Berkeley Talks transcript: How we learn language across communities and cultures

Listen to Berkeley Talks episode #149: How we learn language across communities and cultures.

[Music: “Silver Lanyard” by Blue Dot Sessions]

Intro: This is Berkeley Talks, a Berkeley News podcast from the Office of Communications and Public Affairs that features lectures and conversations at UC Berkeley. You can subscribe on Spotify, Apple Podcasts or wherever you listen. New episodes come out every other Friday.

[Music fades]

Dione Rossiter: Hi, everyone. I always say I’m going to do a little bit of mask removal just to do my intro. It’s so good to see you back. How many of you were at our last lecture? Yeah, it’s nice to be. Oh, hi Tim. I want to thank you all once again for being here. I was noticing in the last talk that I never introduced myself. My name is Dione Rossiter and I’m the executive director of Science at Cal. So, if you want to talk to me about our programs or what we do, I’ll teach you a little bit right now, but we’ll also just go ahead and get started.

The first thing I want to do always is start with a land acknowledgement. We recognize that Berkeley sits on the xučyun (Huichin), the ancestral and unceded land of the Ohlone people, the successors of the historic and sovereign Verona Band of Alameda County. This land was and continues to be of great importance to the Ohlone Tribe and the familial descendants of the Verona Band.

Every member of the Berkeley community has and continues to benefit of the use and occupation of this land since the institution’s founding in 1868. By offering this land acknowledgement, the Berkeley community not only recognizes the history of the land in which we stand. But we also recognize that the Ohlone people are alive and flourishing members of the Berkeley and broader barrier communities today. Thank you for allowing me to make that land acknowledgement.

Of course, at Science at Cal we celebrate science, we bring the wonder and excitement of UC Berkeley research to the public. All of our events and programs are free and geared towards public audiences. We’re here, we’re in our community, we’re at farmer’s markets, we’re at fairs. You’re going to learn more because I’m going to give you a whole mess of save the dates right now and a lot of folks who were here last time were asking if we have any community and family friendly events coming up and last time we could say, yeah but we’re not really sure about them. We have a slew that we have ready to share with you today.

We partner with folks all across the East Bay. If you have an organization and you want to partner with us or bring Cal scientists into your organization or spaces please let us know we’re happy to work with you. I want to thank Berkeley Community Media. Thank you to the folks who are filming today. They go ahead and put it online and we’ll get it out there in about less than a week or a week or so and then once it’s you can share it with your friends and family. We’re super appreciative of them.

I want to thank the Science at Cal Advisory Council, folks who have been working with us for a number of years to make sure that we can be here with you. We are funded right now through the Executive Vice Chancellor and Provost so of course we want to thank them. We want to thank the Lawrence Hall of Science who manages and houses Science at Cal and lastly of course we want to thank our donors.

Folks who donated today or anyone who’s donated in the past, we just had Big Give last week so I hope that you donated. If you’re on our list you should have got a request, but we believe that science should be free and also accessible. Therefore, that’s why exactly why all of our events are open to the public but we can’t do that without you. So if you’d like to donate we’d appreciate it. If you’d like to talk to me about other ways that we could work together to get some funding into Science at Cal. So we can be here for the long run, right? We only have a certain number of years of funding to bring science to you. So if you have some ideas or want to talk more about how to do that please let me know. Here are those dates of our upcoming events, our next lecture which is in exactly one month, our next online lecture which is again virtual.

This is the collaboration with Lawrence Berkeley National Lab. This is in a week or two and then also this is a save the date on April 21st we have the Recyclable Plastics. We are back with our grad student series. That took a little bit of a break because unfortunately our cafe that we were using closed down.

But now we’re working with MudLab, which is near Lake Merritt and so we’ll have two grad students for you coming up on April 6th and this is going to be Earth Day themed as well, too. So we’re super excited, we hope to see you in that cafe. Again, it’s our first time back with our grad students and in a little bit. Starting our family friendly events will be at the Innovation Fair next in exactly one week. So if you want to have friends and families we’ll have our scientists out there at the Quest 4th Annual Innovation Fair. We have Bay Area Science Festival event programming. We’re going to be at Oracle Park in San Francisco with our scientist.

We’re doing an online Spanish language program with some folks talking about careers and what it’s like to be a scientist and then we’re also going to be at the Lawrence Hall of Science. We’re going to have a huge festival on their plaza, if you haven’t been please join us. We’re going to be on our plaza with about 15 UC Berkeley groups and also a bunch of activities from the Lawrence Hall Science. So join us, it’s free outside. Join us and reduce admission inside. So we can’t wait to see you there and I am going to hand things over to Steve. You guys remember Steve Croft is here and he would like to say a few words and we’re so excited to have him.

Steven Croft: Thank you. Yes, I see a couple of familiar faces here. I’ll take my mask off as well. See a couple of familiar faces from before when we were interrupted by the global pandemic. Back in 2009… I’m a researcher in the astronomy department here at UC Berkeley and the International Year of Astronomy was coming up in 2009.

Which celebrated the 400th anniversary of the first use of the telescope to look at the sky and it also coincided with the year of science. Which I think was the 150th anniversary of origin of species. There were some other kind of important anniversaries that year and I sent an email around the astronomy department saying I’m sure there are many things planned for the Year of Astronomy and I’d just like to get involved with this and crickets. There was nothing planned and so I thought, well I better plan something. So I decided to plan a monthly lecture series and then at the end of 2009 I figured I’d run out of astronomers. So, I started asking other people around campus and that’s what became the Science at Cal lecture series that you’re at today.

I managed to do this every month except for April and when we coincided with Cal Day, but every month back since 2009 and then again the last one was in March 2020. I was unable to make the lecture last month that we had funnily enough on astronomy. But I was out of town and a lot has changed in that time during the pandemic and I just really wanted to say. I mean, I know not all of the regulars are here that have been here over that long period of time. I know there are certainly some regulars here who have been with us pretty much since the beginning and I just wanted to say thank you for being here. I’m really glad to be back. I’m really glad to see Dee and the new role as executive director, glad to see Elise here as well and wanted to thank them for really kind of rebooting this lecture series and taking over the running of it.

Which also it became not too much effort for me, but I got a little too busy and then Di has really been coming out of the gates and revving up the engines and this beautiful signage and tables and things that we didn’t have when I was doing this. So just before I bow out, really wanted to say thank you, thank you to all of you. Thank you to Dee and I hope that this lecture series carries on for at least another decade and no more pandemic interruptions in the meantime. Thanks.

Dione Rossiter: We happened to survive through the pandemic so that’s nice, and Steve was volunteering his time. So, we’re incredibly grateful that he started this lecture series and that he continued to be so helpful throughout the years and we’re hoping to get him even more involved with how Science at Cal moves forward. Thank you for allowing us to do those introductions, but you’re all here not to hear me talk. You’re here to meet Mahesh Srinivasan.

He’s an associate professor of psychology and a member of the cognitive science faculty at UC Berkeley. He received his doctorate in developmental psychology in 2011 from Harvard University. On campus, Srinivasan directs the UC Berkeley Language and Cognitive Development Laboratory, which explores how linguistics, cognitive and social abilities arise and interact during human development and across different cultures. Dr. Srinivasan is also co-scientific director of the Psychology and Economics of Poverty Initiative at the Center of Effective Global Action. He has earned numerous awards including the Association for Psychological Sciences Rising Star Award and the World Economic Forum’s Young Scientist Distinction.

Dr. Srinivasan’s work has been published in journals including the Proceedings of the National Academy of Sciences, Cognition, Child Development and Developmental Science, and has been supported by funding agencies, including the National Science Foundation and the James S. McDonnell Foundation and the John Templeton Foundation. So without further ado, Mahesh Srinivasan.

Mahesh Srinivasan: Okay. Can everyone hear me okay? Super. Okay. Alrighty. I’m going to just come down here and… this is a picture of my daughter. It’s nice when you have an actual picture that you can use and not just a stock photo. I’m really pleased to be here today with you all and I’d love for this to be an interactive discussion, so feel free to just ask questions or anything and would love to chat.

