Re: THEORY: Why more than two grammatical relations?

On Thu, 11 Oct 2007 22:04:20 -0400, J. 'Mach' Wust
<j_mach_wust@...> wrote:
>[snip]
>The distinction of arguments and adjuncts is familiar
>to me, though I didn't know the English terms. I
>used "complement" as an ad-hoc translation
>for 'argument' in my last post. The distinction of
>core and oblique, however, is new to me.
These ideas, themselves, are parts of most theories.
But the terms I've used for them are theory-specifc, and not all specific to the
same theory.
I think I mixed RG (Relational Grammar)'s terminology with RRG (Role and
Reference Grammar)'s terminology, and maybe some others'.

>[snip]
>This seems to be a rather fuzzy distinction, not a
>binary one, so there will be a continuum between more
>core phrases and more oblique phrases.
Well, I know that the philosophy behind Keenan's "Subject Properties List" is
that syntactic subjecthood is fuzzy or a continuum; but I'd never heard before
that the "core vs oblique" distinction was.  (Nor that it wasn't; I've just not
been told much about that.)

>Consider that even the semantics of prepositional >arguments may depend
on the context,
>e.g. "Homer trembled with anger" (cause/mode)
>vs. "Homer agreed with Lisa" (comitative?)
>or "the  consisted of one room" (part)
I think somebody's "net nanny" censored a word.

>vs. "the dog died of love" (cause)
>or "Bart benefitted from school" (cause)
>vs. "Bart stole from school" (patiens?).
(Somebody watches the Simpsons.)
The difference between "core" and "oblique" is meant to be mostly syntactic,
rather than largely semantic; just as the difference between "argument"
and "adjunct" is meant to be mostly semantic, rather than largely syntactic.

>And I think the semantics depend more on the meaning
>of the verb than on its form,
I am not surprised that semantics depends on meaning.
The thing is, the semantic _role_ played by a GR cannot be _determined_
without consulting the verb.  Perhaps it can be _almost_ determined.
And, depending on the language and on the verb and on which GR it is, this
can still vary from one form of some verbs to another form of the same verb.
The "form" that tends to matter for this purpose, tends to be
called "grammatical voice".

>compare "the mob gathered" (subject results from
>action),
>"Milhouse sickened" (subject suffers action),
>"Mr. Burns 'excellent' closed the meeting"
>(instrumental subject – not sure whether that's
>grammatical in English; it is in German),
Yes.  (Simpsons much?)

>"a change happened" (subject defines the action),
>"it rains" (empty subject).
>[snip]
>Ah, I can understand these words more easily
Good!

>(though I wonder what the difference would be that
>goes beyond the purposes here...)!
AFAIK there isn't one.  I'm going to assume there isn't one until someone
claims otherwise; but I won't be surprised when (if) someone does.

>[snip]
>After some reading, I guess that the grammatical
>relation of a constituent is the class it belongs to
>with respect to grammatical phenomena such as
>agreement, word order, ommissions in coordinated
>sentences or reflexives?
That's my guess.

>That'd mean that grammatical relations depend on the
>language (that is confirmed by your question that
>their number may vary).
Yep.

>However, I still don't understand the term
>sufficiently because I don't see how a grammatical
>relation differs from a case, or if you will, from an
>adpositional phrase.
Some cases and adpositions mark semantic cases or adnominal cases.  They
don't mark GRs, which are syntactic adverbial cases.

The difference between the syntactic adverbial case and the GR is like the
difference between the street-signs telling people the name of your street,
and its actual space, pavement, kerbs, easements, sidewalks, and street-
lights.

Subjects, or Most Syntactically Privileged (MSP) Arguments, are mostly "zero-
marked" in most languages.  This is especially true of sole participants of
monovalent clauses.  In verb-medial languages, the MSP Object (the 2nd-MSP
Argument) is often shown by word-order; it's often not shown by a case mark
or adposition in these languages.  In verb-initial languages the various Objects
are often shown by some kind of agreement marking on the verb.  In verb-final
languages the various Objects are often shown by some case-mark or
adposition.

>And this is precisely the point where quirkiness
>seems to originate, if I understand it correctly: A
>quirky case is where the overt case marking differs
>from the usual "grammatical relation" of that case.
Close.  It's where the case-mark or adposition used to show which GR that NP
is in, is not the usual case-mark or adposition used to mark that GR for other
clauses.

>An important reason why you believed the examples of
>dative and genitive objects of my first post in this
>thread to be "quirky objects" seems to be that they
>can be transformed into passive voice without
>changing their case:
Somewhere (probably in one of the references I've already given you) I read
that this is an important criterion in some languages and not much help in
some others.  IIRC there appears to be a contrast between German and
Icelandic.

