Re: THEORY: Why more than two grammatical relations?

---In conlang@yahoogroups.com, "J. 'Mach' Wust" <j_mach_wust@...> >wrote:
>On Fri, 5 Oct 2007 10:49:32 -0400, Eldin Raigmore wrote:
>>[snip]
>What's an oblique argument then?
The participants in a clause are split two ways.  Semantically they can be
Arguments or Adjuncts; grammatically they can be Core or Oblique.

The difference between an Adjunct and an Argument is this; an Argument is
semantically necessary for the clause to mean anything.  In other words, for
semantic reasons, such a participant must be assumed to exist even if it's not
explicitly specified, or mentioned "out loud".  An Adjunct, on the other hand,
adds to or modifies the meaning, but even without the Adjunct the clause
would mean something; indeed, probably more-or-less almost the same thing.

I haven't defined "Core".  The "Core" of a clause consists of its nucleus
(usually the verb) and (some of) its Arguments.

One way that works in some languages to distinguish Core arguments from
Oblique arguments is this; if it's a Core argument, its semantic role cannot be
determined just by examining the noun or pronoun or the noun phrase or the
adpositional phrase; instead one must also consult the verb (usually, the form
of the verb (usually, the grammatical voice of the verb)).  But if it's an
Oblique, its semantic role is completely determined by the form of the pronoun
or noun and/or noun phrase and/or adpositional phrase.

I don't know for certain of a way to do it for every participant of every clause
in every language.

"Core" is sort-of theory-specific to a number of theories; but most theories
contain a similar idea, whatever they call it (in some cases maybe they don't
have a word for it, but the idea still exists).

In any case, even if two acceptably-grammatical sentences with the same
meaning are word-for-word translations of each other in two languages, they
can differ as to which participants are "in the Core" and which aren't.

>I thought an oblique case was any case except nominative, the "rect"
>case,
Ancient Latin grammarians of Latin called every case except Nominative and
Vocative "oblique"; I was under the impression that both Nominative and
Vocative were "casus rectus".  But I gather that some people (ancient or
modern? I don't know) refer to "casus rectus" as "governed case" as opposed
to "concordial case" or "agreeing case".

Ray Brown, and also various other people on this List, can correct and/or
clarify and/or add interesting detail to the above.

Many modern grammarians of many modern languages distinguish between
Direct Cases and Oblique Cases.

Most modern grammarians of most modern languages that have Direct Objects
count their Accusative as a Direct Case, along with their Nominative.

Some modern grammarians of some modern languages distinguish between
Core and Oblique cases; or call all Core cases "Direct" cases.

For purposes of interpreting what I'm saying on this thread, you can
assume "Core case" is synonymous with "Grammatical Adverbial case"
and "Oblique case" is synonymous with "Semantic case".  (Also we need to
stretch a point and count adpositions as case-markers.)

Blake and others distinguish "Adnominal cases" from "Adverbial cases", and
distinguish "Grammatical cases" from "Semantic cases".

An Adnominal case specifies its noun's relationship to some other noun; an
Adverbial case specifies its noun's relationship to the verb or the clause.

A Semantic case specifies its noun's semantic role, thematic role, theta-role,
deep case, or case-role.  A Grammatical case specifies its noun's Grammatical
Relation.

So which cases are Oblique is language-specific; and what the Core is is
language-specific as well.  Not everybody likes this terminology, but most
don't stick to the old-fashioned "everything but Nominative is Oblique"
terminology anymore.  Even if I'm wrong about that, I'm using "Oblique" to
mean "not Core" in this thread.

>so both accusative and dative object have oblique cases.
That won't be the terminology I'll be using in this thread.  If the accusative
noun is in the core I won't call it "oblique".  If the dative noun is in the core I
won't call it "oblique" either.

Again, it'll be language-specific.

>>(Isn't German also one of those languages with some "bivalent
>>intransitive" clauses, intransitive clauses with two Core
>>participants? In which case aren't some of those clauses analyzed
>>as having a Subject and an Indirect Object, but no Direct Object?)
>Well, yes, like in the example I gave in the last post. The Duden
>grammar lists more instances of clauses with one additional argument
>that is not an accusative object (apart from subject + predicate +
>dative object and subject + predicate + genitive object):
>
>subject + predicate + prepositional object
>subject + predicate + predicative nominative
>subject + predicate + spatial/temporal/modal/causal complement
Thanks.

