please accept our sincere apologies. The software "feature" which 
allowed this to occur has been fixed, so this should not occur again; but 
that will not, we're afraid, enable us to recover the lost data. We hope 
that all of you have copies and will not have to recompose your messages!

The recent exchange of notes about Martin Joos and his famous dictum about
unlimited variation in language prompts me to the following comment about a-
nother caricature of American structuralism now pretty much taken for granted
by contemporary linguists. It has to do with the idea that the structuralists,
and most particularly the Structuralist's Structuralist Zellig Harris, were 
interested in finding discovery procedures for correct linguistic analyses.
On a certain level this is true, but there's a bit more to the story than 
that. Harris evidently did indeed believe that it was possible to completely
mechanize the process of linguistic analysis; it is not clear that he be-
lieved that this mechanization, once in place, would replace the less ri-
gorous methods then in use. Rather, the results of one's analysis -- however
obtained -- could be CHECKED for correctness by showing that they're what
one would have come up with if one had been applied the mechanical procedures.
>From p. 1 of 'Methods in Structural Linguistics':

 'In practice, linguists take unnumbered short cuts and intuitive or
 heuristic guesses, and keep many problems about a particular language
 before them at the same time ... The chief usefulness of the procedures
 listed below is therefore as a reminder in the course of the origiinal
 research, and as a form for checking or presenting the results, where
 it may be desirable to make sure that all the information called for
 in these procedures has been validly obtained.'

Michael Kac

 Perhaps because the debate is over two decades past, the vigor and
content of the African language classification debate is poorly appreciated
today. Several of Greenberg's conclusions were easily as startling at that
time as his claim about Amerind are today. Classifying Mande and Bantu in the
same language family, or Songhai, Kanuri, and Nilotic together was based on no
claims of sound correspondences. Although some linguists, Welmers, for
example, had suspected the Mande-Bantu relationship, Nilo-Saharan had not even
been hinted at in the literature. Within the close confines of South Central
Niger-Congo, a grouping which is, incidentally, based on a lexicostatistical
study, not a comparative study, the Kwa languages had been classified with
Bantu in Westermann's _Die westlichen Sudansprachen und ihre Beziehungen zum
Bantu_ (1927). However, this classification and his reconstructions were
widely questioned among continental Africanists trained in the Neo-grammarian
tradition because his sound correspondences were not completely consistent.

 Malcolm Guthrie and his colleagues and students in the London school
rejected Greenberg's conclusions on grounds that sound very much like the
arguments Campbell, Goddard, and others put forward today. Their claim was
that relationship can be demonstrated only by consistent sound correspondences
and that such research needed to proceed slowly, with great care, and from
groups where such correspondences could be demonstrated. However, Guthrie's
method, as useful as his results have been, is only superficially like the
comparative method, assuming as it does that language relationship is the
result of borrowing rather than of common inheritance and that therefore the
homeland of a language family is likely to be found in its area of greatest
similarity, not its area of greatest diversity. Guthrie was unwilling even to
call his formulae reconstructions, preferring to call them simply "starred
forms." A more mainline attack on Greenberg's _The Languages of Africa_ came
from Istvan Fodor in 1966 in a book, also published by Mouton, that
painstakingly, and criticized the lack of consistency in sound correspondences
in his cognate sets. This attack was irrelevant since it incorrectly assumed
that regularity of sound correspondence was critical to the method of mass
comparison.

 Westphal and others in South Africa criticized Khoi-San on the same
grounds as Guthrie and Fodor used against Niger-Congo. Westphal had done some
very solid comparative work within narrowly related groups in South Africa that
he calls families. In a very interesting survey article (_Current Trends in
Linguistics, Vol. VII: Linguistics in Sub-Saharan Africa_, Mouton, 1971, pp.
367-420), Westphal breaks Khoi-San into eight language families and claims that
these are not related to each other.

 Westphal, Guthrie, and other British Africanists of the '40s, '50s, and
'60s argued against Greenberg's classification on typological grounds that
comparativists in the Neo-grammarian tradition rejected as irrelevant. In
fact, some of those grounds, especially as the related to such groupings as
Nilo-Hamitic, were based as much on bad theology as on bad linguistics. There
was a wonderful article in the Journal of African History around 1974 that
exposed that line of argument for what it was. I regret that I don't have the
reference handy.

 To try to bring all of this together, I would acknowledge that the
typological traditions that influenced African language classification before
Greenberg are not present in the American language classification debate.
However, the methodological and phenomenological elements of the two debates
are similar. There is, by the way, an unpublished collection of articles that
chronicles the Greenberg African debate. It was assembled 1967-68 by Chuck
Kraft, then at UCLA, and formed the basis for a course on African language
classification that I took with him at that time. If anyone is interested, I
can get the bibliography entered and made available on this list.

 Perhaps the best way to end this too discursive review is to quote
Westphal's _Current Trends..._ article (p. 371): "The result of Greenberg's
super language families has been to force comparison where relationship was
hotly denied...[or where (HS)]...relationship was not so much hotly denied as
simply ignored in [previous (HS)] classifications." That's what our field is
about: not "shouting down" serious proposals but testing them. In African
linguistics much of that testing has been done and still more is going on.
Greenberg's proposals, with some refinement, have held up. The jury will be
out for some time on his American proposals, but given his success in Africa I
suspect Greenberg will turn out to be pretty close this time too.

 Herb Stahlke
 Ball State University


[End Linguist List, Vol. 2, No. 115]