Users think of systems in terms of the functionalities they provide. Any coherent and identifiable bundle of system functionality perceived by users is then a feature of the system. Features cut across the entire system. Thus, aspect oriented languages can be exploited in order to avoid the tangling between the code implementing a feature and the other parts of the system. The problem of feature interactions becomes the problem of discovering aspect interactions and can be studied by analysing the source code units. In this paper we show how programme slicing can be used to build models of aspect oriented units. Every time one wants to reason about an aspect, a developer can use the associated slice instead of the whole programme to build useful models. We believe that this motivates the use of aspect-oriented constructs as an effective tool to manage complexity.