Dever, Josh and Henry Ian Schiller (forthcoming). This paper might change your mind. Noûs.
Work in formal semantics often posits complex structures as conversational scoreboards. As scoreboards components proliferate, though, theoretical underdetermination worries increase. Consider question under discussion (QUD) structure in conversational scoreboards. It’s plausible enough that conversations are often guided and shaped by a dynamically developing sequence of topical questions. But when, for example, we attempt to codify this plausible fact using a question stack as a component of the conversational scoreboard, and then to extract Gricean relevance facts as downstream consequences of the evolving QUD, there’s a real worry that it’s really the relevance facts that are fixing the QUD facts, or that the QUD facts are being ad hoc extracted from incomplete discourse descriptions to fit the relevance facts. In such cases it’s not clear that the QUD is a genuinely explanatory component of the semantic theory, or that the proposed conversational scoreboard will be properly constrained by data. To avoid such worries, it’s then helpful to tie specific scoreboard components to specific linguistic phenomena (p. 36).