A credal network associates a directed acyclic graph with a collection of sets of probability measures. Usually these probability measures are specified through several tables containing probability values. Here we examine the complexity of inference in Boolean credal networks when probability measures are specified through formal languages, by extending a framework we have recently proposed for Bayesian networks. We show that sub-Boolean and relational logics lead to interesting complexity results. In short, we explore the relationship between language and complexity in credal networks.