Toy talk with Pam Hadley

(upbeat music) - Welcome to See Here Speak Podcast Episode 53. In this episode, I speak with Pam Hadley about her work on early identification of developmental language disorder. And I mean early as in toddlers. She also shares her systematic research into an intervention called toy talk that teaches caregivers how to stimulate language skills, particularly at a time when young children are moving from producing single words to sentences. Finally, she discusses exciting new opportunities for artificial intelligence to improve early identification efforts. After listening to this episode, don't forget to check out the website, www.seeherespeakpodcast.com, to sign up for email alerts for new episodes and content. We did transcript of this podcast, access articles and resources that we discussed, and find more information about our guests. (upbeat music) - Welcome to See Here Speak Podcast. This is episode 53, and I'm excited to have Pam Hadley with me today, and I'll start by having Pam introduce herself. - Hi Tiffany, thank you so much for inviting me to your podcast. I'm a fan, and it's great to be here to talk with you today. I'm professor in head of speech and hearing science at the University of Illinois, and I also direct the applied psycho-anguistics lab. I am a clinically certified SLP, an early in my career. My interest was really in classroom and curriculum-based models of preschool language intervention, and over the years, as my research began to focus on early identification of toddlers at risk for DLD. My intervention interests have really shifted towards parent-implemented models of intervention. - Well, I'm excited for our listeners to hear more about your parent-focused intervention and just the toddler space, 'cause we haven't really done that on the podcast much. So it'll be a nice change of pace, but of course we talk a lot about school age, and before kids are in school, they're toddlers. So this is great, and we talk a lot about early intervention, and this is the epitome of early intervention. So I'll just jump right in and get started. So when we're thinking about toddlers, we're thinking about developing vocabulary words, and then they start to, you know, start to talk in sentences, which is such an impressive feat. So, what is a sentence-focused framework that you talk about, and why do you think a new framework was needed to guide assessment intervention with late-talking toddlers around sentences? - That's a great question. So, well, let's talk kind of where this framework came from. And so actually it grew out of a researcher, practitioner partnership that I was involved with years and years ago with Cindy Earle from the Hannon Center, and she was working on this verb intervention, and we were getting ready to do a presentation together. And so she was kind of talking about the sequence, and the PowerPoint had words, verbs, sentences, tens, right? "Tense marking" was a big deal at that point in time. And I remember looking at that PowerPoint, and I said, I don't think it's not vertical. It needs to be left to right. It needs to be developmental from younger to older. And as we were working on this, I pulled my husband in and said, what do you think? Do you think it's bullet points? Do you think it's horizontally said neither? It needs to be stair steps. - Yes. - And you have this great partnership with your husband who's in the field as well, so that's really great. - Yeah, so that really, this notion of stair steps or stepping stones, that's really kind of where it came about. And so that was the figure that we first published in that 2007 paper in seminars and speech and language. But it didn't really have a name. You know, I would go into class and I would, you know, have the students chant words verb sentences tense. (laughs) - Love it. - But, you know, and we could do the chant, but still trying to get people to understand the importance of sentences wasn't really clear in that space. And so I spent more time thinking about how sentences come in and then I started to rebrand this model as the sentence focused framework. 'Cause I really felt like people didn't understand what the role of verbs was in building sentence structure. They knew verbs that children with DLD had trouble with verbs, but how verbs built sentence structure wasn't really a area of strength. I don't think in our student preparation. So I started trying to talk more about sentences and how sentences come in. And so I rebranded it talking about words, verbs, childlike sentences, right? So this period where children are putting subjects and verbs together, you know, the SV and SVO sentences that we know about, but they're not adult like. And so that when tense and agreement come in, they take that childlike sentence and it now is adult like, but we really have to think about those basic building blocks and how we form a sentence in the first place. And so I think it was really when trying to help people understand that, you know, browns, grammatical morphemes aren't properties of words, they're properties of sentences. And so I really had to build this sentence focus to help people understand that. - Let's dig in a little bit more for our more novice listener. So you said, CV, tell us a little bit more about what the childlike sentence would look like versus one that's more adult like. So just some examples. - Sure, absolutely. So a childlike sentence would be something like, I dry it, where I dry it isn't really fleshed out like I'm drying the plate. Childlike sentence might be me go, where you see a pronoun case error, you know, the subject is a me form where it should be an I. A childlike sentence might be bird fly or it fly. And so they have that basic the noun phrase that's in the subject position, they have a lexical verb in that verb position, but they're missing those elements that make it adult like. - Oh, perfect, that's very helpful. And then when you talk about the browns morphemes, which of course I know as a speech language pathologist, but educators may not know, tell me, just could you give us a primer on like what, how many morphemes would be in some of those example sentences for our listeners? - Oh, sure. So like if you had a sentence like me go, that would just have two morphemes. So the me and the subject position, the go, that's the verb. I dry it, would have three, but as a child is starting to produce longer and more complex sentences, they might add that progressive I and G, like I drying it, and we would call that four morphemes, as the auxiliary form comes in, I am, which we contract a lot to I'm drying it, that would be like a five morphemes sentence. And so some of my concerns, but is that what we've focused on more often is how long children's sentences are. And one of the things that we know in our field that is, utterances get longer, they also tend to get more complex, right? That's how typical development works. But when we're talking about kids with language disorders, what we know is that utterance length is