conf A.Bachrach

[Lea Nash]

L'UMR 7023 (SFL)
a le plaisir d'annoncer un exposé


Date : lundi 3 avril 2006
Lieu : Université Paris 8, Bâtiment D,  Salle 143, 93 Saint-Denis
Heure :   11:00-12:30
Métro : ligne 13, St-Denis--Université

11:00-12:30

Asaf Bachrach
Massachusetts Institute of Technology

Right-Node Raising and Delayed Spellout

RESUME :

Consider the following contrast:

(1)	 a. Which animali did John say that Mary knew a man who wrote _, and
	a woman who published _ an encyclopedia article about ti?
   	 b. * Which animali does John know a reporter who made famous a man 
who
	published _, and a woman who illustrated _ an encyclopedia article 
about ti?

Both sentences involve wh-extraction of which animal as well as 
Right-Node
Raising (RNR). In both sentences the wh-phrase has to escape from 
conjuncts
that contain relative clause islands. But (1a) is good while (1b) is 
judged
ungrammatical, presumably because of the additional relative clause 
island
outside the conjuncts. The ability of RNR to feed wh-movement across
conjunct-internal islands (1a) poses a problem for both transofmational
analyses of RNR (Sabbagh 2003) and in situ analyses (Wilder 1999). The
reappearance of island effects above conjunction (1b) is even more
difficult to account for and further constrains the form of an adequate 
theory of RNR. We
present an analysis in which RNR is the result of a general spell-out
procedure interacting with multiple-dominance. The fact that islands 
inside the
conjuncts do not block extraction in (1a) is explained by the assumption 
that the shared
material is not spelled out until it is fully dominated, that is, until 
the
level of conjunction. From that moment, islands have the same blocking 
effect
as always, explaining the ungrammaticality of (1b). We start by reviewing
several puzzling properties of RNR that have been discussed in the 
literature,
as well as some of the proposed solutions. We proceed to present 
observations
similar to those in (1), which have not been discussed before. We show 
that
these new observations pose a problem for all current accounts of RNR. 
We then
present our proposal and explain our background assumptions regarding
spell-out, linearization, and multiple dominance.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: text/enriched
Size: 2935 bytes
Desc: not available
URL: <http://listserv.linguistlist.org/pipermail/parislinguists/attachments/20060327/50948955/attachment.bin>