In this paper, I assess the ontological commitments of frame-based methods of knowledge representation. Frames decompose concepts into recursive attribute-value structures. Attributes are the general aspects by which a category or individual is described; their values are more or less specific properties that are assigned to the referential object. The question is: are these properties to be interpreted as universals or as tropes? Some trope theorists allege that an interpretation in terms of universals is incompatible with frames for individuals in (...) which the values of quality attributes are recursively specified by further attributes. I shall argue, however, that recursive extension poses a challenge, not specifically to the universals realist, but rather to realists about determinables as opposed to determinables nominalists. Moreover, it is suggested that even determinables realists are able to cope with recursively extended frames once they relativize relations between determinables and determinates to particular objects. It is indicated how these findings might bear on adjacent debates in the metaphysics of natural language. (shrink)