This paper is an attempt at providing an account of Long GoN within the system of minimalismbased on case-agreement and single cycle syntax, closely related to the system developed in Chomsky (J998, 1999, 2000). First, we propose that a successful relation of case-agreement can be established not only between full Probes and complete Goals but also between 'split' Probes and complete Goals. We define a 'split' Probe as a (predefined) pair of heads that complement each other and jointly have features [+case, +<p). In the concrete case ofGoN and Long GoN, the two heads involved are F/Neg [+case] and v [+<p). The distance holding between the two half Probes in the cases ofLong GoN and the punitive power of the PIC, force us to revise certain assumptions concerning the status ofphases in the derivation. We propose that for a successful case-agreement relation both the 'split' Probe and the Goal must be placed within the same derivational phase. We therefore submit that neither the vP nor the infinitive embedded under control verbs constitute strong phases. In order to deal with infinitives embedded under control verbs, we reject the idea that they are CPs and propose instead that they are best treated as bare TPs, following Boskovic (1997) or, preferably, Hornstein (2000, 2001). The latter hypothesis, based on movement into thematic positions and treatment of obligatory control PRO as an NP trace, has an additional advantage ofexplaining lack ofdefective intervention effects that a PRO, with its interpretable [+<p] features visible to the Probe, is otherwise expected to cause.
Jacek Witkos