This paper describes an exercise in generic data normalization theory using a data reification calculus base on the categorial approach to datatypes.
We develop a relational data model in a functional language and then use transformations to refine it. The exercise removes data redundancy in close similarity to conventional relational techniques, which extract functional dependences by schema decomposition [Mai83]. Finally the model is extended using the principles of generic programming, suggesting how to scale up normalization theory to arbitrary data.