A better filename (to understand what it is).

[tlg_WH_sluicng.git] / postscriptum-2011-04-24.txt

-*- markdown -*-

Author: Ivan Zakharyaschev <imz@altlinux.org>

I presented this work under the title: 

"Type logic served by co-Merge, Merge and Move: an account for sluicing and
questions of `common European' and Japanese types"

at one of the [predecessor conferences to MOSS], 
namely [Syntactic Structures 2007] in Moscow.

Since people might be interested in how some things I have been exploring
can work out, I'm posting a 2-page [abstract] and a version of the
[handout] for that work of mine. 

I wrote these remarks quite a while after 2007, being encouraged by the
atmosphere and discussions at [MOSS2] (April 2011).

[abstract]: http://hal.inria.fr/docs/00/14/06/68/PDF/synt-Abstr.pdf
[handout]: http://hal.inria.fr/docs/00/14/06/68/PDF/synt-handout.pdf

(Archived at <http://hal.inria.fr/inria-00138201/>; saved locally as
[[imz_TLG-for-sluicing-and-wh-in-situ_abstract-2007-04-02.pdf]] and [[imz_TLG-for-sluicing-and-wh-in-situ_handout-2007-04-09.pdf]].)

[Syntactic Structures 2007]: http://linguistics.ucla.edu/people/grads/korotkova/moss/previous/ss07_program.html
[predecessor conferences to MOSS]: http://linguistics.ucla.edu/people/grads/korotkova/moss/previous/
[MOSS2]: http://linguistics.ucla.edu/people/grads/korotkova/moss/index.html "Moscow Syntax and Semantics 2 (2011)"

Let me make some remarks:

0) I also presented this in a seminar with Mati Pentus, Michael
Moortgat and several Mati's students at Moscow State University in 2007. Based
on their comments, I would make further improvements to the
presentation in the handout, also point to more related literature.
(Vermaat on "movement" in TLG; also, at that time, the suggestions of
a continuations-based semantics for the symmetricized Lambek calculus
had only been worked on by Bernardi&Moortgat, and others, so I have
only a vague statement in the abstract that a continuations-based
semantics must be there for this calculus of types, with no more
references to the research on this. ALso the references on the history
of the employed calculus are missing.)

Unfortunately, I haven't yet recovered my computer from a recent
crash, and haven't yet recovered an updated revision of the handout.
Once I have it, I'll try to make it accessible (to you and others) as
well; perhaps, I'll post it to <http://semanticsarchive.net>. (Unfortunately,
I've been a perfectionist: I thought I would post a version of the
handout onto the web after correcting the technical and stylistic
problems in it.)

In the handout, as compared to the abstract, you'll find the relevant
type derivations spelled out, and also more remarks and discussion.

I was thinking about people not familiar with type-logical grammars
when I wrote this. (This explains some choices I made as to what to
write.)

1) The main interest for me was to capture with a single type some
uses of WH-items that are not interrogative sentences, namely,
sluicing (where the WH-item hangs at the end and refers to preceding
expressions) and "sluicing-based indefinites" (as in "He ate God knows
what" or "He ate we don't know what").

My hope was that the more powerful type system of TLG would be
flexible enough to derive the types that reflect the different uses of
WH-items. This is a problem  for syntactic frameworks with a fixed set
of fixed syntactic categories like the "generative grammar": from the
view of a generative-style approach, a sluicing-based indefinite looks
like a sentence from inside, but behaves like a DP from outside, so
one would need to plug in some covert "adaptor" categories in between
the inner and outer level, which is not nice.

2) Then I had the idea that one could predict the different types of
languages w.r.t. the patterns of how WH-items are used in those
languages: starting from different base types for the WH-items in
different languages, one would derive different patterns of the usage
of WH-items. And I was excited to arrive at such a picture taking
English vs. Japanese as examples. This is depicted with the diagram.

