December 01, 2006

Sprung from a common source

Mark Liberman's latest posting on l'affaire Brizendine follows the diffusion of misinformation from Brizendine's book through a recent review of it by Fiona Macrae in the Daily Mail and then on to (at last count) over 60 media outlets.  Tracking this diffusion is made possible by idiosyncratic errors in Macrae's piece:

Macrae misspelled Dr. Brizendine's first name as "Luan" (instead of "Louann"), and second, she cited the book as The Female Mind (instead of The Female Brain). These scribal errors are as good as a fingerprint or a hyperlink, and they will allow future scholars of media influence to track the flow of misinformation from Brizendine via Macrae to all sorts of places around the globe, simply by text search.

Here we see an echo (surely intended by Mark) of the methods of historical linguistics, and before that, of studies of textual descent.

The crucial step is to use shared innovations to group languages (or texts) together, as likely to have sprung from a common source.  The inference is stronger for a shared innovation that's unusual (no one will be much impressed by languages that share intervocalic voicing of consonants, or word-final devoicing, since these are such common changes; and no one will be much impressed by English texts that share the misspelling of the as teh, or of its as it's, since these are fabulously common errors), and it's stronger when more than one independent innovation is shared.  The inference that takes many recent media reports on Brizendine back to Macrae is supported by both types of evidence.

Using "mind" for "brain" (or vice versa) is probably a reasonably common error, so let's put that aside for the moment.  But "Luan" for "Louann" seems to be rare indeed: removing dupes, I get 25 webhits for "Louanne Brizendine" and 2 for "Louan Brizendine"; there's a huge pile for "Luan Brizendine", but all of them (so far as I can see) from the last few days.

The evidence for grouping the "Luan Brizendine" spellings together as likely to have sprung from a common source is even stronger than it might at first have seemed, since the "Luan" spelling is actually a composite of two separate misspellings: "u" for "ou" and "n" for "nn".  All the other attested misspellings of Brizendine's first name ("Louanne" and "Louan") preserve the "ou" -- "Luanne Brizendine" and "Luann Brizendine" are not attested -- so "u" for "ou" stands out as an unusual error.  As for "n" for "nn", the only moderately frequent misspelling of her first name (before Macrae's review), "Louanne", preserves the "nn" as well as the "ou", so this misspelling, too, is unusual.

So much for the misspellings.  The other error is "mind" for "brain", which is surely independent of the misspelling; absolutely nothing would predict that someone who makes one of these errors would be likely to make the other.  So we have TWO shared independent innovations/errors, and stronger evidence of descent from a common source.

zwicky at-sign csli period stanford period edu

Posted by Arnold Zwicky at December 1, 2006 02:41 PM