Gendered writing
Blogger Liz Henry discusses Herring and Paolillo's (2006) findings that genre, not gender, will predict the "genderedness" of a blog writing sample, as analyzed by the Gender Genie (which uses an algorithm from Koppel and Argamon (2002). They found that "filter blogs" (that is, blogs that focus on a specific topic) tend to be analyzed as male, whereas "diary blogs" are analyzed as female, suggesting that (as Henry puts it) "genre itself is gendered."
However, Schler, Koppel, Argomon, and Pennebaker(2006) believes that gender differences do appear in blog-writing: "The stylistic and content differences we have found are each sufficient to permit reasonably accurate automated classification according to gender and age bracket." Schler et al claim that their algorithm has an 80% accuracy rate for gender or age identification for unknown samples. Interestingly, Schler et al found that writing in both genders grows more "male" with age.
For fun, I put the last 5 posts from this blog into the Gender Genie, but I split them into halves. I put the last 5 "evidence" segments into one analysis, and the last 5 "opinion" segments into another. I was surprised to find that both segments were guessed as male -- although the "opinion" parts had a more androgynous score (M/F = 1.26 vs 1.62). The last 5 posts on my personal livejournal were the most androgynous, with a male/female score of 1.16.
However, Schler, Koppel, Argomon, and Pennebaker(2006) believes that gender differences do appear in blog-writing: "The stylistic and content differences we have found are each sufficient to permit reasonably accurate automated classification according to gender and age bracket." Schler et al claim that their algorithm has an 80% accuracy rate for gender or age identification for unknown samples. Interestingly, Schler et al found that writing in both genders grows more "male" with age.
For fun, I put the last 5 posts from this blog into the Gender Genie, but I split them into halves. I put the last 5 "evidence" segments into one analysis, and the last 5 "opinion" segments into another. I was surprised to find that both segments were guessed as male -- although the "opinion" parts had a more androgynous score (M/F = 1.26 vs 1.62). The last 5 posts on my personal livejournal were the most androgynous, with a male/female score of 1.16.