This is the function tagger first written about in my 2000 paper and continuing through to my thesis.
Build instructions, how to run it, etc, are in the README file. The trained parameters are included; if you wish to retrain, the software for that is included but you will need to acquire the training data (i.e. the Penn treebank) separately.
You will also need to download a recent build of expat if it's not already installed on your computer.
The default data directory is trained on a version of the Penn WSJ Treebank with empty nodes removed, and with some helper verbs tagged with the preterminal AUX, as output by the Charniak parser. If you are dealing with a different data set, you might find the following trained parameters more useful: