We present two pieces of evidence that children's production of questions with medial wh phrases, such as "what do you think what is under the top hat", are the result of issues with performance, not the result of having the wrong grammar. First, such productions are predicted by independent measures of inhibition. Second, such productions are also predicted by children switching the argument structure of the elicited question, a plausible index of taxed executive function. Notably, such productions are also significantly more likely to occur if the question is a subject question. We provide a hypothesis about the mechanism that might underwrite this, and we'd appreciate your feedback on this proposal.