Gender Guessing By Vocabulary (Algorithm)

[devida]

I was fascinated to discover that the way that men and women write (apparently especially in the USA) is so gendered that an algorithm has been developed that will guess the gender of the person writing based on a piece of text 300 words or more. I am not sure how accurate this little algorithm is but it certainly was for me. I checked 6 different posts of mine from here and got either "weak female" or "weak male" with the caveat that I might just be European. Since I identify as non binary and am European that was extremely accurate.

Anyway, not only is the hack fun, but I'd be very interested to know how cross dressing and transexual forum members who identify as male or female, transgender, cisgender, non binary, bigender, or genderqueer do on this. I really still find it hard to believe that vocabulary, especially written, is that gendered.

I am usually gendered female on the phone, though I assumed that was because of my accent, soft and high voice and accent, but maybe it is my vocabulary. Most of the time phone marketers catch themselves pretty quickly and change from ma'am to sir though I did get ma'amed for a whole conversation a couple of days ago.

The site seems perfectly safe. No files need to be downloaded, just your text pasted into a field. There's no login and I haven't been deluged with genderqueer ads on my browser since I used it yesterday morning. But, as always, participate in anything on the internets at your own risk. It is certainly safer than a Facebook quiz where your answers are sold.

Check it out at http://www.hackerfactor.com/GenderGuesser.php#Analyze

Let me know if the algorithm guesses your sex, your real gender identity or is just wrong.

[sandra-leigh]

The large majority of my online posts are on giving technical advice, for one which naturally adopts a low-emotion highly-prescriptive vocabulary that is therefor interpreted as "male". Thus, regardless of your gender, what you write about might drive your writing to be considered one gender or another. At that point the algorithm is implicitly working at the level of "well, males tend to choose these kinds of things to write about, and females tend to choose these other kinds of things to write about."

[mariehart]

Interesting but not consistent enough for me to be totally convinced. But then again I am European and it is weighed for Americans. Generally I come out either female or weak male (European). I used several posts from here but also one from a more technical forum. That one came out male. I put in some creative writing and that came out weak male even though it was something of a boys own tale.

But most interestingly I put in some creative writing from my sister and it came out male! Not even weak! But then she is a strong personality and a manager! I think it shows.

[CynthiaD]

Hmm. This reminds me of being in Vietnamese language school in the army. All of our teachers were native speakers, and most of them were female. When we'd get a male teacher he would constantly correct us saying "Don't say it that way, it's effeminate. Say it like this ..." Apparently, male and female grammar and word choice are distinctly different. I understand that the same thing is true in other languages as well.

[NicoleScott]

That's an "algorithm"? I thought it was when Clinton's VP played the drums.

Added: OK, I took the test. Informal: Weak Female Formal: Weak Male
Could be European (I'm not, but I've been there).

[Jorja]

300 words? I don't think I have ever wrote more than 150 words for anything except in college on theses papers.

[Zylia]

English isn't my native tongue so I guess that slants the results a bit, but it noticed that I'm European anyway. I used more female words in the sample texts I provided, but I used posts from this forum, so there's that.

[Dianne S]

Interesting. I pasted in the text of a couple of short stories I wrote. The first one was written in the third person and came back "male". The second was written from the point of view of a female character, and came back "female".

So I guess the algorithm and I agree. :)

Edit: I looked at the source of the web page. The algorithm is pretty simple and only keys off a couple of dozen words. Nagivate to the page, hit "View Source" (Ctrl-U in Firefox or Chrome) and look for the section headed "function GenderWords(Form)"

[Eryn]

I input the text from my latest post about Persephone's and my adventures. It was 900 words so it should have been a good sample.

Alas, for me it comes out as:

Genre: Informal
Female = 680
Male = 1891
Difference = 1211; 73.55%
Verdict: MALE

Genre: Formal
Female = 871
Male = 1144
Difference = 273; 56.77%
Verdict: Weak MALE

Weak emphasis could indicate European.

Just for fun I cut-and-pasted a thousand words of a Nora Ephron essay:

http://audrawilliams.tumblr.com/post/29354506842/helen-gurley-brown-only-wants-to-help-nora-ephron

into the gender guesser. The results:
Genre: Informal
Female = 382
Male = 1303
Difference = 921; 77.32%
Verdict: MALE

Genre: Formal
Female = 926
Male = 1072
Difference = 146; 53.65%
Verdict: Weak MALE

Weak emphasis could indicate European.

