In everyday talk about language, we distinguish between what someone said and what they implied, or otherwise conveyed. This distinction has been carried over into theorising about language and communication, resulting in much debate about how the notion of what is said should be defined. Against the underlying assumption of these disputes, Nothing is Said argues that it is a mistake to import the notion of saying into our models of basic linguistic communication. Rather than belonging to our basic linguistic competence, the notion of saying is a reflective one resulting from a higher-order metacommunicative competence that is relatively late-developing. This competence allows us to reflect simultaneously on the form and content of an utterance, and hence characterise it as an act of saying. The study shows how this notion of saying can be accounted for without assuming that identifying what is said is a necessary step in basic utterance interpretation. The idea that linguistic interpretation relies on identifying what is said is deeply ingrained. Mark Jary considers the consequences for semantic and pragmatic theory of dropping this assumption, focusing on lexical pragmatics, scalar implicature, assertion, lying, and other topics that have received significant attention in the recent literature. The claims made are supported by reference to empirical data from experimental psychology.