The future of learning to read
Jason Yeatman is an expert in the neurobiology of literacy whose lab is fostering a virtuous research cycle between academia and school communities, aligning scientific inquiry with real-world needs of students, parents, and educators.
His lab has developed ROAR – the Rapid Online Assessment of Reading—a gamified, web-based dyslexia screening tool. ROAR provides fast, precise, and scalable assessments, helping educators identify and get support to struggling students. We’re aligning cutting-edge reading science with the challenges teachers face every day, Yeatman tells host Russ Altman on this episode of Stanford Engineering’s The Future of Everything podcast. Yeatman and his team just launched ROAR@Home BETA, a parent research portal. Any parent who listens can sign up for ROAR here - https://roar.stanford.edu/signup/
Listen to the end to hear a question from one of our listeners for Professor Renee Zhao, as well as Professor Zhao's response.
Transcript
[00:00:00] Russ Altman: This is Stanford Engineering's The Future of Everything, and I'm your host, Russ Altman. I thought it would be good to revisit the original intent of this show. In 2017 when we started, we wanted to create a forum to dive into and discuss the motivations and the research that my colleagues do across the campus in science, technology, engineering, medicine and other topics. Stanford University and all universities, for the most part, have a long history of doing important work that impacts the world, and it's a joy to share with you how this work is motivated by humans who are working hard to create a better future for everybody. In that spirit, I hope you will walk away from every episode with a deeper understanding of the work that's in progress here, and that you'll share it with your friends, family, neighbors, coworkers as well.
[00:00:48] Jason Yeatman: As a cognitive neuroscientist, the fact that learning to read is connecting vision, hearing language in a new way is fascinating. And then of course, once you start studying reading, you can't help but realize that this is critical to success in modern society. So it's a place where scientific insights can really inform advances in society.
[00:01:13] Russ Altman: This is Stanford Engineering's The Future of Everything, and I'm your host, Russ Altman. If you enjoy The Future of Everything, please remember to follow it in whatever app you're listening to right now. That'll ensure that you never miss the future of anything. Today, Jason Yeatman from Stanford University will tell us about learning to read. There's many factors when kids learn how to read. It takes vision, it takes an understanding of language, and the combination of many other factors. It's the future of reading. Before we get started, a reminder to follow The Future of Everything in whatever app you're listening to so you can always be alerted to new episodes and you never miss the future of anything.
[00:01:56] Learning how to read is something that many of us did when we were four, five, six years old. It's a critical thing that they do early in school, and we know that there's a range of skills that kids come in with reading. Some kids can read at three and come to kindergarten ready to rumble. Other kids, it's a bigger struggle. There's a disease called dyslexia or a disorder. It just refers to the kids who are at the lower end of reading efficiency. These can be treated. If we can recognize dyslexia, then there are well-known programs, in fact we'll find out later that this is an almost solved problem once you've made the diagnosis, but making the diagnosis is not that easy.
[00:02:37] Well, Jason Yeatman is a professor of pediatrics education and psychology at Stanford University, and he's an expert at the neuroscience of how we learn to read. He's created and he'll tell us about an app that can relatively quickly diagnose whether there's an issue in visual processing, awareness of language, or a combination of other factors that can then precisely diagnose the issue and allow teachers and others to intervene to make sure we get kids, and even adults, up to the level of reading that they should be.
[00:03:12] Jason, first question is, why have you devoted your life to understanding how reading is acquired and how it works?
[00:03:21] Jason Yeatman: Oh, that's an easy question. It's the most interesting thing the brain does, and it's also really important to society. So I actually started my PhD, I was deeply interested in how children's experience sculpts brain development, and I found reading actually as this interesting way to understand that, right? It's this, um, made up kind of arbitrary invention of modern cultures. We've decided that kids must be successful in learning this skill in order to be successful in society. We spend a lot of time teaching kids how to do this. So as a neuroscientist, I can study how this process of learning to read actually sculpts new brain circuits that do incredible things. As a cognitive neuroscientist, the fact that learning to read is connecting vision, hearing language in a new way is fascinating. And then of course, once you start studying reading, you can't help but realize that this is critical to success in modern society. So it's a place where scientific insights can really inform advances in society.
