That-trace and for-trace effects

(561)0(561

(562)

Who_i do you think (*that) t_i ate the apple? 

That trace violations are blocked by the presence of the following equation:

(562)0(562

(563)

S_r.b:$\langle$assign-comp$\rangle =$ inf_nil/ind_nil/ecm 

on the bottom of the S_r nodes of trees with extracted subjects (W0). The ind_nil feature specification permits the above example while the inf_nil/ecm feature specification allows the following examples to be derived:

(563)0(563

(564)

Who_i do you want [ t_i to win the World Cup]?  (564)0(564

(565)

Who_i do you consider [ t_i intelligent]? 

The feature equation that ruled out the that-trace filter violations will also serve to rule out the for-trace violations above.