ABSTRACT
Brian Loar attempted to provide the Gricean program of intention-based semantics with an account of expression-meaning. But the theory he presented, like virtually every other foundational semantic or metasemantic theory, was an idealization that ignored vagueness. What would happen if we tried to devise theories that accommodated the vagueness of vague expressions? I offer arguments based on well-known features of vagueness that, if sound, show that neither Brian’s nor any other extant theory could successfully make that adjustment, and this is because, if sound, the arguments show not only that nothing can be the content of a vague expression, but also that standard assumptions about semantic compositionality are mistaken. This raises the question of what the facts are about a language whose explanation might seem to require semantic compositionality, and what, really, is needed to explain those facts.
