Re: quadrivalent verb

On Tue, 30 Sep 2008 15:20:11 +0000, Kenneth Asad
<kenneth_asad@...> wrote:
>This of course brings me/us to a bit of a fix...
>Is there any way - a test - to determine
>whether a phrase is a core argument or an oblique argument?
>... I'm not a linguist myself
(Neither am I!  But there are some on the list.)

Here are some helps; some things that are necessary, or sufficient, to be an
argument, or to be in the core.  There is nothing nothing _I_ know of that is
_both_ necessary _and_ sufficient.  And this is not an exhaustive list of the
things that are necessary _or_ sufficient (though I'm not sure such a list
exists).

------------------

About arguments vs. adjuncts.

All adjuncts are oblique, and if they have cases or adpositions (or both), these
show what semantic role they play in their clause.

Also, all adjuncts are optional; if they're omitted the clause still means
something, and indeed means something rather similar to what it meant with
the adjunct included.

Arguments, OTOH, are supposed to be "semantically mandatory"
or "semantically obligatory"; but, in some grammaticist's analyses of some
languages, that doesn't mean "obligatorily explicit" nor "mandatorily explicit".
In other words, they still count something as an argument if both the speaker
and the addressee have to assume it's implicitly there even if the speaker
doesn't mention it in the clause in question.  For instance, it may be specific
(the speaker has one in mind) or definite (the speaker also thinks the
addressee knows which one the speaker has in mind) because the speaker
mentioned it in an earlier clause.

The upshot is; if something is obligatorily explicit -- if it's ungrammatical for a
speaker to speak the clause without mentioning that participant -- then that
participant is bound to be an argument, not an adjunct.

Unfortunately, if the speaker needn't mention it for the clause to be
grammatical, that doesn't mean that that participant is not an argument (is an
adjunct).  That can depend on the language; or even vary within the
language.  AFAIK it could also depend on which theory the given grammaticist
adheres to.

-------------------------

About obliques vs core;

To start with, all adjuncts are oblique.

There are such things as oblique arguments, though; and there are such
things as mandatorily-explicit oblique arguments.

In some languages, and some people think English and many other familiar
languages are included in this, obliques have adpositions and/or case-endings
indicating "semantic cases" (what semantic role, thematic role, theta-role,
case role, or "deep case" the noun plays), while core arguments don't have
adpositions and either have no case endings or have "syntactic case" endings
(case-endings indicating which grammatical relation the noun has to the
clause).

But there are good reasons to believe some languages have "dative
adpositions", like English's "to" (if you believe English has an "indirect object"
GR).  And some languages -- Spanish and Turkish, for example --
have "accusative adpositions", at least for some direct objects (for instance,
definite humans, or definite animates, or specific humans).

There is even some sense to interpreting Tagalog's "ang" as much as
a "nominative adposition" as a definite article.

Anyway:
If the verb has to agree with the noun-phrase or pronoun, the NP or Pron is
probably (almost certainly) in a grammatical relation; or to put it another way,
is probably (almost certainly) a Core Argument.

For languages with polypersonal agreement this is useful because the verb
may agree with two or three or even four participants.

(If the verb agrees with five or more participants maybe this breaks down?  Or
maybe it just means the language has more than four GRs?  I know of some
languages that have five things verbs can agree with, but don't know of any
examples in them of any verbs actually agreeing with more than four of them
at a time.)

For languages in which verbs never agree with any participants this is pretty
useless.

For languages in which verbs have to agree with exactly one participant, this
identifies the Subject GR (so the Subject is part of the Core), but doesn't
identify any Objects; if the language has Objects they have to be identified
another way.  (There are other ways coming up; they often help but are not
always decisive.)

Some languages require the verb to agree with the Subject and with the
Primary (or Direct, depending on the language) Object (if there is one); but do
not require it to agree with any Secondary Objects, although some clauses in
the language do indeed have three (or four?) GRs.  In such clauses in such
languages a Secondary Object is still a Core Argument for that clause, but the
verb doesn't have to agree with it.

Also:
If the semantic role, thematic role, theta-role, case-role, or "deep case" which
the NP or Pron plays, is not indicated by the morphological case or the
adposition, but instead must be deduced from the voice of the verb; then the
NP or Pron probably occupies a Grammatical Relation and so is a Core
Argument.

That's one thing that happens with NPs in GRs; their case (or lack of one) and
their adposition (or lack of one) indicate which GR they have to the verb, not
which semantic role they play; they can play many different semantic roles.
The voice of the verb indicates which semantic role each of its Core
Arguments plays.

---------------------------------------------------------------------------

So;

If an NP is obligatorily explicit it must be an argument; but if it's optionally
implicit it may still be an argument (or may not).

If the verb has to agree with an NP it must be a Core Argument, but if the
verb need not agree with it it may still be a Core Argument (or may not).

If an NP's semantic role is indicated by the verb's voice rather than by its own
case-marker and/or adposition, it must be a Core Argument; if the verb's voice
tells nothing about the NP's semantic role, but its case and adposition tell
everything about its semantic role, it must be an Oblique (either an Oblique
Argument, or an Adjunct).

The above probably do not always settle every question; but they settle
many, and give a lot of hints about most others.