Debates about statutory interpretation typically proceed on the assumption that statutes have linguistic meanings that we can identify in the same way that we identify the meaning of utterances in ordinary conversation. But that premise is false. We identify the meaning of conversational utterances largely based on inferences about what the speaker intended to communicate. With legislatures, as now is widely recognized, there is no unitary speaker with the sort of communicative intentions that speakers in ordinary conversation possess. One might expect this recognition to trigger abandonment of the model of conversational interpretation as a framework for interpreting statutes. Instead, interpreters invent legislative intentions—purportedly “objective” ones for textualists—or purposes. With those inventions in place, judges and theorists then carry on talking about what statutes mean, or would mean to a reasonable person, as if there were a linguistic fact of the matter even in intelligibly disputed cases. But this is a deep and systematic error. Mainstream thinking about statutory interpretation needs a major reorientation. Contrary to widespread impressions, debates about statutory interpretation are not about what statutes mean as a matter of linguistic fact, but about which grounds for the attribution of an invented meaning would best promote judicial and governmental legitimacy. Having recognized that the model of conversational interpretation cannot ground claims about statutes’ meanings in disputed cases, we also need to rethink the role of legislatures and courts in a political democracy. There are limits to what legislatures can reasonably be expected to accomplish. Courts need to play the role of helpmates to the legislature, not just faithful agents. In the interpretation of statutes, linguistic intuitions should matter, but primarily for normative reasons, involving justice and fairness in the coercive application of law, and not because they reveal the legislature’s linguistically clear dictates.