[00:04:21] Russ Altman: Great. Well, thank you. And, and, and so you, you mentioned something about um, uh, it's this weird thing that we've invented for which with great cultural importance, and I did want to ask about, like a very foundational question is, are our brains wired to read? And, and that, that's a problematic thing for me to imagine because we went many, many, many, many years without reading. Um, so tell me about our understanding of why reading and why the visual looking at symbols, and then translate that into words, why does that even exist and like what do we know about the core neuroscientific principles behind reading?
[00:04:56] Jason Yeatman: Yeah. So as you said,
[00:04:57] Russ Altman: Super easy question. Super easy question.
[00:04:59] Jason Yeatman: Lots of questions embedded than the question. Well, as you said, you know, our species was around for a long time before literacy was. Our species was, um, communicating through spoken language for a long time before literacy is around, right? So the prevailing theory is that there's nothing in our DNA that like created circuits that are waiting to see books, right? This is, uh, an invention of modern society. Why was it invented, that I, I don't know. But I think a lot of the theories, um, are actually around like commerce that, um, writing down debts and so on was like the, the first beginning of writing things down. But that's more of a, uh, uh, anthropological question, the neuroscience question, but you know, it, um, the, the first, uh, forms of written language were Sumerian cuneiform tablets, I think, uh, around twenty-five hundred BC. So, um, that's kind of the, the,
[00:05:49] Russ Altman: Yeah, not that long ago.
[00:05:50] Jason Yeatman: Not that long ago, right? And, and the idea of a, of a literate society, like a society where children are all taught to read and math, really a brief blip in the, in, um, you know, the, the existence of the human, human species. So this means that learning to read depends on the brain's capacity for plasticity, right? It's this interplay between, um, circuits that evolve for other things. So, evolved for spoken language, evolved for vision, right? We have, uh, um, we are visual creatures, and we have, uh, you know, a lot of our brain is devoted to processing visual information. And then there's this new cultural invention of literacy that involves tuning these circuits up to do a new thing, basically accomplish a new function through new interactions between these circuits.
[00:06:41] Russ Altman: Really, really interesting. And, and, and I, I presume, I, I guess since we're talking about basics here, um, many things in humanity are kind of bell curves, right? Height, you know, whatever. Um, is, do we see that in reading? Is, and I don't even know how you would measure this, but you do. That's why you're here, that's why we're talking. Um, do we, is there some sense in which reading skill or reading efficiency is distributed throughout the population, like many other traits like height and weight and things like that?
[00:07:08] Jason Yeatman: Yeah. Yeah. Right on. I don't even need to say anything. You, you got it all there. Yeah. So, um, reading abilities lie along a continuum. You know, you have people that are exceptional readers, you have people that, you have a lot of people that are around average, and you have, um, a lot, you know, you have some people that really struggle. When we say dyslexia or developmental dyslexia, um, people in all, people often think that this is, that there's something unique and, um, like some discontinuity, and that's generally not true. It refers to the lower end of this continuum of, of children that, um, persistently struggle with the process of learning to read. Um, and there's great intervention programs and ways to support children who are struggling. Um, but yeah, you know, dyslexia referred to, as you said, the bottom end of this continuum.
[00:07:53] Russ Altman: Great. So I wanted to go to dyslexia 'cause I know you've put a lot of time in and we're gonna talk, uh, uh, in a few minutes about some of the interventions. But, uh, let, let's define dyslexia because you've said some intriguing things there. You said that it's not a, it's not that it, that there's a yes no, but there's more like a, uh, so is dyslexia not really a disorder, so much as a description that, um, that, that, that the reading is not going as well as it would if you were in the normal part of the curve. In the middle of the curve.
[00:08:22] Jason Yeatman: Yeah, you know, but, um, almost all developmental disorders are like that, right? And there's very few, um, you know, unless it's like a neurological insult, like a, a stroke, right? So there are disorders that are caused by, you know, having, having a stroke where, um, one day you are functioning one way, you sustain some kind of injury in your brain, and then after that, so there is, um, acquire, there's acquired forms of dyslexia where, um, adults, you know, who are fully literate, um, have a stroke and then, um, profoundly lose their ability to read, right? So, so that's a case where like, you know, there is this concrete disorder. But you think about anything that we characterize as a developmental disorder.
[00:08:59] Some people like the term disorder, some don't, but, um, you know, things that are typically characterized as developmental disorders almost always lie along a continuum. And my perspective is that it, it's really a policy question of where we want to draw the line. And you can think about this in a number of different ways. So yes, you have, um, a continuum of reading abilities and some children really have profound struggles. They need additional support, right? Um, what's the level on this continuum where, where we wanna say like, oh, this is typical, this is disordered? That is partially a, a question of, um, what sort of resources are we gonna allocate to those that we put on one side of this continuum?
