WEBVTT
00:00:09.029 --> 00:00:10.679
Oh, it is recording, I see the little figure.
00:00:10.710 --> 00:00:11.130
Okay, great.
00:00:11.131 --> 00:00:14.130
I will do my little spiel and then I'll introduce you.
00:00:13.801 --> 00:00:14.669
Nice, okay.
00:00:16.280 --> 00:00:21.050
Hi, I'm Sue from the Salvesen Mindroom Research Center at the University of Edinburgh.
00:00:21.710 --> 00:00:25.219
Um, I'm recording another PsychologiCALL episode today.
00:00:25.579 --> 00:00:38.600
This is our little podcast, we're aiming to make a bit of evidence-based contribution to the conversations that people are having a lot at the moment about child and adolescent wellbeing and development and learning.
00:00:39.229 --> 00:00:52.369
And today I am very delighted to be talking to Courtenay Norbury, who is a professor at University College, London, and a visiting professor at University of Oslo, and Courtenay is also a speech and language therapist.
00:00:52.789 --> 00:01:00.859
And she's going to be talking to me about the SCALES study, which stands for the Surrey Communication and Language in Education Study.
00:01:01.460 --> 00:01:03.560
So hello, Courtenay, how are you today?
00:01:04.159 --> 00:01:06.920
Hello, Sue, I'm really good, thank you, how are you?
00:01:07.640 --> 00:01:14.180
I am not too shabby at all, thank you very much- I'm wearing a glittery top and there's some snow in my garden from last night.
00:01:15.170 --> 00:01:16.159
Sounds wonderful.
00:01:17.450 --> 00:01:22.549
Um, so tell me what are the kind of headline things that you've been discovering in the SCALES study?
00:01:24.290 --> 00:01:29.510
Okay, so I think one of the main findings is that language is incredibly stable.
00:01:29.930 --> 00:01:52.489
And what this means is that if we were to take a whole year group of children and give them a language test and then line them up in the rank order of the scores that they get on that test and then test them on the same test every year, that rank order pretty much doesn't change from the beginning of school, primary school, to the end of primary school.
00:01:53.719 --> 00:01:59.000
So that's quite striking, but that doesn't mean that children don't change.
00:01:59.269 --> 00:02:06.319
So in the SCALES sample we've demonstrated that actually all children increase their language knowledge over time.
00:02:06.680 --> 00:02:16.580
And the exciting, but also head-scratching bit for me is that the rate of language change seems to be similar across that distribution.
00:02:17.120 --> 00:02:32.810
And what that means is that even those children who had the most severe language deficits when they started school- and many of them have other developmental challenges as well- made as much language progress as their peers who were doing really well when they came into school.
00:02:33.530 --> 00:02:42.199
And this is really, to me, really exciting and in my view demonstrates that universal education is a pretty awesome intervention.
00:02:43.340 --> 00:02:58.460
But the downside, the challenge is that that means that any gaps that were there when they started school really persist all the way through, and in our sample it's about a two year gap between the most able and the ones that have the most challenges.
00:03:00.580 --> 00:03:06.490
Well, that's, that is fascinating because it definitely goes against I think what I would predict.
00:03:07.360 --> 00:03:18.759
Um, but before we dig into that result in a bit more detail perhaps we should just turn back time a little bit and you could tell us a bit about what motivated you to set up the SCALES study.
00:03:19.539 --> 00:03:27.340
You have to turn back time quite a lot, Sue[laughs] cause I've been working on SCALES for a decade now, can you believe that?
00:03:28.210 --> 00:03:48.250
Right, so when I started in this field as a speech and language therapist, a million years ago, we talked about specific language impairment and as a clinician I always found that really puzzling because when you see a child who has, has language impairment, it's rarely the only thing that kids, those kids find challenging.
00:03:49.030 --> 00:04:03.460
Um, and as I started my PhD and started to learn more about developmental processes it seemed to me quite unlikely that language would be the only affected system when a child was not developing as we might expect them to.
00:04:03.131 --> 00:04:12.849
So, but I had to be kind of wary of the fact that, you know, most of the kids that I would see were coming through a clinic.
00:04:13.300 --> 00:04:25.540
And so what we call co-morbidity or these co-occurring challenges that kids have, um, might be elevated because the more things you've got going on, the more likely you are to get to a clinic.
00:04:26.050 --> 00:04:34.000
So what I thought we really needed was to look at, um, language profiles in the general population.
00:04:34.870 --> 00:04:37.420
Um, and so that's, that's how SCALES was born.
00:04:38.170 --> 00:04:52.029
Uh, and also we didn't, we didn't have any UK prevalence data about language disorder and the existing prevalence data, um, been collected by the very wonderful Bruce Tomblin, but more than 20 years ago and in Iowa.