I think really language learning is really one of the most amazing things that we do as human beings. Right? What I’m showing you here are examples of a child’s language and how it develops and flourishes across different ages. You can see here at 13 months, the child is really just producing just about one word, right? At 26 months, they’re starting to put multiple words together here, it’s a garbage truck, and then at 42 months, you see these more complex kinds of expressions where there are phrases embedded within the utterance.

Part of what’s so amazing about this is that children are rapidly learning their first languages seemingly better than adults can learn a language and they’re also seeming to learn these languages without any kind of direct instruction. But one important question is what role do parents actually play in this process? So, what’s the role that a parent’s speech to their child actually plays.

I’ve highlighted here for you the different things that the mother is saying to the child across these different ages. These are all publicly accessible corpora of child language and parent-child conversations. You can see in this video here, that the mother is attending to what the child is attending to and providing sort of labels for what the child is looking at. In this case here you can see kind of an explicit sort of question that the mother is asking.

So, “What kind of truck is it?” And then the child says, “It’s a garbage truck.” The mother says, “Yeah.” So, there’s this kind of explicit reinforcement. One question, basically, is how important to the language development process are these kinds of interactions between parents and kids.

I’m going to present a video to you here of a nationwide intervention, kind of a public health intervention. Let’s just watch this video that obviously… it sort of conveys this message that talking with infants and toddlers matters quite a bit. Let’s watch this and then we’ll talk about this a bit more.

(Video clip)

Speaker 1: Say more, say more.

Speaker 2: More.

Speaker 1: All right.

Speaker 3: Michael is ready to experience yet another day filled with love, laughter and language.

Speaker 2: What do you think you’re going to do today buddy? Maybe you’re going to read some books?

Speaker 3: He is surrounded by loving adults who talk with him all day long.

Speaker 4: Michael, are you ready to begin classroom this morning?

Speaker 3: Besides providing the food he needs to grow healthy and strong, his mother, Kyla, and his teacher, Avasia, are providing language nutrition, lots of loving words to fuel his brain so his learning will take off. They know that engaging infants and toddlers in conversation provides a solid foundation from language, communication and reading.

Speaker 4: You want to go over to the other side?

Speaker 3: Avasia is much more than Michael’s teacher, she has partnered with Kyla as her Talk With Me Baby coach to support her in being Michael’s conversational partner. It hasn’t taken months; it’s happening in daily coaching moments where Avasia models talking with Michael, shares an educational message and encourages and supports Kyla’s practice. Their words, smiles and gestures are giving Michael a great start.

Speaker 4: Ready, set, go.

Speaker 3: Every word he hears now from those special adults in his life is making a difference in his future. Building his brain and setting him on a path to language, learning and a world of opportunities.

Speaker 4: Ready? Can you turn the page?

Mahesh Srinivasan: Able to hear that okay? The volume was high enough? Great. So, there’s kind of two broad claims that were embedded within that video. Right? The first was that there really are optimal methods for supporting a child’s language development. There was this metaphor that was illustrated in that video that speaking to one’s child is sort of akin to nutrition for their brain developments, for their cognitive development.

Then further, there’s an additional message in the video that there might be some parents who are not quite aware of these best practices of interacting with one’s child and therefore need to be coached on how to do so. Okay?

So, what I want to do today is to consider each of these claims and what I’ll do first is to just start with the second claim here. Okay? So considering this issue — of whether it really is the case that some parents are just not aware of these methods and therefore need to be coached on how to talk to their kids. Are there any questions so far? Okay.

So, this really has received a lot of prominence. That intervention that was presented in that video, that type of approach was even brought up by then candidate Biden in the Democratic Primary Debates. He was asked a question about what can we do to solve the problem of child poverty and to really help children who are in poverty succeed better and here is his response. Let’s go ahead and watch this.

(Video clip)

Joe Biden: We bring social workers into homes with parents to help them deal with how to raise their children. It’s not that they don’t want to help, they don’t know quite what to do. Play the radio, make sure you have the record player on at night, the phone… Make sure the kids hear words. A kid coming from a very poor background will hear 4 million words fewer spoken by the time they get there. There’s so much we…

Mahesh Srinivasan: Flubbed, but there’s some important things to pick out from that message. Some parts of what Biden said sort of go against the science. Obviously the things about the most important thing to do here is to play the radio, make sure the television’s on, make sure you have the record player and then he said phonograph I think he was going for.

So, those are not things that scientists really say, even with the television it’s generally not thoughts that sort of simply playing kind of passive television is necessarily that beneficial for a child’s language development. But there are other aspects of what Biden said that are much closer to what scientists themselves say.

First of all, let me draw your attention to the bottom part here. A kid coming from a very poor school or a very poor background, what he means is backgrounds, is going to hear 4 million fewer words spoken to them by the time they get to school. This, as we will see, is something that has been shown — the sort of average differences in the number of words that a child from a richer background will hear compared to a child from a poorer background.

Then, the other aspect of this is this idea that parents don’t know quite what to do, right? They want to help but they don’t know quite what to do. So, the correct solution here is to send social workers into people’s homes and to encourage them to speak more to their kids. Okay?

What Biden is referencing here is this really landmark classic study that was conducted by Hart and Risley, the findings were [inaudible] the 30 million word gaps. So what Hart and Risley did is they went into families, homes of different social classes, upper-, middle- and lower-class. They visited these families for one hour each time and they just counted up the number of words that parents would speak to their kid as well as how they interacted and conversed with their kids in other ways.

They essentially estimated that by the time the child was 4 years old, a child from an upper class background would hear 30 million more words spoken to them than a child from a lower-class background. Then they also showed that children’s vocabulary growth showed a very similar trajectory, okay? So, if you measured children’s vocabulary by the time there were 3, by the time there were 4, you also saw a very similar difference, a discrepancy between children from the upper-class backgrounds and children from the lower-class backgrounds. So this is… Yeah, please. There’s a question.

Audience 1: I’m assuming here that’s 30 million total words, not unique words?

Mahesh Srinivasan: Yes. So, the question is 30 million total words? Yes. So there’s a distinction here between word types and word tokens. By word types we mean, imagine I say the word “cat” five times, that would be one word type but it would be five word tokens. So, in this case it would be 30 million word tokens as a difference. Good question.

The reason that people have really focused on this is because this difference in children’s vocabulary is thought to be a key driver of differences in school readiness, so in children’s ability to pick up literacy as well as in their later academic achievements. There have been studies suggesting, for instance, that children’s early language skills are one of the strongest predictors of their later academic achievement. Okay? So, this sort of leads to this idea that really in order to solve the very persistent problems of poverty in our society, perhaps those problems arise and originate in children’s home environments before they even get to school, right?

So, maybe the most effective thing to do would be to intervene on children’s home environments, to get parents to speak more to their kids. Right? So there have been interventions like this you saw the video from Talk With Me Baby just before. But there are other kinds of citywide initiatives. There’s one in Providence Rhode Island — they share this common name — of trying to get parents to speak more to their kids and they are really focusing on coaching the parents to speak more to their kids.

So, they’re focusing and sort of guided by this idea that there might be individual level characteristics of parents that might sort of not quite meeting up to the standards. Right? So maybe parents are lacking in knowledge of the best practices or maybe they don’t have the full level of effort that’s needed to provide this sort of high-quality, rich speech to their kids. Okay. I think I saw a question there.

Audience 2: Yeah. I was just wondering how recent are these interventions? All these initiatives?

Mahesh Srinivasan: How recent are the interventions? So, these interventions are ongoing till the present day. Is that what you mean?

Audience 2: No. More like what year did they start?

Mahesh Srinivasan: The question was when did these interventions first begin, and they’ve been around for quite some time. I can’t say the exact time that they first began. I know that there was some being done in Providence probably at least for the last 10 to 15 years, so it’s been going for some time. A good question. Yeah. Any other questions?

Okay. So, what we thought about in our lab is that there was something that seems like it was missing from the conversation. Namely that parents who are in poverty are facing a number of structural pressures that might actually constrain the ways in which they engage with their children. Okay? So a parent in poverty is likely to be facing housing insecurity, is going to have a scarce financial resources, perhaps inadequate healthcare, food insecurity, other family members that are struggling. These are all things that are highly stressful, right? If you are in poverty and those kinds of stressors, the idea is, might actually take your attention away from your child.

