On (Anti-)that-trace Effects

A fair amount of attention has been paid in previous literature to the asymmetry of the appearance of complementizer ‘that’ in subordinate clauses. In this paper, we will begin with critical reviews on the recent previous literature that attempts to account for the so-called that-trace and anti-that-trace effects. Appealing to the Split C-domain, Spec-to-Spec Anti-locality, and Dynamic Phases, the previous analyses were able to capture why ‘that’ cannot appear when a wh-operator moves from subject position of the complement clause, but the complementizer must appear in subject relative clauses. However, they were unable to account for why anti-that-trace effects are not observed when wh-operators extract out of the object position in both types of clauses. To be more specific, they fail to explain why ‘that’ can be optional in object relative clauses. To explain the lack of anti-thattrace effects, I argue that a phrase that is contained within a phase is still active and can be probed by a higher head. That is, the case feature of object wh-operator in phase edge position is checked but still active, and establishes an Agree relation with C (Pesetsky and Torrego 2001). This triggers the case feature on the C head to be checked by the operator DP, which, in turn, bans on T-to-C movement. Consequently, it leads to no overt complementizer in object relatives.