Gender in text
Posted on July 08, 2003 @ 16:35 in Research
The Boston Globe has an article citing Judith Butler about the performativity of gender. How often does that happen?
Actually, the article discusses the findings of an Israeli research group, who devised an automatic text classification algorithm that will sort English language texts according to the gender of the author with an 80% accuracy. Like most automatic text analyses, the study employs a frequency analysis, finding that it is not the "important" words that make a difference, but the much used personal pronouns, determiners (a, the, that), cardinal numbers, and quantifiers (more, some) that are significant. As the reporter notes, this isn't even terribly exciting news, but apparantly it got written up in the news media because the study kicked up some dust:
When the group submitted its first paper to the prestigious journal Proceedings of the National Academy of Sciences, the referees rejected it ''on ideological grounds,'' Koppel maintains. ''They said, 'Hey, what do you mean? You're trying to make some claim about men and women being different, and we don't know if that's true. That's just the kind of thing that people are saying in order to oppress women!' And I said 'Hey-I'm just reporting the numbers.'''
The article does a good job in discussing the different points of view in gender studies with regard to this kind "I'm just reporting the numbers" reporting. I'm sure the researchers did a decent job and that they found a statistically significant correlation. The problem, however, lies in the fact that the study starts from the premise that there are two (and only two) mutually exclusive genders, that the measuring variables are weighed according to that dichotomy, and that the findings reinforce the premise that there is a significant difference.
It's not so much that there is a difference, but rather what the difference means. That's where the reporter quotes another linguist, saying:
You find what you're looking for. And that leads to this sneaking suspicion that it's all hardwired, instead of cultural.... This whole rush to categorization usually works against women.
And this is exactly what annoyed me so much about the "I'm just reporting the numbers" statement. Numbers don't mean anything by themselves. Even if we for a moment disregard the whole socio-cultural-political framework that produced the research questions, methods and interpretations, and the grant to support the research, then the researchers should still be aware that they have a moral and ethical responsibility to take a stand on what they think their findings mean. "Just the numbers" can be used by anyone to mean just about anything, if only those numbers are framed in a particular way.
How do these researchers want their numbers to be understood? Do they want to reinforce the status quo with regard to gender? What do the differences in uses of pronouns say about the social, economic, and political situation of its speakers? What can be made of the reported fact that, when either a man or a woman writes for a specifically male or female audience, the reported algorithm 'doesn't work anymore'? It's these questions about real, everyday situations and consequences that consistently fall through the cracks of "just the numbers science," and I'm glad that for once, instead of just reporting the numbers, a mainstream publication goes a long way to question just that attitude in science.
The full article can be found on one of the authors' homepage: M.Koppel, S. Argamon and A. Shimoni (2003) "Automatically categorizing written texts by author gender," Literary and Linguistic Computing 17(4).
Comments and Trackbacks
No comments or trackbacks for this entry yet.
Post a comment
Comments and trackbacks have been closed on this site. My apologies.
Since MT-Blacklist inexplicably stopped working I had no other recourse than close comments and trackbacks to stop the spam. I've been meaning to correct this for quite a while, but life got in the way... in a good way I should add.