not a good predictor of their complexity, right? And so they're getting longer without being more complex. And so what we know is that length isn't really the best indicator. And that's kind of why we started looking more at diversity, because I think that as kids are better able to produce more different kinds of sentences, they're actually growing their grammar muscle, if you will, right? That makes total sense. And I want the key to us about some of the findings around sentence diversity. Sure. So in terms of trying to fill in this space about when do kids develop this ability to produce sentences? So one of the things I'm going to actually back up to other people's work rather than mine. So if you look at our textbooks, it will say that kids are really developing these subject verb or subject verb object sentences in sort of that 26, 27 month space. But the way in which we're showing that as an expectation is that we say in 100 utterance language sample, kids can produce at least two different SV sentences and two different SVO. But they could just be sentences like, "I want it, I did it." And one of the things that I worry about when I'm looking at toddlers is those could just be memorized expressions. I did it. Ta-da, I did it. Right. And we know that kids use that I want, and then they fill in the blank with what they want. And so that may not be that they may just be using kind of those high frequency formula and filling in that slide, and their grammar isn't as strong as we think it is when they can just put in long descriptions like I wanna green car, right? And then we think about those as like that general all purpose words too, right? Like they would use those that are more vacuous or more general. - Yeah, so I want the blue one. - Right. (laughs) - Right. So building those longer sentences that could be an example of still having a subject verb object sentence, but what I wanted to know is how many different kinds of subject verb sentences, how many different kinds of subject verb object sentences. And so we kind of went back into longitudinal database that we had to try to characterize when children were starting to, how many different sentences they could produce with different kinds of subjects and different kinds of verbs. And one of the things that is very obvious is that they start out with those eye sentences. So sentences that are about their own wants, needs, and actions. And so they can get lots of different sentences with only that eye subject. It takes a lot longer to get sentences with what I'd call another focused subject. In grammar, we call it a third person subject. Lots of times those third person subjects are it or he, their pronouns or that. And it takes a lot longer for them to start putting in nouns in that position. And so, but when we looked at our longitudinal data, we looked at kids from fast developers to slow developers who didn't have any concerns. But really, that boundary between slow typical and kids we might be concerned about. And all of those kids crossed that full range by 30 months of age were, had at least 10 different subject verb combinations in a 30 minute parent child sample. And of those 10 different kinds of combinations, at least two of that, they had at least two different third person subjects. - Mm-hmm. - So that's kind of what I've been using as sort of this by 30 months of age, kids should be able to produce those other focused kinds of sentences, not just sentences with I. - Well, that's so helpful clinically because we're always looking for something to hang our head on, right? Because we have the late talker kind of definitions, you know, of less than 50 words at two years and no two word combinations, which I think has changed now a little bit, right? Because of the CDC guidelines. - So it's been changed. There's been some talk of that. - Well, the CDC has, I hope to see that come, get revised back down because, you know, at 30 months of age, the average boy is producing about 200 different words based on CDI norms. The average little girl at 30 months is producing 300 words. So 50 words really should be at 24 months. - Yes. - An indicator of concern. And then for 30 months, I think this sort of absence of third person sentences does start to concern me. - Yeah, that makes good sense because we're looking for those markers of risk and that your work really feels the gap in that area between like late talker or 24 months to now I can give you more of a full assessment at like 48 months, maybe more of a standardized assessment and you're really getting that time period. I think it's so helpful and useful clinically. And I think the important thing is to recognize that you can produce a childlike sentence with only two words, right? It's all. - Yeah. - So instead, so the goal from a treatment perspective is instead of trying to make eye sentences longer, we need to create short, diverse sentences. - Very helpful. - Yeah, it's less about the length and more about the diversity. And you mentioned now this idea of what do we do in treatment 'cause of course, such a short gap between, we have assessment, what does assessment tell us and what do we do in treatment. And I know you really tackled both. Do you think this would be a good time to go into some of your intervention work? - Well, can I say just a little bit more about the assessments side of you? - Oh, I would love to. That's why I ask, I love assessments. I could talk about it for hour. That's why I always say, are you ready to move to intervention? 'Cause I could talk about assessment for a very long time. - Well, let's stay there for just a little bit 'cause I think the other thing I wanna say about the sentence-focused framework is that that the reason that I feel we need a different framework in the field is because brown stages, so in our textbooks, when we talk about stages of grammatical development, we go from brown stage one that's talking about these linear semantic relations. And then in brown stage two, we're suddenly talking about morphological development. But there's nothing that connects the linear semantic relations to the morphological development. Right, there's nothing causal between semantic relations and the morphology there. And then in stage three, we're back to different sentence types where we're talking about how a declarative sentence can be turned into a question or turned into negation. Again, nothing about morphological development leads to the different sentence types. - Right. - And so what we've tried to do with this framework is say, what are the early structures that are necessary to go to the next developmental step? So you have to have a vocabulary of single words before you can think about how to build those together. The, when we think about verbs, we're really thinking about how they combine with the object, so eat cookie versus bird fly, like a little subject verb combination, bird flying. And then we have to build the sentence on top of that. So eat cookie is just a verb combination, but I eat cookie or bird