American Scientist

If your computer, smartphone, smartspeaker, or GPS ever talks to you, does it do so in a male or female voice? It's a computer, so how do you know whether the voice is male or female?

Was that the "default" voice of the gadget, or did you change it? If you changed it, do you know why you did?

In the United States, computer programmers typically set (or create) the default voice for talking gadgets to be perceived as female. In other countries, it differs. In Germany, for example, the car manufacturer BMW famously had to issue a recall for making the default voice female, even after explaining to complaining customers that the voice of the built-in GPS system was not a real female voice.

Of course, we have cultural beliefs about what people should (or do) sound like based on their sex, gender, and sexual orientation. And technology both enables us to test our stereotypes or reinforce them. Indeed, would you prefer a helpful voice—that of a digital assistant—to sound male or female? What about a voice giving directions, such as that of a GPS system? If you have access to the technology, I encourage you to try it out.

Research on communication has shown that just digitally altering a voice's pitch—the perceived "highness" or "lowness" of a voice—could be enough to influence our choice of political leaders. That influence occurs whether the candidate's voice was perceived to be male or female, and the results didn't depend on the gender of the research subjects (those judging the digitally altered voices). So it's not overall pitch that seems to matter or the gender of the listener. Something far more complicated is going on.

The researchers in this podcast have been studying the many factors and nuances that make communication gendered, and their findings include the ways science does not support the stereotypes. We all change our voices depending on context, and there are interventions—particularly voice training—that can allow people to change how they sound. Indeed, it can take many months of coaching, but as you'll hear, it is certainly possible for someone who sounds male to sound female even without surgical or hormonal interventions.

A full transcript of the podcast is below.

[music]

Robert Frederick
We all pick up cues about sex, gender, and sexual orientation from speech, whether from a smartphone giving directions,
[clip from a smartphone giving directions]
a fictional character in a movie
[clip from The Revenge of the Nerds movie]
or a robocall
[clip from research on voice pitch]
Yes, there are stereotypes.

Adrienne Hancock
Stereotypes do exist, but science doesn’t support them, and so it’s really time for those stereotypes to die.

Robert Frederick
On this episode of the American Scientist podcast, gendered communication. I’m Robert Frederick.

[music ends]

As you might expect, there are cultural beliefs about what people should or do sound like based on their sex, gender, and sexual orientation.

Adrienne Hancock
So some people and some cultures are more amenable to that sort of gender fluidity and that ambiguity, and accepting of variation.

Robert Frederick
Adrienne Hancock studies voice, including transgender communication at George Washington University.

Adrienne Hancock
For instance, African American women can have a lower pitch and still be perceived as women—or people use the she/her pronouns—even though their pitch is not within the norms of white cis-gender women.

Robert Frederick
But pitch is only one of a whole host of characteristics that convey gender, sex, and sexual orientation. There’s also articulation...

Adrienne Hancock
Sometimes instead of "articulation" we could use the word enunciation, pronunciation. Intonation is how pitch changes across the sentence. I also mentioned volume, and rate. So if we talk really fast, that’s a rate issue.

Robert Frederick
But there’s even more nuance to it. As example, Hancock shares audio clips of a before-and-after voice training of a fifteen-year-old who wanted to sound more female. In the two clips, before and after voice training, the teenager is describing the same picture. Here’s a bit of what she sounded like before training.

Teenager
The water is splashing all over the place and the window is open. Looks like maybe there’s a breeze coming through the window, and that’s it.

Adrienne Hancock
And this is after several months, about nine months, I believe, of training and working on it.

Teenager
Margaret has just made a delicious dinner, but it’s all vegetables and her two children don’t like it at all. So after dinner her two children go up to the cookie jar and steal the cookies behind her back while she’s washing the dishes. And while she’s washing the dishes she looks over and tries to see what they’re doing, but then doesn’t pay attention, and all the water spills out onto the floor.

