🚀 go-pugleaf

Forgive me if this is an obvious question but I've never done any Prolog
programming before and, as I'm working at home, don't have access to any
books (so references will be useless to me). 
How do I implement a natural language parser in Prolog using a context-free
grammar?
I need my program simply to produce all possible sentence trees for a given
sentence.
I would imagine this would be simply a matter of declaring a relationship
for each re-write rule and then giving the program a sentence as a goal.
On a more complicated level, how would I implement a simple feature based
grammar to account for subject-verb agreement, etc.?

Thanks in advance

James Tauber

P.S. e-mail would be perferred. If you know of any ftp sites with parsers in
Prolog, that information would also be appreciated.

jtauber@tartarus.uwa.edu.au