The responsibilities of the English-speaking co-author

In order to disseminate their work as widely as possible, non-English speaking scientists face the dual challenges of both writing in another language, and writing in a rather unforgivingly technical form of that language. My awareness of how difficult I would find it to write a scientific paper in French or Spanish (let alone something like Mandarin) makes me hesitant to criticise the English of papers written by foreign authors too strongly, although – unfair as it probably is – there has to be some minimum standard of comprehensibility.
However, my patience with hard-to-decipher prose is worn rather thinner when I see people who are obviously native English-speakers on the co-author list. By accepting co-authorship plaudits, you become associated with, and thus take some responsibility for, the contents of the paper. As I see it, that responsibility doesn’t just cover the scientific validity of the results being presented, but the paper’s effectiveness at communicating those results. Your involvement in the research might be due to being an expert on a particular technique or field area, and you would be remiss if you didn’t use that expertise to improve any manuscript you were asked to be a co-author on. By the same token, if you are a native English speaker working with foreign scientists, not contributing your “expertise” (if you can call it that) in written English also seems like an abdication of your responsibilities. It also seems rather unsympathetic to the difficulties faced by non-English speakers wanting to publish in high-impact journals.
Of course, I realise that it’s not quite that simple: what if you’ve just provided a dataset, or access to and training for a nifty new bit of kit in your laboratory (of course, if that’s the case, then perhaps you shouldn’t be listed as a co-author anyway)? Plus, given the normal expectation (at least in my field) that the first author is the principal author, there is obviously a balance between helping someone to communicate better in their own words, and replacing their voice with your own (and possible problems with people who regard any attempts at correction in that light).
I’d be interested to hear your opinions on this. In the circumstances described above, is the co-author responsible for a badly written paper?