Many languages exhibit a variety of declarative questions (e.g. You got a haircut?), the key question raised by which is how to principally combine the assertive force associated with the declarative syntax and the question force associated with the final rising contour. Two classical approaches have been proposed: (i) prioritizing the question force and demoting the assertive force, which often gets analyzed as contributing evidence or some form of weakened commitment (Northrup 2014, Malamud & Stephenson 2015, Farkas & Roelofsen 2017), and (ii) prioritizing the assertive force and demoting the question force, which is often attributed to pragmatics (Gunlogson 2001, 2008, Rudin 2018, to appear).
We argue, based on declarative questions and confirmational questions in Cantonese, that both the assertive force and the question force should be represented. However, departing from previous studies that juxtapose the two forces in a parallel manner (e.g., Asher & Reese 2007), we pursue a hierarchical solution. We argue that the question force has its roots in the partial nature of the speech act function associated with the assertive force. In fact, it is predicted that any force giving rise to a partial speech act function can be shifted to the question force, essentially generating a question about whether or not a speech act can be performed in a given context, a phenomenon we call force shift. We test these predictions in Cantonese and English.