AW: SLASH propagation vs. internal merge: Rewind

Tibor Kiss tibor at
Tue Jun 7 02:34:37 EST 2005


I'll try to answer the recent remarks by Bob, Carl, and Detmar.

>From Bob's mail (pseudo-cleft, comparative correlatives):

> Actually I also mentioned the comparative correlative construction, with
the following data:
> No matter how clever the students (are) ___, ...
> The students *(are) very clever.

Correct. Sorry. Do I get it right that the optionality of the copula is what
is at stake here? Or more generally, would one want to provide a
movement-based analysis of the CCC?

>From Carl's mail (pseudo-cleft):

> Consider also:
>   What he may do next is he may just decide to forget the whole thing.
>   What happpened next is (that) it started to hail.
> (Thanks to Elizabeth Smith for pointing out that the postcopular
> phrase in a pseudocleft can be a clause, even if the precopular
> material either has no gap at all, or else has a gap that a clause
> wouldn't fit into.)

Doesn't this point in the same direction, i.e. that one does not want a
movement-based analysis of pseudo-clefts?  (Implying: no movement-based
analysis -> no involvement of CTM).

>From Detmar's mail (partial VP topicalization):

> In what sense is this "an obscure construction" and "not pervasive"?
> Obligatorily coherent verbs are among the most frequent verbs
> selecting verbal complements in German, and we are talking about
> fronting of their VP complement, which is a completely normal word
> order option in German.

As I pointed out, the analysis type is *not pervasive*. The analysis assumes
that something lexical be present at the extraction site, but something
non-lexical identical be present at landing site. The PVP-construction seems
to be the only construction which requires such an analysis.

Matters of frequency are not at stake here. The question is: Assuming the
existence of a particular theoretical mechanism, as e.g. allowing extraction
and landing site type differing w.r.t. lexicality, why is this construction
not attested over and over? (Or at least found in one other construction
apart from PVP?)

>From Carl's mail, referring to the general problem:

> I haven't grasped the overall structure of the argument you are making
> here. Is your position simply that SLASH values should be of type
> _synsem_ (or maybe even _sign_), not _local_, and that position simply
> IS the CTM couched in HPSG terms? If so, then are you advocating typer
> _synsem_, or type _sign_, or just not doing HPSG?

No, my question was: assuming that SLASH values are less than a full copy of
a filler, does anything follow from this partiality? The MP 'advocates' that
SLASH is of type _sign_, while HPSG assumes it is _local_. So one could ask:
What are the differences emerging from this distinction?

>From Carl's mail, re: Chamorro

> 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]].
> >>
> I take you you advocate allowing for the possibility that
> such languages do
> exist?

The problem is that we cannot definitely say that such languages do not
exist. Personally, I would assume the type is unlikely, but I would not
built an argument on top of this assumption.

>From Carl's mail, re: binding

> [1] Which picture of himself_i/j does John_i say Peter_j likes.
> ...
> Does this remain true if LIKES is replaced by STOLE or DEFACED?

I don't understand your question. HPSG-94 took it as a virtue that the
_i-coindexation is not blocked, and I'd say that the _i-coindexation should
be blocked in the majority of the cases in other languages, leaving [1] as a
possible exception to a general rule. [Philological caveat: HPSG-94 did not
speak about other languages than English, I know, but the analysis should
scale up at least a little.]

Finally, Carl on parasitic gaps:

> This is true, but this is not so much a bug in a certain account of
> UDC's as it is is a general HPSG architectural bug, one that was first

Well, it is an architectural bug and a theoretical bug, if you can follow my
considerations re: [3].

[3] [The book]_k,  [[without even reading _k]_j, [I don't know [[how many
reports]_i [Kim [[filed _i] _j]]]]]

In any case, we cannot attack MP by first stating that we have a nice
construction here, which unfortunately leads to peculiar consequences both
from a formal-architectural and from a theoretical-empirical point of view.

In sum, if a movement-based analysis of pseudo-clefts and CCCs are
inevitable and if we can offer a treatment of parasitic gaps which does not
face the problems above (**), then these constructions may be of comparative

(**) I would like to remind you that GPSG did not have the parasitic gap
problem and excluded Carl's initial ungrammatical example as well. In
addition, doesn't [how many reports] in the lower SpecCP already block an
extraction of the adjunct in your initial example?



More information about the HPSG-L mailing list