Abstract argumentation frameworks (AFs) are one of the central formalisms in AI; equipped with a wide range of semantics, they have proven useful in several application domains. We contribute to the systematic analysis of semantics for AFs by connecting two recent lines of research -- the work on input/output frameworks and the study of the expressiveness of semantics. We do so by considering the following question: given a function describing an input/output behaviour by mapping extensions (resp. labellings) to sets of extensions (resp. labellings), is there an AF with designated input and output arguments realizing this function under a given semantics? For the major semantics we give exact characterizations of the functions which are realizable in this manner.