flying, those are childlike sentences. So you have to have some verbs to build the sentence. - Mm-hmm. - And then you have to have that subject verb framework in order to put the grammatical structures and flesh out a childlike sentence into the adult life. So that's really what we're trying to do is say, when we're trying to select treatment targets, how does learning that new structure build and hasten or speed up the next developmental accomplishment? - That makes sense to me. I feel like you're right. The stages are more like a fully formed kind of like, this is where they are, this is where they are, but you're talking about the ingredients to get to the stage. Like how do you bake those ingredients into your invention? So kids can make that next leap and that's not really clear from those stages. They're not, they're descriptively accurate and they haven't changed in 50, 60 years, but they don't help us in intervention. And I think the other thing I try to tell my students is that when Brown did his stages, he didn't intend for it to guide speech language pathology, right? That wasn't their purpose. And so the fact that it doesn't guide us in selecting treatment targets the way we might need to make therapy as efficient as possible isn't Brown's fault, right? That wasn't his purpose. So we need to develop our own framework so that we can make those expectations and make the selection of treatment targets really clear for the populations of these things. - Yeah, that just makes good sense. Building on the shoulders of giants there, giants, and then actually making it something we can turn into an actual treatment that has some evidence behind it. So now we're getting into the treatment part. And I'd love to hear, you know, how has this led to your treatment work? - Yeah, so I think you're setting me up to kind of talk about our toy talk strategies. And you know, I guess that's another story too because in fact, thinking about what kinds of sentences, well, let me back up. The thinking about sentences and pushing children into this third person or other focused space really came out of our early assessment work too. And so what we knew at the time was, you know, these grammatical structures that children at risk for DLD lacked. So we wanted to create opportunities to stand up and see what we can do. how the grammatical structures like that is or past tense ED or third person like Go Go's were coming in. Well, unless you push into this third person or other focus space, you can't actually assess how productive those structures are. And that's not really clean cut is it I always think of like you do a set I think a lot of people think like assessment then intervention, but they're basically the same coin with different sides right and they just buy direction that way so I'm glad you're telling you about this because I think it is important because I have that my own work to is it's not just a slumier process. No, no, and so so we really were teaching our research assistants how to create opportunities to get these structures right and so actually I guess why were we so focused on creating opportunities for these other focus sentences well. But we knew about how the grammatical more themes came in was the very first ones to come in is copula is so kind of like the dog is hungry right so a child like sentence there would be dog hungry where they lead that word is out or the third person singular form like the doggy wants food. So those two structures are the earliest to show up well interestingly they require a third person subject. So to look for the emergence of that grammatical system we had to create lots of opportunities for third person sentences right right and so. And interestingly we in order to see this right we would have 30 minutes of parent child interaction but most of the parent child interaction was in what I call that interpersonal I use space. Yes right so what parents are doing when they're interacting with their kids is asking a lot of questions do you want to are you gonna right or they're using imperatives put it there right which is socially appropriate right. Absolutely absolutely so this kind of descriptive talk is not what's happening in typical parent child interaction. And so we would go in and we would look at parent child interaction and then we would have the arrays come in and push the child into this third person other focus space to say okay so we see what you're doing with mom or dad. What can you do when we push you to this other discourse context as a way to really have them reveal the depth of their knowledge from the child like to adult makes total sense exactly because the question is is that they don't have the opportunity or is it that they can't go here and can't do this right right. And so what that showed us was that you know sometimes we weren't really seeing kids full linguistic capability until we pushed in this direction and then once we did boom we could see all of this third person sentence development showing up but not for all kids right. So then the question kind of so we learned that it was also really really interesting because so I had a thesis student who really really wanted to do an intervention project and so. And you know you know how short a master's program is it's like okay I don't know how we're going to do an intervention project in a short amount of time this is Katie Walsh and so I said okay well what if we tried to teach undergraduates how to push into this third person space. And so we kind of came up with this idea to you know we were going to record we took a video of parents and children playing and then we took away the sound and we asked the undergrads to talk like the parent would. And so and so we had these you know little tiny one and a half or maybe two and a half minute videos right so they would watch the video they would talk for the parent and then we did a very brief intervention and we taught them toy talk. Right toy talk that the emergence of toy talk I love it. And so where did toy talk come from the whole idea was in our intervention literature right we talk about self talk talk about what you're doing we talk about parallel talk talk about what the child is doing right but self talk gets you a lot of input parallel talk gives you you you're building a tower i'm building a tower. Well what we wanted to do was push them to talk about the toy right so we remember the day how do we call this toy talk because so first person I self talk second person you parallel talk third person space toy talk. Now the interesting thing right was that at the same time we were collecting these other language samples and teaching are is how to go in and gather language samples and i'll always remember the day that another student said to me if you had just told me to talk about the toys it would have been so easy but you're telling me to talk about third person subjects. Good point point well taken. Without requiring them to know anything about grammar yes it had we help parents do these things using