[00:09:39] Um, so that's, you know, kind of a public policy or public health question. Um, and others like, what level of literacy skills do we need to get to, to be, uh, uh, I guess for lack of a better term, like a functioning member of society? To be able to be employable. And so, you know, these are actually questions that, um, you know, where should we draw this cut point, I think is much more of a, like, public policy, public health, uh, type question. What do we want our society to look like and what type of resources do we wanna allocate to getting it there? Then it is a, uh, basic scientific question, and I, I can tell you about the factors that contribute to reading challenge. I can tell you about how we measure them.
[00:10:19] Russ Altman: That's what, that's where I want to go next. So, so for, yes, so please 'cause I, because even me, not an expert, knows that it must have something to do with my understanding of my language, whatever my language or language is, are. And it must have to do with my visual system, and then I got nothing. So, so tell me, tell me what, how should we think about reading and the kind of the skills that the kids or the new readers have to come to the task with, and, and, and where they have to be developmentally on all of these different scales?
[00:10:47] Jason Yeatman: Yeah. I'll, I'll give you the conventional view, then I'll give you the, the place where my lab is, uh, contributed specifically. So the, you know, the conventional view is that, um, as you started with, is that, uh, you know, written language is a form of language, just like spoken language. And, uh, disorders of written language are really a, a language disorder that, um, there's a, you know, a lot of evidence for this that, uh, children who struggle learning to read first struggle with something called phonological awareness, which is
[00:11:14] Russ Altman: Phonological. Okay, good. What is that?
[00:11:16] Jason Yeatman: Yeah, phonological awareness. Thanks for slowing me down there, right? So, um, being able to take speech and decompose it into its fundamental elements, phonemes. So take the word cat and uh, Russ, um, what are the three sounds that make up cat, can you hear the first sound? What's the first sound there?
[00:11:32] Russ Altman: Ka.
[00:11:33] Jason Yeatman: There you go. And then what's the sound after that?
[00:11:35] Russ Altman: Ah.
[00:11:36] Jason Yeatman: And then where do we end?
[00:11:38] Russ Altman: Tuh.
[00:11:38] Jason Yeatman: Okay, now I’m gonna give you,
[00:11:40] Russ Altman: You can see I was a pre-med.
[00:11:41] Jason Yeatman: There you go. Yes. Three visual,
[00:11:43] Russ Altman: Eager to please.
[00:11:44] Jason Yeatman: I'm gonna give you three visual symbols to represent those sounds, right? So the, the, what I'm trying to get you to realize that, like, if, if that idea of breaking down speech into its phonemes just fundamentally didn't make, if you said, Jason, whatcha talking about, like what, what the, um, then the idea of saying, well now I'm gonna represent those phonemes with visual symbols, just like wouldn't fundamentally make sense, right? So kids that struggle with, um, this aspect of spoken language, it's really hard to learn to read, and that is kind of the most widely accepted, uh, causal factor in reading challenges and dyslexia. Um, and I would build on that and say, you know, that, that is one factor. And like almost all developmental disorders, you know, we, we scientists start off saying, well, what is the cause? Why do kids struggle? Thinking that we're gonna unlock, you know, the one reason. And there, there are a lot of scientists that said, would say, that what I just told you, that is the one reason, that's it.
[00:12:40] That, that tells you the story, but, and you look at the data and this, you know, explains, I don't know, half the variance, maybe a little less. Um, so if, if you are profoundly struggling with phonological awareness, you can be pretty confident that reading is gonna be a challenge. But you also have a lot of kids who have challenges reading that, that don't. And so this is where the second thing you brought up comes into play. Uh, and, and an area that my lab's done a lot of research, is looking at how individual differences in the visual system, when I say the visual system, I don't mean the eye per se, I mean your visual cortex, like parts of your brain that encode visual information. The individual differences in how your brain encode visual information can also, uh, confer risk for reading challenges. If you have less precision and visual representations or um, uh, like sluggish processing of visual information that, um, these are factors that we can measure and we have found, uh, contribute additional risk for challenges with reading.
