THEORY: What is an active language?

Jörg wrote:

> > As you can see, I have a very vague definition of "active". I think the
> > main thing that distinguishes active langs from ergative and accusative
> > ones is that there is some kind of semantic reason for marking the S
> > argument as either A or P. The most common ones being Control vs. non-
> > control and event vs. state.
> I'd say, "event vs. state" is a bit weak.  The verb "to fall", for
> example, clearly refers to an event, but is its subject an A?
Well, I was being purposely vague there. I don't want to define A
and P. In fact, if I could do without A and P, I would. The Chickasaw
way of using I and II is _so_ much better, since you get rid of the
connotations of Agent and Patient. My definition of an active language,
is a language that has "active alignment". At least I think that's
my definition. I mean, if some intransitive predicates get marked
in way A and some itr predicates get marked in way B, that's when you
have an active language. Then the fun part kicks in: you get to
investigate why some predicates get A-marking and some get B-marking.
Most of the time -- at least in my experience -- there seems to be
a valid semantic reason, whatever it may be.

Let me give you some examples:

A) Georgian: if [+ control] then AGT.

B) Central Pomo: if [+ control] then AGT -- UNLESS you are also
                 significantly affected, then PAT.

C) Eastern Pomo: if [+ control] then AGT -- unless the speaker also
                 chooses to express *empathy* with you, then PAT.

D) Guaraní: if [+ event] then AGT.

E) Lakhota: if [+ performance/instigation/effect] that is NOT
            necessarily [+ control] then AGT.

Etc, etc.

> > I don't know if my view on things really matters in this discussion
> > (Marcus doesn't quite agree, having a narrower definition of "active"
> Which seems to include the absence of a case system ;-)
Okay, we get that now, I think. I would like to hear Marcus' comments
on this one. I think Marcus has a point.

> > and Matt having a very broad definition (as it seems) and Jörg basing
> > his definition more on animate vs. inanimate)
> Well, I'd rather say it is based on volition, which of course implies
> animacy.  But a non-volitional animate is still a P (or an INST, or
> whatever) and not an A, at least as far I understand it.  An inanimate
> entity, in my personal model, can NEVER be an A, though.  But I am not
> so strict so say that anything that doesn't strictly adhere to this
> model is not active.
This is the way Johanna Nichols thinks of active languages as well.
Or rather, she says that this _usually_ is the case, not that it
always has to be like that. She has a table of lots of criteria that
often applies to active languages, but not always.

> Hence, I call languages where this holds "strictly
> active", while "active" alone leaves some leeway for things that are not
> allowed in a "strictly active" language, e.g. treating an inanimate
> subject of an active verb (as in the sentence "The stone breaks the
> window") as an A.  I am not sure by myself where to draw the line
> between "active" and a more general "split-S".
I think you have a good point here. And as I said before, split-S
_could_ be thought of as a strictly syntactic alignment, but IME
that is not usually the case, since there is some semantics to it.
IME almost all active languages have some fluid predicates. I don't
see the point in making a distinction between fluid-S and split-S
as Dixon does. If anyone can explain that point, I'd be grateful.

I think an important part is not to get hung up on Agents and Patients
as semantic roles. The reasons for marking A and P differently for itr
predicates vary too much, IME.  [ Someone give me a digital kick in
the butt if I write "IME" again... :) ]

> But one thing is certain: it doesn't matter whether the language uses
> cases or verb agreement marking to distinguish A and P.  If one says
> "No, this is not active; the examples you gave are simply unergative
> verbs/irregular forms/whatever" merely because the language marks nouns
> for case, I cannot take that seriously.
Well, I agree. But agreement marking is _so_ much more common, due to
its being the verb carrying the meaning and distributing the semantics
to the arguments. All the active langs I've looked at are head-marking,
except possibly Guaraní. (I don't know about Acehnese, the active
alignment is expressed by pronominal clitics on the verb).

Nonetheless, active alignment can still be expressed by case.

> > but at least you know what I think about active langs (at least
> > for the moment, my view changes all the time).
> So does mine.
And probably mine too, when you and Marcus and Matt has replied
once more... :)

> Heck, I don't really care whether a particular language is active or
> not under theory XY.
I don't care about any theories at all. I still don't know if that's
a good thing or a bad thing. Prolly both.

> Every language has a somewhat different system.  And I
> have seen languages being labeled "active" where I asked, "Why?", and
> others labeled "not active", where I asked, "Why not?"  And when I
> designed Nur-ellen, I didn't knew that there is a linguistic term for
> such a system.  It just sprang to my mind ans I liked it; hence I just
> did it because it felt right to me.
Yup. That's what conlanging is all about.

daniel, sleepy. Very, very sleepy... zzzz....

--
<> Daeselaidh goddi mis giall! <> daniel.andreasson@telia.com <>
<>              Lwodadh giall! <> Daniel Andreasson           <>