SLASH propagation vs. internal merge: Rewind
tibor at linguistics.ruhr-uni-bochum.de
Mon Jun 6 14:38:44 EST 2005
I would like to clarify what I had in mind when I posed my question.
One point was to look at the empirical reasons for assuming that SLASH does
not contain NONLOCAL information. The other point was to ask whether the
distinction between partial filler-gap-identity vs. full identity could be
made a crosstheoretical issue.
It seems -- somewhat surprisingly -- that a competetive comparison of
theories is not possible even in areas where the other theory (humbly
referred to as a 'program') is lucid and clear, be it simply for the reason
that we are short of empirical phenomena to justify a theoretical point.
The following arguments in favour of a partial filler-gap-identity have been
1) Partial VP topicalization and the LEX +/- contrast (Stefan Müller, Detmar
2) Principle A and Principle C reconstruction (TK, Stefan Müller)
3) Pseudo-Cleft (Bob Borsley)
4) Parasitic Gaps (Howard Gregory, Carl Pollard)
5) Chamorro-type SLASH marking (Carl Pollard)
I think that the following remarks are ok:
Ad 1) [and 5)]
The analysis suggested by Stefan and Detmar assumes that something is a head
(LEX +) at its extraction site but has to be a phrase (LEX -) at its landing
site. The problem is that the phenomenon is not pervasive, i.e. we know of
very little cases where the landing site type is forced to be different from
the extraction site type. The partial VP analysis seems to cover the facts
but that's it. In addition, it rests on a dubious feature (LEX), which is
NONLOCAL to make the analysis work and for no other reason. [** Fn: People
who know my work on German infinitives will recall that I explicitly
preferred not to have an analysis of PVP instead of adopting one akin to
Stefan and Detmar's solution.]
In any case, I don't believe that the copy theory of movement can be
seriously challenged with an obscure construction and the difference between
LEX+ and LEX-.
I think that the Chamorro-case should be taken in the same way: The
phenomenon is not pervasive and in fact, we don't know whether languages
exist which actually mark an XP[SLASH Y[QUE ZP]].
These cases show that the 'standard theory of binding' in HPSG (PS 1994) has
problems with reconstruction and nothing more, as has already been hinted at
by Stefan, who offers the following example:
> (2) a. Kennt er_*i Karls_i Freund?
> b. [Karls_i Freund]_j kennt er_*i _j.
While this example could be discussed away by assuming that Principle C does
not exist, we get the same problems with Principle A, witness
 Which picture of himself_i/j does John_i say Peter_j likes.
At first,  seems to show that Principle A in HPSG does it right, because
both indexations are surprisingly possible. Note, however, that structurally
identical examples in German and other languages (one could call it *the
majority*) do not allow the _i-coindexing. So an appropriate cousin of 
runs into the same problems as Stefan's (2b).
I have never worked on pseudo-cleft, but in order to make the contrast work,
one would have to assume that pseudo-clefts are derivationally related to
the ungrammatical examples provided by Bob.
> What he may do ___ is go home.
> *He may do go home.
The parasitic gaps seem to be the most interesting candidates. However,
Carl's recent email made it clear to me that they seem to be as problematic
to HPSG as they possibly are to MP. Carl points out the ungrammaticality of
 can be derived from the observation that "there is no way for the gap in
the upper filler (i.e. _i, TK) to get linked with the object gap."
 *[Without even reading _i]_j, [I don't know [[how many reports]_i [Kim
[[filed _i] _j]]]]
But how can this example be excluded in HPSG? There is no general --
possibly non-local -- condition to the effect that unlinked
filler-gap-relations are illicit in HPSG (there could be a coindexation
_i/_k, where all the information about _i is identical to all the
information about _k).
One could add a constraint requiring that roots do not contain SLASHes
(while  does). However, such a constraint misses the generalization
(which is taken seriously in MP work, but originally led to things like the
Subcategorization Principle) that roots do not like any open dependency
(i.e. a root with open SLASH is as out as a sentence with SUBCAT <NP>).
Even if we ignore this and include such a constraint, how do we block
something like ?
 [The book]_k, [[without even reading _k]_j, [I don't know [[how many
reports]_i [Kim [[filed _i] _j]]]]]
In sum, if I see this correctly, while one may have a hunch that the copy
theory has its problems, it seems that we are unable to take advantage of
these. How unfortunate.
With kind regards,
Prof. Dr. Tibor Kiss
Sprachwissenschaftliches Institut - Ruhr-Universität Bochum
+49-234-3225114 // +49-177-7468265
More information about the HPSG-L