Re: Fishing for Complements?

This brings up the interesting issue of "syntactic" vs. "semantic"
roles.

My extremely amateurish explanation follows.

The classic "syntax independent of semantics" position has two
different *kinds* of roles -- "subject" and "object" are syntactic
roles and are a part of grammar, not of meaning.  "Agent" and
"patient" are semantic roles and are part of meaning, not of grammar.
You can tell them apart because things like passivization redistribute
semantic roles among syntactic roles, and some verbs (like "suffer")
have an inherently unusual distribution of semantic roles among
syntactic roles.  (If you step away from highly isolating languages
like English, you get an additional layer of complexity -- in addition
to syntactic and semantic roles, there are also *cases*, like
nominative, accusative, and so on, which are usually but not perfectly
correlated with the syntactic roles, just as the syntactic roles are
usually but not perfectly correlated with the semantic ones!)

Given this formulation, it would be perfectly possible to use
"subject" and "object" for the language's two syntactic roles... but
one hesitates to because "object" is so strongly tied in our minds to
"patient."

By the way, this separation has been challenged by grammarians who
don't believe in the complete independence of meaning and grammar.
People like Ronald Langaker and George Lakoff and Charles Fillmore
prefer to suggest that "subject" *is* a semantic role, but a more
subtle and general one than "agent" -- even though they often, but not
always, coincide.  I don't remember, off the top of my head,
Langacker's exact semantic definition of "subject," but I believe it
involves the "trajector/landmark" distinction, which shows up in
things like the semantic difference between "above" and "below." (The
meanings of the two are identical except for a trajector/landmark
reversal.)  I believe Langacker identifies the subject as the
"trajector" in the image which is the meaning of the sentence.

This distinction would be invisible to a semantic theory which was
based on truth conditions rather than cognition, since it is a
distinction in how one *views* or, as Langacker says, *construes* a
situation, rather than an objective difference in the situation
itself.

So the difference between "the man hit the dog" and "the dog was hit
by the man" would be parallel to the difference between "the star
above the moon" and "the moon below the star."

Ed

Ed Heil
edheil@postmark.net
=========================================================
Worship the potato? The idea seemed silly to me. But then
I thought, what else is more deserving of worship? It's simple,
it comes from the earth, and it can kill you if you disobey it.
   -- Deep Thoughts by Jack Handey (Saturday Night Live - NBC)

Nik Taylor wrote:

> Jim Grossmann wrote:
> > However, the subject can be an agent or a patient or a bunch of other
> > things, depending on the clause-initial particle.
> [snip]
> > QUESTION:    What do I call the argument that comes before the subject?
>
> I don't understand how "subject" even applies to what you're calling
> "subject".  If it can be patient or agent, it seems rather a stretch to
> call it subject.
>
> --
> "[H]e axed after eggys: And the goode wyf answerde, that she coude not
> speke no Frenshe ... And then at last a nother sayd that he woulde haue
> hadde eyren: then the goode wyf sayd that she vnderstood hym wel." --
> William Caxton
> http://members.tripod.com/~Nik_Taylor/X-Files
> http://members.tripod.com/~Nik_Taylor/Books.html
> ICQ: 18656696
> AIM Screen-Name: NikTailor
>