As I have understood from the papers on Japanese syntax I looked into,
Japanese sluicing is not really sluicing because--among other
things--it is island-sensitive; so this is accounted for by showing
that the usual sluicing mechanism shouldn't be available for the type
I suggest for Japanese WH-items.

I should have been accurate and should have put an example and a
reference for this very fact (island sensitivity) into the handout. (A
todo for me.)

3) I hope you won't be confused with 2 different uses of the word
"type" in the texts. (Actually, for my presentations, I would
substitute the word "category" for one of the uses of the word "type"
-- in order not confuse the audience, which wasn't familiar with TLG.)

4) A better name for "sluicing-based indefinites" might be
"sluicing-based XPs" ("XP" hints that there are not only DPs, but
there are also "how", "where", "which", and so on).

5) Please don't be confused by the fact that I denote the type of a
complete interrogative sentence as "q" -- some people were inclined to
think that "q" is a notation for the single WH-items (or perhaps
quantifiers). (What could be a better notation for the type of a
complete interrogative sentence then?..)

6) Actually, I also should not have a single "q" type, but rather I
should "index" the "q" type with the type of what is asked for ("dp"
for questions about entities, introduced by "what" or "who"; a
different index for questions introduced with "how"; with "where"; and
so on).

7) Here, my designation of only the "movement" and "merge" module as
"(low-level) Syntax" is not something essential. Of course, one can
include also the TLG module into "Syntax", or perhaps call it
"Syntax-Semantics interface". It's just a matter of words.

8) I consider this to be only a preliminary interesting idea, which is
not very convincing until, first, we have a corresponding semantics
which would make sense, and second, have an understanding of
"movemeant" (particularly, of "WH-movement") in TLG, and third, know
how to make sense of the bad stipulation I'm making there (the
conversion from the dual types).

Of course, these three things are interrelated.

It's better to think about the analysis ingredients in a "modular"
way: although I had to suggest a particular mechanism, say, for
"movement" in order to be able to present the overall view on how the
WH-based constructions can be worked out in a type-logical fashion, I
do not insist that we should stick to these particular ancillary
suggestions. (Simply, the presentation would become too abstract
otherwise.) I expect they can be replaced by other analyses someone
comes up with.

A more tricky type for the fronted WH-items is to be expected, but I
don't know what it should look like exactly; it's oversimplified for
now.

(Yesterday, in [Daniel's talk][Gutzmann2011] on modal particles in German, we saw a
simplification of the word-order issues -- I think, its role was
similar: it is done in order to be able to express the ideas about the
particles independently of other complications, but still formal
enough.)

[Gutzmann2011]: http://linguistics.ucla.edu/people/grads/korotkova/moss/files/2011/a_gutzmann.pdf "Modal particles: Deriving syntax from semantics"

9) Also, there is the question of how to treat inflectional
morphpology (particularly, on the WH-words) in TLG in a language like
Russian, but also in German, and perhaps even English (with some
remnants of cases). I'm putting this issue aside.

10) (A longer note on my guess about the possible similarity of the TLG
analyses for Sluicing-based XPs and for the "sau"-phrases from [GutzmannTurgay2011]
has been extracted from this file to [[sluicing-based-NPs-and-sau_Gutzmann.mdwn]].)

[GutzmannTurgay2011]: http://linguistics.ucla.edu/people/grads/korotkova/moss/files/2011/a_gutzmann_turgay.pdf "Some (in)definiteness effects of expressive intensifiers"


I'm writing this just in the hope that some of this can turn out to be
interesting for you; if it is so, then it's great. If you have
questions, corrections, ideas, please write me; I'll try to reply.

Ivan

Copies of these notes ("mirrors") are at:
* <https://gitorious.org/tlgrammar/wh_sluicing_tlg/trees/master>;
* <http://repo.or.cz/w/tlg_WH_sluicng.git/tree>.

<!-- Local Variables: -->
<!-- Local IspellDict: english -->
<!-- End: -->