00:04:52.569 --> 00:05:11.860
So I really thought that it would be useful to know how many children were starting school in the UK with lower than expected levels of language, what other developmental challenges they had at the time, and then how those things play out over time and what impact that has on other aspects of development.
00:05:14.680 --> 00:05:18.610
And so, so tell us a bit about the method that you're using in SCALES.
00:05:18.620 --> 00:05:23.500
This is a longitudinal study- you're following up the same children over time, is that right?
00:05:23.529 --> 00:05:28.540
Yep, it is a, it is a longitudinal study- very longitudinal study.
00:05:29.339 --> 00:05:35.949
So what we did is we, um, we have very great partnership, uh, in Surrey county council.
00:05:35.951 --> 00:05:41.470
So they were very willing to, uh, let us do this and, and really helped us do it.
00:05:41.471 --> 00:05:43.240
We wouldn't have been able to do it without them.
00:05:44.009 --> 00:05:58.790
And what we did was we, we tried to get every state maintained school in Surrey that had a reception class to take part, and just over 60% of all of those schools did- so we had about 180 schools taking part in all.
00:05:59.149 --> 00:05:59.839
That's amazing.
00:06:00.529 --> 00:06:11.839
[laughs] What we did was we bought reception class teachers out of teaching for a day and made them sit down and do some online questionnaires for every child in the class for us.
00:06:11.870 --> 00:06:14.480
So, bless them, they are amazing people.
00:06:14.899 --> 00:06:28.730
And in these questionnaires, we asked about, uh, the child's language and communication, their behaviour...um, and then we have the early years foundation stage profile which is, um, early learning attainment goals.
00:06:29.810 --> 00:06:42.290
So we started this in, it was children who started a mainstream reception classroom in 2011 and we had screening data for over 7,000 children.
00:06:43.220 --> 00:06:49.279
Um, and then we took, uh, a subsample of that big group and we followed them up over time.
00:06:49.281 --> 00:06:51.980
So we started off with about 600 children.
00:06:52.339 --> 00:07:04.699
We saw them in year one, year three, year six and we tried to see them last year when they were in year eight, uh, but COVID then decided to make that particularly challenging for us.
00:07:05.089 --> 00:07:07.639
So we were only able to see about half of them.
00:07:08.209 --> 00:07:19.819
Uh, but we do have, um, data on their language, their reading, their problem-solving skills, their social skills, um, and more recently aspects of their mental health.
00:07:19.821 --> 00:07:22.459
So it's a really rich data set.
00:07:24.589 --> 00:07:32.420
And so for those follow- ups your- this isn't just teacher report now, this is more face-to-face work as well with the kids.
00:07:32.810 --> 00:07:47.329
Oh yeah it is, um, so we had a whole army of wonderful research assistants who would go into schools and they would assess children directly on standardised tests of language and cognitive ability.
00:07:47.689 --> 00:08:02.689
We had child questionnaires- they tell us about their experiences of school, we did experimental tasks on their recognition of emotion or their executive control skills, so a whole range of things.
00:08:03.680 --> 00:08:06.019
Um, yeah, very rich dataset.
00:08:06.620 --> 00:08:08.839
Yeah, it's amazing, it's completely amazing.
00:08:09.620 --> 00:08:17.480
Um, so I can imagine that the, I mean, the options, the things to explore in that are absolutely, you know, there are so many.
00:08:17.480 --> 00:08:26.930
But if we think about this, this, um, finding that you flagged at the beginning about sort of stability of language profile across development, right?
00:08:27.230 --> 00:08:37.460
Could you tell us a bit about how you analysed that stability, what, what kind of techniques you were using to, to chart that change or lack of change?
00:08:38.299 --> 00:08:48.950
It's, it's very complicated and I am eternally grateful that Andrew Pickles is a collaborator on the branch because he is possibly the world's greatest statistician.
00:08:49.820 --> 00:09:05.909
Um, so one of the reasons that it's complicated is because, uh, you know, we have the smaller sample of, of 500 or so kids that we've seen all the way through, but we want to use that sample to say something about the bigger population that they came from.
00:09:06.659 --> 00:09:11.370
Um, and there are statistical techniques that you can use to do this.
00:09:12.330 --> 00:09:18.299
Um, so one of the ways we started off was to use weights, and that takes account of our study design.
00:09:18.301 --> 00:09:31.019
So for example, in our study, um, teachers rate boys as having more language difficulties than girls, but we recruit, we brought into our smaller sample, equal numbers of boys and girls.
00:09:31.500 --> 00:09:40.769
So you can use a weight to kind of say, well, the boy data should count for more because there are more boys with this profile in the population.
00:09:41.429 --> 00:09:41.700
Right, right, right.
00:09:41.940 --> 00:09:58.559
Um, but this becomes trickier to use when you've got longitudinal data because as you go along you not only have to account for the initial study design but you also have to account for why people drop out of the study, because people always do drop out of the study.
