Jacek Witkos


On Reconstruction and the Preference Principle.


In the MP the Preference Principle makes sense because of the Copy Theory of Movement, there is 
no countercyclic displacement but a partial deletion of the Copy in the target position, and 
Economy conditions: minimal movement (of Q/Wh) is required to satisfy a [+wh] feature. The 
whole phrase moves up only because of language specific pied-piping requirements. Compare:

1)
a.	How many books did Mary read t?
b.	[How many x] did Mary read [x:books]
c.	Ile ksiazek Maria czytala t?
d.	[ile (x)] Maria czytala [(x):ksiazek]?

	(1b) is an illegitimate spell-out structure in English but (1d) is a possible spell-out structure 
in Polish. This confirms a common view that Slavic languages tend to wear their LF on their 
sleeves.   
	The Preference Principle is supposed to cover the following cases:
2)
a.	John wonders [which book about Bill] he read t ?
b.	John wonders [which x] he read [x:book about Bill]

This Principle is not an inviolable Economy Condition but can be overcome by other more crucial 
principles such as Theta Criterion:

3)	John wondered [which picture of himself] Bill saw t?
a.	John wondered [which x] Bill self-saw [x:picture of t]?
b.	John self-wondered [which x, x:picture of t] Bill saw t?
 

	The Preference Principle seems adequate to cover the following cases in Polish, (4b) is the 
reconstructed representation of the question at the interpretive level of  LF:

4)
a.	[ile ksiazek o Janie] on przeczytal t ?
b.	[ile x] on przeczytal [x:ksiazek o Janie] ?
c.	[ile ksiazek o sobie] Maria czytala t wczoraj ?
d.	[ile x] Maria czytala [x: ksiazek o sobie] wczoraj ?

	In (4) Binding Principle C requires disjoint interpretation of 'on' and 'Jan' which is computed 
on the reconstructed representation in (4b). In (3) Binding Principle A requires a c-command 
relation to hold between the subject and the reflexive pronoun which is established in the 
reconstructed form in (4d).


0.2.		Problems with the Preference Principle.


Problems with the Preference Principle come from two areas:

A.	The displaced category must be in two positions at the same time. This affects Chomsky's 
(1993, 1995) system where one copy is left for interpretation at LF. These cases are discussed in the 
literature (Brody 1995, 1996); Epstein, et.al. 1998):
 
5)
a.	Mary wondered [which claim that pictures of herself disturbed Bill] he made
b.	Mary self-wondered [which x, x: claim that pictures of t disturbed Bill] he made [x: claim 
that pictures of herself disturbed Bill]

6)
a.	[which paper [that he gave to Mary]] did every student think that she would like t?
b. 	[which paper [that he gave to Mary]] did every student think that she would like [which 
paper [that he gave to Mary]]

B.	Why does Principle C require reconstruction w.r.t. the subject position but not necessarily 
w.r.t. the indirect object position? Consider this structure, adapted from Willim (1989):

7)
a.	Która pocztówkę Jana z Paryza Maria oddala mu przedwczoraj? 
b.	[która x, x: pocztówkę Jana z Paryza] Maria oddala mu przedwczoraj?

	Note, that there is no apparent reason why, on the strength of the Preference Principle, there 
is no full reconstruction of the direct object into its thematic position, that is none of the more 
general grammatical principles such as the Theta Criterion in (7b) appears to trump complete 
reconstruction. More importantly, which principles of grammar allow for partial reconstruction or 
lack thereof?
	Now, consider more complex cases in Polish which suggest that more than one copy of a 
displaced category should be accessible to interpretation at LF:

8)
a.	[ile ksiazek Jana o sobie] Maria oddala mu t po wakacjach?

	Consider the possible interpretation of (8) in which 'Jan' has written several books about 
'Mary', thus the interpretation involves the following coreference relations: 'Jana' and 'mu' are 
coindexed and 'Maria' and 'sobie' are coindexed. These two pairs of coindexed nominals impose 
mutually exclusive demands on reconstruction:

9)
a.	[ile x, [x: ksiazek Jana o sobie]] Maria oddala mu [x] po wakacjach?
b.	[ile x] Maria oddala mu [x: ksiazek Jana o sobie] po wakacjach?

	The relationship between 'Jan' and 'mu' based on the Binding Principle C prohibits 
reconstruction and overrides the Preference Principle: the restriction is interpreted in the position of 
the quantifier (9a). On the other hand the relationship between 'Maria' and 'sobie' requires 
reconstruction (9b).
	There seems to be a way of solving both problems (A&B) facing the Preference Principle, 
thus salvaging the single copy approach and providing rationale for the Preference Principle. 
Assume that the structure of the clause accommodates a phrasal projection called Reconstruct 
(Phrase). In terms of its prominence in the clause it occupies the following position:

10)	{Topic/Wh-phrase {Subject 1 {Reconstruct ({Object-indirect) {Object-direct {Subject 
2}}}}}

	The advantage of (10) is that the restriction of the quantifier can be said to reconstruct at 
least into Reconstruct in all the Polish cases above. 
	Reconsider (9) with respect to the structure in (10):

9)
c.	[ile x] Maria [x:ksiazek Jana o sobie] oddala mu t po wakacjach?

	In this configuration the conflicting requirements of coindexation are met in a single copy 
approach: 'Maria' c-commands 'sobie' in keeping with Binding Principle A and 'Jan' is out of the c-
domain of 'mu', in line with Binding Principle C. Reconstruction of the restriction into Reconstruct 
is the most parsimonious option. The other option is to save the restriction in the object position in 
which case 'mu' and 'Jana' refer to different people.