Arguments vs. Adjuncts

[Andy Kehler]

I just spent some time teasing apart my intuitions about John's
examples, so I thought I would forward them to the list.  Exerpts from
John's post are indented, my comments are not.    

   I am trying to understand the difference between argument and adjunct PPs.  One
   of the purported differences between them is that you can only have one
   instance of an argument PP whereas you can have multiple instances of adjunct
   PPs.  However, I don't have any problems with sentences like the following:

	   (1) John relied on Bill, on his ability to get things done.
	   (2) On whom did John rely on his ability to get things done?

The only interpretation I get for these sentences is that in which the
meaning of "on" is paraphraseable by "based on", i.e.,

  (1') John relied on Bill, (based) on his ability to get things done.

In this case, the two `on's have different meanings, and thus are not
both arguments to "rely".  I take it that John may be getting this
reading too, because it is this reading which is supported by the
assumptions that he postulated in order to make sentence 5 (awkwardly)
acceptable:

   There are some restrictions on the arguments of multiple PPs.  One restriction
   is that they must be "unifiable":

	   (5) #John relied on Bill, on Tom's ability to get things done.

   (I can only get this if I postulate a relationship between Bill and Tom's work,
   say, that Bill is Tom's boss and can tell him what to do.  Even then it is
   awkward for me.)

That is, you need a context in which it makes sense for someone to
rely on Bill _based_ on Tom's abilities.  So I don't think there is
anything having to do with unifiablity here.

   What is interesting is that adjunct PPs have the same restriction:

	   (6a) Bill panhandled in New York City in Central Park.
	   (6b) #Bill panhandled in New York City in Golden Gate Park.  (Golden
   Gate Park is in San Francisco.)
	   (6c) In which park did Bill panhandle in New York City?

   Notice that put, which does subcategorize for location, has exactly the same
   behavior:

	   (7a) Bill put the statue in New York City in Central Park.
	   (7b) #Bill put the statue in New York City in Golden Gate Park.
	   (7c) In which park did Bill put the statue in New York City?


   So this would suggest that adjunct PPs may not be adjuncts at all but rather
   multiple PPs that together fill a single slot in the semantic frame (in the
   case of (6a), the location slot).

The only way in which I interpret sentences (7a)-(7c) is if "in New
York City" modifies "statue" (unless there is an exaggerated pause
before the second "in", as with an afterthought).  Then (7b) is good
too, although semantically odd unless I assume that "statue in New
York City" means essentially "statue that was located in New York
City".  This is in contrast to (6a) and (6c), which are fine assuming
there is a single panhandling event (or habitual event) being
described.

However, it is true that if you are going to fill the same slot in the
semantic frame with different values instead of with different
descriptions of the same value (for instance, if there is more than
one event occuring in different locations), then you need to conjoin
with adjuncts just as with arguments.  For instance, if (6b) is meant
to mean that Bill panhandled in two different places, then you need to
conjoin the locatives just as with (7b), in which the locations are
arguments.

Just my two cents, minus the pennies.

  -- Andy Kehler