Even though the vocal tract is the same regardless of a child’s sex before puberty, researchers have found time and again that adults can perceive gender differences in speech as early as four years of age. In this episode, we talk to Priscilla Fung, lead author of “The development of gendered speech in children: Insights from adult L1 and L2 perceptions,” which appeared in the January 2021 issue of JASA Express Letters. We talk about how her team’s research about gendered speech shows that gender differences may be apparent even earlier than previously studied, how acoustic measures play into those differences, and the role the first language of the child and the listener will affect those perceptions.
Priscilla Fung, Jessamyn Schertz, and Elizabeth K. Johnson. “The development of gendered speech in children: Insights from adult L1 and L2 perceptions.” JASA Express Letters 1, 014407 (2021); https://doi.org/10.1121/10.0003322.
Read more from JASA Express Letters.
Learn more about Acoustical Society of America Publications
Music: Min 2019 by minwbu from Pixabay. https://pixabay.com/?utm_source=link-attribution&utm_medium=referral&utm_campaign=music&utm_content=1022
Kat Setzer (KS)
Welcome to Across Acoustics, the official podcast of the Acoustical Society of America’s Publications office. On this podcast, we will highlight research from our four publications, The Journal of the Acoustical Society of America, also known as JASA, JASA Express Letters, Proceedings of Meetings on Acoustics, also known as POMA, and Acoustics Today. I'm your host, Kat Setzer, Editorial Associate for the ASA.
Today we are joined by Priscilla Fung of the University of Toronto. We'll be discussing her article, “The development of gendered speech in children: Insights from adult L1 and L2 perceptions,” which appeared in the January 2021 issue of JASA Express Letters. Thanks for taking the time to speak with us today, Priscilla. How are you?
Priscilla Fung (PF)
I’m doing well. Thanks for having me.
So first, tell us a bit about your background.
So my name is Priscilla, and I’m a graduate student at the University of Toronto. I work with Dr. Johnson from the Department of Psychology on speech development and language acquisition in children. And as you can imagine, being in the speech and language lab, we collected recordings from children for different kinds of studies. And one thing we noticed from these speech recordings was how hard it is to tell apart children’s gender from their voices alone. So I’m going to let you try. I’m gonna play you two voices. One is from a boy and one is from a girl. See if you can tell them apart.
[first sound recording of a child saying “turtle” and “bunny”]
So I’m gonna play the second child.
[second sound recording of a child saying “turtle” and “bunny”]
Can you tell which one is a boy and which one is a girl?
It’s funny. We did this in the rehearsal, and I still can’t tell them apart, even though I should know.
So the answer is the first one is a boy and the second one is a girl. So I think we can all agree now this is quite a difficult task. We found this really interesting. So we started a project with Dr. Schertz from the Department of Language Studies to look at when boys and girls start to sound different from one another.
Okay, so what was known about gender and children’s speech prior to your research? In particular, how is it that children’s speech can sound gendered long before they reach puberty?
The truth is, we still know very little about gender and children’s speech, which is one of the reasons the study I’m going to talk to you about today is so fascinating. When you think of adult speech, if you’re asked to identify gender from adults’ voices, it’s a relatively easy task. For example, men typically sound lower and deeper than women. And one of the reasons is because during puberty, males would develop this longer and thicker vocal cord than females, which results in a lower-pitched voice. But as you can tell from the recordings I played just now, identifying gender from young children’s voice is a much difficult task. But interestingly, previous study has shown that adult listeners can actually tell apart gender from children’s voices as young as four years of age, which is long before puberty, when sex differences in vocal tract anatomy emerge. And the most plausible explanation as to why young children speech can sound gendered even before puberty is that children learn these gender differences in speaking style through socialization from their own language community.
That is so, so cool. So what were your goals with this study? What did you hypothesize?
So one of the goals of our study is to look at how early gendered speech patterns develop in children. In other words, at what age listeners can start to tell apart boys and girls from their speech. We also wanted to look at whether listeners’ sensitivity to gendered pattern in speech is dependent on their own language experience. So we thought listeners who share a similar language background to the children would be better at classifying gender in these children than those who had a different language background. This is because children presumably learn these gendered speech differences from their own language community. And because these differences can vary across languages and dialects, so people who share the same language background as the children may be better at identifying the gendered speech patterns.
Okay, so it kind of ties into their culture, essentially.
Okay. How do gender differences and speech vary across languages or dialects?
Well, some speech differences between male and female are generally consistent across languages. And some of these differences are related to the systematic differences in males’ and females’ physiology. For example, another reason for why males tend to have lower pitch than females on average is because males are also larger than females on average. So they will have this lower natural frequency range. But because we have evidence that gender can be classified from children at four years old, this means that gender differences in speech cannot be completely accounted for by physiology alone, but must in part be learned. And for this reason, these gender differences are not always the same across languages. And as I mentioned earlier, men generally have lower pitch than women. But these differences are much bigger, for example, in Japanese speakers than in Dutch speakers. And pitch isn’t the only thing that differs in men and women either. There is some evidence that the pronunciation of certain consonant sounds like the s sound, as well as voice quality, like the breathiness of the voice, can also cue gender differences. But which cues are used and how much they’re used can differ based on the different languages or dialects.
