There is a new parallel corpus available at:
http://www.statmt.org/setimes/
I hope this helps SMT applications, especially online tools like yours. I had also compiled a list of documents available in English and Turkish which could help build a large corpus in those languages. The list is available at:
http://denizyuret.blogspot.com/2009/02/erguns-english-turkish-machine.html
By the way, I have observed that translation quality of Microsoft's tool far exceeds its competitors for pairs including SOV languages like Turkish. I think this is a result of your unique hybrid approach to MT. Congratulations and keep up
the good work!
Regards
Murat