the simplest language possible yes right and same thing with teachers. And and students right and so what Katie did was she would have the under read talk for the parent watching this silent video then she gave them a 20 minute coaching session right where she instructed them on what toy talk is she modeled it then she coached them on it reviewed it and then we played the videos again right so in one hour session we could do a. Pre test do the brief intervention post test and we got big changes in how many noun subjects the parents for the undergraduates were using the two strategies we taught them was talk about the toys and give the object its name right so talk about the toys gets you to talk about the tower but you can still say it's really tall. Yes right so when you're talking about the tower you could still refer to it as an it or that or a he right he's jumping mm hmm so talk about the toys and give the object its name now those two very simple strategies end up giving you nouns in the subject position. So powerful in such a short period time so since then now you're working with parents yes so then so following that that was sort of our feasibility study can we do this it was a little awkward right because there wasn't that natural back and forth piece and so sometimes they overtalked or they have these really long sentences but then that got us to an exploratory R21 where we actually taught parents how to do this. So again we started with so we focused on slow but typically developing toddlers because what we wanted to know first was if we teach parents how to use toy talk again in a brief kind of coaching model. What will the impact the on children but we but since we didn't really know whether this would facilitate language in typical kids we couldn't go straight to kids who are at risk because then we wouldn't know is it because the strategy doesn't work or is it because the learner is a disadvantage learner. So we focused on children who are developing slowly but typically we focused on kids at 21 months so they're about at the language level of our 30 month old kids that were so good right so they're essentially language matches yeah. We used basically a group instructional session first where we would bring in three parents at a time or so we would teach them about typical language development normal language milestones and then we would introduce at the very very end so maybe they got 30 minutes of toy talk and we did the same thing here's the information let's model it for you let's coach you a little bit in the group. And then then they then they came back with their child individually and we did two coaching sessions which wasn't very much because they would first come in we would record them interacting with their child. Then we would bring the video in and play back the video and stop it and you know talk about here is a great example what could you have done in this. situation right and that took so we replayed and watched and coached through video coaching for about 30 minutes and we did that twice. So it was a 90 minute group session the one hour coaching session that was a lot of you know really only about 30 minutes coaching. A second time was repeated and then that was it that's how brief it was. Wow. And then we followed those families at three month intervals so they came back at 24 months at 27 months and a 30 months. Okay. So all that coaching took place between 21 and 24 months. Okay. And you know we followed up to with some email suggestions. So here's what we talked about. Now here it is in writing. But in doing that what we wanted to see was we wanted to see parent change and then start to use more of these toy talk sentences. By the time the kids were 24 months of age but we didn't think that would have you know sudden impact for the kids or a development takes a long time yeah or grammar development does and then by 27 months of age what we wanted to see was that they were starting to use more of those third person sentences. And we had hoped that by 30 months of age we would start to see kids' sentences have an impact on their learning of those tense and agreement more themes and that's exactly what we found. Oh great that's fantastic. So that allowed us to be able to say okay when we let me emphasize how rare these toy talk sentences are not in a naturally occurring sample. Right so before that kind of intervention these are happening less than 3% of the time. Oh wow. Yeah it makes sense. Right they just aren't typical kinds of ways in which we're talking with kids. And even with the intervention we only shifted parents and intervention group up to about 10% of their sentences or toy talk sentences. So it's just a tiny little shift. Right and what was interesting was you know I didn't want them to do it all of the time and make the way they were talking feel unnatural right. So I would say just try to do it once or twice every minute. Sorry once or twice a minute. And what was so interesting was our average was 40 times in a 30 minute sample. Which is about once or twice a minute. Yeah it's perfect. Well and also you're saying it is a small percentage to keep it kind of like making social sense I guess but it wasn't if I heard you correct like almost three times more though. Yes that's true. Right so even though it's still small which makes it kind of socially appropriate it's still a big change relatively speaking. Right. Yes. Yeah and so what we so in terms of looking at the growth and change what we know is that the number of these toy talk sentences is in fact a significant predictor of how many different third person sentences the children are using. Right. Right so we we saw a lot of variability in the parents and so we didn't get a real clean intervention group effect but what we got was the individual effects right. You imagine yes because not all the parent you know it's it's a mediator model it's not we're not doing the intervention we can't control the dosage so it all depends on how much the parents do it. Right so you do the intervention portion we train them but ultimately it was how many times they used I guess it's almost like it's a fidelity right. Yeah. The fidelity to the treatment quote unquote that really predicted how well children did. Yeah and so and what's interesting or important to is that we're when we're looking at the growth in their sentence diversity we're controlling for their overall lexical diversity too so we control for how many different words they have before we look at how this tweak in in using third person sentences and specifically with nouns as subjects is growing that sentence diversity. And then what we think is happening what brings along the grammar is that when you have a strong sentence foundation right that's now you're able to kind of catch the grammatical elements in the input sentences because you got the strong sentence foundation to understand and make sense of the grammatical patterns that you're exposed to. And that that ties right back to in the beginning where you call