When Elizabeth Warren announced her creation of an exploratory committee to consider a 2020 presidential run in late December, echoes of 2016—and 2008—reverberated in the news: “Is she likable enough?” Even as I write this from a café in my small town on the north shore of Massachusetts, the (women) baristas are abuzz over a Globe piece on our home-state U.S. senator: “I mean, you’re drinking that beer looking like you’ve never drunk a beer in your life!” Another chimes in: “I love her as our senator, but stay here! Don’t do it! You’ll get eaten alive!”

With a record number of women entering Congress this year and major efforts taking shape to unseat a president infamous for misogynist language, the hurdles that women in politics face have taken center stage in the media. With analysts weighing in on who could defeat Donald Trump in 2020, and multiple women candidates in the mix, it is high time to reexamine the double bind of language and gender. First and foremost, political buzzwords like “likability” and “authenticity” that abound in the polls need to be deconstructed: these concepts are actually just amalgamations of impressions that people make primarily on the basis of how politicians speak, and how they convey their identities as politicians, and necessarily, also as women.

Linguistic research has demonstrated that our impressions of people’s language are deeply influenced by other assumptions—and prejudices—related to factors like race, age and gender. For instance, the linguists H. Samy Alim and Geneva Smitherman have shown that descriptions of President Obama as “articulate,” in one case infamously uttered by Joe Biden, are derived from underlying racist assumptions about the language of black men in America. Assumptions about gender influence our perceptions of language in a similar way. The conventional language of power, when it comes out of a woman’s mouth, is often described as sounding “bossy,” “shrill” or “bitchy.” These impressions are so ingrained, in the minds of both women and men, because we are socialized into them from a very young age. It’s an integral part of what sociologists call the “gender order.”

The gender order affects both our own behavior and our impressions of others’ behavior, which includes speech. For instance, research has shown that young boys and girls differentiate their vocal pitches, with girls speaking at the higher end of their vocal range than boys, starting around preschool. This is long before the onset of puberty, which creates a physiological reason for pitch differences between the sexes.

On the other hand, our impressions of the speech of children—even preverbal infants—is equally informed by gender norms. One experiment showed that when adults viewed a video of a crying baby, they were likely to interpret the cry as expressing fear if they believed the baby was a girl. They interpreted the same cry as one of anger, however, if they were told the baby was a boy.

I’ve informally replicated this study in a course I teach on language and gender, with an eye toward how these gender-differentiated impressions carry over into the political sphere. In this exercise, I show some students a clip of a male politician making a speech at a slow pace that includes many pauses. I ask students to jot down their impressions of the speaker. They respond with qualities like “thoughtful,” “emphatic” and “serious.”

 I show other students a clip of a female politician speaking in a similar way. While some impressions overlap with the assessment of the male, other descriptors of the female include “hesitant,” “unsure” and “unprepared.” While it’s true that slow speech can indicate that the speaker is either thoughtful or unsure, it is our socialization into the gender order that makes us more likely to interpret a woman’s speech as insecure.

Research has shown that even nonverbal communicative devices like laughter are interpreted through a gender lens, as Tanya Romaniuk’s study of 2008 presidential candidates illustrates. While men’s laughter goes unremarked in the media, Romaniuk documents how Hillary Clinton’s laughter gets replayed, scrutinized and recontextualized as a “cackle.” The obvious connotation of this verbal choice is not haphazard: while a laughing man projects likeability, a laughing woman conjures up the image of a witch.

The language of emotion is especially important when people are concerned with the “authenticity” of a politician, and we have witnessed differential treatment of emotional politicians in recent years. When President Obama teared up as he introduced new gun laws in 2016, his emotional display was widely interpreted as an impassioned plea. But when Hillary Clinton teared up during the 2008 New Hampshire primary, she was described as finally displaying her “softer” side. And while many assessed this moment as a positive for the candidate, since it presented a more “human” side to her campaign, others wondered whether a woman who cries on the campaign trail could be tough enough to handle the job of commander-in-chief.

But even the positive reactions to Clinton’s emotional display in New Hampshire are troublesome: Why does a woman need to demonstrate that she’s human when running for office? Herein lies the double bind of language and gender. The concept of the double bind, originating in the work of the anthropologist Gregory Bateson, was first applied to language and gender by the linguist Robin Lakoff in her pioneering essays published in the 1975 book Language and Woman’s Place.

Expressed in shorthand as “damned if she does, damned if she doesn’t,” the double bind is a two-part rule that is internally contradictory. When applied to women and language, it goes like this: In order for a woman to be valued in a position of authority, (1) she must talk in a style that is associated with authority; and (2) she must talk in a style that is associated with femininity. The internal contradiction is that authoritative styles of speech, at least in Western society, are based on a long history of only men holding positions of authority.

So, when the “guy you want to have a beer with” factor comes into play in choosing a candidate that appears most “presidential,” is it any wonder that Warren backfired so miserably when she appeared drinking a Michelob Ultra on social media? The quip of one communications professor who deemed her more of a “chardonnay senator” illustrates just how damaging the gender order can be for woman in politics. In this case, a woman tried to navigate the double bind by appealing to the practices of men (beer drinking) but was rebuffed and relegated to the feminine, buttery world of white wine.

So, if a woman can’t drink a beer to achieve presidential authenticity, and she risks presidential credibility by displaying her feminine side, then how can she overcome the hurdle of the double bind? One thing we know for certain is that an individual cannot transform the gender order alone. However, we do have some evidence that as women enter the highest ranks of their professional fields, they are finding ways to creatively use the gender order to their advantage. Judith Baxter, who has studied the language of business executives in the U.K., has found that women leaders are subverting old stereotypes of women in power (“role traps,” as the management scholar Rosabeth Moss Kanter called them), like the “Iron Maiden” or “Mother” figure, and using them to powerful ends by leveraging aspects of these identities that intersect with transformational leadership identities.

My own research on the language of presidential debates has found that family plays an important role in projecting political leadership as well. In primary debates, men frequently introduce themselves as good fathers and grandfathers—in other words, as competent executives in the social unit of the family. Given the nation-as-family metaphor that dominates political talk in the United States (we have “Founding Fathers,” “homeland security,” etc.), showcasing one’s fatherhood proves an effective strategy to convey presidentialness.

It remains to be seen whether women can leverage motherhood in a similar way. Until now, in the spheres of both business and politics, motherhood has been seen as more of a liability than an asset. But as the presence of women in politics becomes less marked, and motherhood slowly becomes visible in the political sphere, women are carving out new paths so that rather than getting trapped in the double bind, the new generation of women leaders can sidestep the double bind altogether and slowly transform the gender order and the language of leadership.