[00:13:37] Russ Altman: So this is really fascinating. And for both of those two things, the phonological awareness, my new favorite phrase, rolls right off the tongue, uh, as well as, um, the visual system, do we understand, I don't know. I'm not even sure this is an important question, but it might be interesting. Do we understand how much of that is, uh, nature and how much of that is nurture, how much of it is genetic, and how much of it is your environment and what you're exposed to as a little kid?
[00:14:01] Jason Yeatman: That's a great question. I mean, so this is, um, now we're gonna get into some measurement and screening stuff, right? So, um, for almost every aspect of language, it's really hard to distinguish those factors because at, you know, um, the, uh, I mean, it's always both. I mean, that's the simple answer. It's like, basically any, any, is it nature, is it nurture debate, um, the answer is it's both, right? But, but the question is kind of is how much, right? How, how much is a, you know, a phonological awareness is a, you know, intrinsic trait of the child versus like the language environment and the schooling environment they're, um, they're, uh, brought into like, you know, the, the educational opportunities that society affords them.
[00:14:44] Um, and, um, so the answer is both. But one of the challenges there when we think about, um, assessment and screening is, um, trying to develop equitable measures that, you know, really are measuring something can be a persistent challenge as opposed to just measuring the lack of opportunity that, that, uh, our educational system has given to a child. And, um, that is more and more where I think non-language based measures, such as measures of visual processing, can really be useful. Those are ones that, uh, we have more evidence, um, are less, you know, are, are less affected by experience.
[00:15:23] Russ Altman: Right. That kind of makes sense. Whereas your environment, what your exposure to language and your exposure to words and, and people talking, that's gonna be very much a function of the situation you find yourself in with your family and your home. But everybody, unless you've been profoundly. Yeah.
[00:15:38] Jason Yeatman: Phonological awareness again is one, it, so it's, it's written into every dyslexia screening legislation and, you know, many parents know this is a foundational skill of reading, so, I, my youngest daughter is five and, um, you know, just kind of getting to that point where she's learning to read and we're already starting to play all kinds of games around phonological awareness and dissecting speech into a phoneme. So like when she enters kindergarten, how much of her strength or weakness in that skill is gonna be about her, or about the fact that like, this is our bedtime routine. So, we play phoneme games, and, um, uh, but when we think about like the, the speed and fidelity that your visual system can, uh, encode information, um, we, we tend to think at least at a young age, actually, there's, there's, um, evidence just like video game playing, other things actually, like improve, um, functions like visual spatial attention. So, you know almost everything about the brain is influenced by experience, but, um, when we think about screening for young kids, I think there's a lot of promise of adding visual measures on top of language-based measures.
[00:16:40] Russ Altman: Great. So I want to get to ROAR, you know very well what I mean, so tell me about ROAR and where it comes from.
[00:16:46] Jason Yeatman: Yeah, so ROAR, rapid online assessment of reading, um, has been a big transition in my lab's research. This started as a pandemic project, so my lab studies the neurobiology of reading, how, basically the, the factors that shape individual differences in learning how, how learning shapes brain development. And then came March of 2020. So every study we ever run always starts off with, um, someone in my lab sitting in a room, a small room, usually one-on-one with a kid, uh, administering assessments, often screening for dyslexia or, um, doing, using diagnostic assessments to, um, you know, assess whether the child qualifies, for example, for a study of kids with dyslexia.
[00:17:24] And as we all remember that era, suddenly we couldn't do that anymore. So, um, my lab had a long tradition of developing open-source software. A lot of that like kind of interplay between engineering and science has, uh, played out in, uh, widely used packages for processing brain imaging data. So now we're all at home and we started asking the question like, well, can we develop other ways to assess reading skills that don't require this one-on-one interactive approach? And so we started, you know, leveraging ideas from the cognitive science literature about tasks that tap into a core factor of reading. We developed like a first web application that's kind of a fun gamified, um, version of a task called lexical decision task. Very simple task. Flash text on the screen and, uh, your job just to say whether it's a real word or not. So, you know, you can do that with a swipe on an iPad, um, could do it on a phone, but we, we disabled phones for variety of reasons.
