«  |  »

Confront your inner butch

Le Génie du Genre (entendez: sexe) est un petit logiciel que prétend savoir déterminer si un texte (anglais) a été écrit par un homme ou par une femme. Résultat: soit ça ne marche pas, soit je ne suis pas une vraie femme.

  • 2004-07-21

“That is one butch chick,” is the Gender Genie’s unchanging comment when I once again inform it (him? her?) that I am, in fact, not male. Its verdict has a very funny side since I am not butch either, though several of my dearest friends are, and proud of it, and I of them. But that’s another story.

The Gender Genie eats texts, digests them and comes up with an opinion about the writer’s gender. (I’m inclined to write “sex” instead of “gender” whenever appropriate, but here “gender” does fit better.) The program behind the input box searches the text for “masculine” and “feminine” keywords, assigns a weight to them, and calculates a male and a female score. Whichever is higher determines whether, to the genie, your texts look masculine or feminine. Its success rate, according to user feedback, is only about 62.5%, but this number can’t be very reliable: Are you more likely to tell the genie that it was wrong or that it was right? Is there any indication that the likelihood of a user giving feedback is the same in both cases?

My eggcorn post, declared to the genie as a blog entry (which it is), turns out to have a female-over-male score ratio of 0.49 — a long way from the 1 I’d need to be even considered androgynous. Assuming a very simple linear probabilistic model, the probablity that my texts were written by a woman is 34.8% — and the likelyhood for me being male 65.2% (probf = scoref / (scoref + scorem)).

Why does it think I’m a man? The main culprit is the definite article. The contribution of my use of the to the male score is nearly as high as my entire female score. (I re-read my texts, and found there are a few definite articles there that might be grammatically questionable. But they are in the minority: all in all I stand by my thes.) Then follow the masculine keywords this, in, is and some. Small traces of femininity can be found in my using but, like, so and out. When I pretend the post isn’t a blog entry but non-fiction, with, not and and count towards my feminine side, but the overall result is even worse (31.5%). I fare a little better with other posts. This one is a little less macho (maybe the genie has looked at the picture?), but I’m still clearly male (probability for being female: 43%).

So do I write like a man? Maybe. One reason might be that I’m modelling my writing, which is done in a foreign language and certainly heavily influenced by native English-speakers’ writing, after writers that happen to be male. Or that I’m feeling some inscrutable affinity with male writing. At least I’m in good company: I submitted random articles by Polly Toynbee and Mary Ridell, two Guardian journalists, and they, too, seem to prefer a masculine style, as does Naomi Klein. Even better, the first chapter of A Room of One’s Own was apparently (probm: 56.5%) written by a man. The genie didn’t declare Virginia Woolf a “butch chick” though; it defensively replied, “I didn’t write the algorithm. Blame the scientists.”

What to think of this? I have no idea.

Incidentally, I’m having an interesting email conversation with someone, a man, who referred to me with a masculine pronoun after reading this blog. I’m completely convinced he (no, I won’t tell who; this is not about pointing fingers) is usually perfectly careful about making assumptions. I’m not at all offended — the error occurred in a quickly penned post, and my first name is gender-ambiguous after all — and got a charming apology. Still, there’s apparently nothing in my writing that points to my real sex. I was a bit surprised that I hadn’t revealed it explicitely at least somewhere on this site, but I checked and, yes, I only once used a feminine form to refer to myself, in a French text on one of my (badly out of date) photography pages. I even remember choosing “la photographe” instead of “le photographe” (which in France a lot of people still consider as gender-neutral).

But maybe there is something to it. I tend to prefer being judged on my own merits to being pideonholed from the outset, even if the pideonhole is quite large, like the one labelled woman. On the other hand, I make a point of not using fake “gender-neutral” masculine forms and even employ ugly constructions like double suffixes (eg mes ami/es) in French, the morphology of which makes gender-inclusive writing rather difficult. In France, this alone marks my writing as that of a woman. Marina Yaguello, who is a linguist at Université Paris 7, has written two books (small ones, for the general public) about the women and/in language question, among many others.

