Do women and men write differently?

The title of this post is a question I’m frequently asked. Usually, what the questioner wants is (a) confirmation that there is, indeed, a difference, and (b) a list of the main stylistic features that distinguish women’s writing from men’s. If you’ve read this blog before, though, you won’t be surprised to hear that my actual answer is not that simple.

When people ask questions about male-female differences, they’re rarely motivated just by idle curiosity. They may formulate the question as a neutral inquiry into the facts of a given matter (‘how do men and women do X?’), but often the underlying question is more like ‘why do women have a problem doing X?’, or ‘what are women doing wrong when they do X?’ In relation to language, that last question is perennially popular: it’s the starting-point for all those ‘521 Verbal Bad Habits Women Really Need To Fix’ pieces. Recently, the critics who write this stuff have been preoccupied with the way women speak. But there’s also a long tradition of criticism which focuses on the way women write.

One commentator who has managed to link the two is Naomi Wolf. In the article she wrote last summer criticizing young women’s use of uptalk and vocal fry, Wolf suggested that these ‘destructive speech habits’ had also infected women’s writing. Among university students, she claimed,

Even the most brilliant [women] tend to avoid bold declarative sentences and organize their arguments less forcefully [than men].

As I pointed out at the time, this claim does not stand up to scrutiny. But it’s exactly the kind of claim that doesn’t usually get scrutinized, because it repeats something we’ve heard or read a million times before. The (spurious) connection Wolf makes with women’s speech gives her argument a modern twist, but essentially she’s just recycling a very traditional view of women’s writing–that it differs from men’s in being less forceful, less daring, less logical in its structure and less individual in its style.

That view was already familiar in 1922, when the Danish linguist Otto Jespersen made one of the earliest attempts to survey what was known about gender differences in language-use. Women, according to Jespersen, were linguistically less innovative and less adventurous than men:

Women move preferably in the central field of language, avoiding everything that is out of the way or bizarre, while men will often coin new words or expressions. …Those who want to learn a foreign language will therefore always do well at the first stage to read many ladies’ novels, because they will there continually meet with…everyday words and combinations.

Jespersen also had doubts about women’s capacity to organize an argument using the complex syntax which is typical of formal writing (21st century readers should note that the word ‘period’ in this quotation means ‘sentence’):

A male period is often like a set of Chinese boxes, one within another, while a feminine period is like a set of pearls joined together on a string of ands and similar words.

What he’s saying is that men use subordinate clauses which allow them to specify the logical relationships between points (‘because…’, ‘however…’, ‘therefore…’), whereas women just string their points together in any old order using the all-purpose co-ordinator ‘and’. Actually, Jespersen seems to have thought that any kind of sentence-construction posed a bit of a challenge to the average female mind:

Women much more often than men break off without finishing their sentences because they start talking without having thought out what they are going to say.

If you’re wondering what evidence Jespersen had for these sweeping statements, the answer is, very little, and none that would pass muster today. But then as now, dodgy propositions about male-female differences often went unquestioned so long as they resonated with popular sex-stereotypes. And if they seemed to be at odds with the stereotypes you could always find a way to make them fit–as Jespersen ably demonstrates in this comment on an experiment which found that women read faster than men:

But this rapidity was no proof of intellectual power; some of the slowest readers were highly distinguished men. …With the quick reader it is as though every statement were admitted immediately and without inspection to fill the vacant chambers of the mind, while with the slow reader every statement undergoes an instinctive process of cross-examination.

In 1977 the researcher Mary Hiatt attempted a more systematic study of male-female differences in writing style. She picked 100 passages from works of popular fiction and non-fiction produced by male and female authors, and subjected them to linguistic analysis, plus a bit of basic number-crunching. Her main conclusions were that women used shorter, less grammatically complex sentences, had a less ‘authoritative’ style and were less likely than men to write in a way that stood out as ‘noteworthy’ or individually recognizable. In other words, she basically agreed with Jespersen. But this being the 1970s, she favoured a different explanation:

The chief reason is doubtless that women are a minority group, more likely to conform than to dare. …they seem unsure that anyone will believe them, reluctant to arrive at conclusions, a bit over-determined to present a cheerful face…

Hiatt’s methods don’t meet today’s standards either, most obviously because her data sample was so small. Since the 1980s, technological advances have enabled linguists to work with much larger samples. And I mean MUCH larger. One resource that’s often used by linguists in the UK, the British National Corpus, contains a hundred million words of authentic English, and was designed to offer a ‘balanced’ sampling of written genres, from scientific articles to tabloid editorials.

