The vast majority of approaches to the liar paradox generate new paradoxes that are structurally similar to the liar (often called revenge paradoxes). There is a complex group of issues surrounding revenge paradoxes, the expressive powers of natural languages, and the adequacy of approaches to the liar. My goal is to provide a precise framework against which these issues can be formulated and discussed. The centerpiece of this framework is the notion of internalizability: a semantic theory is internalizable for a language if and only if there exists an extension of the language such that (i) the theory is expressible in that extended language, and (ii) the theory assigns meanings to all the relevant sentences of that extended language. The framework is applied to three examples from the literature: Reinhardt and McGee on theories that require expressively richer metalanguages, Field on revenge-immunity, and Gupta on semantic self-sufficiency.