Power rarely manifests as anything so obvious as a barked order, or a smirk that grows nonchalantly into a maniacal laugh. More often than not, though, power is evident in a smattering of pronouns.
Five recent studies led by Ewa Kacewicz and James Pennebaker found that during conversations, high-status individuals use fewer of the first-person pronouns I, my, and me—and more personal pronouns like you, we, and us—than lower-status individuals. This proved to be the case whether these conversations took place in person or in print, and whether status was determined organically (as in one study, which analyzed actual email exchanges between, say, a student and her professor) or whether status was manipulated by the researchers, who randomly assigned one member of a group to serve as its leader.
Why are people in power less likely to use first-person pronouns? Well, quite simply, because they are less likely to talk about themselves. The reasoning goes something like this: having a higher status than the people you’re chatting with allows you to take your own position for granted. You don’t have to waste as much mental effort second-guessing yourself, or preventing a misstep, which frees you up to focus on other people. (This is not the same thing, it should be noted, as acting in their best interest.)
In his book The Secret Lives of Pronouns, Pennebaker argues that a wide variety of social categories traditionally associated with power are also associated with a distinctive pattern of pronoun use. The studies he describes are particularly interesting because there is nothing about participants’ explicit roles (as a student or a professor, say) that can explain the patterns. For instance, he finds that students from higher socioeconomic backgrounds use fewer first-person pronouns in their college admissions essays than students from less prestigious backgrounds. Older people use fewer first-person pronouns than younger people, and men use fewer than women. Pronoun use within individuals receiving testosterone injections even changes predictably as levels of the masculine hormone fluctuate with time.
Plenty of things besides pronouns can indicate who holds the power. Excessive politeness signifies lower status, as do hedges like “so okay, here goes …” There are nonverbal tells too, in posture and tone. But analyzing pronouns is really, really effective—at least for the computers that do the tallying.
What makes this research so different from most studies on language and power is that it doesn’t deal with what sociolinguists call “variables”—different forms of language that all refer to more or less the same things in the world. When I say ask but you say ax, these are variables. Ditto when I say yes in formal situations but yup in informal ones. If two forms share the same referential meaning, any differences in how they are perceived—namely, the social categories and situations they’re associated with—can be attributed to social meaning.
But of course you and I don’t share referential meaning. These pronouns correspond to two entirely different people, and you have no choice but to use I if you want to say something about yourself. In some sense, then, counting pronouns doesn’t tap into how we communicate so much as what information we are compelled to share.
This makes it tricky to pinpoint what, if any, social information these pronouns actually convey to others. When I hear my neighbor say, I think I’m going to buy myself a snow-blower because I need one, does his social status momentarily drop—my feelings about purchasing snow-blowers aside?
Certainly not consciously. Our intuition about function words and what they signify is pretty terrible. “When I give talks, I routinely ask the audience about who uses pronouns, articles, etc. at higher rates—men or women,” Pennebaker told me in an email. “Most people get the majority wrong. We have these set theories such as ‘men are more arrogant and self-confident than women and use I-words at higher rates than women.’ Well, men MAY be more arrogant but arrogance is associated with fewer I-words.”
Still, people may well process these pronouns at an unconscious level, and use them to accurately perceive others, Pennebaker tells me. In one study, he analyzed 100,000 blog posts written by nearly 20,0000 bloggers. Participants were asked to guess: Was a given post written by a male or a female? Their accuracy generally fell between 55-65 percent, which is better than chance.
But in that same study, a computer that simply tallied the relative frequencies of different function words matched 72 percent of bloggers to the correct gender. At the very least we’re not as sensitive to the social information conveyed by pronouns as we could be. An open question: Why not?
Psycho Babble will be on hiatus until January.
Permission required for reprinting, reproducing, or other uses.