But even with massive amounts of data and powerful computers to process it, the question of whether men and women write differently is not a straightforward one to answer.

The methods used in corpus linguistics are pretty good at identifying what’s distinctive about the writing of a specific individual. It was these methods which revealed that ‘Robert Galbraith’, the obscure author of a moderately successful crime novel, was actually J.K. Rowling writing under a pseudonym. The analyst compared features of the Galbraith text to the way the same features were used in texts already known to be by Rowling. The match was close enough for him to conclude that ‘Galbraith’ must be Rowling (a conclusion Rowling then confirmed).

But it’s easier to identify an individual’s linguistic ‘signature’ than it is to do the same thing for a whole social group—especially one as large and internally diverse as ‘women’ or ‘men’. That diversity means that any generalization based on group averages will be false for large numbers of individuals.

The question is also complicated by the fact that the relationship between gender and language is often not direct, but mediated by something else. For instance, since writing is something people typically learn through formal instruction at school, men and women may write differently because they didn’t have the same access to education. If so, it’s somewhat misleading to call this a ‘gender difference’: there’s a connection with gender, but what’s producing the differences isn’t gender as such, it’s the related educational inequality.

Another thing that influences writing style is the genre someone is writing in (and, relatedly, the subject they are writing about). You don’t find the same linguistic patterns in academic articles and novels; you don’t find exactly the same patterns in history and physics articles, or in romances and action adventure stories. This kind of variation may also have a gendered dimension, in that many written genres are either male or female-dominated. If you find differences between men and women in a sample of fiction where the male texts are mostly thrillers and the female texts are mostly romances, it can be hard to disentangle the effects of gender from those of genre.

In one study of the language of blogs, the researchers found what appeared to be differences between male and female bloggers; but on closer inspection they turned out to be more closely related to the distinction between ‘diary’ blogs, containing the author’s personal reflections, and ‘filter’ or content-sharing blogs, where the author comments on the links s/he recommends. This looked like a gender difference because more women in the sample produced diary blogs, and more men produced content-sharing blogs. Of course that in itself is a gender difference; but it’s not a gender difference in writing style, it’s a gendered preference for different kinds of blogs.

The blog study was partly inspired by some research from the early 2000s which claimed to have found a way to determine an unknown writer’s sex with 80% accuracy. The researchers took a 25 million word chunk of the British National Corpus and counted the frequency of a large number of linguistic features, looking for the features whose relative frequency most reliably distinguished male from female-authored texts. They found that some of the best discriminators were

  • personal pronouns (especially forms of ‘I’, ‘you’ and ‘she’)
  • the articles ‘the’ and ‘a’
  • quantifying expressions like ‘a lot of’ and ‘fifty-seven’
  • phrases containing ‘of’, like ‘a shelf of books’.

Higher frequencies of pronouns correlated with female authorship, while higher frequencies of articles, numerals and ‘of’ phrases correlated with male authorship.

You’re probably thinking: but what does it mean, and why does it matter, if women use more pronouns and men use more articles? When someone claims that women ‘avoid bold declarative sentences’, or use more commonplace vocabulary or fewer subordinate clauses, we know why that’s meant to be significant. Anti-feminists can interpret it the way Jespersen did, as evidence of women’s intellectual limitations, while feminists can interpret it the way Hiatt did, as evidence that women’s potential has been limited by sexism. It’s not so obvious what deeper truth about men and women we might infer from differing frequencies of articles and pronouns.

But the researchers had a theory. They speculated that male writers were most interested in specifying the properties of objects precisely, while female writers were more interested in constructing a relationship with the reader. OK, it’s a stereotype (men are into things and women are into people), but it isn’t as blatantly sexist as ‘women’s writing lacks logic/boldness/force’. And at least these researchers, unlike Jespersen or Wolf, had solid statistical evidence for the pattern their theory was meant to explain.

Yet if we ask what these male and female ways of writing actually look like, the answer is a bit of an anti-climax. In one of their academic papers, the researchers illustrated the differences by comparing the opening paragraphs of two linguistics books, one written by a man and one written by a woman. The man’s book began: ‘The aim of this book is…’. The woman’s book, in stark contrast, began: ‘My aim in this book is…’. The difference is significant in the statistical sense (i.e., not just there by random chance), but it’s hard to invest it with the kind of deeper symbolic significance that a lot of people want gender differences to have.