[00:18:16] Russ Altman: Swipe right if it's not a word
[00:18:18] Jason Yeatman: Exactly. That actually, there you go. So then we ran a first study where we had all these kids that came through the lab and we asked, you know, we created, we had this new hypothesis, right? This, just a scientific hypothesis about a new way to assess reading skills. Um, we have all this assessment data 'cause our lab had just shut down. We sent this web app home to all the families, sent just a link, URL, um, and then we linked these two types of data, the individual assessment data and the, you know, scores from this new web app. And, you know, it's not often in my scientific career that hypotheses are, are, have such a clear answer. So, um, the correlation between the two was point nine four, right? So this, you look at a scatter plot and this is one of those few ones where there's just a bunch of points along the line. Exactly, so that was the first version of our single word reading measure. And from there we went through,
[00:19:06] Russ Altman: And just, this is a technical question, but is that over all levels of kind of skill or efficiency? Or is it at the mostly on the kids who were struggling? Uh, or, or, or all of the above, or none of the above?
[00:19:17] Jason Yeatman: That's a great technical question. Yeah. And so that was one of the other, um, you're getting to like figure seven of the paper here. Um, so, uh, that was one of the other things we looked at is asking, you know, is this just about detecting kids who are struggling or, um, and basically along the whole reading continuum, it was equally precise.
[00:19:33] Russ Altman: So you're getting me my whole bell curve that I asked you about in the first question.
[00:19:37] Jason Yeatman: Exactly. Yeah. That we could, um, and so the first version of, of ROAR, single word recognition, was a long, you know, basically a long task, but it was automated so we could send the link home to a bunch of people, they could take it, we'd get the scores back. Um, and we went through kind of an iterative research and design process where we tried to make it more interactive and gamified so that we could bring it down to younger kids. And we ran another study and found that, um, even in kindergartners, um, which is a really challenging age, by the way, to think about like building an interactive web app, right? Because these kids are just learning their way through technologies. One, make sure it wasn't just sensitive to, you know, how much time had you spent on an iPad, but actually, so that was the kind of next series of studies.
[00:20:19] And then, um, we started building better computer adaptive testing algorithms that would like run, uh, client side in the browser. So basically this is all like running on, you know, on your computer, um, or on a Chromebook in a classroom. Um, and got to the point where, from our original study, to where ROAR single word reading is now it's about ten times faster. Where we can in three to four minutes through this computer adaptive testing approach, really zoom in on a kid's single word reading ability. Whether they are a kindergartner, first grader, second grader, you know, kind of at this early stage of learning to read or if they're a high school student. And you might or might not be surprised to know that actually there are a lot of high school students around the United States that are struggling across all their classes because they never mastered the foundations of reading.
[00:21:14] So, yeah. That's where ROAR has gone now. So we have this suite of measures that, um, that span the continuum from kindergarten up through twelfth grade, that look at different aspects of reading skills. We have, you know, a decoding measure, a single word reading measure. We have a measure of sentence reading efficiency. So this other thing kind of alluded to like how quickly and efficiently can you read for understanding. We have a measure, as you might guess, of phonological awareness. There's a whole suite of different measures that kind of all weave together through computer adaptive testing approaches.
[00:21:45] Russ Altman: This is The Future of Everything with Altman. More with Jason Yeatman next. Welcome back to The Future of Everything. I'm Russ Altman and we're talking to Jason Yeatman from Stanford University about learning to read. In the first segment, Jason told us about how reading goes across a bell curve in the population, and that the people who are at the slower end of the bell curve with respect to acquiring reading can be helped and can be assessed with his new tool called ROAR. In this segment, he'll tell us how he's pushing ROAR out into the world to help schools, kids, parents understand if there are problems that need to be addressed, and he'll talk about a new partnership he's creating, not through a company, but through an academic partnership where the readers, parents, schools can be participants in deciding what questions to ask and how to answer them in collaboration with Jason's lab.
[00:22:54] In this segment, I wanted to ask, ROAR sounds extremely exciting. This is the tool that your lab has built. It has many dimensions and can measure, measure many things. How actionable is this information? You mentioned two really important groups, really young kids who are just learning how to read, you know, around kindergarten. You also mentioned teenagers who might not have ever been, um, diagnosed previously. Can we do things to help these folks and, and what can we do?
[00:23:20] Jason Yeatman: Russ, let me ask you a question. Are you a glass that's half full or half empty kind of person? 'Cause I'll give you an answer.
[00:23:25] Russ Altman: Half full for sure.
[00:23:26] Jason Yeatman: You're half full. Okay. Okay. Well then this is, I, I think, you know, um, I would say reading interventions are kind of a solved problem scientifically. Yeah, so, so we know what to do. Like, uh, we like the broader we of like the, um, there are many exceptional evidence-based, highly effective intervention programs for students with dyslexia of all ages, you know, be that, uh, young students. So, right, the, the ideal, the ideal scenario is that you're screening kids when they're young. That, you know, right at the beginning of kindergarten, kids that need extra support are getting a little additional support. Some of those kids, that's all they need is a little additional support. Others need a more targeted intervention.