And as a reader? I’ll admit it, I’m fond of detective fiction (my US friends will be shocked to hear that everything I know about Chicago is learnt from reading Sara Paretsky (well, okay, not everything)), and I find myself reading almost exclusively mysteries written by women. It’s embarrassing; I really don’t discriminate intentionally. Still, in what I’ve read over the last year or so I’ve made an exception for precisely two male mystery writers, Ian Rankin and Iain Pears. What’s strange is that when I recommend, say, Laurie King or Manda Scott or Frances Fyfield or Anne Rambach (who used to be an acquaintance), or even classics like Dorothy L. Sayers to fellow mystery lovers, someone turns up his (yes, his) nose and says, “Well, this is nice but more, like, for women.” I have no idea why. The same happens with movies, typically those who feature more women than men among their main characters. It can’t be a lack of violence (Anne’s books are really quite bloodthirsty). Is it just the sex of the protagonists?

Or maybe I do have an idea after all, but only a vague one, unbacked by knowledge, research or even google searches. The elusive difference between men’s and women’s speach (and writing) many researchers try to frame and explain is not just an invention. In everyday reading for enjoyment, I wonder how much of this difference is language, and how much subject matter, characterisation (in particular of female fictional characters) or the world-view that underpins a work of fiction. Things get even murkier: like all presumedly feminine and masculine traits, differences in language surely follow a wide statistical distribution. There are, I think, poles, but no sealed walls between them.

During the email exchange I mentioned, I was pointed to theories about how and why uses of and attitudes to particular uses of language (may, and I think do) differ between men and women. Reading my way through explanations in terms of cultural differences (men’s vs women’s culture) or male-over-female dominance leaves me unsatisfied. I’m fundamentally sceptical of “two cultures” theories: they tend to be applied beyond their scope, thus getting out of hand, and battle with the slippery slope that leads towards essentialism. Male dominance sure exists (anectotal evidence alone is overwhelming) but is is a model that can explain every part of what it is supposed to? And why should we need to decide between the theories? If a group is assigned a socially inferior (or, for that matter, superior) position, tokens of membership will most likely develop, including linguistic markers. Conversely, cultures cohabiting inside the same social and linguistic context, however they may have formed, will rarely be considered of equal rank.

I’ve recently been on the receiving end of what could be described as (institutionally sanctioned) bullying by people who I professionally depended on and was practically, if not officially, subordinated to. All of them were women. A lot of the way they interacted with me felt insidious and even cruel, but was couched in terms of care and concern. It looks like the flip side of solicitude and nurturing to me. I believe research into women’s speech needs to keep well clear of loaded terms that connote moral judgements. Male-over-female domination doesn’t make women morally better, not even on the average. (This is also a pitfall with respect to minorities. Whenever a member of a minority commits a major crime, some members of the majority group start pointing fingers. I don’t see any reasons why there should be proportionally fewer criminals, idiots, or otherwise unpleasant people among a minority. The reason discrimination is wrong doesn’t derive from the one discriminated against being a better person.) (I am not talking about behaviour. Physical aggression directed against others, eg, seems to be much more prevalent among men, a fact obscured in much journalistic writing and even some research on crime and violence — maybe because men being “by nature” more violent than women is just taken for granted. I’m not sure this is a good idea.)

Well, two parentheses and an aporia are as good a place to stop my rambling as any.

(Note: according to the Gender Genie, the probability of the author of this text being a woman is 39%.)


2 commentaire(s) pour 'Confront your inner butch'

  1. (Comment, 2004-07-22 01:41 )
    #1Root

    This is really interesting. Can we tell guys and girls apart from postings, comments, emails etc?
    And if so why? Can it be faked? I do not know, but generally I have found it very wise not to make any presumptions at all. And I dare not test my own writing :)

  2. (Comment, 2004-07-24 01:13 )
    #2chris

    If it were just as easy as the genie thinks… I’m quite sceptical about the possibility of ever getting the author’s sex correctly identified for all writing, or nearly. (For the moment I’m unable to find any text the script considers “feminine”. Hm.