So, I'll take heart that Nora Ephron wrote less female than I do.

How about Mark Twain? I cut and pasted some of his "advice to youth" essay.

http://grammar.about.com/od/classicessays/a/adviceyouth.htm

Genre: Informal
Female = 921
Male = 2008
Difference = 1087; 68.55%
Verdict: MALE

Genre: Formal
Female = 1331
Male = 1199
Difference = -132; 47.39%
Verdict: Weak FEMALE

Weak emphasis could indicate European.

I think that their algorithm needs work!

[kimdl93]

I randomly selected a long passage from an old post and ran the analyzer. Verdict: Informal-weak female, possibly European, and Formal-weak male...just like yours. I'm American, second generation. Of Norwegian descent and from the upper Midwest, so I don't know what to say.

It will be interesting to see the results for others.

BTW, I also used an except from "for Whom the Bell Tolls" by Hemmingway. Verdict: informal - male, formal-weak female. I doubt Pappa would have well received the latter characterization of his writing style!

[Dianne S]

Try the following text:

actually am because but everything has him like more out since so too and be her if myself not should when with


Now try:

as ever good in is now some something the this well a are around it many these what

Pretty simple-minded.

[Georgia_Maine]

I wonder if my background in scientific writing - short, concise, clear and unemotional - skews the test. My results were all male.

[Eryn]

I wonder if my background in scientific writing - short, concise, clear and unemotional - skews the test.

That's my problem too, but I'm not going to change my writing style at this point! :)

[Carol Elizabeth]

I wrote a letter to myself about 22 years ago and had saved it on my computer. The letters intent was to preserve some memories my son who suffered of a tragic death at the age of 17. I chose to use this in the analyzer because it was not something written for the analyzer that would possibly skew the results.

Genre: Formal
Female = 590
Male = 628
Difference = 38; 51.55%
Verdict: Weak MALE

Weak emphasis could indicate European.

Genre: Informal
Female = 866
Male = 912
Difference = 46; 51.29%
Verdict: Weak MALE

Weak emphasis could indicate European.

I wonder if the "Weak" part of the writing was a result of the loss - or was the "Male" part because I was trying to be strong for the sake of the family.

CE

[Carol Elizabeth]

Genre: Formal
Female = 1874
Male = 1707
Difference = -167; 47.66%
Verdict: Weak FEMALE

Weak emphasis could indicate European.

Genre: Informal
Female = 1740
Male = 3117
Difference = 1377; 64.17%
Verdict: MALE

About 3 years ago, I was asked to say a few words at the funeral of an Aunt (by marriage). She died of natural causes at the age of 103 so the death was not exactly unexpected.

The results above show two quite different results. Makes me understand how both men and women gave favorable comments on what I had said.

CE

[devida]

I guess being a lit major makes me uncomfortable with the idea of gendered writing. I can certainly accept gendered speech. I am not convinced, on the basis of the replies so far that the algorithm is actually that accurate. Maybe it is too simple or maybe most writing, especially formal or technical, just is not gendered. I really don't think you could say that, for example, poetry is gendered. But I admit to being prejudiced against the idea.

[Beverley Sims]

I do speak and use language that fits my presented gender.
I laugh, talk with my hands and make the appropriate facial expressions.

[sometimes_miss]

unfortunately, when people are aware that their work will be observed for something in particular, they will often consciously or subconsciously tailor their responses to get the results they want if they can. This is one of those cases, just like all the other online gender tests. So I took it, out of curiosity; pasted a 3000+ word sample from a comparison study that I did at work a few years ago. Got weak female. Again out of curiosity, pasted a sample from a Tom Clancy novel. Again got weak female. So much for the algorithm. I'm guessing again, but I suppose that the writers of the algorithm used specific samples of which they felt were male and female writing. Or maybe their algorithm includes what previous websites the person being tested had previously used.

[Dianne S]

See my comment #11. The "algorithm" is brain-dead simple. It looks for maybe 25-30 words, each of which has a score of "maleness" or "femaleness" and then it combines the score. IMO, it's snake-oil without any scientific basis.

[Claire Cook]

What fun!! I like to use ellipses (like this...) and lots of !! even in "male mode". So...

Genre: Informal
Female = 236
Male = 126
Difference = -110; 34.8%
Verdict: FEMALE

Genre: Formal
Female = 60
Male = 85
Difference = 25; 58.62%
Verdict: Weak MALE

I'll take that!!