Wh-questions in many languages are formed by moving the wh-phrase to the beginning of the sentence, away from the verb or preposition with which it is interpreted as a unit. The movement of a wh-phrase is subject to syntactic constraints that are usually referred to as "island constraints." For example, in the question in (1), which is judged to be unacceptable, the wh-phrase "which bill" is intended to be interpreted as part of a relative clause "who supported." Relative clauses are thus considered to be "islands" that block the movement of wh-phrases. Previous research has identified systematic exceptions to these island constraints. In this talk, I discuss an exception (2) first systematically discussed in Chung and McCloskey 1983. While sentences like (2) seem to be acceptable in informal introspection, I show using formal ratings data that these sentences are rated worse than their non-island counterparts. The lower ratings suggest that some constraint is violated, and that these sentences are not genuine exceptions to the relative clause island constraint. I also discuss several hypotheses about why the Chung and McCloskey sentences are judged and rated as more acceptable than regular relative clause island violations.

(1) Which bill did the president meet the senator who supported? (2) Which bill does the president need to find a senator who will support?