Adrienne Hancock
Okay, so you can see see just in that brief picture description--she's describing the same picture both times—Obviously pitch changed, resonance, or tone quality of voice changed, and those are definitely the most salient gender markers. And that’s what we primarily work on changing, pitch and resonance..., or the forward tone. So this is a very throaty resonance, there's a lot of tension back there. This is a very nasally resonance, or sound.... And then this is a very forward resonance, where I'm almost over-articulating, I'm really using the entire vocal tract to resonate and deepen the sound.

Robert Frederick
But there’s no recipe for sounding female, no recipe for sounding male, or heterosexual, or homosexual, or bisexual, or transgender, or intersexual, or asexual...

Adrienne Hancock
Stereotypes do exist, but science doesn’t support them, and so it’s really time for those stereotypes to die.

Nic Palomares
Our stereotypes are mostly inaccurate. Sometimes they’re true—or sometimes they manifest, I guess I should say—but most of the time they don’t.

Robert Frederick
Nic Palomares researches what causes gender-based language differences. He’s from the University of California, Davis.

Nic Palomares
Whether that be intonation, whether that be sound of “s,” whether that be the level of tentativeness, whether that be how many apologies you get—like, all sorts of behaviors, even like non-verbal behaviors, but like where do I sit in a table... who gets the head of the table? That is highly contextually dependent.

Robert Frederick
And Palomares says that highly contextually dependent context is why most artificial intelligence voices are female,
[clip from a smartphone giving directions]
at least those in the United States.

Nic Palomares
You can change it, but the default is usually—at least in the United States—a feminine-sounding voice. And that I think has a lot to do with the fact that there’s this interaction between sex of the receiver and sex of the sender. And we anthropomorphize everything... even if you knowingly realize that it’s completely non-sentient, right?

Robert Frederick
And as you might expect, these interactions between the sex of the receiver and the sex of the sender begin as children, when boys' and girls' voices are both still higher pitched.

Benjamin Munson
Early in life, boys and girls display speech that can be very different from one another.

Robert Frederick
That’s Benjamin Munson, who studies speech perception and production in children at the University of Minnesota, where he also trains speech and language pathologists and therapists.

Benjamin Munson
And children really make a sophisticated sort of social parse—let’s use a fancy term here—social parse of speakers that they encounter during language acquisition, which is a way of saying, you know, they learn preferentially from people who are truthful, people who are honest, people who are consistent, and we know that in things, in terms of things like learning word labels.

Robert Frederick
Word labels—so which words are associated with which physical objects.

Benjamin Munson
What we don’t know is whether or not those same cognitive, social-cognitive mechanisms underlie children’s learning of different speech variants. So... a woman asked me, you know, if she had been raised by a group of men, would she speak like a man... Probably not, because men speak different ways in different contexts. And what the child would likely do is listen to the men around her and identify different aspects of their speech that correlated with different, sort of transient states that they had.

Robert Frederick
So being excited, you might talk with a speech style that is excited. And if the child is an excited person, then she’ll use that speech style. So even if this hypothetical girl were raised by a group of men, Munson says,

Benjamin Munson
Given that the men are going to be producing diverse speech styles to convey their diverse identities in different circumstances, she still could end up with something... perceived as a female speech style because there’s nothing about a female speech style that has to do with the biology of being female.... The differences are differences on average that relate as much to if not more to the social circumstances that we find ourselves in than to the biology that we are endowed with.

Robert Frederick
All of which complicate transgender people’s efforts to learn a different speech style that matches their gender identity. Because even though testosterone lowers voice pitch and there are surgical options to raise pitch, Adrienne Hancock at George Washington University says outcomes are better when speech therapy is paired with hormone therapy or surgery.

Adrienne Hancock
In America, most insurances do not cover gender-related services. In other countries they do. So Sweden, for instance, everyone who goes to a gender clinic gets, first of all to go, and secondly it includes a voice and communication evaluation. Now back to America, it highly depends on your socioeconomic means.

Robert Frederick
Barriers also include not even knowing that such vocal training exists or how it could help—however a person wants to sound or for whatever social context.

[music]

You’ve been listening to a podcast from American Scientist magazine, published by Sigma Xi, the Scientific Research Honor Society. I’m Robert Frederick. Thank you for joining us!

[music ends]