This paper shows, drawing data from Sinhala, that the strategy of a null Qu-operator movement is a available in wh-in-situ languages, and attempts to verify the 'null operator movement' hypothesis, forwarded by Watanabe (1992) and Aoun and Li (1993). In Sinhala, wh questions display some characteristics which serve to present a clear diagnostic for the 'nun operator' movement; in Sinhala, a Q element associated with a wh element appears either in a sentence-medial position or in clause final position. When a Q element shows up in sentence medial position, the wh question induces a particle-predicate concord. If it appears clause-finally, then the particle-predicate concord does not surface.
I argue that a Q element is generated by agreeing with a null Qu-operator, indicating the position of the null Qu-operator in overt syntax. If the null Qu-operator movement is induced in overt syntax, the movement yields the effect of canceling the special verb marking, and as a result, the Q element appears in sentence final position without particle-predicate concord. On the other hand, if the null Qu-operator is not moved overtly, but in the LF component, the Q element appears sentence-medially, without canceling the special verbal marking, and the particle-predicate concord emerges.
The Sinhala data gives us a clear indication that a null operator associated with a wh-in-situ may be moved either in overt syntax or in LF, contrary to the claim by Watanabe (1992) and Aoun and Li (1993). On the basis of the Sinhala data, we can also ascertain that in a wh-in-situ language, a strategy to evade island effects (when a wh phrase is deeply embedded) is available irrespedive of whether the null Qu-operator movement is invoked in the overt component or in the LF component, and that there is absolutely no asymmetry between overt and LF movement in regard to constraints on movement.