Social Text received Sokal's piece, found it amateurish and unreadable, asked him to make changes, he got cranky and insisted it run as is. They sat on the piece. Later, they had a special issue on The Science Wars coming up, figured an amateurish and unreadable contribution by a scientist would be better than no contribution by a scientist at all, and ran it as a non-peer reviewed contribution.

http://linguafranca.mirror.theinfo.org/9607/mst.html

Retellings by Sokal fans make it seem as if they received it breathlessly as great work and rushed to put it into publication. That is not what happened. The Sokal Affair is less evidence for the fuzzy-headedness of litcrit and more evidence that you can often get what you want just by lying, being a dick, and having some kind of finished product for people to work with.

And Sokal's submission wasn't some fluke, it was in the ballpark of a lot of the heady self-referential fluff in that field.

Only if you're very insensitive to tone, style, vocabulary and methodology does the Sokal piece resemble any legit article in Social Text. From the descriptions of it you'd think Social Text was full of maniacal gibbering about how nothing is true, everything is permitted, and quantum mechanics is a tool of the phallologocracy. But in fact, the quality of work is far more pedestrian, if a bit wordy. Here's the table of contents for the latest issue (articles behind paywall, abstracts are free):

http://socialtext.dukejournals.org/content/current

Is it self-referential nonsense? Maybe! Does it bear a reasonable surface resemblance to "Transgressing the Boundaries", if we're not being deliberately obtuse? No to me.

the terminology of science (like the term "theory")

The term theory pre-dates empirical science by a good bit, and the litcrit usage is not really orthogonal to the scientific one anyway.