Harris (forthcoming in Mind and Language) defends two familiar claims and argues that they are related in a special way, the first entailing the second. The first claim is that semantic processing is the work of a module encapsulated from general cognition. The second is that the semantic value of an expression must be a constraint on what we could say in uttering that expression, contrary to the standard view of semantic values as contents. According to Harris, it is because semantic processing is encapsulated from general cognition that semantic values must be constraints: content resolution requires central cognitive processes, like mindreading, and therefore cannot be the work of an encapsulated module. However, this argument trades on an equivocation concerning the content of a sentence versus the content of a speech act. The workings of an encapsulated semantic module certainly do not result in attributions of speech act content. However, the workings of an encapsulated semantic module could result in resolution of sentential contents—at least, of the sort proposed by semantic minimalists. There are other reasons to prefer an alternative account, but semantic minimalism is not ruled out by the observation that semantic processing is modular.