|Title:||Burzio's Generalization, Markedness, and Constraints on Nominative Objects|
|Abstract:|| Recent research on Burzio's Generalization converges on a surprising conclusion: what
blocks accusative Case in unaccusative constructions has nothing to do with the Case or theta
assigning abilities of unaccusative verbs; rather an overriding principle requires sentences to
have a nominative Case. But there is little consensus as to how to formulate the relevant
principle, and most proposals fail to predict a large range of counterexamples in the form of
sentences with no nominative. It is argued here that the relevant principle is markedness: when
there is a choice of licensed Cases for a DP, the grammar selects the less marked Case. There are
exceptions because markedness is violable and overriding principles may require a more marked
Case. Markedness accounts for the lack of accusative Case on unaccusative subjects, and also
for the presence of nominative objects in dative or ergative subject constructions. However,
nominative checking on objects is sometimes blocked: in Icelandic, it is blocked when the
subject has lexical accusative Case, and in Faroese, it is blocked in all active constructions.
The claim is that such blocking is due to constraints on Case checking domains. The ideal Case
checking domain contains no DP whose Case is not checked by the head of that domain.
However, some languages tolerate deviations from this ideal, with an additional
partially checked Case being worse than a completely mismatched Case.
This work is relevant to the general issue, important in syntax and phonology, of whether
it is an intervening potential source, target, or both that matters in situations where like blocks
like. This paper shows that Case checking can be blocked by a closer potential target (DP), just
as movement and binding can be (Rizzi 1990). Since a closer potential source (head) can also
block Case checking, we can conclude that both are relevant.