It is quite cognitively demanding to engage with your child in these kinds of ways. Moreover, it might have impacts on your mental health which we know can in turn spill over into how you interact with your child. Okay?

We also know from research and behavioral economics that simply the experience of resource scarcity can actually affect cognitive processes in a very direct way. There’s a really interesting study that was done with sugar cane farmers in south India. What the study found was… Well, what they did first is they tested the farmers on a battery of cognitive tests. One test was a Raven’s Matrices test, which is kind of an intelligence kind of test where you are asked to sort of complete the pattern — this is the test — and then you have to choose which of these sort of complete the pattern. Another kind of task test called the Stroop Task, where you might see these numbers here and you’re just simply asked how many numbers were there.

The correct answer here would be three, but that requires that you inhibit the number five, which you’re automatically reading, so this is all sort of measures of your cognitive inhibition or your fluid intelligence. They tested the farmers before the harvest, right? So, before they had a lot of resources versus after the harvest. After the harvest, they presumably have more resources and are experiencing less scarcity.

What they found was that the farmers performed better after the harvest compared to before the harvest. Okay. So, that, again, has suggested that resource scarcity can directly impact cognitive processes. So, what we were interested in is whether that might also impact parent-child interactions, okay? Does that logic make sense? Are there any questions about anything? Yeah, please.

Audience 3: How would you know that it was specifically resource scarcity and not just like worrying about all the things you have to do with your harvest, right, because I might have a resource scarcity before I go to work, and I know it’s a busy day, but I might not be as focused?

Mahesh Srinivasan: Yeah. So, the question is, how do you know it’s resource scarcity as opposed to a myriad number of other factors, right? That was the question.

That’s a great point. So, our general hypothesis that a person, a parent who is lower in socioeconomic status is facing a whole variety of factors like this. Resource scarcity may just be one such factor. So, we’ll look at, in the actual studies that I’m going to present, we can consider whether those are isolating, this variable of resource scarcity or not. Okay. So, that’s a great question.

Okay. So, first of all, why is it actually important to consider these structural explanations? I think it’s important for a variety of reasons. The first is that by considering them, we take the blame off of parents who are trying to do their best in a difficult situation. So, we take the blame off of them and attribute it instead to the unequal structural systems that they’re operating within, right? So, we know from other studies that parents are feeling sort of blamed for their perceived shortcomings, so doing research on this actually could be helpful for that.

There are also maybe implications for the existing interventions, right? So, a lot of these interventions like the ones that I showed you, they do show short-term beneficial effects during the intervention, but they often do not carry long-term effects. One way of explaining that perhaps is that when parents are done with the intervention, they’re still subject to the same structural constraints and pressures, right? So, if you don’t relieve those structural pressures, you might not really get beneficial outcomes in the long term. Okay.

Finally, I think that by illuminating the structural forces here, it could potentially highlight the need for actual changes, policy changes, etc., that address inequality, so things like the child tax credit, for instance. Okay.

Okay. So, I’m going to just show you some of the research that we’ve done that’s tried to get at this topic. This has been led by graduate students in my lab, Monica Ellwood-Lowe and Ruthe Foushee.

So, in the first study, we randomly assigned participants who are from higher socioeconomic backgrounds. These are parents from these backgrounds, and we had their 3-year-old children with them. So, there were 84 of them. They’re randomly assigned to a scarcity condition or a control condition. So, the reason we chose parents from this type of background is that they should have all of the individual sort of traits and characteristics that are needed to provide this sort of high-quality engagement with their kids according to the previous research. Okay.

So, we’re interested, basically, in whether when we prime them to think about instances or times when they didn’t have enough of something in their life, would it actually affect their subsequent interactions with their child? This is a priming study, and I’ll explain this a bit more.

In this scarcity condition, parents were asked basically to fill out a survey to talk about three or four times that they didn’t have enough of something in the past week. They could talk about anything, but then they were just asked to elaborate on whatever it was that they talked about. Choose one thing, and elaborate on it. The control group were simply asked to just write about three or four things that they did in the past week, okay? So, that’s all happening. A child in the meantime is just completing some unrelated experiment with the experimenter. Okay.

Then, once the survey is completed by the parents, the experimenter essentially said that they need to load an additional survey onto their iPad, and they forgot to do that, so they need to leave the room. At the last minute, just before they leave the room, they “remember,” scare quotes there, that there is a toy in that cabinet that the child can play with while they wait. Okay. This is one of these puzzle sorting toys. Then, the experimenter leaves the room for 10 minutes. The parent and child are left to their own devices, and we just record their actual language interactions. Okay. Any questions about the setup of the study or how that works? Okay. Yep.

Audience 4: Do the parents know it’s being recorded?

Mahesh Srinivasan: They don’t know. So, do the parents know that they’re being recorded? They don’t know, so this involves some deception. Actually, one thing that we had to deal with is sometimes, the parents would leave the room because they’re wondering where is this researcher, right, but it was necessary for our purposes that they didn’t know. Okay. So, this is an example of the parent and child together. So, let’s watch this.

(Video clip)

Child: This or this one.

Parent: What is it? How many corners does it have?

Child: Five.

Parent: Five? Let me see. Let’s count together, okay? One.

Child: Two, three, four, five, six, seven, eight.

Parent: Eight, yeah. So, eight so that’s…

Child: Nine, ten.

Parent: Eight, so that’s an octagon. What is the octagon? Can you say octagon?

Child: [Inaudible].

Parent: Oh, what is it? You have to find something with eight. Eight corners. Wow. Good job.

Mahesh Srinivasan: That would be considered high-quality, child-directed speech, right? So, the parent is attending to what the child is doing, they’re providing a new label, octagon, right? They’re eliciting conversation. So, these are the kinds of things that practices that parents are coached on doing. Okay. So, what I’m going to show you now is the data, and so we are comparing the results from parents from the control condition and the scarcity condition here. The Y axis is the number of parents’ words. So, these are word tokens to use that concept that we talked about before.

Initially, when we look at the data, we don’t actually see statistically significant differences between the control and scarcity condition, right? So, it’s not the case, actually ,that the parents in the scarcity condition spoke significantly fewer words to their kids than those in the control condition, considered overall. There was a difference. It was lower, but it wasn’t statistically significant, but we did do a post-hoc analysis. So, this isn’t something that we preregister, but it was post-hoc and exploratory where we compared parents who specifically wrote about finances compared to parents who wrote about some other scarcity, okay?

The parents who wrote about financial scarcity basically said they didn’t have enough money, they’re running out of funds, something like this. The parents on the other hand in the no finances condition might write about kinds of scarcity that were often less persistent and chronic, so things like not having enough milk in their fridge or something like that. Okay. So, when we look just at those parents who wrote about financial scarcity, we saw actually that they spoke significantly less both than parents in the control condition, as well as parents in the scarcity condition who did not write about finances. Okay. Yep, please. Question?

Audience 5: I thought you said these were all fairly high socioeconomic.

Mahesh Srinivasan: Yeah.

Audience 5: There were people who still had severe financial problems.

Mahesh Srinivasan: So, the question is, so these are fairly high SES people, they still have severe financial problems. Well, I think the question … Yeah, the issue is they’re living in the expensive Bay Area, right, and maybe they have mortgages to pay, etc. So, I think even in this context, people were not immune from financial scarcity, but that’s a good question.

Yeah. So, obviously, this is post-hoc. We didn’t expect this type of distinction necessarily, and it’s possible that we didn’t really see priming per se. Instead, what happened is that maybe those parents who came into our study who were already experiencing some sort of financial problem in their life, we sort of gave them the opportunity to reflect on that, and we were therefore able to identify those parents, but still, this is suggestive evidence.

In our next study, we wanted to ask, what about parents who are actually experiencing financial scarcity in the moment? How might their speech actually vary as a function of that scarcity? So, here, we actually took advantage of a well-known fact that if you look and survey people at different times of the month, they report experiencing different levels of financial scarcity.

