[Pred the Penguin]

Code: [Select]

Total words: 15
Too few words.  Try 300 words or more.

Genre: Informal
  Female = 0
  Male   = 25
  Difference = 25; 100%
  Verdict: MALE

Genre: Formal
  Female = 47
  Male   = 35
  Difference = -12; 42.68%
  Verdict: Weak FEMALE

Weak emphasis could indicate European.


[watsisname]

OH SNAP, ANOTHER WEAK FEMALE!  GET OVER HERE, WE GOTTA COMPARE OUR EUROPEAN SHOULDERBAGS!

It's like its got this hint of sexism but you can't quite pinpoint exactly where!

[Flipside]

Oh nice, a program that manages to combine racial and sexual stereotypes in one easy bundle....

[Enigmatic Entity]

I think I get a formal average of 75% male and informal percentage of 50% based on lab reports and silly little stories and lists that I have written...

[Nuke]

so i took a sample out of one of my old colledge reports and tested it. mind you this is circa 2002. and here are the results:

Total words: 1093

Genre: Informal
  Female = 1121
  Male   = 2375
  Difference = 1254; 67.93%
  Verdict: MALE

Genre: Formal
  Female = 1158
  Male   = 1748
  Difference = 590; 60.15%
  Verdict: MALE

not satisfied i looked at my most recent forum posts, found the biggest one on the first page and got this:

Total words: 472

Genre: Informal
  Female = 788
  Male   = 1078
  Difference = 290; 57.77%
  Verdict: Weak MALE

Weak emphasis could indicate European.

Genre: Formal
  Female = 496
  Male   = 470
  Difference = -26; 48.65%
  Verdict: Weak FEMALE

Weak emphasis could indicate European.

the first sample was obviously formal, since it got me an a in my psychology 101 class. the second sample just being a long forum post on gd from one of the recent space threads (so slightly technical), sans spelling errors. so id call it informal. using the right genre preference, id say their algorithms are adequate though far from perfect.

[Lester]

Weak emphasis could indicate European.

Code: [Select]

Total words: 481

Genre: Informal
  Female = 558
  Male   = 877
  Difference = 319; 61.11%
  Verdict: MALE

Genre: Formal
  Female = 183
  Male   = 591
  Difference = 408; 76.35%
  Verdict: MALE
Not bad for a European, eh?

Conclusive proof that EU >>> US. Both in masculinity and otherwise.

[Nuke]

for the record i am not european.

[Colonol Dekker]

Proud English.

[-Sara-]

When I wrote something simple and normal about daily things, I get this:

Code: [Select]

Genre: Informal
  Female = 725
  Male   = 285
  Difference = -440; 28.21%
  Verdict: FEMALE

Genre: Formal
  Female = 676
  Male   = 350
  Difference = -326; 34.11%
  Verdict: FEMALE
But if I use more serious text about games or aliens like the shivans I said in the BP forum, it says something else again. So I'm not sure about these silly things, I guess it gives words a score or whatever which then says how likely it is a guy or girl wrote it. If I write formal, I write the way my old teachers at school told me to write. And ofcourse because Dutch standards were not so high back then for learning English, I used what I read on forums and have seen on television and what I saw in games. Having to write scriptions for college and artschool also makes you look at native english texts to get your sentences just right. A good example is my old teacher I had the last few years who was a britton himself. He saw I had a keen interesting in improving my english and asked if I wanted to go a step further by learning to use british english. I say colour and not color and Americans and Europeans raise eyebrows when I pronounce the word 'either', while the British crack a smile and give a thumb up. It sure shows that when you are not english, you instead go with what you learned. Monkey see, monkey do?

Taking a text from my post history that I wrote, trying to be a lot more formal, gives:

Code: [Select]

Genre: Informal
  Female = 2628
  Male   = 7291
  Difference = 4663; 73.5%
  Verdict: MALE

Genre: Formal
  Female = 3926
  Male   = 4735
  Difference = 809; 54.67%
  Verdict: Weak MALE

Weak emphasis could indicate European.
So I say, shenanigans! I might be an alternate rock baitch, but I'm sure not butch, lol.

So here comes the interesting part.. I tried playing around with that tool a bit! And the results? Amazing(ly wrong), but really funny too! Some really funny facts I collected by playing around with the tool  :

• When I used the words hon, honey, love, sweetie and dove a lot, I got a predominantly female score!
• Upon pasting texts written by a 12 years old Yahoo!Answers user who lacks punctuation marks, it almost always says female!
• I pasted a text from a feminist blog, then another text and yet again another text. They were all rated very Male (over 75% of the score)!
• Also I pasted some rants, complaints and generic writings by Conservative Republicans which I googled up and despite the poster's gender, almost all were rated as weakly female!
• If I paste texts written by people with a learning disability, it says they are female!

