Actually, I am dealing with form submission software, where it scans the internet for forms, stores the form vars, and then submits crap. In my case, every 3 minutes. So since this is form submission software I don't think robots.txt can do anything for me. The bot has already come by.
It uses random proxies so I can't block by IP. It has a user agent, but likely faked and made to appear common.
So that is why I am investigating the accept language thing I have heard about.