So, this is a result from Pew Research Center. Earlier in the month, people are more likely to say that they have money in savings. They’re more likely to say that they’re financially secure, more likely to say that they could replace one month of income. But at the end of the month, this figure really plummets, which you can see here. Okay.

Okay. So, given this, we were interested in whether we could actually see whether the same parent speaks less to their child if we are getting measures from them later in the month compared to earlier in the month. So, what’s powerful about this is that in this way, we’re holding constants the parents, right? So, their individual level traits are technically the same, theoretically the same, right, but what is changing potentially is the external conditions that they’re operating within. Okay.

So, in order to test this, we took advantage of this corpus called HomeBank, which has a bunch of recordings of conversations between parents and their children and these recordings. There are multiple recordings for the same parents. So, there are 192 families, but they contributed over a thousand data points. So, there’s multiple data points from the same family. We can basically look at the time of the month in which those recordings were taken and see if there are differences in the parent-child conversational speech during those different times. Okay.

So, the predictions, basically, were as follows. We expected that early in the month during the, usually when the first paycheck is there, there would be more conversational turns as well as around this week of the month, but toward the very ends of the month, we expected that there would be fewer conversational turns between the parents and the child. That’s what we sort of preregistered in our analyses. Okay. Are there any questions about any of it? Yeah, please.

Audience 6: What do you mean by conversational turns?

Mahesh Srinivasan: Okay. So, the question is, what do I mean by conversational turns? Great question. So, basically, this is a recording device. I’ll say a little bit more about that. This is a recording device called a LENA recording device that basically sits in the child’s pockets. It records all of the language in the home environments for up to 14 hours. Then, that recorder produces these automated estimates of the language environments including the number of words that the adult speaks but also this metric called conversational turns. What that means basically is a conversational turn is when a parent might say something and very soon after, a child vocalizes. So, there’s kind of this contingent back and forth.

What’s nice about that is we have better evidence that this is speech that the parents is directing to the child because it’s sort of occurring in this contingent manner. So, we’re looking at the number of conversational turns, meaning one thing could just be two conversational turns or it could be six. If it was six, it meant that the conversation went for a longer number of turns. Okay. Does that make sense? Yeah, please.

Audience 7: So, in your previous experiment, I thought it was told if you do a post-hoc analysis, the statistical analysis is a lot more demanding in the sense of making things statistically valid. Was that true in your case that you extracted the financial people and that then, I think makes a lot higher standard on making that difference valid. Was that the case of, even with that post-hoc aspect, there still was this key difference?

Mahesh Srinivasan: So, yeah. The question is, were we sort of penalized because we were doing a post-hoc statistical comparison? Yeah. I mean, we were in the sense that there were very few parents in that group that wrote about financial scarcity in particular, so as a result, there’s more variability in that group.

Audience 7: That wasn’t real hypothesis going on.

Mahesh Srinivasan: Yep. It wasn’t. Yeah.

Audience 7: My understanding is if you do a statistical analysis after the fact with new hypothesis, take out the financial, are they different?

Mahesh Srinivasan: Yeah.

Audience 7: Then in fact, you have a much higher bar to…

Mahesh Srinivasan: Yeah. So, we controlled for things like multiple comparisons in our space. So, yeah, that’s a good question. Okay.

Audience 8: I did have a question.

Mahesh Srinivasan: Yeah, please.

Audience 8: It’s more of a broad question though.

Mahesh Srinivasan: Yeah.

Audience 8: I’m curious, if you’re measuring scarcity financially, I’m curious if trauma would work in a similar way or if you guys have thought of, I don’t know, done studies around that as well.

Mahesh Srinivasan: Yeah. So, the question is like, we’re focusing on scarcity, but what about other things like trauma, right? Yeah. I think at this point, everything is still very open. There could be many kinds of stressors. We don’t think necessarily financial stressors are necessarily special. In that previous study, I think maybe they were in the sense that the kinds of other, the non-financial scarcity that parents were writing about didn’t seem like it was as severe, but yeah. I don’t think necessarily financial scarcity is special. I think there’s all sorts of different kinds of stressors that might sort of be operative. I’ll talk about another study in just a bit that might get at other kinds of external pressures. Good question.

Okay. So, are the predictions pretty clear here? So, basically, the idea is that in the last week of the month, maybe parents will be speaking less to their kids as measured by the conversational turns compared to earlier in the month. So, what I’m showing you here is in fact that we did find this effect. So, on here is a prediction of a regression model. So, you don’t need to worry about it too much. Just what I’m showing you here on the X axis are the days of the month. The very last week of the month, we do see that there’s a noticeable dip in the number of conversational turns between the parents and the child. Okay. I’ve sort of put up some statistics on the right here in terms of what this actually amounts to in terms of magnitude. So, 306 fewer conversational turns that week, thousands fewer that year.

This isn’t a huge effect, but the idea here is that what we’re looking at here is the financial fluctuations that are affecting parents just in a relatively small sense toward the end of the month, but if you are considering parents who are always in poverty, who are constantly facing financial pressures and scarcity, that might help explain why there are these large gaps and differences in how parents are speaking to their kids. Okay.

So, part of what was really interesting to us about this, right, is that this is providing a different kind of explanation for this word gap, right? It’s not necessarily that parents are lacking in their knowledge of how to speak to their kids, right? The same parents speaks less to their child when they’re experiencing more financial adversity than when they’re experiencing less financial adversity. Okay. Questions about that before I move on? Okay.

I’ll just tell you a little bit about how we’re sort of pursuing this hypothesis going forward. So, obviously, that last study that I mentioned is purely a correlational study, right? We were just looking at time of the month. We don’t even really know that time of the month is … that financial scarcity is really what’s explaining that correlation. So, we’re testing this in a causal way by looking at the impacts of cash transfers on how parents interact with their kids. Okay.

We know that parents, families that are receiving SNAP benefits, experience this cyclical type of effect, where by the middle of the month, they sort of run out of their SNAP benefits. So, that’s what’s plotted here is the average daily expenditures for food. You can see that … and then the X axis is days since the SNAP benefit receipt. Very early on in that period, they’re using most of their SNAP benefits, but they run out very quickly.

So, what we’re doing in this new study is we’re looking to see what happens if we give some of these parents a $400 sort of influx of cash, right? Does that actually change the way in which they interact with their child in any meaningful way? So, we’re going to be giving parents these LENA recorders, and we’re going to be giving some parents this cash transfer, and others in the control group will not get the same magnitude of the cash transfer. We’ll see if there are any effects on the parent-child conversations as well as we’ll measure other kinds of factors or mechanisms that might explain the change, so things like changes in the parents’ mental state, in their mood, in their sleep levels, etc. Okay.

Another thing that we are doing ongoing is we’re looking at how families are reacting to the pandemic as a case study of how adversity is affecting parenting. So, the way in which this works is that we wanted to get a sample of parents’ speech to their children and to collect that in multiple points of time, right? So, a repeated measures design, it’s called.

So, basically, what we ask parents is to record, give us an audio recording every time that they bathe their child for 30- to 60-day period. So, we just take out their iPhone and do an audio recording during that bath time interaction. So, the reason we chose bath time is we did some surveying before, and we found that across socioeconomic strata, this is a context that often appears. Mealtimes are much more variable. Book reading is much more variable, but bath time is quite regular across socioeconomic classes. Okay.

So, parents are giving us 30 to 60 of these recordings. Then, they’re also, at the same time, completing a survey where we ask them about a number of things, such as their mood, sleep, what they’re worried about, as well as any kind of external pressures they’ve been facing, right? So, have they lost a job? Do they have a family member who is unhealthy? Things like this. Okay. So, just showing you sort of the design here. We have these repeated measures over a 30- to 60-day period of the bath time interaction, takes a very long time to manually transcribe all of these things, but we’re in the process of doing that.

Then, we have all of these other kinds of data. So, we have data on the parents’ mood and how that’s changing over time, data on their worries as well as their sleep, their child’s wellbeing, the time that they’re spending doing different kinds of activities, and then also, these more external, structural kinds of shifts, like whether they received a government check or whether they lost a job. We really started this study during the height of the pandemic back in 2020, so we’re interested to see how the data turn out. Okay. Any questions about this or any of the follow-up studies?

