I've rejected papers for bad English, but not because I assume sloppy language will result in sloppy research. Rather, I'm reviewing for conferences which don't have a "major revision" option, and I have to assume that accepted papers will not be heavily revised. So if the submitted paper has language that is sloppy and hard to understand, then I assume the published paper would be the same. And that's not acceptable, because no one benefits from a paper which is hard to understand because of its presentation (not its content).