A discussion of Stern's 2000 book postulating a metaphoricity operator 'Mthat' modeled on Kaplan's 'Dthat'. I focus on Stern's claim that we need to adopt a semantic analysis of metaphor because metaphor exhibits interpretive constraints which cannot be explained on a pragmatic view; I argue that in each case the 'constraint' is merely defeasible, and that a pragmatic analysis can accommodate the data more parsimoniously and in greater generality than Stern's theory can. (Elisabeth Camp)