Audience 9: Okay. The numbers on that are what?

Mahesh Srinivasan: The numbers on the X axis, you’re asking? Yep. So, good question. So, the numbers on the X axis here are the number of days, right? So, basically, each day, the parents contribute a recording of the interaction, and then they also complete a survey, right? Part of what we’re interested in are what are the correlations, right, within an individual family.

For instance, maybe for one family, there’s a very strong relationship between mood and child-directed speech, how much speech they direct to their child. For another family, there might be other kinds of relationships. So, this sort of design gives us the opportunity to look at relationships both across our whole sample but also within individual families. Okay.

Now, we’ve sort of considered this second claim here that some parents are not aware of these methods and need to be coached. I’ve instead suggested that simply the experience of financial scarcity and perhaps other kinds of pressures might impact how anybody would engage with their child. So, perhaps parents, this sort of idea that parents need to be coached isn’t quite on the right track, although we could discuss that some more.

Now, what I want to turn to is that first claim that really that there are optimal methods for supporting a child’s language development and that these kinds of practices are kind of akin to nutrition for the child’s brain and for their language development. Okay. So, there is a very long literature, very robust literature on looking at parents’ speech and how that relates to children’s language outcomes. That literature has identified a number of features that seem to be supportive of children’s language development. So, I’m just going to mention a few of them here.

So, the first is sort of this exaggerated intonation and sing-song prosody, the rhythm variation that you see in parents’ speech to children. So, when you speak to a child, you will say, “Look at that toy. That’s so nice, isn’t it?” Right? That’s the sort of exaggerated intonation. You probably wouldn’t do that to an adult to the same extent, right? Maybe to a romantic partner, maybe to a dog or something, but it’s different between adults and kids.

Audience 10:  Is that universal in cultures?

Mahesh Srinivasan: Yeah. So, that’s a good question. Is that universal across cultures? It seems to be present in many cultures, although I don’t think it’s quite the case that it is universal. There’s differences, though I don’t think it’s quite the case that it is universal, there’s differences in how much parents modify their speech when they’re speaking to kids in across different cultures. But it does seem to be a case that early in life infants seem to show a preference for listening to this kind of speech compared to adult directed speech. Okay. So there is that type of difference there.

Audience 10: The reason I ask is there are languages like Chinese where tone is a major part and English, that’s why it’s just a dramatic change child speak adult speak because we don’t change our tones all over when we speak to adults and I’m just wondering whether different in Chinese, do they have some other way of changing it to facilitate learning the total aspect?

Mahesh Srinivasan: Yeah, that’s a good question. So the question is, in Mandarin, there are tones in Cantonese, as well. And so, would it still sound different if you were speaking to a child versus to an adult? I think it still would, there’s still this sort of even more exaggeration. And so people think that maybe this is helpful in maintaining the child’s attention. Maybe it’s also helpful in allowing them to figure out where the boundaries are between sound categories by exaggerating them. So, that’s one idea. It’s also thought that language that follows the child’s attentional focus, so if you imagine a parent who’s paying attention to what the child is looking at, and then says, “Oh, do you know what that is? That’s a garbage truck.”

It’s thought that type of input, that type of language from the parent, is facilitative of their vocabulary development. So, that’s another thing. More words spoken to the child. We’ve talked about that, that’s thought to be predictive of vocabulary development and also grammatical development and also more diverse words. So, it’s not just hearing “cat” 10 times; it’s hearing a diversity of different word types, as well as rare words, so words that are not as frequent hearing, words that are not as frequent, like octagon is another example.

And then, finally there are qualitative differences. So, in addition to using a variety of words, there’s also qualitative differences like using strategies that elicit conversation from your child. So, instead of asking them to put something in a place, giving them an imperative like, put the piece in this puzzle, instead you can say, what is that? That’s an octagon. So, these are all things that have been linked to beneficial outcomes, language outcomes for children.

But critically, most of these studies have been done in Western middle-class educated context. There are some exceptions to that, but by and large, they’ve mostly been done in those contexts. And the reason that that’s important is that those kinds of practices are not universal. So, when we look across different cultures, there are lots of ethnographies that have been done as well as cognitive science studies. We see that these kinds of parenting practices, the ways in which children are socialized to language can vary dramatically.

So, I’m just showing you a few different contexts that have been studied here. One thing that is shared in all of these contexts is that early in life, children are actually not spoken to directly very much. Particularly, before children can themselves start to speak language, they’re not spoken to very often. It makes sense, why are you speaking so much to something that can’t really speak back to you?

And the idea, the intuition, is that these kids are going to learn through observation. You don’t really need to be teaching them the language. And yet, even though these kids are not spoken to a lot, they’re still embedded in very rich social environments. They have the opportunity to observe interactions around them. They are hearing lots and lots of speech around them, so this is speech that we would call overheard speech. It’s not speech that’s directed to the child, it’s speech that an adult might direct to another adult in the environment, or an adult might direct to an older child than this particular child. Okay. Any questions about this idea of the cultural variation, please?

Audience 11: I’ve always been intrigued. I don’t have children, but I’ve always been intrigued by my compulsion to talk to kids that I know can’t understand what I’m saying to them. And so that’s also, that’s very culturally embedded that, because not having kids, I don’t have any reason to talk to a 9-month-old child, but it’s always struck me as odd that I found myself doing that, but that must not be reasonable.

Mahesh Srinivasan: Yeah. So the comment was just, we often feel that it’s very intuitive to speak to a child. And I have a young daughter and I talk to her all the time — she’s only babbling right now — so, I’m embedded within this culture. I have this experience and that’s what I do, but it’s definitely not universal. And, in fact, it might be what people call a WEIRD practice. So WEIRD is this acronym — Western, educated, industrialized, rich, democratic — and so, it may be a bit more of a WEIRD practice. Other questions? Yes, please.

Audience 12: In this slide, you’re talking about speech that’s happening around a child. In the last part that you just presented, you were talking about vocabulary and what comes to mind for me is the difference between single word, recognizing how the letters fit together, saying it. But whether they understand it or not, the understanding is not necessarily speech because does a child’s vocabulary imply that the child has learned and knows what the words mean? Or is it just they can recite these words? I think about hyperlexia or conditions like this where a child may have an incredible vocabulary, but may not understand much of it at all.

Mahesh Srinivasan: The question is: What do we mean when we say that the child’s vocabulary size and vocabulary growth? There are many different ways of measuring this and one way would be to say that the child produces the word, and you can look also at corpora of these spontaneous conversations to see is the child producing the word or not. But as you rightly point out, simply saying a word doesn’t mean that you know that word or understands that word. So, there are other kinds of measures that people use.

One popular kind of measure, which is an explicit measure, would be to present four pictures and then you say to the child, can you point to the dog? And then there’s three other distractions there and you see if they actually point to it, that’s some measure of whether they understand the meaning of the word. And then, there’s other kinds of more implicit measures that can be done with younger infants, one of which I’ll show you in a bit that is actually asking the infant to look at something and then you monitor their eye movements to different things.
Other questions? I think I saw one here, please.

Audience 13: I’m just interested, is there a correlation between this observational method of learning language and a communal style of living like an environment where there are multiple families that interacting closely rather than the single-family [inaudible] normal and weird cultures.

Mahesh Srinivasan: That’s a great question. The question is: Is there a correlation between this type of observational learning and household family structure?

And yeah, I think you’re right that typically in the more observational learning, it’s a larger family context, joint families. The other thing that I think is an important shaper of this style of observational learning versus formal learning is the degree to which the parents have had formal education, formal schooling.

So, when you go to school, you get these practices inculcated within you — that you have to focus on what you’re paying attention to, that a teacher’s going to ask you the question that they already know the answer to. And then, we carry these practices over to our kids. We, basically, are testing our kids, we’re providing them language lessons. And I think there have been studies done, for instance, that show that parents from the same backgrounds who have had more or less formal schooling experience interact with their children in quite different ways. So that’s another type of factor that’s important here. Great questions. Any other comments or questions? Yeah, in the back.

