The syntactic literature discusses a number of phenomena in which a constraint that rules out a certain class of syntactic dependencies fails to rule out structures containing both an ill-formed dependency and a well-formed dependency; well-formed dependencies seem to be able to “help” dependencies that would be ill formed in isolation. In this article I attempt to provide a unified account of these phenomena. I postulate a principle that allows the computational system to “ignore” parts of a syntactic structure that have already been checked with respect to a particular constraint.

This content is only available as a PDF.