>[examples and discussion snup -- thanks]
OT:
Turkish allows "double passives", where you passivize a verb that's already
passive.
Hindi(? some Indian language anyway) allows morphological "double
causatives", where you morphologically causativize a verb that's already
morphologically causative.
Some languages allow you to passivize or anti-passivize a monovalent or
intransitive clause; if the original was monovalent, this results in a clause with
no core arguments; it may result in a clause with no arguments.

Back on topic:

>From that I'd say that there's nothing quirky about
>German dative or genitive objects with intransitive
>verbs except that they are the verb's only object,
Well, unless the genitive marking or adposition is just like some GR's marking or
adposition, the genitive is always a quirky way to mark a GR, because the
genitive is an adnominal case, and GRs are usually marked with adverbial
syntactic cases.

But the impression of "quirkiness" in an "S E" clause (one with a subject and an
indirect object but no direct object) is probably due to lack of familiarity
with "S E" clauses.  They're not so much "quirky" as heretofore ignored.  So
you're right about the dative.

>but that seems to be a poor justification for
>quirkiness
Yes, but for the reason I gave above, not for the reason you give below.

>because for the same rationale, every adpositional
>argument would be another quirky object,
No, not those that were oblique arguments anyway.

If the adposition or case-marking that marks an NP in a particular clause as
occupying a particular GR, usually is an adnominal or semantic case-mark(er)
and usually doesn't mark its NP as occupying that particular GR for other
verbs, then it's "quirky".  It's also "quirky" if it usually signals a different GR
than the one it signals in this event.

>so there would be a quirkiness inflation.
I don't think it would be as bad as you feared, but I can't rule out that there
might be some.

>But I don't claim I have much insight in that
>subject, so please correct me if I got it wrong.
I don't have enough insight, myself, to correct you.


>I just have sneaked around in the discussions here,
>and then I've tried to understand some
>of "Quirky “subjects” and other specifiers" by
>Gisbert Fanselow, though I had a hard time
>understanding any of it since my knowledge of
>generativist terminology is very limited:
Farrell's "grammatical relations" book examines GRs, including the quirky ones,
from several different theories' standpoints; it should help.

Aikhenvald&Dixon&Onishi's collection's summaries are theory-neutral; so they
should help too.

< http://www.ling.uni-\
potsdam.de/~fanselow/d/files/fs_fanselow.pdf >

Thanks for the link!

>[snip]
>Well, there. I'll have another try anyway:
Thanks.

>It seems to me that the German dative is not much
>different from the English prepositional phrases
>with "to". The semantics are quite predictable. Both
>can occur as the second argument of an intransitive
>verb, for instance "she listened to him" – "sie hörte
>ihm zu" she.NOM listened him.DAT [detached part of
>the verb]. Incidently, both languages may use these
>as a replacement of constructions with double
>transitive
("ditransitive")

>verbs (in German only in special cases, though), for
>instance "she teached him a trick" – "sie lehrte ihn
>einen Trick" she.Nom teached him.ACC a trick.ACC. The
>involvement in grammatical phenomena is quite low, so
>I'd say both the English to-phrases and German dative
>would be more "oblique" by your definition than
>nominative and accusative.
I didn't know English and German were so similar in this regard.

Recipients are human goals, so the semantics of the dative and the allative
are similar; so it's no surprise that one of the ways of indicating the recipient
in English and German is by using the allative adposition ("to" oder "zu").

In English, besides the "S V DO to IO" construction, there's also the "S V IO
DO" construction; and you've said that German is similar.

In most languages that have a third grammatical relation it is more similar to
the oblique arguments than the first and second GR are.

Many people consider the question of whether or not English has a third GR to
be an unsettled one.  Since English is a well-studied language, chances are
the question is destined to remain unsettled until the language changes.

>The "datives" of the romance languages of Southern
>and Western Europe would be a little less "oblique"
>because they often agree with the verb.
Good point.  That seems so.

>I guess that in languages with many cases, most would
>be very "oblique".
Indeed.  There are (usually?) at most four or five grammatical cases, of which
(usually?) at most three or four are adverbial; so in languages with more than
six or eight cases, most cases are semantic adverbial cases.  In fact, only a
few semantic adverbial cases are not locational or directional; so in languages
with many cases, most cases are locational or directional.

>Or like this:
>The more grammatical relations, the more oblique.
Not sure what you meant.
Perhaps something like:
"The more GRs a language has, the less distinction it makes between the 'later'
GRs and oblique arguments."?

>But that's just some speculation of mine.
Well, I'd bet that way, too.

>[snip]
>Sure it's fun!
Thanks!

>---
>grüess
>mach