As Crystal was requesting the intuitions of his readers, and as anyone had yet to post, I thought it a reasonable opportunity for increasing my notoriety, and quickly drafted a few remarks. These stated my intuition that "for instance" was perhaps more likely to be used in spoken English, than "for example," and showed that "for example" was more common, citing evidence from the Google Books corpus. I also remarked that my hunch about spoken vs written usage would be testable, given access to the right corpus. Happily I managed to express all this, and still get the first post, which I'm sure are more read than others 'below the line'.
Even more happily, I'm now five weeks into the most excellent Future Learn MOOC on corpus linguistics, which has given me access to the British National Corpus, and some good teaching about to how to use it. Anyway, as we all now know on the course, the BNC comprises corpora of both transcribed speech and written texts, enabling me to test some of the intuitions I offered Crystal two years ago against hard data. Here's a table showing the relative frequencies of both terms in the spoken and written sections of the BNC:
|(freq/10^6 words)||(freq/10^6 words)||(ratio)|
|Data: British National Corpus hosted at http://bncweb.lancs.ac.uk/|
As some of Crystal's other commentators (Lucy, Rick Sprague) remarked two years ago, fr'instance does slip more felicitously off the tongue, or, if you like that kinda thing, "is preferred for phonotactic reasons."