The precision of many type based analyses can be significantly increased
given additional information about the programs' execution. For this
reason it is not uncommon for such analyses to integrate supporting
analyses computing, for instance, nil-pointer or alias information. Such
integration is problematic for a number of reasons: 1) it obscures the
original intention of the type system especially if multiple additional
analyses are added, 2) it makes use of already available analyses
difficult, since they have to be rephrased as type systems, and 3) it is
non-modular: changing the supporting analyses implies changing the entire
type system. 

Using ideas from abstract interpretation we present a method for
parameterizing type systems over the results of abstract analyses in such
a way that one modular correctness proof can be obtained.  This is achieved
by defining a general format for information transferal and use of the
information provided by the abstract analyses.  The key gain from this
method is a clear separation between the correctness of the analyses and
the type system, both in the implementation and correctness proof, which
leads to a comparatively easy way of changing the parameterized analysis,
and making use of precise, and hence complicated analyses.

In addition, we exemplify the use of the framework by presenting a
parameterized type system that uses additional information to improve the
precision of exception types in a small imperative language with arrays.