Re: More thoughts about S11 grammar

On Tue, Mar 22, 2005 at 07:39:20PM -0500, Paul Sherrill wrote:
[...]
> Your goal is to get rid of the arbitrary line between arguments and
> adjuncts, but does using suppletive verb forms get in the way of this?  I
> think you said that, for example, there'd be one verb for "to read" and then
> a second for "to be read".  Neither of these requires the other, but if one
> wishes to express the transitive concept "x reads y", one would have to
> phrase it "x-reads y-is.read" and not, say, "x-reads y-is.bought".  But if
> one then wanted to express, for example, "x reads y before dinner", one
> would presumably use the construction, "x-reads y-is.read dinner.follows",
> where the verb "follows" isn't stipulated by the verb "read".  For far,
> there's not much of a problem, until you approach a case that's somewhere in
> the middle, for example "x gives y to z" = "x-gives y-is.given z-receives".
> Contrast that with "x sends y to z", or something, = "x.sends y-is.sent
> z.receives".  At this point, you have to choose whether the two verbs for
> "receive" are different (and therefore presumably take part in the
> suppletion of the other verbs; i.e. you have one set
> give/be.given/receive.gift and another send/be.sent/receive.mail) or not.
I'd say, the choice here can be more than two-way. You can draw a
distinction, for example, between:

	x-gives y-is.handed.over z-receives

and

	x-gives y-is.thrown z-receives

and

	x-gives y-is.mailed z-receives

Having separate verbs for each role gives you *much* more flexibility
in describing the relevant action.


> In the first case, what you end up with is essentially a trivalent verb set:
> to express all three entities that take part in the action, you have to use
> specifically those three verbs, whereas to show the involvement of other
> entities you are more free in your verb choice.  You then have an arbitrary
> distinction between nouns that must be paired with verbs from the suppletive
> set (the equivalent of arguments) and those that need not be (the equivalent
> of adjuncts).
I'd say this need not be the case.


> Even if you decide to have a catch-all verb for "to.receive"  and
> all other such roles, all you've done is made the decision that all
> your verb sets are at most ditransitive.  There's still an arbitrary
> line between which verbs must complement each other and which
> needn't.
Not really. I'd say under Henrik's system, any verb can go with any
other verb; the only requirement being that it must make semantic
sense. For example, you can say that green ideas sleep angrily, it's a
grammatical construction but has no semantic meaning. Similarly, you
can say "x-gives y-follows z-runs", it'd be grammatical but
meaningless. But I think the logical thing to do is to allow such
mixing of verbs where it *does* convey a coherent meaning, such as
"x-sends y-carried z-receives", or "x-sends y-mailed z-receives", or
even "captain-sends report-carried king-convinced" - the captain sends
a report that convinced the king. Notice how the English requires two
clauses whereas the S11-equivalent accomodates it in the basic
sentence structure.


> So although arguments, a given number of NPs required by a given verb, don't
> exist as such, there seems to be an equally arbitrary set of verbs that must
> come in pairs/triples/whatever (assuming one wants to express all the
> relevent roles).
[...]

Semantically speaking, verbs have a LOT more roles than what one might
normally consider. Semantics is a very complex realm. For example, we
normally think of "to see" as a divalent verb involving a seer and the
thing seen. But semantically speaking, there are a lot more roles that
can accompany "see".

For example, the time of the seeing, the manner of seeing (by
accident, by conscious watching for the event/thing, by mental
realization, etc.), the place of seeing, the sight itself (as opposed
to the object seen, e.g. "I saw that the mountain was green"), the
time interval (I saw it for a moment), etc.. And who's to say that the
two "main" roles, the seer and the thing seen, and the most relevant?
What if I only wanted to convey the fact that I have seen something,
but do not wish/need to say what it was I've seen? What if I wanted to
convey the fact that something has been seen (the secret documents
have been seen by somebody), but do not know or do not wish to specify
the seer? What if we were discussing where the secret documents were
uncovered, and I wanted to say "the seeing took place in Room 15", but
do not need to specify the seer or the thing seen?

The relevance of any role, IMHO, is context-dependent. Some roles may
occur very frequently, so to-see and to-be-seen might occur together
most often, but that's merely a consequence of the frequency that that
particular pair of semantic roles is needed. The grammar itself does
not prefer to pair those two verbs above any other; it can pair any
verb with any other verb. It's just that many verb combinations either
don't make sense semantically, or just aren't required very often.

This is my analysis of it, anyway. We'll see what Henrik says. :-)


T

--
Klein bottle for rent ... inquire within. -- Stephen Mulraney

Reply