Perspicuous Analysis of Wager-class verbs in ECM Constructions

This study is an attempt to examine whether wagger-class verbs can exceptionally assign Case to the embedded subject of the infinitival complement.　To our surprise, although, like believe, wager-class verbs allow passive raising (e.g., John was wagered to be crazy)and exceptionally Case-mark wh-trace (e.g., Who did Peter wager to be crazy), they cannot exceptionally Case-mark lexical NPs (e.g.，*Peter wagered the student to be crazy).　With respect to wager-class verbs, I assume here that Topicalization of the embedded subject of the infinitival complement has the following structure， as in (2) but not as in (1).
(1) [_Top Mary_i][_CP null-Op_i [_IP Bill wagered [_IP t_i to have won the race]]]
(2) [_Top Mary_i] Bill wagered [_IP t_i to have won the race]
↑　　　　　　 　　 　 　 ￨
└────────┘(Th/Ex rule)
Under the phase theory， my assumptign is that lexical inertness NPs， which are to be given the Case-value from the matrix verb wagger，must be raised "The matization/Extraction rule" (Chomsky 2001: 23).