So I guess that debunks it. It embraces sexism to the point that everything written by someone with not such a high IQ is considered female, while formal and elaborate texts are rated as male! And that by an Israeli university? We must release Herr Battuta on this topic at once!

[Snail]

Code: [Select]

Total words: 933

Genre: Informal
  Female = 180
  Male   = 2607
  Difference = 2427; 93.54%
  Verdict: MALE

Genre: Formal
  Female = 888
  Male   = 1406
  Difference = 518; 61.29%
  Verdict: MALE

[Luis Dias]

.....and to call europeans "WEAK FEMALES". I mean just gimme a gun and an address, will you?

[MP-Ryan]

Am I the only one who actually opened the cited paper that spawned this thing?  If one looks at the abstract...

This paper explores differences between male and female writing in a large subset of the
British National Corpus covering a range of genres. Several classes of  simple lexical and
syntactic features  that differ substantially according to  author gender are identified, both in
fiction and in non-fiction documents. In particular, we find significant differences between
male- and female-authored documents in the use of pronouns and certain types of noun
modifiers: although the total number of nominals used by male and female authors is virtually
identical, females use many more pronouns and males use many more noun specifiers. More
generally, it is found that even in formal writing, female writing exhibits greater usage of
features identified by previous researchers as "involved" while male writing exhibits greater
usage of features which have been identified as "informational".  Finally, a strong correlation
between the characteristics of male (female) writing and those of nonfiction (fiction) is
demonstrated. 

...one will find the experimentally-derived rationale for the formula.  Don't criticize the tool for sexism; it's merely reporting on writing tendency correlations.  It has nothing whatsoever to do with IQ, stereotypes, or sexism.

(As an aside, it is unsurprising that much formal/technical writing returns a male result as it tends to be used in fields that, until very recently, were/are dominated by men.  Successful women tend to emulate male writing styles in those scenarios, and in doing so conform to the expectations of the reader.  Nowhere is this more evident than in journalism - a point the paper touches on).  As for weak emphasis indicating European, non-native English speakers should show fewer markers in general because they tend to think in their own native language - most of which actually incorporate gender directly into several aspects of language (while English generally does not).  French is an excellent example.

But don't let me get in the way of silly outrage, please do continue. *eyeroll*

[Nuke]

this implementation may be bogus but it makes you think that this is a thing computers can do given the proper algorithms. no doubt the basis of these algorithms is stereotypical data. you would have to to a lot more research into how males and females use language, identify other cues than simply whether words used were weighted male or female. take it a step further and you could probably identify country of origin as well. id bet the cia has something that can do this.

[MP-Ryan]

no doubt the basis of these algorithms is stereotypical data.

Looks observationally-derived from a scientific methodology to me.  See my post above.

[Nuke]

no doubt the basis of these algorithms is stereotypical data.

Looks observationally-derived from a scientific methodology to me.  See my post above.

sorry we posted at exactly the same time there. and i wasnt gonna edit

[MP-Ryan]

sorry we posted at exactly the same time there. and i wasnt gonna edit

Fair enough. =)

[Luis Dias]

It mostly seems statistical rubbish to me, conveyed to demonstrate the stereotypes that the researcher already had in his mysoginistic mind.

[MP-Ryan]

It mostly seems statistical rubbish to me, conveyed to demonstrate the stereotypes that the researcher already had in his mysoginistic mind.

Maybe you should have read the paper before leaping to conclusions.  If you had, you'd see that the principle is based on 30 years of research and that the methodology used here is based on rigorous AUTOMATED (i.e. no human input) statistical analysis of published documents (including journal articles), half from writers of each gender, from a reputable source collection (British National Corpus).  Page 5 of the PDF, if you'd like to stop making silly assertions without a shred of evidence.

Or carry on, whichever.

[Luis Dias]

Every statistical analysis is "automated" in the sense of being a computation, so I really don't get where you are going there with that word. The paper is fine and dandy, in that sense. It would only get interesting if it really predicted anything outside of its own database.

[Enigmatic Entity]

Your little deviations produces this result, you guys:

Genre: Informal
  Female = 107
  Male   = 317
  Difference = 210; 74.76%
  Verdict: MALE

Genre: Formal
  Female = 230
  Male   = 160
  Difference = -70; 41.02%
  Verdict: Weak FEMALE