Simplifying the evidential condition on asking polar questions

In classic accounts of polar question semantics, positive polar questions like "Did Mo sing?", low negation questions like "Did Mo not sing?", and high negation questions like "Didn't Mo sing?" all denote the same set of answers: {that Mo sang, that Mo didn’t sing}. At the same time, it is well known that these three question types have different distributions. In particular, they have different requirements with respect to contextual evidence for the answers, the Evidential Condition on polar questions. Despite widespread discussion of this fact, no universally accepted explanation has emerged. In this paper, I make the novel argument that high negation questions do not have an Evidential Condition, and so only the conditions for positive and low negation questions need to be explained. I then argue that an explanation can be given based on general principles of markedness and information structure, even while maintaining a classic {p, not-p} semantics for both positive and low negation questions. I discuss ramifications for polar question semantics.