Simple versus rich language in disclosure games

This paper studies strategic information disclosure when the sender may not observe the payoff-relevant state, and the receiver may interpret messages naively. We characterize equilibria as a function of the language available to the sender. The language is simple if an informed sender can either fully disclose the state or nothing. The language is rich if he can disclose any closed interval containing the true state. We show that an informed sender and a strategic receiver get a higher ex-ante equilibrium payoff when the language is rich. The reverse holds for a naive receiver and an uninformed sender. Overall, our work suggests that the design of language is key in situations where disclosure is voluntary.