it a framework because it's like it gives a frame. Yeah it is in there. We build a sentence frame and then you can put in all of the little extra elements. Right so you're moving from the sentence frame of a bit of the eye focus, the self-talk, the parallel talk, and now you're giving them the sentence frame which is right for subtle for jumping to the next stage to have more more themes and more tense marking. Exactly. And so that makes that then instead of building longer sentences we're saying make your sentences more diverse. Go from the first person to third person. Yes that makes yeah it makes a lot of sense. And this is so important you talked about your training parents how have you disseminated the work such important work for parents to know about but parents aren't really journal articles. I mean we barely have time to do that. How are you separating that to those that that want to know about this? Well it's great question. So first I'm talking to you. Yes, first. Which is it's really great to be able to have this platform to share some of that. Well I guess one of the first things I do a lot of presenting to to clinician groups and you know I really really love presenting to season to clinicians because they engage with this material in in such a great way and I get great questions when I do those kinds of clinical workshops that I go home and I chew on and really think about oh that's a great question. How am I going to solve that problem? But I guess you know one of the things that that we started doing actually was during COVID my youngest daughter got sent home from college and she has a lot more media skills than I do. So she started really helping with building a website for us and so I started out at during that time when we all had a little bit more downtime. I started trying to put all of the clinical kinds of features that have been written about this work in one place and so there's now a resources page that on our lab website that people can get to if they find it. Another thing you mentioned my husband Matt, Matt Rispoli, he recently retired and so what he's doing in his extra time now is he's doing a lot of YouTube videos and these are like three minute videos that are trying to explain these basic concepts and so he calls it I mean it's called sentence focus framework and so you can google that on YouTube and that should come up. We really are we feel really motivated to fill the gap in textbooks so you know if you just pull out an intro textbook and you look for sentence development it may not even maybe a sentence on sentence development right and you know right there so it's like we go from vocabulary to grammatical morphology without really understanding or building this bridge on how do you build sentence structure and so we're that's what we're really trying to do is give give people a way to think about sentences in a way that we really haven't given it much attention and I think I think it's also interesting because the fact that our textbooks especially in our field don't really spend that much time talking about sentences tells you something about how effortlessly typically developing kids go through this phase. That's a good point such a critical juncture for children who struggle with language development. We have the development language disorder and other difficulties but for children who are typically developing you're right it seems it's it's magical in many ways right it just seems like it's there. Yep and and you know if you talk to typical language researchers you know we talk about the magical third year of life right because what happens is just between two and three kids go through this rapid acceleration but it's not a rapid acceleration for our kids and where is that that they're stumbling they're stumbling in getting in assembling sentences right and so if you go back and think about you know what are those those difficulties that kid that young kids at risk for DLD have they have trouble with verbs okay that's the core for how we build a sentence they have difficulty with pronouns subject pronouns in particular that's the building block of sentences they have trouble with tensing agreement that's how we get from childlike sentences to adults like sentences. And all of that is there before they start trying to produce complex sentences, but we don't spend any time on simple sentences. You know, it's so interesting though, because I often think about when you're, you know, you're working with clinicians and you're trying to train, you know, graduate students, undergraduate students, educators, like what does DLD look like, right? And if you work with seasoned clinicians, they can tell you, oh, I know it very quickly, in a young child, when I'm concerned, I do a quick deal to play with them and they'll say, I know. And when you ask why, it's exactly these things you said, they're looking at, maybe intuitively, maybe unconsciously, a lot of times, maybe consciously, they're looking at this idea of how much tends do they have? What is their diversity? You know, what, it's just, you're putting it so clearly articulated and into this framework and then you've translated from something assessment, which we know we need, but then to the intervention part in such a critical period that we talk so much about early language development, but to have something to do about it is so, so important. And I love how you talked us through the story behind it, because there's always the story behind how it develops. And I love how you have, you know, you talked about getting this grant, the R21, which is an exploratory grant from NIH, where you can set the foundation for bigger work. And now you have the clinical trial to evaluate this in the most rigorous way of, you know, having the study and some children who are, I assume not, maybe you can tell me a little bit more about your clinical trial. You know, who are the participants? What are you doing? And then about some of the highlights and challenges of that clinical trial. Well, so we, we start, this is a clinical trial that is in collaboration with Anne Kaiser, Advanced Rebuild, and Megan Roberts at Northwestern. And it's real, so the intervention is called EMTSF, so for enhanced milieu teaching, sentence focused. And so what we've done is we've taken their teach model coach review process that really teaches parents the methods for how to use naturalistic language facilitation strategies and really focuses on responsivity and matching balance turns in early vocabulary. But the vocabulary targets that we're really focusing on are really building the word classes that we need to form sentences. So in addition to the core noun vocabulary that we always think about, we also have target words that are prepositions, adjectives, and the verbs. But we actually are really building those adjectives, what we call descriptive words when we talk with the parents, and then the prepositions which we talk about is location words when we are teaching the parents. So that's the vocabulary foundation. And