But such is the popular fascination with its subject, this highly technical piece of research was soon repackaged in a more user-friendly form. Some enterprising person made an interactive program based on it, and put it up on a public website under the name ‘the Gender Genie’ (the site was later taken down, but something similar is available here). If you pasted some text into a box on its homepage, the Genie would guess whether the author was male or female. I monitored the site for three months, and also tracked a sample of blogs where people had posted a link to the Genie and commented on their own experiences with it.

What people invariably did with the Genie–in most cases it was the only thing they did–was paste in a sample of their own writing. Obviously they already knew if they were male or female, so presumably what they were trying to find out was whether their writing was gender-typical. And when the Genie told them it wasn’t (which happened frequently: while I was monitoring it its success rate never got above 68%), their reactions were instructive. Almost no one concluded that there was something wrong with the program, or with the basic idea of gendered writing styles. More commonly they fell to pondering why they, as individuals, did not match the profile for a ‘normal’ male or female writer.

Women who’d been misidentified as men often put this down to being ex-tomboys or geeks who had no truck with ‘girly’ things: none of them seemed offended by being told they wrote like men, and sometimes they appeared to be flattered. Men who were miscategorized as women, by contrast, more often expressed bafflement, annoyance or discomfort. They also got teased by other people in the comments: had they been writing poetry again? Were they secretly gay?

These contrasting responses underline the point that gender isn’t just a difference, it’s a hierarchy. As Caroline Criado-Perez notes in her book Do It Like A Woman, to do something ‘like a woman’ usually means to do it badly, or less well than a man would do it. It’s your basic deficit model, in which men set the standard of excellence and whatever women do is somehow deficient, weak and inferior.

Women’s writing, on the face of things, is not an obvious candidate for this treatment. If we consider writing as a basic skill, it’s one on which girls outperform boys from an early age, and if we consider it as an art, it’s one that women have excelled in for centuries. And yet the idea has persisted that men do it better. Only yesterday, I heard a male writer on the radio explaining why he preferred to read other male writers: one of the reasons he gave was that men’s writing gets to the point (while women’s by implication beats endlessly about the bush). Had he ever, I wondered, opened Finnegan’s Wake, or any of the novels of Henry James?

But that question is a bit of a red herring. When someone voices a general objection to women’s writing, you can be pretty sure that what they really object to isn’t the writing part, it’s the women part. And if that’s the problem, you can’t solve it by tweaking your prose style. There is, though, one time-honoured solution, used by writers from the Brontes to J.K. Rowling: don’t let anyone know you’re a woman. Write under a male pen-name, or use your initials, and don’t appear in public until your talent has been acknowledged and your gender no longer matters.

But won’t your writing style give the game away? Well, if you’re J.K. Rowling posing as ‘Robert Galbraith’, a statistical comparison between ‘his’ style and an authenticated sample of yours will show that you’re J.K. Rowling. But it’s a different matter if you’re an unknown woman pretending to be an unknown man.

When the writer Catherine Nichols was looking for a literary agent, she put this to the test by sending out exactly the same manuscript under her own name and a fictional male name. She found that what readers said about her language depended on whose work they believed they were reading. Whereas Catherine’s sentences were described as ‘lyrical’, those of her alter-ego ‘George’ were ‘well-constructed’. It was ‘George’ whose writing was more positively received: with seventeen expressions of interest to Catherine’s two, he was, as Nichols drily observes, ‘eight and a half times more successful than me at writing the same book’.

Nichols’s experience suggests that what causes writing to be perceived as ‘male’ or female’ may have less to do with the objective characteristics of the language a writer uses, and more to do with the tendency of readers to select and interpret data in a way that reflects their expectations. As Carol Ohmann put it in an article about the reception of Wuthering Heights (a novel whose language suggested to one reviewer that its author ‘Ellis Bell’ (aka Emily Bronte) might be ‘a rough sailor’):

There is a considerable correlation between what readers assume or know the sex of the writer to be, and what they actually see or neglect to see in ‘his’ or her work.

And that, in a nutshell, is the problem. It’s not easy to persuade people of the virtues of a ‘female style’, but it’s even harder to convince them that in reality, there’s no such thing.

Advertisements