This paper examines the contradictory demands of using language expressively and still qualifying as language, proposing a functional explanation for the form of words in a linguistic word category. Being expressive requires expending more energy, emitting a more robust signal to convey additional information about the speaker, the perception of an event, etc. Doing so requires violating the common linguistic constraints of everyday language, yet to be recognized as language requires that one’s speech obey these same rules. How speakers satisfy these demands tells us about language in both its function and form. The resolution of this dilemma requires the use of suprasegmental rather than segmental features, e.g., a wider range of and more varied use of F0. Because prosodic features are more susceptible to manipulation, they provide the resources for expressivity. Segmental parameters cannot be so easily violated, though manipulating phonotactics remains fair game. Thus one sees that there are constraints on violating constraints.