[Corpora-List] ad-hoc generalization and meaning

[Mike Maxwell]

Rob Freeman wrote:
> My assertion is that the fact you can find generalizations which are 
> both true _and_ not true in this way among word associations in corpora, 
> may tell us something interesting, fundamental, about natural language. 
> ...
> *(The observation that "supported" = "accompanied" but also "supported" 
> != "accompanied" a couple of messages back was one. The observation in 
> the last thread that "done" and "made" are in the same class in the 
> context of "study": "do/make a study", but they are not the same class 
> in the context of "attempt", was another.)

But people have known these things for centuries, and codified them in 
the form of multiple senses in dictionaries--at they same time as they 
assigned the *same* part of speech to these multiple senses (or not, of 
course, depending on the word!).  Now you can argue that the parts of 
speech they used were too course, or too close to Latin, and you would 
have a point--a point that linguists have made for decades.

In sum, no one doubts that 'support' has (at least!) two meanings.  At 
the same time, it is also clear to most of us (OK, I'll say "to some of 
us" :-)) that 'support' can be a noun or a verb, and still have these 
senses; and that it is useful, for a number of reasons, to view part of 
speech and meaning as at least somewhat orthogonal and distinguishable. 
  That is, syntax and semantics, while interacting, are distinct.

One advantage of splitting apart syntax and semantics in this way is 
that you begin to see how you can understand long distance dependencies 
of meaning, for example
   The support that John brought was helpful to making our argument.
'support' means one thing; whereas in the following sentence
   The support that John brought was helpful for bracing the building.
it means something quite different.  Syntax allows you to relate the 
choice of sense in each case to words at an arbitrary distance away, in 
a way that collocation *by itself* cannot do.

(Since this is the corpus list, I must confess I made these examples up. 
  I have no problem with that, and if you do, then wait until they get 
archived somewhere--at which point they will become part of someone's 
corpus :-).)

In sum, I think we all agree that there is more to language than syntax 
(and I'll point out that no generative linguist that I know of ever 
thought otherwise, unless you count generative semantics as 
"otherwise").  At the same time, it seems--at least to me--that there 
are clearly reasons for making a division between syntax and semantics. 
  Dividing the problem up is at least a priori a useful strategy for 
finding a solution.  You might find later on that the division you 
started with was wrong, but that's life.  And in this case, I haven't 
seen any reason thus far to think that the division between syntax and 
semantics is wrong, although there are clearly individual phenomena for 
which it's not clear which side of the boundary they lie on (like maybe 
the unergative vs. unaccusative distinction).
-- 
	Mike Maxwell
	maxwell at umiacs.umd.edu
	"Theorists...have merely to lock themselves in a room
	with a blackboard and coffee maker to conduct their business."
	--Bruce A. Schumm, Deep Down Things

_______________________________________________
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora