Last month, a variety of parenting blogs were in an uproar over the story of a Canadian family that didn’t feel like sharing the sex of newborn Storm with the rest of the world. The media had a field day with the notion of raising a “genderless” child, even after Storm’s mother published an explanation making it clear that their goal was to buffer the child against the relentless gender stereotyping we foist on infants from day one. From garish pink onesies that proclaim “Daddy’s Little Girl” and powder blue “Little Man” t-shirts, to letting our girls’ hair grow out and cutting our boys’ hair short, to offering our girls a doll and our boys a ball, we indicate to our children through subtle and overt actions what their future role might be in society: girl or boy, woman or man.
Within this discussion about de-emphasizing gender norms for the most vulnerable members of our culture—those who are unable to think for themselves—a lot of attention has paid to bucking gendered trends in toys, clothing, and hair style, but only one news piece that I saw brought up the subject of language:
"It is very courageous to challenge [the world] on adjectives that you use on children," [Cheryl] Kilodavis [author of the children’s book My Princess Boy] tells ParentDish. "Instead of saying what a strong boy what a pretty girl, they are saying what a strong or beautiful child."Language is the most important tool that humans ever developed. It allows us to collate and categorize information to make sense of our world, and it allows us to pass on that information to succeeding generations. But language differs around the world – not only in the words used to describe something, but in the number of words used to describe something. That is, the words used by a group of people generally reflect the interests and concerns of those people – so people in cold climates have a larger range of words for cold-weather phenomena than do people living in warm climates, who may have a larger range of words related to their own environment.
This means that language can also differ along gender lines. In a paper that is often assigned in introductory anthropology courses, Daniel Maltz and Ruth Borker discuss the reasons for “male-female miscommunication.” Rather than looking to psychological differences between the sexes to explain differences in communication styles, Maltz and Borker think we should be discussing sociolinguistic subcultures, or the culturally-influenced differences between men’s and women’s approaches to communication. They suggest that women tend to use language to negotiate and express relationships; we tend to use a lot of personal and inclusive pronouns, interject questions and comments in order to show interest; and we are concerned with making segues between topics. On the other hand, jokes and stories are highly valued in men’s speech; loud and aggressive speech is common; and put-downs and insults are normal ways of talking with friends.
What about actual gendered words and phrases? Sure, English, like many languages, has masculine and feminine pronouns, as well as gendered nouns for various relationships and occupations. But we also have more subtly gendered vocabulary, as illustrated in the quote above: we praise our strong boys and our pretty girls. Two researchers at the Center for Mind/Brain Sciences at the University of Trento (Italy) recently decided to empirically test the question of whether there is a gender bias in what women and men talk about. Their goal was not anthropological, but rather computational - to find a way to model “common sense knowledge” as part of the eventual perfection of artificial intelligence (Herdağdelen & Baroni 2011):
Common sense knowledge consists of the simple facts that nearly every person knows but almost never states explicitly because of the very assumption that it is already shared by everyone. Some examples are that mountains are taller than buildings, grocery has a price, or rivers flow downhill. The assumption that common sense knowledge is shared is what allows us to communicate with other people and interact with our surroundings in an efficient and natural way. Therefore, an AI system needs to possess common sense if it aims to interact with people in a natural way.That is, we have all been enculturated into a particular way of life, and we expect people of different ages, occupations, and genders (among other qualities) to interact with us in different ways:
Prejudices and stereotypical knowledge present an intriguing aspect of common sense. As human beings, we rely on (and possibly suffer from) stereotypical expectations. Obviously, we would not want to engineer an AI with its own prejudices and stereotypes, but on the other hand, if an AI system is to relate to humans, it should know about the stereotypical expectations as well—whether it is right or wrong, an AI should know that (we expect that) women like shopping and men like football. Without an explicit knowledge of the stereotypes, such beliefs can be implicit, hidden, and intermixed with other “objective” facts in a knowledge base.The authors, Herdağdelen and Baroni, analyzed a data set consisting of over ten million tweets broadcast from the U.S. in English over Twitter from November 2009 to February 2010. Cross-referencing each Twitter user’s first name with the database of male and female infants’ first names put out by the U.S. Social Security Administration, the authors isolated 5.2 million tweets belonging to men and 5.9 million tweets belonging to women. And they did find gender bias in certain phrases. For example, “[want to] make money” ranked numbers one and three for “masculine” phrases. On the “feminine” side, they found “go [to] bed” and “feel like.” The coolest thing about this research, though, is that the authors set up a nifty online widget – at www.TweetOLife.com – where you can put in any word or phrase you want, to see how it falls along gender lines.
