Marantz (1991) argues that case is assigned to an NP due to the c-command
relationship with another NP in a given domain and not by a dedicated head.
The author contends, furthermore, that case assignment and nominal
licensing are dissociated. One empirical argument in favor of this claim is
provided by NOM objects in Icelandic ECM constructions, where there is no
dedicated finite T head assigning NOM, but the object is nonetheless
licensed. Following Branan:to appear, I argue instead that dependent case
and nominal licensing are not necessarily contradictory with each other:
dependent case can, in fact, be a licensing strategy for a nominal. The
empirical basis is provided by pseudo noun incorporation (PNI) in Wolof,
specially with regards to the conditions under which a PNI-ed nominal in
this language is not adjacent to the verb, a property which is otherwise
usually found in PNI.