[00:24:07] Um, and you know, so ideally what you want early in elementary school is an effective, multi-tiered system of support where kids are kind of progressively getting more support as data shows they need it. And that's like one place where we work with ROAR, is trying to help schools build an effective, multi-tiered system of support. This goes all the way up through middle school and high school. And I, you know, I mentioned a lot of middle school and high school students are still struggling and, and that's the one where, you know, you can be half full or half empty on that point as well. Where on one level it's quite sad that many students are having struggles across their classes, right?
[00:24:43] Think about middle school and like math is, math is word problems. A lot of information is gained from text, right? Students are expected to be able to go home and read a book and analyze, you know, produce, um, you know, reason about it. So if you still are struggling to decode words then everything is a challenge. And you can say it is quite sad that, you know, a large portion of, of, um, upper school students are still struggling with something that's actionable. Or you can say, well this is great. You know, we could create an educational system that provides the resources schools need, and, um, really gives every learner what they need. So I, I stick on the optimistic side. But I'll, I'll let our listeners choose where they want to go.
[00:25:25] Russ Altman: Exactly. So, so just one little refinement there, question, follow up is, um, when you were talking about ROAR, you mentioned several tasks. I think one of them was single word recognition, but then you had other ones, phonological awareness. So that implies to me that you're gonna be able to make a somewhat precise diagnosis so that every, it's not that every kid will get exactly the same prescription, but that based on the results of the screening and the diagnosis, you'll know where to put the emphasis. Is that right?
[00:25:53] Jason Yeatman: Yeah, that's the exact idea, right? So, so, um, you know, data informed decision making can be used in a number of different ways, right? One is just this idea of a multi-tiered system of support, seeing which students just need more directed instruction. About fifty percent or so of kids, um, they're given a decent reading curriculum, like really kind of just make their way through it and don't have challenges. There's a, you know, we're on a continuum here, right? So there's a lot that like do need a little extra support, more, more guidance, more systematic and directed instruction. Then there's some that like really have profound struggles and need a more targeted, often when we say intervention program 'cause more than just, you know, a little more exposure to a general education curriculum, but more kind of targeted support developing skills.
[00:26:36] Russ Altman: Great. And so go, so going back to ROAR, rapid online specimen assessment of reading, I believe. Uh, this is a, is, is a game, my word's not yours, a game changer because now we can do this at home. It's easy. I'm gonna guess it's a lot cheaper, it's a lot easier for the family, it's easier for the kid. It's probably less stressful. I'm, I, I'm not paid by you. Uh, and I don't even know if this is true, but when you are now thinking about pushing it out to the world, does this happen in a company? Or does this happen with you as an academic working at Stanford? And how does one decide the best way to impact the world with this kind of tool?
[00:27:14] Jason Yeatman: That is our, uh, continual question. I'll give you my answer. I don't know if I'm right, but here, here's my answer. Um, so, so we've decided to scale ROAR through a research practice partnership model. We decided to keep this as a, um, as an organization within Stanford. We currently work with, uh, hundreds of schools around the country. I think we're now have been implemented in thirty-one states, maybe, I think it's actually thirty-two today. We just added on another state.
[00:27:40] Russ Altman: Congratulations.
[00:27:40] Jason Yeatman: You can look online and see our map of, uh, of which states have schools that have adopted ROAR. So we, we work through a research practice partnership model with schools all around the country. And what this means is that, um, teachers, superintendents, administrators, families, communities, students are all a part of shaping the core research questions and guiding the direction of the technology. So it's not just us in the lab saying, well, here's the pressing challenge that we're gonna solve, and now we're gonna, you know, bring this to society. Instead, the, the, the development of the research brings in the voices of the community. And when I say the community,
[00:28:15] Russ Altman: So partner, the partners with a capital P and they really are partners.
[00:28:18] Jason Yeatman: They really are partners. Yeah. And, and you know, to me it's really important that we hear the voices from around the country, right? And, and this is for research, and this profoundly shapes our research now where the conclusions we're drawing about reading developments, when I'm telling you specific things that the visual system does that are important for reading, this is not just an inference based on a couple hundred kids in Palo Alto who came into my lab. This is based on, you know, data from, at this point, thirty states, hopefully soon, fifty states. So, um, you know, our, our team, uh, would like to envision a new model for the role of academia and society.
