Anyone run across this? - it's called the Gender Genie and it's supposed to 'guess' what gender you are by analysing a segment of your writing - to call it flawed is an understatement as in the context I couldn't see why my writing was considered female. But it was guessing other male authors were female too. lol.
I've heard of it, but I've never used it. It sounds silly, but I'm curious as to how it decides what sounds masculine and what sounds feminine.
I noticed a lot of it were using certain prepositions, conjunctions and to be verbs. Actually it seemed like the less clear you are - a at it - instead of she, and and me is considered masculine. and around rather than with is considered masculine. Very weird.
I just used it, it said my writing was feminine too. I don't see how you could base gender off key words. I mean maybe you can, I just don't agree with it.
I tried three pieces, and each time it said mine were feminine. Key feminine words included "and." Even a story I wrote with multiple different references to mean an erection was considered feminine. But it was correct, so I can't really argue.
Hmm. On my first, shorter, sample, it said that it was 'completely confounded' - my male and female scores were identical. On my second, longer, sample, it correctly guessed that I was female.
Maybe it just figures women are better writers than men so if it's written well it must be feminine. So... I tried it and by my theory my writing must suck because I was way male!
For my short sample (200 words), it said female; there was a difference of 28 points. My next sample was ~450 words, and it said male, though the scores were 12 points apart. The third one was ~530 words, and it said male; the difference here was over 70 points. This really is a silly program. Using articles ("a" and "the") is considered masculine? Since when? It was a fun program to use, but I don't take the results very seriously.
Hi, Which fits well with Andy being the most feminine of all names! (I suppose there is Andie MacDowell). Cheers, Greg.
I got interesting results. The gender of the character I was writing in the perspective from was the gender it ruled the text to be from! Weird. haha
Heh, so I run my excerpts, then my husband's, and we're both females. According to the points, he's actually even a bigger female, but then again, while the passages were about the same size, there must have been a different ratio of "feminine words" and "male words."
i just got 'male' assessments for a short story and 2 essays... the biggest gap was 752 female to 1018 male on the second essay... the first and the fiction piece were slightly lower numbers, but still with at least a 200+ differential... and i'm not only female, but a mom of 7 who was married twice and had too many long term 'love affairs' and shorter 'liasons' to count! [all heterosexual]... so, either the whole sytem is flawed, or i write 'ambigenderally' with a male-ish slant... judging from previous feedback both human and otherwise, i have to assume it's the latter...
Really? I would assume the former. But then again, I believe that gender stereotyping is all so much rubbish anyway.
Just curious, if you don't mind my asking, what your digit ratio is? Are you a girlie girl? I'm also hetero but definitely not a girlie girl and I have shorter index fingers than ring fingers. Two paragraphs from an earlier draft of my book: Two paragraphs from a later draft: And a section with more dialogue: The paper the algorithm is based on appears to be based on valid science: Gender, Genre, and Writing Style in Formal Written Texts. I wonder how long of a submission is needed for accurate analysis. My guess is one needs more than a couple paragraphs.
Valid science? All they are doing is performing a signature analysis based on a partitioning of a population. It shows that there is a statistical difference between the two partitions, and that has some limited predictive value across those two partitions in a similar sampling environment. So in the sanme social setting, it will reinforce that a socialized difference in behavior exists. As the social environment evolves, the validity of the model will change as well. And to be fair, if there were inherent differences in ways of thinking and writing between the genders, elements of the signature may persist. But the existence of the signature difference does not prove that there is an inherent difference in the way men and women think and write. In point of fact, true scientific method never proves the truth of a theory. It can only disprove theories, leading to increased support for one or more other theories. But these statistical sampling methods are the smoke and mirrors of modern science,more often misused than applied meaningfully.
And? What's your issue, that it reflects on social and cultural differences instead of biological differences? Are the researchers saying anything else? Biology suggests at least some differences will persist. If you define "inherent" as purely biological, sure. But even then, with this one study you cannot rule biology out either. There is plenty of evidence that there are biological differences in how men and women think. The gender bias mythology comes when that difference is claimed to affect more than it actually does. If the sample is appropriately randomized, and large enough, then the study shows there is an objective measurable difference. If you find the methodology flawed, you'd have a point. But I don't see that you are addressing the methodology of this study, rather you are lumping all statistical analyses into one pot, as if the statistical analysis was inherently flawed. It isn't. It just happens to be rife with misuse. Are you claiming anything using statistics is inherently bad science? As for science 'proving' theories, it doesn't. Anyone with the most basic science knowledge knows that. No one said anything about "proving". Please re-read what I said: "The paper the algorithm is based on appears to be based on valid science."
And my point is, that's great hand waving. But you can't just dismiss findings you don't like or that you don't believe. How about some specific criticisms of the study's methodology.
Looking at the study, the actual indicators are extremely bland. All it does is say that men are more likely to use determiners and quantifiers, and women are more likely to use pronouns. There is no magical style-analyzing algorithm. The gender genie doesn't really tell you if you write like a boy or girl, just how many determiners & quantifiers, you use as compared to pronouns. If there are differences between genders, is it because of biological differences, or because of societal molding of gender norms? That is to say, is it more nature or nurture? This question is far more interesting to me than, "Do I write like a female or male," and this study makes no progress toward that end.
If the study methodology is sound, then the indicators are likely going to be valid. Let's start with the abstract: (bold is mine) The researchers found specific objectively measurable differences that correlated with gender. You can object to the sample. Was it random? Was it large enough? Do the results only apply to the Brits? As with any good research, the limitations in the scope of the conclusion are noted: Next they address criticisms of past studies: So the sample size is large, diverse and objectively measured. I find no flaws there. The sample and size: I can't say that the 604 documents represent a random sample but the researchers go on to describe how they assured randomization: The methodology is further described: As I continue to look at the study, the only limitation I see is the fact the sample is from a British source. So obviously additional studies need to be repeated using samples from other countries. What I see are people dismissing the science without looking at it. By all means, let's hear what is wrong with the methodology besides one doesn't believe the results.
ginger... my index and ring fingers are exactly the same length on both hands... so, what does that make me?
Ginger, You misread my post. I agree that the indicators are valid. I never said they aren't – I said they are bland. The "Lexical and Syntactic Features" are nothing more than how many determiners {a, the, that, these}, quantifiers {one, two, more, some}, and pronouns {I, you, she, her, their} you use. *Yawn*
Given that, the genie only really determines the content you're writing about. If you're talking about things, you're not going to use he/she pronouns, and if you're describing the relationship between two characters, you'll use more. And what samples were they analyzing? Conversations? Blog entries? Magazine articles? The only thing you can say about the study is that it found that, in general, women talk more about people and men more about things, but it doesn't mean women can't write about things or men can't write about people.