Audience 14: Is this sense of this observational learning, wouldn’t that be the evidence that for optimal parent-child language interaction, you don’t need directed speech to children because clearly they were centuries, for generations, been able to learn the language without having child-directed speech.

Mahesh Srinivasan: Yeah. Great question. So the comment is that: Wouldn’t this be evidence enough that primarily it is overheard speech, observational learning, and we know that these kids are learning their language. Why is child-directed speech necessary?That’s definitely an intuition that I think makes a lot of sense, yet I think still what people are thinking here from the other perspective is that even though this type of child-directed speech, this practice of speaking a lot to your child, even though that might not be necessary for language development, perhaps it helps. And the idea is that maybe the best way to learn, the most helpful way of learning and the way that would speed the course of language development and make it the strongest possible form, would be to speak a lot to your child. But it does run into these very thorny issues of are we considering practices that are founded within our Western cultural experience, normative and optimal everywhere. And so this is an important tension in the literature. Any other questions or comments? Yeah, please

Audience 15: Question then is, do these kids start talking later?

Mahesh Srinivasan: Yeah. That’s a great question. So, do these kids actually start talking later than children who have been exposed to lots of child-directed speech? So this is the other area. There’s a literature that’s primarily from linguistic anthropology that is focused on these kinds of communities. And then, in very different literature from developmental psychology that’s focused on the outcomes from hearing a lot of speech and they don’t always meet up. So that anthropologists will say, these kids are arriving at language milestones at a very similar age to the children in America, middle-class, upper-class context. So they’re producing their first words at a similar age. They’re combining words together at the similar age. And then, they’re doing that consistently at a similar age. So, from that observation, it seems as though maybe there is nodes away, but that’s something that I’m going to get to in just a second.

All right, so we’ll keep going. What we did in this study is we were working with a community in Tseltal Mayan, which is in the Chiapas state of Mexico. And this was led by my former graduate student Ruthe Foushee. What we were interested in is whether infants in this community who are receiving very little speech directed to them, still show evidence of early knowledge of words. We are working all with infants who are being carried on their mother’s back, which is a very common practice there before the infants are walking, the mothers carry them on their backs. And usually during this stage, the mothers are not speaking a lot to the child because there’s very little speech directed to the child.

However, these kids have a very front row seat of what’s going on in the environment because they’re right behind their mother. They’re right there. They can peep out and look and see lots of different things. So, we are interested in basically whether these infants would show evidence of knowledge of some words. There have been some studies done in the U.S. with young infants, 7-month-old infants, I believe, who showed evidence of understanding some words. And I’ll give you a little bit of information about how they did that.

This was the measure that they used in previous studies. It’s an implicit measure that we adapted to this community. In this case, basically, the experimenters are here and they’re controlling the presentation of these stimuli. And then there are two pictures that will appear on different screens. The infant is here and is seated on their care caregiver’s lap. And then, the caregiver is hearing something over their headphones and they’re basically told what to say to their infants.

For instance, they might say, if this was in English, they might say, “Where is the baby?” And then what we do is we monitor the infant’s eye movements and see whether they look at the matching screen or not. So, fairly simple, the caregiver during this period is just keeping their eyes closed so that they’re not giving any cues to the infants. We adapted this to include words from Tseltal, the Tseltal language, including words that we thought the infants would be likely to have heard quite a bit and likely to have learned by this age. Any questions about that?

Okay. I’ll also just say, in terms of the methods you might expect that infants might have a preference for looking at one of these images over another. So, if I say, “here is the baby?” Maybe the infants would do really well at looking at this because they already want to look at the baby. So, to actually control for that, what we have to do is to compare how much they look at this when they’re asked, “Where is the baby?” to how much they look at this when they’re not asked about to look for the baby, when they’re instead asked to look at the corn. “Where is the corn?” So, that’s what we do here. We have, this is called a pared picture method, and we have in this case, this trial where there’s a baby and corn, and they’re asked to look at the baby, as well as this case here, where they’re asked to look at the corn and we just subtract across these two trials, the proportion that they’re looking to the picture when it’s the target of the word versus when it’s not the target.

Does that make sense? That’s a way of controlling for the baseline preferences toward these two different images. So, here is a video that just shows how this goes. The mother is here and then the child is on their lap. And you’ll see the images that the child is seeing come up on either screen here [Tseltal language] I think more time looking at the horse.

So, here are the results of this study and what I’m plotting here is the increase in looking toward the target after that word was uttered, after the critical word. In the last case, it was the word for “horse.” So after that word is uttered, how much more are they looking toward the target compared to the non-target? And so anything that’s positive on this graph, so above zero would indicate that they have a preference for looking at the target picture. And then these are all the different contrasting words and pictures that we showed in the experiment.

And so, what we see here is that if you look across the different trials, you see some evidence that the infants are looking preferentially at the target across the different items. And if we also look at the subject data, so at the individual children, the large majority of them are doing this where they’re looking more at the target than at the non-art. And we also compared these data to data from US infants, American infants, and we see a very similar type of quality to the data, where infants in both cultural contexts seem to be understanding words. So this is interesting, again, because these are infants in Tseltal who have not heard a lot of speech directed to them, but it seems as though they may have been able to pick up these words probably from speech around them. Question here.

Audience 14: Would they do better, then, if you take the pairs where they are farthest above the lines and you take baby corn, will they favor, that one works and you take another one that works. So, then you say, “baby,” “fire” or “dog” or “corn,” would those consistently all would even predicted that those would be above the line because that indicates that they do have some knowledge of those words. Maybe the ones that weren’t above the line is they didn’t know the words. If you put from this, they did know the words and then made new combinations with those that they seem to know or take ones where they didn’t seem to know. Would that be predictive of whether those would be above the line or not? I just wonder if it went that next step.

Mahesh Srinivasan: Yeah. We haven’t done that in this. The question is, basically, what do we make, for instance, of the fact that some of these are above the line or not others? Is it necessarily because they don’t know what these words are here and if they do succeed in this case, what does it mean about their knowledge of the meanings of these words? So, I want to stress that what this is showing here is not that you really totally understands the meaning of the word. All this shows you is that you can differentiate “dog” and “fire.” And so, you have some partial meaning of this word, potentially, but you might not really know it completely, but this is still all that was shown for the American infants too. So they’re still on equal footing here, but that’s a good question. And I think we still need to know more about what’s the… How sophisticated and complex are these meanings that infants have early in life. Other questions? Yeah, please, in the back.

Audience 15: Just out of curiosity, how do you recruit friends and how do you recruit families from… Do you have someone in the research team who knows how to speak the language [inaudible]?

Mahesh Srinivasan: Yeah. So, how do you recruit people in this community? My former graduate student, Ruthe, had established some relationships there even before she came to graduate school. She went there, really did all of this work and she has contacts there and worked with folks there who did the actual experiments on the ground. It’s a lot of work, but she did all of it. Okay. Any other questions?

Audience 16: Well, I don’t quite know how to read it because take the pair dog fire, so when they’re shown fire and when fire is a target and dog isn’t and then you switch them, how do I know which was which in that?

Mahesh Srinivasan: Yeah. The way this works, the way to read this, is just that, so there’s going to be a picture of a dog and fire on two different trial trips, but there’s always a picture of a dog and a picture of fire. On one trial type you’ll hear, “Look at the dog,” and then we’ll measure if you’re looking toward the dog. That’s the one value. And then on the other trial type, you measure looking to the fire when you’re told, look at the fire and you measure that’s one value is how much are you looking at the fire? And then you average those together. So both of those values should be positive if you’re seeing, understanding.

So, moving on from this, it’s possible that these infants, although they aren’t spoken to a whole lot, it’s possible that they still learn those words that we presented to them from speech that was directed to them because it’s not the case that they’re never spoken to. So maybe with whatever little speech was directed to them, they actually just learned from that speech.

And so, what we looked at in the next study was whether there’s some kinds of language that the infants could only have learned through overhearing, through listening to language around them. And so, we made use of the facts that Tseltal has this interesting system of honorifics or greeting terms. These are terms that are very frequent and very distinctive, they’re by definition never addressed to the child because these are terms that are greeting term for an older man, a younger woman, things like this, they would never be addressed to the child.

