Is there a somewhat reliable (like, for example 70% success rate) way to determine whether a paragraph in the English language was written by a man or a woman (adult male/female)? Any credible references to the study, and perhaps a how-to?

If this is possible, what about the general case of generally telling the difference between the writing style of a male and female person (the scope is increased to any English speaker that are 10 years or older).

Of course, professional writers may be excluded as they're probably good enough so that they can deliberately pick a writing style to suit the occasion.


3 Answers 3


Yes, this is possible. Deitrick et. al. (1) describes a model that yields 95% accuracy in classifying the gender of the person based on the e-mail that he/she wrote. Apparently men and women tend to have differing word choices. This research is based on the Enron corpus, and thus meets your criteria of non-professional writers (all of them were Enron employees, hence the percentage of journalists or fiction writers would probably be small).

  1. Author Gender Prediction in an Email Stream Using Neural Networks, 2012

I don't know how accurate they have got but there are ways of telling the gender of a writer based on word usage - at least for fiction.

The famous author V S Naipaul claimed that he could easily tell based on women's more emotive language - there is a test based on his claim in the Gruniad (UK newspaper)

There is a website based on some CS research that uses word choices to guess

  • 1
    I really stunk at guessing the writer's gender in Naipaul's test. (And was told I need to read more books by men -- probably true.) And the other link guessed wrong when I plugged in some writing samples. So, concluding anecedotally, I would say it's pretty hard to determine gender by analyzing writing.
    – JLG
    Apr 27, 2012 at 2:54

In a word, no.

Are you, by any chance, doing a Literature degree?

Look at 'Conley, J. M., W. M. O’Barr & E. A. Lind 1979 ‘The power of language: Presentational style in the courtroom’. Duke Law Journal, 1978, 1375- 99.' This describes the style of language used by the accused in a court-room, and they found that it had all the claimed characteristics of women's speech listed by Robin Lakoff etc, regardless of the gender. The conclusion is that if women's speech does differ from men's, it is down to the power relationship rather than the gender.

More indirectly, look at the work of Simone de Beauvoir (The Second Sex) or, later, Judith Butler (Gender Trouble). Simone de Beauvoir is summed up as the famous "one is not born but becomes a woman" sentence, meaning that 'female' and 'woman' are not the same thing, and Judith Butler went on to argue that 'woman' is a performance and not genetic. On that basis, 'woman' exist basically as an idea, a socially accepted role-play, and as such cannot really be attributed with any inherent language characteristics.

  • yeah it is social/ cultural phenomenon (same to accents) that enforces all males and females to use a definite 'idealism' in speech, hobbies, dressing, color-preference, etc. Oct 22, 2016 at 14:37
  • As proven by Johnny Tsakiris, this statement is now incorrect.
    – annawie
    Jun 4, 2017 at 4:17
  • Only if you think correlation = causation.That report detected a social group that is predominantly female, and then leapt to an invalid conclusion that because the group is mostly female, being female must 'cause' the group. Language characteristics can also detect age groups quite accurately, but that doesn't 'prove' that being a particular age causes those characteristics. For example, 'her husband' is (arguably) associated with female writing, but the cause is social pressure to be 'manly' preventing men from discussing third-party's husbands. It is a social thing, not a sex thing. Jun 4, 2017 at 5:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.