[00:28:55] You know, a model where research really is a conversation where we are aligning our academic questions, you know, the core research questions we're addressing, are aligned to, uh, real world challenges that, you know, a teacher, a family, um, that can, can speak to. Um, and, and we're hearing their voices. And at the same time also, this is creating a pipeline where, uh, we hope to see deeper systemic connections between research and practice. You know, where it isn't a ten, twenty or never-ending year long cycle to get new innovations into schools, but that this is, you know, done in real time through these conversations.
[00:29:34] Russ Altman: But this does mean, so that, that sounds great and, and I'm sure you're aware of this, but this means that the communication channels from your laboratory to your participant, um, you know, uh, colleagues, you, you need to tell them what's up. You need to tell them what questions we're asking this week or this month, or this year. Um, you have to negotiate with them if their favorite question is not the one that is the one that you're decided to work on this month or this year. So I'm guessing that you have to build up a huge communications infrastructure. Uh, and how do you do that?
[00:30:05] Jason Yeatman: That's great question. Yeah, I mean, that's, you know, it's team science, right? So, uh, the composition of my lab has changed a lot over the, the years where we've been moving into this model, right? It, um, this type of work can't be done by, you know, a professor, a grad student, and a postdoc, right? So we have,
[00:30:22] Russ Altman: All of them have exactly the same training.
[00:30:24] Jason Yeatman: Yeah, exactly, right? And um, so, you know, there, there is, there's a lot of grad students who are actually, I think, kind of exciting for the grad students, right? So you have, you have grad students that are running randomized controlled trials, running experiments, um, with N of ten thousand participants. Which, when you think about, you know, for a grad student, we're often, like, you know, the years of grad school are building up to like recruiting the fifty kids that are, can come in for your study. Um, and they take that, you know, they, they, they take the responsibility there very seriously. See, they see the opportunity, but also the, um, uh, you know, the expectation to like, do things kind of at a different level. And we have an engineering team. So I think right now we have seven full-time engineers that are all work here at Stanford and are part of developing ROAR. 'Cause we have, you know, we have to maintain the highest level of data security standards.
[00:31:15] We're integrated with the, um, major student identification systems that are used around the US. So for a school, this often comes from school. A school wants to partner with us on a new research question. And you know, one question is like, well, do we have funding to actually tackle that? You know, I have to have like a postdoc that can lead the research, um, you know, postdoc or grad student to lead the research agenda. But often when there's alignment, you know, there's this exciting place where we can, um, Russ, if you said that your school district wants, is excited about our current research, uh, goals, um, and wanted to deploy ROAR to the twenty thousand students, I don't know, actually know where you live, um, but that are in your district. I'd say, uh, yeah, we can do that on Monday.
[00:31:54] Russ Altman: Wow. Wow. Okay, so in the last minute, um, ROAR is clearly a work in progress because I looked at, you know, I, I looked at your CV right before and there were papers about ROAR four years ago, and then you started adding little letters, you know, it was ROAR, then it was ROAR dash this and ROAR dash that. What is the future of, of the needs for assessment, which will then feed into these kinds of experiments that you just described?
[00:32:18] Jason Yeatman: Yeah, I mean it, you know, it's extensive 'cause there's many contributing factors that go into success, right? So we started off really thinking about, uh, dyslexia, which is a challenge in kind of the foundation of reading, accurate and efficient decoding of words, right? So that was our first, we measured, we're focused on this, but you know, of course reading and academic success is so much more, right? There's comprehension, there's vocabulary. So, um, we've expanded to kind of thinking about, um, to get nerdy for a second, the multidimensional space of, of reading development and the variety of factors that contribute at different points along the learning trajectory. Um, and then, you know, thinking about how these kind of combine together to set kids up for success or challenges and, you know, what can be done.
[00:33:05] So there's, um, you know, many questions. Some of them are basic academic questions about, you know, does this aspect of your visual system contribute to reading development or does learning to read actually influence this aspect of your visual? And then there's others that are more, you know, directly practical questions like, you know, what's the interplay between, um, the speed with which you read and like the depth of comprehension, right? So one level, they're all measurement questions. Can they come down to having high quality measures that can, uh, look at this, but then, you know, it's an, once again, an interplay between the academic research questions, the postdoc, professor, the student that wants to drive the research along with the school partners.
