On the problem and/or impossibility of formalisation of metaphor within Relevance Theory. among many others, Robyn Carston (2010; 2018) has repeatedly attempted to bridge the ongoing divide between primarily semantical account of metaphor and ‘image’ theories that focus on its non-propositional content or effects. However, Wilson and she (2019) noted a persistent ‘problem of formalisation’ inherent in trying to provide an account of metaphor both as complete and as formal as possible. In this paper, I will argue that this is not a contingent or resolvable practical problem in theory formation, but a constitutive characteristic of metaphor. To do so, I will draw on two arguments of Derrida’s regarding the impossibility of a definition of metaphor. The first point is that every definition of metaphor would necessarily exclude, or be unable to explain, at least one term or metaphor: metaphor itself. This formal argument leads him to his second proposition that metaphor is grounded in but at the same time undermining important metaphysical and linguistical concepts usually taken for granted, like meaning, truth and the opposition of metaphorical and conceptual. Hence, if Derrida is right, this would have important implications both for the (im)possiblilites of formalisations of metaphor. To illustrate Derrida’s argument and its implications, I will analyse Carston’s struggle with accounting extended or imagistic metaphor and what I believe are potential shortcomings: First of all, it confuses a functional, empirical explanation of metaphor with a concept of it. More to the point however, her account is internally consistent iff one believes that metaphor has a semantical or propositional dimension only or primarily – which it clearly does not, as she herself repeatedly acknowledges. Thus, this paper shows the need to rethink how to formalise metaphor and how to avoid the demonstrated inconsistencies.