It’s generally assumed that women in American culture distinguish among more color words than men do, possibly as a result of the myriad colors in clothing and makeup. Our parents and our friends likely train us to be aware of these subtleties. Let’s examine this using Tweet-O-Life:
Whereas “red” is basically 50/50, slightly more women than men used the word “maroon” and many more used the word “scarlet.” It’s not a perfect test, of course – those women may be talking about the Scarlet Letter or Scarlet O’Hara. The brilliance of this widget is that you can click over to “detailed query” and find that, while the men are tweeting about “scarlet” with “red,” “knight,” “fever,” and “sin,” the women are tweeting about it with “letter.”
How about language relating to children and childcare? Our “common sense” tells us that women still do the majority of child-rearing.
The term “infant” is the only one that more men say than women, and “toddler” is disproportionately said by women. Interestingly, whereas men used the word “toddler” with words like “autism,” “grandmother,” and “craft,” women used the word with “bed,” “nap,” and “scream.” The diversity of names for children may not be split too heavily along gender lines, but the words used with “toddler” suggest that women may be the primary (naptime?) caregivers.
What if we try something like “computer”? As with “red”, we get basically a 50/50 split between men and women. The really interesting differences come in the detailed query:
Men talk about computers as if they’re actively engaging with them or at least bragging about them. They tweet about their processors, the gigabytes they have, the operating system they’re running, and how skillfully they can manipulate them. Women, on the other hand, talk about computers just as often, but mention their aesthetic appeal and express defeat at technology they can’t control. Even a seemingly gender-neutral word like “computer” is not, as men and women describe it differently.
I tasked some of my friends with finding the most “masculine” and “feminine” words possible. One tried “drywall” and “chainsaws” (per The Oatmeal ) to no avail. After some trial and error, this is what we came up with:
Only 10% of the tweeted mentions of the operating system Linux were by women, and only 11% of the tweets that mention Justin Bieber were written by men. (If you can find a word that is more gender-biased, please leave it in the comments!)
With such strongly gendered words in evidence, how could NPR ask on June 23 if we’re nearing “The End of Gender?” The article ends with a quote from neuroscientist Lise Eliot, who suggests that:
if parents did not buy into the gender stereotyping of children's toys and clothes, kids would stay open-minded longer during childhood. The goal is to keep girls physically active, curious and assertive, and boys sensitive, verbal and studious.Pink and blue onesies are problematic, as are marketing campaigns aimed at encouraging a strict gender division in kids’ toys – as one of my favorite blogs, Sociological Images, is often pointing out -- but refusing to buy gendered toys and clothes only goes so far when we still call our girls “pretty” and our boys “strong,” or praise our boys for not crying and our girls for being quiet.
The way we speak is conditioned in our society by geographical area, education level, race, class, ethnicity, status, and gender. We’ve been trained since before birth to pick up on these differences, as they give us a world of information about the person we’re talking to. One of the salient conclusions of the Herdağdelen and Baroni study is that men and women on Twitter tend, by and large, to tweet about what we expect them to tweet about. They perform the gender roles we expect of them, and their language reveals that.
The question remains, why do men talk about Linux and women talk about Bieber? Is language informing our outlook on the world, or is culture informing our linguistic patterns? The answer is probably a bit of both.
A. Herdagdelen & M. Baroni (2011). Stereotypical gender actions can be extracted from web text. Journal of the American Society for Information Science and Technology.
Maltz, D. & R. Borker. 2007. A cultural approach to male-female miscommunication. In: A Cultural Approach to Interpersonal Communication, Essential Readings, L. Monaghan and J. Goodman, eds., Ch. 20, pp. 161-178. Blackwell Publishing.