That is so interesting. So tell us about your experimental setup. How did you test the age when gendered speech becomes apparent?
So what we did was we asked adults to listen to recordings of children saying various words. And after hearing each word, they had to decide whether they were listening to a boy or a girl. So just like what you did in the beginning. And the recordings were from children at three different ages: at two-and-a-half, four, and six years old. And because previous work had found that children's gender was identifiable by four years old, we expected that our listeners would be able to accurately classify the recordings from the four- and six-year-olds. But what we're especially interested to see is if they can do so even earlier with the two-and-a-half-year-olds.
How did your experiment allow you to determine the effect of listeners’ language experience on perceptions of a child's gender?
So the children we recorded were all native Canadian English speakers, meaning they learned English as their first language here in Canada. And to test the effect of listeners’ language experience on their ability to tell apart child's gender, we asked both native Canadian English adult speakers and non-native English speakers to classify children's gender. So the native speakers are those who learn English early in life, and non-native speakers are those who learn English later in life. And we expected that the native English speakers would be better than the non-native English speakers at this task, because they shared a similar language background as the children, and therefore they might be more familiar with the language-specific characteristics of gendered speech that the children might be developing.
Okay, yeah, that makes sense. So what did you end up learning about the age at which a child's gender can be perceived in speech?
Well, we found that some children as young as two-and-a-half years old already speak in a style that is identifiable with their gender, which is earlier than what was previously found. And this finding is fascinating, because at this age, there are no anatomical differences in the vocal tracts of boys and girls. This means that children are already picking up gender-typical speech patterns at this early age.
That is crazy. So what did you find out about the effects of a listeners’ language experience on their perception of a child's gender based on speech signals?
So interestingly, the ability for adults to detect gender from children’s speech is dependent on the listeners’ language background, that is the native English listeners were better than the non-native listeners at identifying children's gender based on how they speak. So this supports the hypothesis that certain characteristics of gendered speech in children are learned, and may be specific to their own language community.
Okay, that makes sense. It's-- they're familiar with the culture, and so they can pick up on it a little bit more.
Okay. So you did some analysis of various acoustic measures of the children's speech samples to see if any of those measures could predict the listeners’ response. How did you do that? And what did you find out?
So in order for us to test how the actual acoustics of how boys’ and girls’ speech might differ, we measure the frequency properties of the stimuli that we played for the listeners. We took two acoustic measures that are known to show strong and reliable gender differences in adults. So first, we measured the fundamental frequency, which corresponds to the pitch of the voice. We also measured the formants, the resonant frequencies of the vocal tract, which characterize different vowel sounds. In adults, male shows systematically lower values for both of these measures. And what we want to see was to what extent these gender-specific patterns also held in children. If it turns out that the same patterns occur, this would be an explanation for why listeners were able to classify gender at above chance accuracy. But surprisingly, we actually found little difference between boys and girls in terms of these acoustic measures. There are slight differences in the expected directions, but only for the oldest age group. And the differences are not very reliable. This makes the listeners’ performance even more impressive and intriguing because they must be using other sorts of information to figure out the gender of the child that they're hearing.
Yeah, it's not the physiological effects then. R
Okay. Right. So given the weak of relationship between your acoustic measures and child gender, how did listeners perform above chance in the gender classification task?
Well, we think it's likely that listeners are using other cues, the ones that we did not measure, to make their classifications. We tried to measure the acoustic properties that we thought would be the most indicative of gender differences. But speech does vary in many different ways. And it's not possible to measure every type of difference. For instance, there may be other more holistic properties of voice quality that our analysis didn't capture, and uncovering what these other cues might be and how listeners use them is a question we would love to explore more in our future work.
So on that note, what do you see the next steps in research regarding the develop of gendered speech in children?
So currently, we actually do have couple of follow-up projects going on. We're interested to know how children learn these gendered speech patterns. In particular, we focus on the influence of parents, we asked how gender speech might be related to parents’ attitudes towards gender, as well as gendered behaviors in their children. We also have another project not only looking at children's production, but also children's ability to detect gender from their peers’ speech. And because the studies we've done so far have been with cisgender children, another project we are currently working on right now is to look at whether gender-diverse children also adopt these gendered speech patterns. And we're collaborating this project with Dr. Vanderlaan, who does cross-cultural work on gender development. And we are currently planning some fieldwork in Thailand to investigate how gendered speech development differs in cisgender and gender-diverse children.
These are al so fun! I can't wait to hear how these all these different studies turned out. Thank you so much for taking the time to speak with us today. I will now spend the next few days wondering if my son's voice makes him sound like a boy or not. And like over listen to it. Have a great day.
Thank you. It's my pleasure to be here. Thanks again for having me.
Thank you for tuning into Across Acoustics. If you'd like to hear more interviews from our authors about their research, click subscribe and find us on your preferred podcast platform.