Some linguists have developed algorithms that purport to predict a writer's gender by analyzing writing samples. It's called the Gender Genie and you can find it here. The algorithm ignores content and focuses on key words consisting primarily of conjunctions, prepositions, pronouns, etc. Intrigued, I tried it out with various samples from my blogs. I began by entering a couple of book reviews - and discovered that I'm a male!
Figuring that book reviews follow particular conventions that may reflect a more "masculine" approach to writing, I tried out the genie with several routine blog entries, family news, etc., that I've made over the past several months, from August through December. Two of them were written by a female and the other four were written by a male! YIKES!!!! Don't tell Dave about this - he thinks he's been married to a woman for almost 28 years!
So, what's the deal here? Am I about to undergo an identity crisis? Or is there a flaw in the algorithm?
Well, having read some of the theoretical background (there's some information available in the New York Times and Nature), it turns out that the algorithm is based on some tired old cliches about female and male communications styles. I know you've heard or read this before, so say it along with me: females are more comfortable talking about people and relationships, males are more comfortable talking about things. Methinks perhaps the algorithm is flawed. Regardless, the genie is still kind of fun, so you may enjoy trying it out. Maybe you'll discover your inner male or female!
(PS: I entered this post into the genie - it was written by a male.)
6 comments:
What a hoot...I married a guy before the courts moved in that direction. Would this tool still say we are on the cutting edge by having two children. LOL
No matter what, your interesting relationship is legal here in Canada!
I must try that genie sometime. I'm having a hard enough time keeping up with the family blogging, especially mine, never mind self analysis! It would be a hoot, though.
i'm 100% girlie! i was a little worried though when kevin's first test came up girlie too (so was he) but the other 3 were boy. he's happy that his masculinity has been redeemed
well ... you are the mother of two boys. Maybe it's just all that testosterone that's flying around your house. I'll be interested in seeing what mine comes up as.
In the Netherlands research has shown that women tend to use particular words far more often than men do, and vice versa. Examples: "schattig", "beeldig" ("cude", "gorgeous") are typical female words, also other words relating to "care". Interesting, that women also tend to use identifiable languge patterns. If you don't fit into the pattern, don't worry, I think that there is only a probability that any given individual will fit in it. Just as not all smokers die from lung cancer (I'm a smoker, hence the horrible example)
Erik - The research to which you refer looks interesting. I think the differences in "male/female" vocabulary should be tied to differences in conversational subjects rather than gender. Talking about people and relationships requires a different subset of language than does talking about things and events. Women and men may have particular conversational tendencies, but they also cross those boundaries a great deal of the time. The analytical algorithms are not robust enough to account for those crossovers. Essentially, they are based on the probability, determined by the examination of a very small set of words, that a writer is male or female. As with any probabilistic equation, there will always be exceptions to the "rule."
Post a Comment