Filler-gap mismatches

[Ivan A. Sag]

[LONG REPLY]

Hi Ron and Carl,

Ron:
> > 6)a. Before you do that, please think of the possible consequences
> >      and that you might be wrong.
> >
> >   b. *Before you do that,  please think of that you might be wrong
> >      and the possible consequences
>
> This pattern is pretty easy to explain, given the way coordination is
> set up in LFG (e.g. Kaplan & Maxwell, 1988, reprinted in Dalrymple et al
> 1995, from CSLI).  The only categorial matching is defined by a general
> c-structure rule schema that requires, for example, that the category of
> only the left-most conjunct be identical to the category of the
> coordination as a whole, so the NP object of "of" can be realized as the
> sequence
> 	NP and/or S
> but not
> 	S and/or NP
> >>

Carl:
> Did I misunderstand what you meant? What you said seemed to imply that
> the second conjunct of a complement isn't subject to any categorial
> restrictions at all from the head that governs the complement.

Right. So, for example, be allows any predicative XP as complement,
but become allows only AP or NP:

Kim became unhappy/a churchgoer/*in love/*talking to the family.

And these category selections are preserved in coordination:

Kim became a churchgoer and *in love.
Kim became unhappy and *talking to the family.

If you free up selection into noninitial conjuncts, how does this get
captured?

Ron:
> I haven't thought about  7), and it doesn't strike me as particularly
> good.
>
> > 7) ?Sandy could think of only that he might be wrong, and not of
> >    what the actual consequences might be if he were right.
> >>

Carl:
> Try replacing THINK OF with TALK ABOUT?

I tend to agree with Ron about this, but if I really work at it, the
following seems to just about make it acceptable for me:

   Sandy was finally able to talk about not only that he might be wrong, but
   also that his entire system of judgment might be flawed.

But I'm struck by the amelioration effect of parallelism, etc. Remember that
Chomsky (1955) cites passivization examples like the following as
unquestionably ungrammatical:

*England was died in by Kim.

But of course coordination and parallelism can upgrade such stuff. The
following sounds like Churchill might have said it:

Out beloved country, that has been lived in and died in by so many for so
long, ....

Is this processing complexity rescuing an ungrammatical example from
unacceptability?

Ron:
> An argument from somewhere in the literature (I talked to Annie about
> this and we couldn't remember where this kind of thing was discussed)
> involves the contrasts:
>
> 9a)  That the Earth is flat is obvious.
>  b)  *Is that the Earth is flat obvious?
>  c)   Is the fact that the earth is flat obvious?
>
> 10a)  Whether John left isn't clear.
>   b)    Is whether John left clear?
>
> The contrast in (10) suggests that the subject position in an inverted
> sentence requires a true NP, and that-clauses don't qualify.  But (11)
> shows that a "whether-clause seems perfectly OK, and thus that a
> whether-clause is more nouny than a that-clause, maybe is in fact an NP.
>  In other words, whatever explains the acceptable of 10b) would explain
> the acceptability of 8).
> >>

Carl:
> This account seems based on a conception of category where the
> category of an expression is determined externally (by what
> environments it can occur in) rather than internally (what its
> internal structure is). So then, on that conception, are the verbal
> gerund DELIBERATELY SLAUGHTERING RARE WARBLERS and the PP UNDER THE
> BED actually NPs in
>
> UNDER THE BED sure beats the commodities market if liquidity is what
> matters to you.
>
> I prefer UNDER THE BED to under the dresser.
>
> I have always thought of UNDER THE BED as a stupid place to store
> the good crystal
>
> DELIBERATELY SLAUGHTERING RARE WARBLERS sure beats the commodities
> market if irrational exuberance is what matters to you.
>
> I prefer DELIBERATELY SLAUGHTERING RARE WARBLERS to spectator sports.
>
> I have always thought of DELIBERATELY SLAUGHTERING RARE WARBLERS as
> an efficient way to offend liberals and other losers.
>
> If so, this is a quite different conception of syntactic category
> from the one that has been used in most HPSG work that I'm familiar
> with, a difference that goes beyond the matter of which features, if
> any,  are part of the syntactic category (e.g. case,  verb form,
> and syntactic selection of valents in HPSG but not LFG).

External distribution is surely part of category determination.  And I think
the analysis of these things is still an open issue (in any
framework). Categories need to be more fine grained than standardly assumed,
as Pullum and Zwicky have argued. So, for example, they point out that
possessives are not totally free. Only certain `NPs' appear there:

 Kim's book
*[[Kim's leaving]'s possibility] is negligible. (cf. [The possibility
of [Kim's leaving]] is negligible. )

 *There's being five people in the room was obvious.
 *It's raining had been predicted.

English grammar must distinguish between various things that are commonly
referred to as `NPs'. Note also that a plausible approach to gerunds -- in
terms of fine-grained category grain -- is developed by Rob Malouf in his
recent CSLI book.

So the issue Carl raises might be stated in terms of questions like these:

Given a more fine-grained category space,

(1) what category exactly is the post-aux element in the SAI construction
(Ron's example) restricted to?

(2) what category exactly is the notion `verbal subject' restricted
to (Carl's example)?

(3) what category exactly is the notion `raised argument' restricted
to (Carl's example)?

(Of course, the answer to (3) might be determined by one's answer to (2).)

So an analysis with a NP[...] --> PP construction in it might well be part
of this picture, depending on the details of the [...] and what
one decides is the right [...] for SAI constructions. I don't think
we can rule it out.

Ron seems to be assuming that there is a simple notion of NP that will be used
in the grammar of inversion constructions and Carl seems to be assuming that
NP[...] --> PP analyses are incorrect/unmotivated. I'm not sure that either
assumption is right. Of course, I don't have a completely worked out
analysis of these problems either....

HTH
Ivan