Quoting "SMITH,MARCUS ANTHONY" <[log in to unmask]>:

> On Thu, 25 Jul 2002, Daniel Andreasson Vpc wrote:
> > Tom Wier wrote:
> >
> > > "Active" is a kinda old-fashioned way to describe these kinds of
> > > languages.  Nowadays, it's more usual to call them "split-S", when
> > > certain Ns always take certain case agreement, and you just have
> > > to know which it is, or "fluid-S", where some verbs may take take
> > > either patient or agent marking depending on the context of the
> > > sentence.  Dixon wrote at length about this in _Ergativity_
> > > (Ch. 4:  "Types of Split System").
> I'm not sure I agree with this characterization. I hear/see the term
> "active" far more often than I do "split S" or "fluid S". Perhaps that's
> because I work with a literature that is firmly entrenched in Sapir's
> terminology, I couldn't say. I've never heard a Muskogeanist call the
> languages Fluid S, for example. Other people talking aboug Muskogean
> languages do so, though. My brief forrays in Siouan languages give
> me the same impression.

FWIW, my experience is mostly the opposite.  One of my professors,
Amy Dahlstrom, is an Algonkianist who works on grammatical relations
and switch-reference systems.  Whenever the topic comes up, she's
referred to it in Dixon's terminology, not the traditional "active"
or "stative".  It it well to note that Dixon himself (_Erg._ p. 77)
quotes her as using "active" in a work of hers from 1983; therefore,
at least *she* has given up the older term. The one apparent exception
that I know of, Howard Aronson (who works on grammatical relations in
Georgian), refers to "active" languages but only in the context of
his debate with Klimov, whose theory of progressively and teleologically
systemic change in languages he rejects (or at least disfavors).
So I cannot say that I have heard anyone of late favoring the use
of "active" in current linguistic theory.

> > I agree with him that one could consider truly split-S languages to
> > have syntactic marking just
> > like accusative and ergative languages. The
> > problem as I see it is that the division into AGT
> > and PAT predicates is still based on the semantics
> > of the verb, unlike ergative/accusative langs.

I think this constitutes a misunderstanding of what Dixon was
getting at.  Speaking of the case-system of Mandan, he says:

  "Note that in a Split-S language like Mandan each intransitive
  verb has fixed class-membership -- either S-a or S-o -- generally
  on the basis of its prototypical meaning.  If one wanted to use
  a verb with deals with a prototypicaly  noncontrolled activity
  to describe an activity done purposely, then it would still
  take the So marking (and something like an adverb "purposely"
  could be added)" (p. 72).

Understanding his use of the word "generally" here is crucial.
Although Dixon believes that a verb's prototypical semantics are
important in the development of Split-S systems, he does not
believe that it is the determining factor in its synchronic
use.  He goes on to clarify:

  "There are Split-S languages where the two intransitive classes
  do not have as good a semantic fit as those in Mandan and Guarani.
  Thus, in Hidatsa, another Siouan language (Robinett 1955), the
  S-a class includes volitional items like 'talk', 'follow', 'run',
  'bathe', and 'sing', but also 'die', 'forget', and 'have hiccups',
  which are surely not subject to control. And the S-o class includes
  'stand up', 'roll over', and 'dress up', in addition to such clearly
  nonvolitional verbs as 'yawn', 'err', 'cry', 'fall down', and
  'menstruate'" (74).

So, while it is clear that Dixon is acknowledging a diversity of
behavior in terms of how clearly S-a and S-o marking can be predicted
from semantics alone, I think it is fair to say that he is drawing
a useful distinction by classing all such languages together in one
group since precisely because *all* verbs can only take one set of
markings.  In contrast, Dixon's "Fluid-S" verbs are in principle not
restricted to either A- or O-marking, but do so according to
principles of narrative structure or other pragmatic factors.  The
trick here is that there are often activities that discourse-functionally
almost never are used with one marking or the other -- Dixon
cites 'tremble', 'be hungry', 'be ripe, grow up' in Tsova-Tush, for
example, as ones that usually take S-o marking. He notes:

   "The results Holisky [who worked on Tsova-Tush] obtained were
   determined partly by world-view and by other pragmatic factors.
   She mentions 'when I constructed the first-person form of the
   verb "get poor" in Tsova-Tush using [S-a] marking, my consultant
   did not say *categorically* that it wasn't possible.  She said
   it isn't possible because you would never want to be poor'
   (Holisky 1987:115)" [his emphasis] (80).

So, it appears that speakers really do treat verbs in different
ways in languages belonging to different classes.

> Dixon established an idealized description of the facts. Like most
> idealizations, they are very rarely met in practice. The majority of
> Ergative languages are not purely ergative, for instance. There are some,
> but they are in the minority. The same is true to a fair degree of
> Nominative languages: most of them have deviations from a purely
> nominative-accusative pattern, including dative or genitive objects, or
> dative subjects, just to name a couple. Most deviations are semantically
> motivated, just like the deviations from Dixon's idealized model.

Have you read his book? Dixon spills considerable amounts of
ink discussing exactly such problems.  I think it is rather
unfair to question his motives in this respect, since he is
usually very conscientious about when he is citing empirical
facts and when he is making generalizations (cf. his discussion
of the nominal hierarchy, p. 90, first paragraph after the

> > I'm interested to hear other people's thoughts on this
> > (especially from Tom, since I'm merely an amateur
> > on these matters, and you seem to be more up-to-date
> > (not to mention more professional), especially on Dixon's theorys).
> I do agree with you in general that Sapir's original "Active-Stative"
> terminology is superior to Dixon's. The terminology at least gives you
> guidelines about what to expect in general. It doesn't evoke images
> of the "control" vs. "non-control" that is frequently found, but at least
> it isn't as vague as "Fluid" and "Split".

(1) I think it's rather questionable to claim that terminology
would ever be very helpful in telling you much about complex
phenomena.  Terminologies are sets of labels, and precious
little more.

(2) That having been said, if you've read Dixon's book, you know
precisely what he means when he talks about "Fluid-S" and "Split-S"
systems.  In the context of the current study of ergativity, these
are far more insightful to me than "active" and "stative", since
(if one is going to insist on 'meaningful' labels) they at suggest
(correctly) that the case system is sensitive only to the behavior
of the subjects of intransitive, and not transitive, verbs.
(Examples of languages whose intransitive constructions are
determined syntactically but whose transitive constructions are
sensitive to semantics like a Fluid-S are apparently entirely

Thomas Wier          "...koruphàs hetéras hetére:isi prosápto:n /
Dept. of Linguistics  mú:tho:n mè: teléein atrapòn mían..."
University of Chicago "To join together diverse peaks of thought /
1010 E. 59th Street   and not complete one road that has no turn"
Chicago, IL 60637     Empedocles, _On Nature_, on speculative thinkers