>I for one was taught that grammatical categories originally have
>semantic explanations.
That's bound to be true.  The explanation I want, however, may not
necessarily be the "original" explanation, in some cases.

Grammatical relations, IME and IMO, tend mostly to have syntactic uses; most
of the rest of their use is pragmatic (specifically, most of it is information-
packaging).

But nearly all GRs probably originated as grammaticizations of semantic and/or
pragmatic functions/relations/categories.  For instance, "Subject"s are
obviously grammaticized "Topic"s.  But they're not the _same_ as Topics.  For
example, in _every_ language _every_ Topic _must_ be definite.  In some
languages, every Subject must be definite too; but in some languages,
Subjects are allowed to be indefinite.

>I am not familiar with the words "grammatical relation" and
>"quirky case", so I don't see their benefit yet.
As for GRs;
look at the references I've already given;
look at
< http://www.us.oup.com/us/catalog/general/subject/Linguistics\
/SyntaxMorphology/?view=usa&ci=9780199264018 >
or
< http://tinyurl.com/27cgj9 >
; and Google for "Grammatical Relations", which gives 1,870,000 hits, the first
ten of which are pretty good; six of those are online and four are books.

For you to answer my question I suspect I do need you to understand GRs.

As for "quirky" or "non-canonical" case or subjects or objects, there are some
threads on this CONLANG List that discuss them, and I've participated in some
of those threads; look them up and read them.

Also look at "Non-Canonical Marking of Subjects and Objects", edited by
Aikhenvald, Dixon, and Onishi.
< http://www.benjamins.com/cgi-bin/t_bookview.cgi?bookid=TSL%2046 >

And Google on "quirky case", "quirky subject", "quirky object", "non-canonical
case", "non-canonical subject", "non-canonical object".

I don't think you really need to understand "quirky case" to answer my
question, but it's kind of fun anyway.

>When you're asking, what's the advantage of additional cases, but
>you don't want a semantic answer,
Well, I'd love a semantic answer; but I don't expect one.

The additional cases I'm asking about are not Semantic Cases; they're
Grammatical Adverbial Cases.

In the articles in Chris_Bates_Maths_Student's dokuwiki I referred you to,
you'll see that many of the uses of Syntactic Subjects are indeed syntactic.  I
expect the same to be true of the various Objects; not only the syntactically
most-privileged Object, but also lesser Objects.

There is a proposed information-packaging use to the grammatical relations.  It
has been proposed that all the participants in every clause are strictly linearly
ordered by empathy; and that the empathy hierarchy is the same as Comrie
and Keenan's "Noun-Phrase Accessibility Hierarchy".  Furthermore it has been
proposed that every subordinate clause's ordering of its participants must
match that of all of its containing clauses, including its matrix clause and the
main clause (if they are different); so if a subordinate clause has two or more
participants in common with one or more of its containing clauses, the order of
empathy in the subordinate clause must match that in its containing clause(s).

>then I don't know an answer.
If you think of one, I hope you'll tell us; or, at least, tell me.

>Why is it not *"I think you" but "I think of you"? If I don't
>consider semantics, then I must say, no idea, it's just quirky. If I
>consider semantics, then I can say, maybe there's a semantic
>difference between "thinking" and plain transitive verbs, as the act
>of "thinking" does not have a simple goal, but evoques an
>impression, and I could go on searching for similar cases in English
>and in other languages.
Aikhenvald Dixon and Onishi's book shows that there are several features (at
least many of them semantic, probably most of them semantic) that are more
common among "quirky" clauses than among "semantic" clauses; and
conversely "quirkiness" is more common among clauses with these features
than those without.  But none of these features, and no combination of these
features (whether "anded" or "ored"), is either necessary or sufficient to
generate "quirkiness".

One of them applies to "think".  Dixon divides verbs into Primary-A verbs,
Primary-B verbs, and Secondary verbs.  Secondary verbs are those verbs
which _must_ have a sentential or clausal argument.  Primary verbs are those
verbs which can get away without any of their arguments being clauses or
sentences.  If I remember right, Primary-A verbs are those Primary verbs which
_cannot_ take a clause or a sentence as a participant.  Primary-B verbs are
those Primary verbs which can have a sentential or clausal participant.

Clearly in English "think" is a Primary-B verb; and its glosses in other languages
are usually going to be either Primary-B or Secondary.

>---
>grüess
>mach
Thank you, very much, for all the effort you've already put in.

I hope I'm not asking too much of you.

But in any case, thank you in advance for any future effort you put in to
answering my question.