Re: THEORY: Incorporating Agents vs. Patients in Verbs (was: 'Yemls Grammar)

On 9/27/05, tomhchappell <tomhchappell@...> wrote:

> [QUESTIONS]
> 1. How synonymous are the terms "incorporating languages"
> and "polysynthetic languages"?
"Polysynthetic" is a term I avoid, because different writers use it in
such different ways, but most incorporating languages would qualify as
polysynthetic whatever your definition.

There's no especial tendency the other way.  Certainly polysynthetic
langs are much more likely to incorporate than analytic ones, but
incorporation isn't like a special notable feature of polysynthetic
langs.  A lot of polysynthetic langs have noun incorporation, and a
lot don't.

Actually this may have been one of the first big arguments about noun
incorporation, and is discussed famously (albeit with different
terminology) in Kroeber (1910) and Sapir's seminal response to it in
1911.  They were wondering whether object incorporation was the same
or related to what they called "pronoun incorporation"; they both came
to the conclusion that it wasn't.

>
> 2. How co-extensive are the two types of languages?
>
Well, as before, you're not going to find much prototypical noun
incorporation in a more isolating language.  You might, however, find
noun stripping, which is a very similar process and also fairly
common.  If you can't find anything of use by googling that, email me
and I'll see if I have any articles about it.

> 3. Do ergative incorporating languages lean toward agent-
> incorporation the way accusative incorporating languages lean toward
> patient-incorporation?
No, it doesn't seem to have much effect.  The semantic roles that may
be incorporated are pretty much the same regardless of the case
alignment of the language.  That is, the "incorporability" of a
participant depends mostly on (1) its semantic role and (2) its
animacy (Animacy in the sense of "likelihood of being an agent") and
very little on the "surface" case it would have taken had it not been
incorporated.

>
> 4. What questions do you think I should have asked instead of these?
>
One interesting question regards the other semantic roles (especially
Instrument) that are incorporated.  Incorporation of the instrument is
fairly common, and I think you also might find beneficiary, location,
etc.  One question I wonder about: "Can you have incorporation of role
X without incorporation of role Y"?  I have a feeling there might be a
hierarchy of incorporability along the lines of "patient > instrument
> location" or something like that.  Meaning "every language that can
incorporate an instrument can also incorporate patients; every
language that can incorporate location can also incorporate
instrument, etc."  I know of no big cross-linguistic study of this.

> 5. a. How common are natlangs attesting both agent-incorporation and
> patient-incorporation?  b. How common are these two phenomena (should
> we call them "processes"?) in these languages?
Agent incorporation of the sort you mention is very rare, if it does
exist.  (If you can give me an example in a specific language, I would
be pleased; it might bear on my current research.)  You might be
confusing it with the fact that in some langs the grammatical subject
may be incorporated, but it's only those subjects with a low degree of
agency.  Let me try to think of a few examples.  Here's a Totonacan
example off the top of my head:

?Min-kaa'kni 0-ka'¢an.
My-head     it-hurts.

Ik-kaa'k-ka'¢an.
I-head-hurt.

This is possible in Totonac because of the semantically stative nature
of "hurt" -- it doesn't require the subject to have any agency.  (The
question mark is because I'm not sure this "excorporated" example can
actually occur in this dialect.  Any mistakes above are mine; this is
just from memory.)

I'll need to fish around for a more prototypical example -- Totonacan
languages (somewhat exceptionally) lack the ordinary "Type I"
incorporation, so I can't use them to illustrate that.  (If you need
to know the different "types", send a quick reply and I'll explain
them further.)

Here's one I'm using an example sentence in an upcoming paper.  I
misplaced the book it came from at the moment, but I think it's in the
Chukchi chapter in Spencer's Handbook of Morphology:

tirk@-tir amecat-g?e
sun-ABS appear-3sg.s
The sun appeared.

terk-amecat-g?e.
sun-appear-3sg.s
"It sun-appeared"

This example is actually a pretty prototypical subject-incorporation
cross-linguistically; natural forces like the sun lack agency but
still "do" things, making them natural candidates for subject
incorporation.

Hmm.  In Malagasy the agent of a passive or circumstantial verb takes
the genitive/ergative case and then compounds with the verb, but I've
never seen this treated as agent incorporation.  Maybe it is; I've
never really thought of about it like that.  Nah, I think it's a
different sort of process.

> c. What languages have
> a lot of verbs which can have both the A and the P incorporated?
You mean simultaneously?  I'm going to estimate none, due to the
vanishing rarity of agent incorporation and the fact that usually only
one participant may be incorporated.  If they exist, compounds like "I
knife-foot-cut him" are pretty rare, even though patients and
instruments are commonly incorporated.  The only attested examples
I've seen of double-incorporation are ones where (1) the "first"
incorporation has become lexicalized, and is not synchronically an
incorporation and ones where (2) the "second" incorporation has become
grammaticalized as some other sort of affix, like a causative or
locative marker.

For an example of the second, I believe the following form would be
grammatical in Totonac:

Ik-puu-kaa'k-piS "bathtub"
I-LOC-head-bathe bathtub.
I bathe my head in the tub.

(I have no idea how you would say "bathtub".)

The affix "puu", which licenses the locative object "bathtub", is
diachronically related to the incorporated form "puu" -- "insides".
(And I wouldn't be surprised if the causatives "maa-", "maqa-" and
"maq-" are derived from "maka/maqa" -- "hand".  It wouldn't be the
first time that change occured.)  But "puu" is not synchronically a
case of incorporation; it's now a locative affix.

But if someone did find a language that could incorporate two
participants, my jaw wouldn't drop and cause me to tear up all my
work.  It seems at least possible.  I just haven't come across it.

-- Pat

Reply