Movement's just another word for nothing left to merge

[Stefan Müller]

Hi Tibor,

> One crucial idea within HPSG was that not all information is available in
> the trace-filler relation, but only LOCAL information. If this idea yields
> empirical predictions (I don’t remember), it could be used as a weapon
> against the full-identity copy theory of the MP.
>
> Do I would like to ask: Do we find empirical consequences of the idea that
> SLASH only carries LOCAL information?

Well, yes. There is a problem in German, if you assume full reconstruction.

(1) [Ein Märchen erzählen]_i wird er ihm _i müssen.
     a   fairy.tale tell     will he him    must
    `He will have to tell him a fary tale.'

Usually `müssen' takes only simplex verbs or verbal complexes as
arguments (it constructs obligatorily coherently). In (1) the filler is
a phrase so it does not fullfill all requirements of `müssen'. If the
information about the phrasal/lexical status is not shared between
filler and gap, sentences like (1) are unproblematic.

This solution was first suggested by Tilman Höhle during a GGS talk and
can also be found in several papers by Detmar Meurers and me.

The most recent publications are:

Book Chapter (English):
http://ling.osu.edu/~dm/papers/hpsg-volume98/pvp-revisited.html

Book (English)
http://www.cl.uni-bremen.de/~stefan/Pub/complex.html

Article (German):
http://www.cl.uni-bremen.de/~stefan/Pub/satz-lb.html

If one believes that remnant movement analyses are wrong (evidence for
this is provided in Kordula de Kuthys dissertation and in an article of
Kordula and Detmar), an analysis in terms of argument attraction is a
nice alternative. This can be combined with the partial reconstruction
of nonlocal information mentioned above and everthing works nicely.

Another phenomenon I remember was the Binding Theory. PS94 argue that
HPSG has an advantage since traces do not have daughters. I am not sure
though whether this is really an advantage since for sentences like (2)
reconstructions seem to be needed:

(2) a. Kennt er_*i Karls_i Freund?
    b. [Karls_i Freund]_j kennt er_*i  _j.

The BT of PS94 predicts that the binding in (2a) is out, while no such
prediction is made for (2b), since `er' does not o-command `Karls'.

The full story can be found in Chapter 20.2 of:

http://www.cl.uni-bremen.de/~stefan/Pub/hpsg.html


> T. (not A.)

Where does this A come from?

Best wishes

	Stefan


--
Stefan Müller

Universität Bremen/Fachbereich 10      Tel: (+49) (+421) 218-8601
Postfach 33 04 40
D-28334 Bremen

http://www.cl.uni-bremen.de/~stefan/

http://www.cl.uni-bremen.de/~stefan/Babel/Interaktiv/