But will it work for Ebonics?
It seems to work with data that has a consistent set of grammar rules. ...so maybe not. :-)