then it transitions into really using toy talk to build the sentence foundation. The trial is designed for kids who have, are at risk, toddlers who are at risk for DLD, they need to have low comprehension when they come in. So your low comprehenders at a younger age, the tricky part is that we don't have good assessment tools for picking, you know, for measuring comprehension in this period. And so right now the kids, I guess a really interesting design element also a challenge is that all of the kids come into the study exactly at 30 months of age. And so that's really kind of a characteristic of how we've done our work trying to look at what typical trajectories are starting at the same age point and who is growing rapidly versus who is growing slowly. It's made it really hard in terms of recruiting, but the design when we're done is going to be so strong because we're going to be able to control age and say how are kids accelerating or not. Right, which in contrast is like the randomized control trials I do, we typically start at a certain grade level and they're just a range of ages and we try to wash that out, you know, in our analyses or just even just acknowledging that's variability that's going to be there. But you can't really do that in your design, but I can imagine that's so tricky because when I enroll, it's like I enroll a cadre each year and you're having to enroll each kid, kids are enrolled at all times. And the thing with the, with most late Tucker interventions, right, is that if you look at those studies, sometimes kids will come in, you know, in that 2021 month range and we know a lot of those kids excel, you know, are late bloomers. And so now was it the intervention or was it the fact that they really were to at risk at all, right? We're just, right, a late bloomer. Yeah, and so by waiting until 30 months, we're trying to wash out the late bloomers, right, and truly get the kids who really, really have a DLD profile or are at risk at that point in time. But, but so when you look at the late Tucker studies, a lot of times you'll pull in kids anywhere between 18 and 30 months of age. And so that when you ask who's responding to the intervention or not, you've got this huge age effect that we know about. So anyway, right now we've got, so the intervention is in three phases. The first phase is a vocabulary phase where the parents are really learning the basics of EMT. The second phase is really this toy talks switching to sentence focused targets. And that's a six month phase. And then the third phase begins when the kids are 42 months of age. And we shift our intervention towards a once a month kind of parent education model. And we focus on the transition to decontextualize talk. And so vocabulary sentences decontextualize talk. Notice that we're not actually targeting grammatical morphology at any of these places. So one of our questions is to what extent if we build the sentence structure in this way, can we actually get generalization to the other grammatical targets? And I just want to back up with a woman and have you defined decontextualize talk for our listeners that might not know. Yeah. So decontextualize talk is how easily kids, it's talking about things in the not here and now. So the in our textbooks, we might talk about it as the there and then. So outside of the immediate context and not in this time or this exact moment. And why are we interested in that? Well, one of my colleagues here, Cynthia Johnson was on one of my master's thesis committees. And she says, Pam, what you're doing with toy talk is just a form of contextualized literate language. And I'm like, yes, you're exactly right. It's contextualized literate language. Right. And so what I've always been curious about is if we can help kids learn how to talk about the objects in the environment in the here and now, will they be better at being able to talk about objects and events that are not present when they make that transition to preschool? As such a great segue to thinking about children's abilities to tell stories, which we know is so late to future outcomes for reading comprehension, listening comprehension, academic knowledge. It's just that whole bridge of development that you're setting that foundation to really take that leap into these contextualized language. And so, you know, if you can't talk about, so if you think about even our science lessons, right? So we're talking about does the boat sink or float? Well, if we want them to go in and talk about objects and make comparisons about them, you've got to learn how to talk about them first. Yeah, absolutely. Absolutely. So we're and the other thing I think we're excited about is that, you know, when kids go into preschool, parents kind of get pulled out of the intervention context. And so what we're trying to do is start at this boundary between part C services and moving into school and keep the parents involved by teaching them these strategies early, but then helping them make that transition and see how they can stay involved and learn more. Yeah, that's such an important part because going from early childhood, you know, or birth to three, any type of intervention, it's really all about the parents. And then when they go to preschool, you're right. It's like now the kids with a different language simulator that the teacher, most of the time, and their parents, and then well, then they go to school and they're really kind of out. So it's great that you're bridging that gap to have them still be a strong supporter of their language development, even as they move into early childhood education. So we're really we're excited because we have these kids for 18 months, right? They come in at 30, they exit the study at 48 months, we have language assessments on them every three months and standardized measures on them at 30 months when they come in at 36. We have vocabulary measures because we want to see vocabulary growth change. We have grammatical measures on them at 42 months and then we have the whole battery in our primary out endpoint at 48 months. So it's going to be, we have over 100 kids enrolled at this point. And so we're about to enroll our last kids. But they'll be with us for another 18 months. Right. Yeah. The long period. Yeah. It's a lot. The hardest part for me in running randomized control trials is you have to wait so long to see what you find out. And a lot of times you're not even the ones, you know, reviewing the data because you're blinded and the randomized control trial and safeguard against bias. But so I assume you're you're rating for findings right now. So you're kind of in that mode or like, okay, you're enrolling, you're rating. But what are some of the kind of, you know, challenges you found and highlights from the study we talked about a few already? Yeah. So, well, certainly the fact that we started the trial before COVID and had to pivot to tell you know, it's been, you know, it was certainly a challenge. The blessing in it is that we've been able to recruit a national