The aim of this chapter is to make progress in understanding what is common and what is not across canonical and non-canonical questions. The framework adopted is that of the Table model in Farkas and Bruce (2010), and the view of declaratives and interrogatives in Farkas and Roelofsen (2017). After laying out the theoretical assumptions, I argue that the commonly assumed properties of canonical questions follow from the semantics and the basic conventional discourse effects of unmarked interrogatives. Turning to non-canonical questions, whether marked or unmarked, weaken, I argue that they override or, in some cases, reinforce default assumptions characterizing their canonical sisters.