Yet, they have very distinctive pitch contours. They could be quite attention getting for the infants. These are very good case studies of language that could only be learned through overhearing and might be quite learnable. And I’ll show you an example of this. This is what might be an older man and if a younger woman saw this man, they might say the following to them as a greeting, [Tseltal language].

I don’t speak Tseltal, but that’s my imitation of it. And then, the man might say in response, he might say, [Tseltal language].

Okay, so it’s a different greet that’s being addressed to the younger woman. What we did in this next experiment is that we used these greeting terms and took a very similar approach. So there’s one greeting term for an older man, [Tseltal language], one for a younger woman. And so, we would present [Tseltal language] and we would see the degree to which they look at the older man compared to the younger woman and similarly [Tseltal language] and see if they look at the younger women more than the older men and similarly for this other pair of honorifics here. Any questions about that?

Part of what’s interesting about this, and why this is actually quite challenging, potentially, for a child is that these are all pictures from the same semantic meaning category, they’re all faces of people. This is actually quite difficult, potentially, for the child to differentiate between between these meanings. It’s not a very large difference, like a corn to a baby, this is a subtle, small difference.

Okay, so, here’s how this looked. The child is going to hear [Tseltal language], which is the greeting for the older men. Let’s watch this.

Speaker 5: [Tseltal language]

Mahesh Srinivasan: The child seems to be looking at the older man. Okay. So, here are the results. So, this is the result from before, for the nouns. And then here’s the results on the right for the honorific terms. And so, what we see here is again, positive increases to the target for both of these pairs of honorifics. And so, what’s interesting about this again, is that infants are showing knowledge of language that would never be addressed to them. So, they are capable for sure of learning from overhearing, these studies demonstrate that.

And part of what we’re interested in here, is the idea that maybe infants are adapting to what is the structure of their local environment. So, our results raise this possibility that what is really truly optimal really is something that has to be considered in context. So, children in all environments might adapt their learning strategies to the structure of their environment. So, on the one hand, kids in Western, middle and higher SES context, they’re spoken to quite often, they might come to be spoken to often, they might develop that expectation. They might develop the expectation that their attention will be managed very tightly by their parents. On the other hand, children in other communities, outside of the “weird” context, might come to learn from interactions around them. They might start to pay attention to interactions around them and learn language and maybe other things from that. I saw a question. Yeah.

Audience 17: I was just wondering if you used a completely different word in your dog fire experiment, like cloud, what would the reaction of the baby be and would that be indicating of something?

Mahesh Srinivasan: Yeah, so, the question is, if you use a completely different word like cloud in the study, would … so are you saying if we contrasted cloud with one of those words or-

Audience 17: No, you have the pictures of dog and fire.

Mahesh Srinivasan: Oh, I see. Yeah.

Audience 18: When you ask clock, what is the clock?

Mahesh Srinivasan: Yeah, so that’s an interesting question. One thing that is related to that, I think, is this idea that if children hear a word that they don’t know the meaning for, they might make an inference that it doesn’t have the meanings of other words. So, for instance, if they don’t know what cloud is, they might think it doesn’t have the meaning of dog and fire.

Tthis is a phenomenon called mutual exclusivity, where children have this expectation that every object is going to have only one word associated with it, such that if you hear a new word, it’s going to label a different object or a different meaning. So, that’s one thing. If they did know cloud already, if they already did know the word for cloud and they were presented with dog and fire, it’s a little hard to know what they might do, but it would be interesting if they looked at what was semantically closest. Like, maybe the fire would be closest. I’m not sure. Good question. Yeah, please.

Audience 19: So, I’m curious from your last slide, whether the children who were good at, you got some that look like they’re really good at distinguishing between these. Are the ones that are good at distinguishing the words on the left also the ones that are good at distinguishing the words on the right?

Mahesh Srinivasan: That’s a good question. The question is, are the ones who are good at distinguishing the ones on the left as good at the ones on the right. I don’t know those results off the top of my head, but that would be good for us to look at, for sure. We did look to see whether kids who were better at this in general, whether there was an effective age. I think there might have been an effective age with that, but still it was quite comparable to what you see in America. But it would be interesting to see if it’s applicable across domains, like in the honorifics case too. Good question. Anything else?

Okay. All right. So, I just mentioned this idea that maybe there’s adaptations to the local environment. Maybe some children, they learn by starting to observe others around them. And there’s an interesting extension of this beyond the domain of language, simply to learning from interactions around. So, the capacity for observational learning. So, what I’m going to show you is a video here that compares Mayan children with European middle-class children in America.

And what happens in this experiment is that these kids are siblings, and the teacher is instructing this child on one sort of task, like how to make some sort of object, and then the question is, what’s the degree to which the sibling is going to observe this demonstration and actually learn from that demonstration? And they’re comparing that between the Mayan children and the European American children. So, this is a test of the degree of observational learning. So let’s watch this.

Speaker 6: [Tseltal language].

Speaker 7: [Inaudible] in Mayan children were twice as likely to pay attention to a nearby demonstration of how to make a toy using a rubber band for energy. They were attentive while they waited for their turn to make a different toy.

When middle class European American children were in the same situation, they showed much less keen attention.

The Mayan children were more attentive and they also learned more. They were later able to make the toy that their sister or brother had made was less help from an adult than middle-class European American children. The researchers have found a similar pattern of keen attention and skillful learning among Mexican heritage children in the United States, especially those whose families were from Indigenous heritage communities.

Indigenous communities, specifically in Mexico and North and Central America, children are often engaged in learning situations where they are involved in the activities of their communities and expected to observe, even when they’re not told to pay attention to the events that are taking place around them. And this contrasts with patterns found in middle-class European American communities, where children are often segregated from adult activities.

Mahesh Srinivasan: It’s a really interesting example of these sort of differences in how learning is structured across different cultures. So, in the case of these Mayan communities, the kids have the opportunity to observe sort of what’s going on in the local context and actually it’s a pitch in and help, whereas often we’ll give our children different kinds of activities, different toys that are segregated from what’s going on in the home environment. And it seems as though this contributes to this overall tendency of the Mayan children and even these Mexican heritage children, with Indigenous heritage, to observe interactions around them. Any questions about that?

Okay. So, one thing that I want to point out just is that these are strengths. The ability to learn from observation really is a strength, but it’s not really a strength that is typically utilized within our school system. So, within our school systems, we really emphasize the ability of a child to selectively attend to what they’ve been given. So, look at this homework in front of you, look at this reading, pay attention to that. We really downplay the ability to collaborate, to learn through observation in a more informal way. And so, this raises the possibility that children adapt to their home environments in different kinds of ways. They develop different kinds of skills. A Western middle-class child might develop one kind of skill from a European American background, a child from another context might develop a different kind of skill.

But the school system might privilege one kind of skill over another kind of skill. And that then helps to explain why we see these differences in academic outcomes. So, you really need to think about how can we tailor and change our curricula in a way that actually leverages all children’s strengths, as opposed to just focusing on the strengths of some children. All right, so I just want to sum up and wrap up here. I started out with these two popular broad claims, which you saw in that video, that there are optimal methods for supporting language development, second, that some parents are not aware of these methods and need to be coached. And so, what I’ve suggested and said is that children actually do seem to learn across environments that deviate quite a bit from the normative Western practices.

So, these environments that don’t have a lot of speech directed to the child, where instead observational learning is more of the norm, children do seem to learn language, obviously, in those contexts, but they don’t seem to be delayed in learning words in those contexts. And then, I also suggested in the first part of the talk that if we really want to have interventions that increase low-income parents’ speech to children, we really must consider the structural pressures that they face. It’s not simply a matter of lacking knowledge of parenting, lacking efforts, being in a context of poverty can really constrain the ways in which you interact with your child. So, to the extent that it really is a goal for a parent to speak a lot to their child, we have to consider the context that might limit parents from achieving that goal.

