LINGUIST List 3.316

Sun 05 Apr 1992

Disc: Ergativity, Agentivity and Classification

Editor for this issue: <>

Directory

1. Matthew Dryer, Ergativity in Africa
2. "Michael Kac", Re: 3.301 Responses: Agent, Natural Language, Its, Right-Dislocation
3. , Language Classification (Greenberg's)

On ergativity in Africa, see Andersen, Torben. 1988. Ergativity in Pari,
a Nilotic OVS language. Lingua 75: 289-324. Also Mechthild Reh gave a
paper at the recent African Conference in East Lansing discussing
ergativity in Anywa, a language closely related to Pari.

Matthew Dryer

Message 2: Re: 3.301 Responses: Agent, Natural Language, Its, Right-Dislocation

Another possible case of agentivity coded as locativity: In Latin, the usual
case for location is the ablative, which is also the case for the 'agent' in
the passive construction. (Please note my use of quotes here: the 'agent' of
a passive construction can in fact be playing any of a number of semantic
roles, including that of experiencer).

There is also in Latin a different case form traditionally called 'locative'
but which, to my recollection, is found only with certain nouns (e.g. *domus*
'house, home') and is indistinguishable from the genitive.

Michael Kac

Some time ago, we had an extensive discussion on LINGUIST of
Greenberg's classification of the languuages of Native America.
One point that was raised at that time in defense of Greenberg's
proposals which no one was really able to respond to was that,
after all, his methodology had worked so well in his work on
Africa.

Since then, I have looked at two languages without close relatives
which Greenberg classifies as Amerind: Tonkawa (which he says is
Hokan) and Zuni (which he says is Penutian). I then tried as best
as I could to use Greenberg's own method and see if I can arrive
at the same classification.

In the case of Tonkawa, it became apparent that, especially as
far as the "grammatical" morphemes are concerned, it seems to
belong with the Na-Dene family rather than Amerind (in the three-way
classification proposed by Greenberg). For example, the first
person pronouns of Amerind are supposed to be n- and m-, Tonkawa
has s- and n- (exactly like all the Athapaskan languages in the
Na-Dene group). The lexical evidence is less clear, but consistent
with a Na-Dene classification.

In the case of Zuni, it became apparent that Greenberg cites
not a single "grammatical" morpheme of this language, and
the entire corpus of "lexical" elements which he cites as
linking this language to Penutian and ultimately to Amerind
can be equally well linked with Indo-European, or, I suspect,
with any arbitrary language family.

And one can certainly make a case for "grammatical" resemblances
between Zuni and IE, if one wants to. For example, 1sg and 2sg
are ho' and to' in Zuni and *'egh(oo) and *tu in IE.

It is my contention, therefore, that Greenberg did not apply
his method (whatever one may think of it) properly in his work
on the languages of the Americas. Rather, it would seem that
he ASSUMED Sapir's proposal that Tonkawa is Hokan and Newman's
that Zuni is Penutian. In other words, he seems to have taken
for granted earlier "lumping" proposals, without examining them
independently.

Thus, I discovered, quite to my surprise, that it was not so
much the method itself, as its application that seems to be
the problem. Presumably, in the case of Africa, the situation
was different. I would add that I found the method itself less
objectionable that I thought I would. Although it still is
in need of testing (and, of course, of a more precise definition),
it does now seem to me that it is not completely vague and useless
(as has been argued in the literature).
----------------------
Greenberg, J. Language in the Americas. Stanford U. Press.

Manaster Ramer, A. "Is Tonkawa Na-Dene?". Submitted to IJAL.

Manaster Ramer, A. "Is Zuni None of the Above.". Submitted to IJAL?