sample because, you know, and so, so that wasn't originally planned, but that's been been great. I think, you know, we're going to have so much to learn about how we're using the PLS as our eligibility assessment. And there's such a range of kids' abilities who all have the same score on the PLS, you know, they all have you. So it'll be really interesting to kind of pull that apart and have this much data. I think one of the, you know, I, we don't know who's, you know, when we're looking at the progress data, my lab actually is doing all of that data processing and they're all naive. I work with the intervention team, so I don't get to touch any of that data, but I think, you know, it's, but every once in a while, I, you know, without knowing who they're talking about, you know, I'll hear them comment on, you know, oh my goodness, this kid's just killing it at, you know, 42 months of age. So I can't wait to get in and really be able to look at the range of responses to the intervention. Yeah. That's great. I know we have the same thing where we have a randomized control trial where we don't, you know, we don't analyze the data at our site, but we pull the kids out from their classroom and then the teacher will say, I can just see so much progress. You're like, I hope you are, you know, but you don't know. You just see it. I really hope you're in that intervention. Yeah, exactly. You just, you hope for the best. You're putting so much time and effort, but the truth is, even if there's not the expected finding, I mean, you have so much rich data that you won't only have treatment data, you have so much assessment data. Like you said, that you're going to, I'm sure have some really interesting assessment papers to dig deeper on that front too. Well, speaking of projects, you have a new project I read about it. It was advertised, got a lot of press, you know, because it involves artificial intelligence, which is all the rage. So I was excited to see it. You have a new project, artificial intelligence, the scale of access to preschool speech and language screenings and to provide ability based interventions for school age children. And I would love to hear a little bit about that project. Oh, you know, let's just talk about leaning in and stepping out of your comfort zone. It's a dinosaur like me involved in an AI project. But it's, so, so the overarching project is really about trying to develop use inspired tools that will be useful to SLPs in our public schools. And so what an important project to be involved with if I can make anything useful for practitioners given the caseload sizes and, you know, the burdens in terms of reporting and progress monitoring and some of those time consuming things that will be a huge, huge, it's a great opportunity, right? But the two tools that are envisioned feels kind of futuristic at times, but you know, dream big. One is an AI screener. And that's really what I'm involved with. And the goal is really to make it so that we can pull authentic language sampling into the universal screening process. And, you know, so I mean, I've lived my life in language samples and that's something that we just don't have time to do in our day to day kinds of screening processes. And so I'm really a big piece of this is developing automatic speech recognition that can go down from adult into children into preschool. And so that will be kind of that process. And we envision this as something that will be able, you know, by constraining the kinds of sentences that we expect kids to be producing in the screener that we can get the ASR to recognize not only children's speech, but ideally children with speech and language disorders. And then do automated kinds of analyses. And so, you know, the engineers and the computer scientists are envisioning that not only will we be able to do this in quiet, but we'll be able to push it into a classroom and get, you know, speech samples in an authentic kind of environment. And if we can do that, wow. Wow. Well, that's just amazing. I'm picturing like a Siri type that says, you know, child ABC is at risk. But they're imagining that that dashboard that will be able to separate out, you know, child A from child B. And, you know, be able to collect critical snippets of, you know, here's a little piece of interaction that you want to review. But I think what I'm most excited about is that that the other thing that AI is going to be, I hope, really useful for capturing, is the things that we we're kind of leaving on the table right now. And that's the effort involved in producing a sentence. Right. So we all perceive the hesitations, the filled and silent pauses, the repetitions, but we're not using that data. And yet that's the difference between a child who's got strong language skills and weak language skills. And if we can help AI, if AI can help us quantify effort. I think it's a may, you know, it doesn't feel that far fetch for me when I moved here 10 years ago, one of our students was a linguistics major at MIT who worked on this kind of, what was machine learning at the time. Wasn't AI was machine learning, which is, you know, kind of cousin there. And she came over and she looked at my speech samples and she did do an analysis with machine learning and just looked for pauses. And the pauses alone, as you said, the pauses alone were pretty high in predicting our samples. Restricted, there's a lot of limitations. She did publish that paper in, I think it was like machine learning something or maybe she maybe it was in our speech, enter inner speech, which is a conference about, you know, kind of pulls in machine learning computer, all these things. But then you know what she got and I was so excited, I was like, oh, this could be something cool, but she got scooped up by Google. We took her away. She's working happily for Google now. But it was just, I was so skeptical in the beginning, it was just a short master's thesis. That's what it was, just a short thesis. And I couldn't believe that work. And a lot of it was based on the speech sound science work that Jordan Green has done and others. Because a lot of this work is so similar to what's being done in Parkinson's, ALS, like early detection of dementia. I think that we as child language researchers could really learn from those efforts that sound like you are, which is so great and you're working with the right people. So I'm so excited to hear about this. I think it, you know, well, now I feel like before chat, GBT, the AI, I feel like people would have thought like, oh my gosh, no way. But now I feel like people are like, oh yeah, that's fine. And you know, so the other side of the other tool that's envisioned is this AI orchestrator. And, you know, I mean, I'm sitting around with colleagues who are talking about, well, you know, but why can't we just have AI generate the pictures for our tick, you know, like generate a list of, right, which shouldn't be hard. They just have to help us with the prompt, right? Or, you know, so how can it generate therapy