This article takes Propp’s work on Russian fairy tales as point of departure for a general theory of character. It is widely held that oral narratives lack complexity in characterization; but it is argued that all narratives can imply psychological traits. It is the case that Propp’s analysis is primarily in terms of ‘functions’-- generalized constant actions performed by various personages. However, one aspect of his study is devoted to how functions are distributed among the dramatis personae into ‘spheres of (...) action’. Seven are recognized, including hero, villain, helper, among others. These are comparable to roles in semantic (case) analysis of sentence structure, such as agent, patient, etc. Propp’s analysis is marred by a number of inconsistencies, with the consequence that his narrative roles are in need of extensive revision. Greimas has complained that Propp presented the roles as a simple inventory, stopping short of a systematic inquiry into relations among them. He offers his alternative ‘actantial model’. However, his model is inadequate in that it fails to recognize that two separate systems of narrative roles need to be recognized, one that holds for ‘dramatic’ narrative structure, the other for ‘instrumental’ narrative structure. (shrink)