Re: Functions of Classifiers (in a conlang)

I'm coming back from a weekend away, so I hope I don't repeat
anything...

Chris wrote:
<<
(i) As the quoted text from Aikhenvald mentions, many languages use
their noun classifiers as pronouns. This is a natural extention.

(ii) Another possibility is to use the classifiers as THE marker of
nominality, ie nominalizations of finite verbs can be carried out
simply by giving the verb a classifier, and the lack of a classifier
can turn any stem into a predicate (thus making nominality explicitly
marked rather than associated with certain stems).
 >>

This is what Zhyler does exactly.  However, they're noun classes
(or "genders", the old term I don't like).  I suspect that combining
these two ideas results in noun classes, and not classifiers.  But,
to cite examples, Zhyler has only first and second person pronouns
(singular and plural); no third person pronouns.  It does have 17
noun classes, though, so you get:

/pal/ (v.) to wear

palka (class i, human) one who's wearing something
paldM (class v, human professional) model
palkMz (class viii, land animal, non-mammal) snake
paller (class x, long things) cloth
palSa (class xii, object) dress, suit, outfit
palmos (class iv, water animals) barracuda

So (i) is completely transparent; (v) seems obvious enough;
(xii) is somewhat related to stuff that you wear, and (x) is
somewhat related to that; you can come up with a story for
(viii), since snakes "wear" their skin, and shed it every so often;
and I don't think anything can be said about (iv).

This is how they work.  The noun classes occur with all nominals,
without exception (except borrowings, which must be fit into a
noun class, nevertheless).  The forms are bound, and you can take
a verb and fit it into a noun class to get a noun.

However, the markers themselves have pronominal forms that
are used in particular circumstances--in particular, relative clauses:

matkalarum sexa rudZbanar uslar.
/see-(i)-PAST-1sg. man(i) bread(xvii)-ACC. eat-PAST/
"The man I saw ate the bread."

Here the verb in the embedded clause has to agree with the
target of relativization.  Otherwise, these markers never appear
on the verb.  In this situation, they can also be used as pronouns--
for example, if they occur as an argument which can't be marked
on the head (verb):

kas tSelvener ellerym sexa rudZbanar uslar.
/(i)PRON.-DAT. dog(ii)-ACC. give-PAST-1sg. man(i) bread(xvii)-ACC.
eat-PAST/
"The man I gave a dog to ate the bread."

So this seems like an example of what you were talking about,
and it seems like it doesn't lend itself to noun classifiers.

Chris continues:
<<
MAN.CLS man come and MAN.CLS angry shout

Is this:

The/a man came and he shouted angrily

or...

The/a man came and the angry (man) shouted?
 >>

As Patrick Littell pointed out, ambiguity is something that
conlangers and linguists worry about--not natural languages
(or their users).  This was amusing in fieldwork.  A friend of
mine working on Moro (or several, actually) were constantly
trying to come up with sentences that were ambiguous.  For
each one they'd come up with, he'd ask for the context, then
give the meaning.  They'd ask if it was ambiguous, and he'd
say, no, it has that meaning.  When pressed, he could see that,
yes, two interpretations were possible, but one always lent
itself more readily to the given context, so it really didn't matter
that they were technically ambiguous.  For example, in Moro,
if you say "I give x to y for z", there's no way (neither word order
nor marking) to distinguish between "to y" and "for z".  It's
always ambiguous.  For any given context, though, it never is,
so there are no problems.  (And, of course, if one arose, the
speaker could simply explain.)

Given your example above, in order to solve the ambiguity,
you can do one of the following:

(1) create some form (no matter what it is) ex nihilo to resolve it

(2) do nothing

(3) vary the word order

(4) use intonation

(5) use "pro-drop" (in this case, classifier drop) vs. present
classifier

(6) use classifier vs. full NP

(7) some combination of the above options

The pro-drop solution would be kind of like English intonation:

(a) "Bob came in and he was angry." (he = Bob)

vs.

(b) "Bob came in and *HE* was angry." (he = someone else)

With your example, the one without the classifier would likely be
(a) and the one without would likely be (b).  With the full NP
solution, the classifier would be (a), and the full NP would be (b),
most likely.

For a combination, you could simply do nothing, and then use
one strategy for disambiguation if there's confusion.  For example,
just use the sentence you gave in all occurrences, and if someone
gets confused, use intonation, or the full NP strategy.

Doing something like this would be working with the morphomes
(no misspelling) that your language has already.  Creating new
forms seems unnecessary, unless those forms are going to be put
to use somewhere else in the language.  It certainly can be done,
but usually a language tries to solve its problems with the stuff it
already has as opposed to creating new stuff out of thin air.

-David
*******************************************************************
"sunly eleSkarez ygralleryf ydZZixelje je ox2mejze."
"No eternal reward will forgive us now for wasting the dawn."

-Jim Morrison

http://dedalvs.free.fr/