All right. So, I will wrap up there. I want to say a big thank you to the Ph.D. students who really did the bulk of this work. So, Ruthe Foushee and Monica Elwood-Lowe, to our lab manager, Grace Horton, and then really a big thank you to all the families that we worked with, the children that we worked with, the research assistants, both here in Berkeley, but also in the the Tenajapa area in Chiapas, including these cute kids over here, all of the members of our language and cognitive development lab, and then our funding supports here. Thank you all. And I would love to hear any questions that you have. If there are time for questions. Yeah, please.

Audience 20: One question I have is I’m curious about the difference between cultures and the topic of speaking about emotions or having emotions be taught in English. I don’t think it’s a simple answer, but I’m curious if you know where…

Mahesh Srinivasan: Like having emotions taught in different languages you’re saying?

Audience 20: Like some cultures and languages preference, emotional language and some don’t. And then, I don’t know, I just noticed that there’s a variation there, and I’m curious if you know anything about that.

Mahesh Srinivasan: So, the question is about differences in how languages sort of express emotion. And I think also just cultural norms, and I would say like in how acceptable it is to give these displays of emotions. So, like to say things like, “I love you,” things like that, we know that there can be some big differences across different cultural groups and languages.

So, we’ve done some work comparing, well, actually looking at Chinese parents, immigrants, who are bilingual in Chinese and English and looking at the different ways in which they express emotion when they’re using Chinese versus English. In this case, it was Mandarin. And we do find that they’re more likely to use these kinds of positive emotional expressions when they’re using English, compared to in Mandarin.

And then, we also see differences kind in their facial expressions as they’re using one language versus another. So, there is an interesting literature on this. One possibility that we looked at in our work is whether the practice of code switching is a way in which parents might regulate their emotions. So, if they’re feeling a lot of emotions, might they switch from their first language to their second language? Or might they switch from Mandarin into English? How does that affect their emotion regulation and their facial expressions and so on? So, I’d be happy to talk more about that. Yeah, please.

Audience 21: Not a question, but an observation, just looking at those videos, it just really drove home to me how hard it is to be classroom teacher in this country that you’ve got to make each child feel like my attention is focused on you, you know?

Mahesh Srinivasan: Yeah.

Audience 21: It just must be so hard to be in a classroom.

Mahesh Srinivasan: Yeah. The comment is, it must be so hard for a teacher to manage all of these kids in a classroom. And in California, actually, we’re expanding transitional kindergarten to all 4-year-olds very soon. And so, that’s going to be a very … yeah, it’s a great possibility, but there’s also a lot of concern about how prepared are we going to be for that. There’s a lot of teachers that need to be trained, the classroom sizes are too large right now. So, having one teacher manage 25 kids is crazy. I have a lot of difficulty with just my one child. So, it’s a lot.

The thing that I found really interesting about that video is obviously you saw that the European American child was completely tuned out when the teacher was instructing their sibling. And the other child was looking quite a bit at what their sibling was doing.

And if you think about that in the classroom context, that child who’s looking at the other child’s paper, is going to be thought of as cheating. It’s devalued. So, this is what I mean about how can we think about these practices and really be culturally sensitive in our educational practices. Yeah, please.

Audience 22: All right. I’m not totally sure how to articulate this question. I noticed that all of the hypotheses are about the parents, but it seems to me that a lot of what you’re discussing is the assumptions and biases that are baked into the initial premises, that are just from highly situated points of view, and I wonder where does that conversation come in to reflect back the questions and mirror to those initial articulations of what’s best and so forth and who they’re coming from? Just to change the lens over time and not only focus on the … does that make sense?

Mahesh Srinivasan: I think so. Yeah. So, I’m not sure how to repeat the question, but the question basically is … we’ve been talking about the initial biases and focused on parents and what’s best, how can we come back and think about, maybe what’s best based on all of this? Or that was a bad imitation of your question.

Audience 22: So, I’m like, who is saying upfront, those initial things that you just proved-

Mahesh Srinivasan: Who says those things?

Audience 22: What are their experiences? What are the biases and assumptions that are leading to those things, that are being disproved here? But one of the focus is actually on the parents, and I’m curious how we can also be critical of how those folks who are pausing those things in the first place are situated.

Mahesh Srinivasan: Yeah. I see what you mean. So, we were talking about the parents mostly here, but how do we take this back to think about the researchers, the interventionists and so on? Yeah, that’s a great question. And I think that researchers are totally well-intentioned here. They’re seeing sort of this socioeconomic academic achievement gap, which is a real problem. And if you’re doing really poorly in school, obviously that’s going to have major downstream consequences for your later life outcomes. And so, people are trying to make a difference here. And they really think that this word gap is really one of the primary drivers of that, so they want to address that and change that and shift that. But I do think it has some unfortunate consequences.

Like, just simply even calling it a gap thinks about things in a deficit mentality. It takes us away from the strength-based mentality that I’ve been talking about here. I think people are coming from this perspective because they themselves have … most of us, like myself included, have been in school pretty much most of our lives. And so, this is what we’re used to, and this type of educational practice and way of learning is what we think is best, and it’s very hard to sort of get out of that mindset, I would say.

And still, I think if you do get out of that mindset, you still need to operate within the structures that exist. So, we can’t say that it’s not important for a child from a lower SES background to hear lots of speech directed to them, because we know that it is in fact really critical for their academic success. I’m not contesting that. But I’m just saying that the ways in which the school systems are set up privileges that, and it need not privilege that.

But a child does need the … when are those structures going to be changed? That’s not going to happen instantly. And so, the child does need to develop those skills to thrive within the existing structures. I don’t know if that helps, but yeah. Other questions or comments? Yeah, please.

Audience 23: Thank you for all the highlights of the importance of the cultural awareness [inaudible]. Do you it’s to consider that in future programs for education, to teach to teachers about that and how to change the dynamics in the culture in class?

Mahesh Srinivasan: Yeah. So, great, great question. So, is it important to teach teachers about these cultural differences and to really ensure that those dynamics are reflected in the classroom? Yeah. I do think that’s super important. I’m not an education researcher, but I think it’s important that teachers really facilitate that relationship and understanding. And one way of doing that, we know that the teacher-child relationship is a very important predictor of how they’re going to thrive in school.

And starting that even before the child gets to school, so having the teacher visit, for instance, the parents’ home, and to really see how things are there. Having the parents come to the classroom and volunteer, if possible. Having lots of these kinds of conferences. Things like that can really make it feel like a more inclusive environment, but also facilitate the knowledge on the part of the teacher and where the child is coming from, I think can make a big difference.

So, I think that would be my answer, but I think there’s a lot of things I think that could be done. Also, the curriculum can be more tailored toward observational learning and not just this formal style of learning.

Dione Rossiter: Why don’t we take one more question if there is one.

Audience 24: Just had a quick question on your research in the Mayan communities — is that novel? Has that not been done before?

Mahesh Srinivasan: The research with the Mayan communities? Yeah so, the question is, has that research been done before or not? So, research like this that actually takes a quantitative approach to characterize language development within those communities, as opposed to ethnographic descriptions, hasn’t really been done, it’s just starting to be done now. There is some other work with these communities as well as other communities outside of the weird context, but it’s still relatively new within our field. So, there is the linguistic anthropology work. But as I said before, it doesn’t quite meet the developmental psychology work. So, now people are starting to use these more quantitative methods so that we can start to make these different literatures actually meet with each other. Thank you.

Dione Rossiter: All right. Let’s thank Dr. Srinivasan one more time. So wonderful.

Mahesh Srinivasan: Thank you. Thanks very much.

Dione Rossiter: Thank you for coming out, everyone. There are so many good questions. So, please come again next month.

Mahesh Srinivasan: If anyone has any more questions, feel free to come and chat with me. Thank you.

[Music: “Silver Lanyard” by Blue Dot Sessions]

Outro: You’ve been listening to Berkeley Talks, a Berkeley News podcast from the Office of Communications and Public Affairs that features lectures and conversations at UC Berkeley. You can subscribe on Spotify, Apple Podcasts or wherever you listen. You can find all of our podcast episodes with transcripts and photos on Berkeley News at news.berkeley.edu/podcasts.