Re: THEORY: languages without arguments

I've just joined the list, and wouldn't you know it, this topic is what
I've been working on for the past couple of months.

At 4/17/00 06:12 AM +0000, Tim Smith wrote:

>I've read in several places the idea that in radically head-marking,
>polysynthetic languages like Mohawk, all full noun phrases are adjuncts;
>the only arguments are the agreement-marking affixes on the verbs.  (If I
>understand correctly how these terms are defined, an argument is _required_
>by the verb, whereas an adjunct is optional.  Thus _all_ full NPs in
>languages of this type are optional, and many sentences consist of nothing
>but a verb.)
The theory makes other predictions as well.  The easiest to see is that
word order should be completely free.  That is, any word should be allowed
to appear in any position.  Sentences can be VSO, VOS, SVO, SOV, OSV, OVS.
Obliques and adverbs show the same range of variation.  It also permits,
but does not require, that demonstratives and quantifiers can be separated
from their noun.  The following Mohawk sentences exemplify this:

akweku wa'etshAri' ne onhuhsa'.
all          she-found  ?? egg
"She found all the eggs."

kikA wahiyena' ne kweskwes.
this  I-caught-it ?? pig
"I caught this pig"

(If "pig" and "this" or "all" and "egg" change places, the sentences are
ungrammatical.)

Also, since the agreement morphemes are acting as the arguments, everything
you need to interpret the sentence should be contained in the verb.  In
many of these languages (Chickasaw [the language I work on], Warlpiri,
Ainu, etc.) the third person does not have an agreement affix.  The bare
verb stem is interpretted with a third person argument.  The claim is that
the affix is there, just silent.

>Has anyone ever proposed a similar analysis of languages like Japanese, in
>which all NPs (except predicate nominals) must be objects of adpositions?
>In other words, the immediate constituents of the verb phrase are
>adpositional phrases rather than noun phrases.  To me, this makes it sound
>like all NPs are oblique, which in turn makes it sound like they're all
>adjuncts rather than arguments.  (Especially since in Japanese, I have the
>impression that basically all NPs _are_ optional.)  But then, are there any
>arguments at all, given that there's nothing in Japanese even remotely
>corresponding to the verb-agreement affixes that are arguably the "real"
>arguments in Mohawk?
Except that you can get nominals without the "adpositions".  In colloquial
Japanese, if the grammatical function of the noun can be determined from
context, the particle is frequently left off.  These sentences were offered
by a native Japanese speaker.

Masao kinoo         kita yo.
Masao yesterday came
"Masao came yesterday."

Dareka     Hanako tazunete-kita?
someone  Hanako come-to-visit
"Did anybody come to visit Hanako?"

This makes it look to me like at least -ga and -o are case markers rather
than postpositions.  These are also the same adpositions that may not be
used before the topic marker -wa.  Also, they usually cannot be used
directly before the copula.

Japanese also does not have the free word order of these head marking
languages.  NPs rarely come after the verb, and when they do it is under
very specific situations.  Also, any word moved out of its "typical"
position has to have an adposition.  When they are left off, the word
cannot move.  In the head marking languages, the NP can move whether there
is agreement or not.

>In some ways, Japanese seems like the exact opposite of Mohawk: its verbs
>have no agreement morphology at all, and its nouns all have postpositions
>to mark their case roles, while Mohawk nouns have nothing to show their
>case roles except the very elaborate agreement morphology on the verb.
Actually, Mohawk sentences quite frequently do not mark the role of the
direct object, especially if it is inanimate.  In the sentences above,
"egg" did not get agreement on the verb, but "pig" did.  Even with a lack
of agreement, it is obvious what is intended  if the NP isn't present,
because agreement is only missing in specific circumstances, such as an
inanimate third person.  You can't do that in Japanese.  If I say "Miru"
you must rely on context to tell whether I meant "I see" "you see" "she
sees" "they see" etc.

>Japanese is radically dependent-marking, while Mohawk is radically
>head-marking.  Yet somehow they both end up with noun phrases that are
>never mandatory and don't look like arguments.
An interesting point, although I would have to disagree about not looking
like arguments.  There are restrictions in both of the these that don't
make sense if all NPs are adjuncts.

>ObConlang:  The reason I'm wondering about this right now is that I've just
>started thinking about revisiting one of my older conlang ideas, which
>started out as a sort of "inverse Japanese" (basically like Japanese only
>verb-initial and with prepositions instead of verb-final with
>postpositions).  (When I first got this idea, many years ago, I didn't
>realize that lots of natlangs, such as Tagalog, more or less fit this
>description; I thought I was onto something really new.)
I'd say go for it.  Many linguists (misguided in my opinion) think this
type of language is real, so I don't see why you shouldn't try it out.  I'd
be interested in what you come up with, because the conlang I am working on
is a blend of Mohawk and other languages (mainly Chickasaw and Ainu).

Marcus Smith