materials? How, think about, it could generate a picture description task that we could use as a probe, right, for pre-test and post-test that have comparable complexity. Amazing. I mean, it could help us generate social stories. You know, so you start thinking about all the things it could do to make our lives easier. - I don't know. But if we can crack the ASR nut too, then can we have AI help us with progress monitoring or being able to tell us what level of support we need or whether a child is modifiable in the course of therapy. So, you know, drinking. - That speech recognition work is amazing. And that is gonna be such an important part of it. - Well, I'm mindful of time and I love what we're sharing with all the listeners here from assessment intervention to, somehow, you know, now you're telling me about AI. (laughs) But we're just going, all gamut. I will link the YouTube videos as well into the show notes. So, listeners can grab those easily and all the resources we've discussed. But my final questions are always, the first one is, what are you working on now that you're most excited about? So, I'm gonna say that it is a new assessment tool with Wendy Crock. We call it the sentence diversity priming task. And we also fondly refer to it as serendipity, SDPT. (laughs) Awesome. And that's because, in fact, on another project with late talkers, a colleague challenged me to figure out how to create a remote assessment. So that we didn't have to bring the kids into the lab for every, you know, to look at growth. And I knew that Wendy had developed these awesome animations as part of her dissertation. Her dissertation was focused on a priming task to see how kids could learn that auxiliary is structure. But she had designed it with 24 different subjects and 24 different verbs. And I was like, we could just morph this into a great structured protocol. And so we've been working together on that. She's been advancing that through a foundation grant. But that's what we're using to gather the structured probes now for the clinical trial. So that's what we have in looking at these kids over seven different measurement points. And that's actually what we're going to be using to look at the effort in this AI project. And she's actually working right now to, you know, get that put on a website so that we can make it available to others as we keep moving it forward. So I am so excited about it because in such a short amount of time we can get diverse sentences. We can see individual differences in kids. So some of the kids can just do a single word noun. Some of the kids are doing, you know, verb phrases. Some of the kids are doing childlike sentences. Some of them are producing adult like sentences effortlessly and some of them struggle. And we can see it all on this little task. And it's so efficient compared to a language sample. And so that's great. And so accessible, like you said, because it's you can do it online. You can just, it's quick to access. I can't wait to see that. It's feeling such an important gap in the. So what we did was we developed it kind of like as a parent child shared book reading. So if we're zooming into the home, right? The parent is actually, you know, looking at they do a model. So they model what's happening. Then they see another animation. Say what's happening here? If the child can't answer it, they say, tell me about the bunny. And so you get this nice little dynamic assessment. That's amazing. I have to tell you that that's one silver lining, as we know from the pandemic, is this pushing into assessment, tell the health. And even I see my doctor sometimes in the clinic sometimes, you know, it's easy to just jump on zoom. It's so nice. I remember before the pandemic having a meeting with, this has been a bear-sing true story. Meeting with the local researcher asking about tasks she has. And she says, oh, I do it online with parents. And I was like, what? I just remember being kind of dumbfounded. Like, what do you mean? Like, how do you even know their, are they paying attention? I just remember being totally dumbfounded by this. And then I remember when she left the meeting, I was like, we're not going to do that task. I don't know. I can't even imagine. I don't know what's how to do that. And then fast forward, like, two months, the pandemic hits. And I'm like, tell my lab manager, get her on the phone right now. Like, it's a doctor. Now we need to do everything she's doing. And it was so helpful, because she did meet with me. And we did talk, and it was great. But now everyone's really pushing that forward. It's so accessible. For most people, we still deal with accessibility issues around internet and such. But it's amazing. And you're really, you're leaning into that so much with your clinical trial and the assessment. It's fantastic to get that work out of there. Wow. I'm super excited. We move forward. AI. I'm just going to keep a search that says, like, Pam, happy AI. I'll just know everything that's happening right away. I'll be so high tech. Okay. So the last question I asked is, what is your favorite book from childhood or now? All right. So, I'm going to say corduroy. Do you have a corduroy? Love corduroy. Love corduroy. And, you know, I think it's because, you know, maybe I was just a child of Mr. Rogers, but that theme of, you know, of liking someone just the way they are is just the central message of corduroy, right? And so I still, like, I still tear up when I read that to my kids when they were little. My grand nieces and nephews. Especially especially. I love that one too. Oh, Pam, I do. It feels like yesterday. And I know you have little ones and there aren't little anymore, but I'm sure you just think about, I don't know. I agree with you. Corduroy is so special. I can't believe no one's mentioned corduroy before because I don't think I've ever had a guest mentioned corduroy. And if you such an important one, yeah, I, you know, I just, when I, yeah, I'm just very sentimental about it. I don't know if there's an early childhood memory of cuddling up and reading about it and thinking about it, but always, always a favorite. Oh, that's great. Oh, well, I'm glad you said that because that's a big one. I'm lying too. Well, Pam, thank you so much. I really appreciate your time in getting this information out to the listeners. Like I said, I'll get all the resources in the show notes and I'll just be looking forward to the results that you'll have in the next several years. Well, thank you so much for inviting me on. It's been such a privilege to be here with you today. Check out www.seahearspeakpodcast.com for helpful resources associated with this podcast, including, for example, the podcast transcript, research articles, and speaker bios. You can also sign up for email alerts on the website or subscribe to the podcast on Apple podcast or the other listening platform. So you can be the first to hear about new episodes. Thank you for listening and good luck to you. Making the world a better place by helping one child at a time.