state methods continue to dominate in describing the morphology of
languages. Due to restrictions in the commercial use of
some of the finite state platforms, here we describe an alternative
method, which can be implemented with open source tools. In this
method, the morphology is described in two phases. The first phase
gives a result, where various morphemes are identified and categorized.
In the second phase each morphological item is further analyzed. As a
result, each word-form gets all grammatically correct interpretations.
The output can be formatted to meet the requirements of the further
processing phases. In the implementation on Swahili, the output was
formatted to be suitable as input for the Constaint Grammar
parser. More behind the link