Is it possible to provide a theory of truth that is capable of distinguishing the semantic status of paradoxical sentences from that of other ungrounded sentences without bringing meta-linguistic resources into play? We explore an account that extends Kripke's theory of truth with two primitive operators, one standing for the notion of paradoxicality and the other for the notion of hypodoxicality. Our results are mixed. While the paradoxicality operator behaves nicely, a number of restrictions need to be imposed to accommodate (...) the hypodoxicality operator. (shrink)