A long-standing generalization about parasitic gap constructions is that they are not licensed by in-situ wh-phrases, even when those wh-phrases are taken to move covertly. We present novel evidence from Bùlì which presents a counterexample to this generalization. We discuss the consequences of these facts for theories of parasitic gap licensing. Finally, we develop a theory of parasitic gap licensing, building on the work of Nissenbaum (2001), which captures languages that conform to the generalization and ones that do not. We note that this theory leads us to expect languages that do not conform to this generalization to lack a number of constructions involving relative clauses, and show that Bùlì behaves as expected.