September 29, 2004

Plagiarism and coincidence

Joseph Bottum has an interesting recent article on scholarly plagiarism, and the related phenomenon that some are calling "plagiaphrase" (that's where you borrow all the main threads of the fabric of someone else's writing but you re-weave it a little). The issue relates very clearly to themes in statistical linguistics.

How frequent is the phrase "the vagaries of the Electoral College"? It gets about 50 webhits on Google (whG), but that's not very many in 4.3 billion pages.

What about the phrase "in the Holmes mold"? It gets about 19 whG. But more specifically, how frequent is it in contexts where the reference is to Oliver Wendell Holmes? There are no Google hits for that at all. It's a very unusual locution with that reference.

Finding both phrases in the same book already has a ridiculously low probability. But when unusual phrases such as these are repeated in a later-published book in just the parts where they discuss a point that was discussed in an earlier-published book that also used those phrases, you really have to stop and take a closer look.

Bottum finds evidence of this sort in abundance in a book by Professor Laurence Tribe (Harvard University), who he claims has been guilty of constantly repeated plagiaphrase of a book by Henry J. Abraham (University of Virginia). Bottum supplies enough quotations — dozens of them —that you can make up your own mind whether you think it goes far enough to beyond the boundary of accidental phrase similarities, and into the land of dishonest appropriation.

Bottum did ask Abraham for comment, and the 83-year-old legal historian seems justifiably rather miffed. He talks of laziness, careless use of young assistants, and the desire to get a book out and make a quick buck.

Posted by Geoffrey K. Pullum at September 29, 2004 01:55 AM