And Rosta wrote:

  > #What grammar does is give you cues for the interpretation. So, in:
  > #
  > #(2) How many could she?
  > #
  > #HOW MANY might invoke the concept *cookies*, and uncomplemented
  > #COULD invokes the concept *eat*. These are not syntactic or
  > #phonological elements per se, except through spreading activation.
  > #I don't deny that when structures are compared, you can "fill
  > #in the missing pieces" but that's a bit post hoc.
  >
  > But would you have the grammar generate (2) as a syntactically
  > complete unit?


Yes, why not?

  > Some sort of rule that says a filler needn't match
  > up with a gap, if there's been VP ellipsis?


Strictly speaking, there's no gap or elipsis here because COULD
works like a pro-form. I don't see the problem.


  > How would you deal with
  >
  >   How much umbrage would he <be likely to take>?
  >
  > -- assuming that UMBRAGE must be complement of TAKE.


Wait a second: I don't think you can say:

*? How much umbrage would he?

But of course it would be predicted that TAKE would appear
in the sentence--that's just plain statistical learning.



  > #The problem with invisible words can only be seen when they are
  > #claimed to occur before or nonadacently after the valence carrying
  > #element. For example, we know that people move there vocal aparatus
  > #when they read silently. I doubt you get any movement after SHE in
  > #(2). If it were to happen, then I'd change my mind. Other findings
  > #like this might make me change my mind too.
  >
  > This seems silly to me. Surely it would be the case that when people
  > move their vocal apparatus during silent reading it is the PHONOLOGY
  > of the sentence that determines that movement -- exactly as happens
  > when speaking aloud. So if a word has null phonology, then exactly
  > what you'd expect is that it *wouldn't* trigger movement of vocal
  > apparatus.

I had a feeling you would say this. And maybe it's true (but,
then again, who would have ever said that the McGurk effect and
the Stroop effect were true--and they are).

So, what I'm trying to say is that I want empirical evidence to
bear on this issue. As far as I can tell, most empty-category
analyses are arguments from elegance, but that's not what I'm
interested in.

Joe
_________________________________________________________________
Home page:        http://lingua.fil.ub.es/~hilferty/homepage.html
__________________ http://www.ub.es/filoan/hilferty/homepage.html