[00:33:44] Russ Altman: You know, I, I just gotta say, I, I have a, I have a six-year-old grandson who's doing two things for, this is from Grandpa's point of view. There's two things happening. He's learning how to read. And he is learning how to hit a baseball. And I can't help but wonder, are those things correlated? Like is, is success? And, but, but the point is, what I'm hearing you say is that you're looking at a much bigger now set of potential inputs into the eventual kind of performance of the reader. And, and that'll be done in partnership with all of these, um, new partners that you have across the country. Uh, uh, one other quick question. I, we're over time, but the, is the issues of reading that you see in English, do we, do we see that across all languages or is there, or is there differences in different languages? And I'm sorry to lay this on you so late, but I,
[00:34:30] Jason Yeatman: You got, you got the big question right at the end. So I'll try to give you the quick answer, but actually, funny enough, uh, right after this I'm going to teach and that's the topic of class today.
[00:34:38] Russ Altman: Alright. Awesome.
[00:34:39] Jason Yeatman: So, um, the quick answer is, yeah, I mean, reading, reading disabilities are, um, across all languages. There's some languages that they're more prevalent. The one that's most prevalent is English. Why is that? Because it's incredibly complicated. We have a very, uh, complicated orthography where there's not a one-to-one mapping between, um, letters and sounds. So that, that, you know, increases the probability that the kids are gonna struggle, but they're everywhere. Uh, the contributing mechanisms are broadly similar across languages. And one of the current like big research questions that actually been shaped by our partnerships is how to think about this for multilingual learners, right? How to think about the interplay between language development, reading development, development of the visual system, um, for kids coming with incredibly diverse linguistic backgrounds.
[00:35:27] Russ Altman: Fantastic. Well, thanks very much. You've been listening to The Future of Everything. Thanks to Jason Yeatman, that was the future of reading.
[00:35:34] Hey, before we conclude today, I'm excited to share with you that we're trying a new Q and A segment. If you wanna send us a question about one of the guests, something that came up, a question that you think others might have too, you can send it to TheFutureOfEverything, all one word, @Stanford.edu. Today's question comes to us from a listener, Sam Lewis, in Lancaster, Pennsylvania, and it's for Professor Renee Zhao, one of our recent guests. Let's take a listen.
[00:35:59] Sam Lewis: Hello, professor. Um, I was, I love the show. I'm a big fan, but I was intrigued by, uh, Dr. Renee Zhao, the guest that you had on with these, uh, micro, very micro robotics. I think that's an amazing idea. Just think about the hospital time that we could save, patient time. But I have a question though. Those robots, how smart is the AI? In other words, do they guide themselves? Say you wanna go after a blood clot or a uh, tumor for example. And so you have to program them to attack that particular enemy, you know, in a matter of speaking, that microbiotics. So can you do that on the outside? Do you control it? Uh, you know, with sensors, I would think if it's AI, um, you program it the correct way and off it goes. I'd love to hear about it though, but I appreciate it very much, professor. Thanks.
[00:36:53] Russ Altman: Sam, thanks again for listening and thanks for submitting a voice memo question. We had a chance to check in with Professor Zhao, and here's what she said. When it comes to controlling the mag spinner robot, AI will work alongside fluoroscopy or X-ray imaging, which helps identify the clot location and enables AI to program the robot's trajectory and its interaction with clots. Fluoroscopy will essentially be the vision for the robot and for the AI to help make this all possible. Thanks to Renee for providing that answer. I hope that helps to answer your question, Sam, and thanks again for sending it in. To everyone else, don't hesitate to send us a question. Again, it's TheFutureOfEverything all one word, @Stanford.edu. We'd love to know what's on your mind, what's really tickling your interest. And we're curious what conversations you want to have.
[00:37:46] Thank you for tuning into this episode. Don't forget, we have zillions of old episodes in our back catalogs, and you can spend hours and hours listening to them and learning about a wide variety of topics. If you're enjoying the show, please remember to share with your friends, family, colleagues, acquaintances, anybody who you like. Because personal recommendations are a great way to spread news about the show and help us grow. You can connect with me on many social media platforms. I'm @RBAltman, or sometimes @RussBAltman on LinkedIn, Threads, Bluesky and Mastodon. You can follow Stanford Engineering @StanfordSchoolOfEngineering or @StanfordENG.
