Mrs. Malaprop was prone to making linguistic mistakes of a special sort, saying that someone "is as headstrong as an allegory on the banks of the Nile", when she meant that the person was as headstrong as an ALLIGATOR on the banks of the Nile. In my talk, I will argue that such malapropisms, and more generally, linguistic mistakes pose a major challenge for Dan Sperber and Deirdre Wilson's influential Relevance Theory (RT). Their book Relevance: Communication and Cognition (1986) marks the inception of RT, which has since been widely discussed and developed. According to RT, the gap between linguistic meaning and what speakers mean on occasions of utterance is far wider than philosophers of language (like H. P. Grice) have traditionally thought: to bridge the gap audiences employ in interpreting speakers involves complex inferential processes guided by considerations about relevance. In more recent work, "A Deflationary Account of Metaphor" (2008), Sperber and Wilson argue that the inferential processes RT posits for literal utterances can account for metaphorical utterances also: "There is no mechanism specific to metaphors, no interesting generalisation that applies only to them"; literal and poetic metaphors occupy opposite ends of what is a continuum of cases in each of which audiences employ essentially the same inferential process to interpret what the speakers have meant. My aim is to show that the inferential processes posited by RT are missing some key details. Without those details, the literal-metaphorical continuum extends to malapropisms and other linguistic mistakes, an unwanted result. And once the missing details are included, we have reason to abandon the literal-metaphorical continuum.