Both the sokal paper and the first paper by the Boghossian group were published in open-access journals, of the same standard as the other fields. The only papers to be published in peer-reviewed journals are the last few by the Boghossian group, and those don't really prove that it's a problem unique to social sciences, either. The Boghossian group didn't seek to establish a control study, so we don't know if a similarly motivated group couldn't get papers published in other fields.

What makes the predatory journals relatively non-serious? What merit is there in your distinction between a "hoaxed" journal and a journal that publishes faulty papers on purpose?

Please see the "Bogdanov affair", where "serious" peer-reviewed journals accepted bogus theoretical physics works. I already linked to the relevant Wikipedia page in a nearby comment.

I agree that the Bogdanov Affair is shameful! I tend to lump that in with string theory more generally, which looks like an embarrassment to me.