Making the rounds today, I came across a couple of different references to the Gender Genie, a little online app that asks you to submit some text, does a keyword scan, and claims to be able to identify your gender with something like 80% accuracy.
So I dropped the first four pages of my book manuscript in there, and found out that I'm actually female. Interestingly, based on 35,000 responses in the past week, the site can only lay claim to just over 50% accuracy, making it about as accurate as flipping a coin. Of course, as soon as your poll audience knows why you're asking the question, the possibility of skew renders the question itself useless.
The July issue of Nature magazine has a brief article on the thing here, although I must confess to some question about why. For a much more interesting and nuanced discussion of similar issues, I recommend Author Unknown by Don Foster, the English professor who outed Joe Klein, testified in the Unabomber case, and discovered a new work by Shakespeare. Foster makes the point that this kind of statistical analysis, by itself, is useless. It's how you use the information that makes a difference, and his book is (in part) about how he came to use these methods to accomplish some interesting results.
Contra jill/txt, then, I don't find this especially scary, particularly since I've known since Deborah Tannen at least that my own language use is often "characteristically female," whatever that means.
This entry, however, tested male, clearly a result of my self-consciousness about the book manuscript...
I tried it with the first few pages of the diss (not *my* diss--to use possessives marks me as female....) and I'm apparently male. Call me Amos, I guess.