A Preference Semantics for Imperatives

There is a rich canon of work on the meaning of imperative sentences, e.g. Dance!, in philosophy and much recent research in linguistics has made its own exciting advances. However, I argue here that three kinds of observations about imperatives are problematic for approaches from both traditions. I contend that these shortcomings stem from an entirely standard assumption of these semantic theories: a symbol’s meaning is the content it ‘refers to’ or ‘denotes’, e.g. a particular entity, a property, a proposition. In response, I offer a new analysis of imperatives that takes a different form, namely a specific formulation and interpretation of dynamic semantics. Here, the meaning of a symbol is the characteristic role it plays in changing users’ mental states. I propose that imperatives change what’s mutually preferred in the conversation. Preferences have a well-established role in decision theory and artificial intelligence where they are key to understanding how rational agents decide what to do. Analyzing the semantics of imperatives in terms of preferences therefore makes it possible to understand their role in guiding action using well-developed tools from related disciplines. This leads to a precise integration of the semantics and pragmatics of imperatives that makes clear predictions and improves on existing accounts.