00:09:59.460 --> 00:10:20.309
Um, and so that's, that's where Andrew has come in recently and he's been looking at some really cutting edge techniques that will look at variables that you have for the entire population, so our screening questionnaire, and then use those to predict what your missing data would look like.
00:10:21.269 --> 00:10:26.850
Um, and that allows you to just, uh, make use of the whole sample that you started off with.
00:10:27.629 --> 00:10:30.539
Um, so that's a really tricky statistical one.
00:10:30.840 --> 00:10:40.559
But I think one of the things that, um, has really become apparent to me over time is is the issues and challenges that we have with measurement.
00:10:41.009 --> 00:10:53.070
So, um, so for example, we have a study where, um, we want to know, um, how language influences attention skills and vice versa.
00:10:53.549 --> 00:11:01.799
So for language, we have a standardised test- it's very reliable, it's the same test, the same items at every point in time.
00:11:02.370 --> 00:11:10.529
But the attention measure is a teacher questionnaire, um, which kind of intrinsically has less reliability.
00:11:10.740 --> 00:11:13.620
And also you have different teachers at different time points.
00:11:14.309 --> 00:11:32.879
So you have a much higher stability estimate in language than you do in attention and, because of that, you've got more room to explain variation and attention because there is more variation than you do to explain any variation in language.
00:11:33.210 --> 00:11:36.360
So that, that has made interpretation really difficult.
00:11:37.139 --> 00:11:48.509
Uh, and it, it has made me think a lot harder about how we measure the things that we're interested in and what other factors can influence that measurement.
00:11:50.159 --> 00:11:51.779
That's such an important point.
00:11:52.299 --> 00:12:13.779
And of course even you weren't using a teacher questionnaire, you know, I've found this with studies, um, just where we have quite a wide age range of children involved, it's very hard to find an experimental task based measure of attention that is easy enough for say a five year old and hard enough for a, for a 10 year old, right?
00:12:14.169 --> 00:12:15.820
That is the same measure.
00:12:16.029 --> 00:12:17.860
You know, it's, it's just incredibly difficult.
00:12:18.419 --> 00:12:19.230
Absolutely.
00:12:19.230 --> 00:12:43.740
And we hit that wall with our work on emotion recognition because not only do you have to have something that basically means the same thing over time, but we have huge variation in the language and cognitive skills within the population, so you have to have something that, at age five, everyone can do because if you've got floor effects that also affects the interpretation.
00:12:44.610 --> 00:12:45.570
Um, so yeah.
00:12:45.571 --> 00:12:53.490
We, we have done, we, we've tried to be upfront about that and, uh, you know, journal editors have varying degrees of sympathy with that.
00:12:53.490 --> 00:12:58.049
But it's, it's been quite an eye-opener.
00:12:58.051 --> 00:13:07.350
And I think the other thing is that, you know, a lot of our experimental measures are designed to show group differences, not individual differences.
00:13:07.679 --> 00:13:17.669
And so when you then try to, uh, take an experimental measure and relate it to one of your measures of individual differences, doesn't always work.
00:13:18.269 --> 00:13:19.799
Um, and so that's quite tricky too.
00:13:20.669 --> 00:13:22.590
Uh, and that's just in the psychological field.
00:13:22.591 --> 00:13:27.059
Once you get onto education assessment there's many more challenges with that.
00:13:26.850 --> 00:13:31.799
So, measurement has become increasingly important to me and that's a surprise.
00:13:33.720 --> 00:13:35.220
Yeah, no, but I think it is.
00:13:35.220 --> 00:13:52.440
And I think this is one of the reasons why people maybe sometimes get impatient with how slowly research moves, but quite often you come up against new discoveries and questions, but before you can address them, you need new measures or, you know, more broadly validated measures.
00:13:52.470 --> 00:14:00.149
So one of the reasons why we are so, so slow, I'm afraid people out there who were impatient with research, we're doing our best!
00:14:01.860 --> 00:14:02.669
Absolutely.
00:14:04.220 --> 00:14:17.340
So I could talk to you about measurement for ages, but let's go back to the, to the original kind of headline finding that, um, children are staying broadly in the same rank order over time.
00:14:17.370 --> 00:14:23.639
But also, I guess we could think of it in terms of how far away they're standing from their neighbors, right?
00:14:23.640 --> 00:14:26.129
Those spaces are about the same as well.
00:14:26.879 --> 00:14:54.980
So this is, I find this very counter-intuitive because I think my assumptions about language and development is that if you are, um, you know, at a kind of language disadvantage early in your life with a small vocabulary perhaps, or, um, something like that, that's going to compound and actually increase the gap between you and your peers, you know, and you're going to fall further behind.
00:14:55.940 --> 00:14:58.340
Was that, was that what you expected as well?
00:14:58.341 --> 00:15:00.210
Am I being very naive and ignorant with that expectation?
00:15:02.120 --> 00:15:23.600
No no, um, well I think, I think that is a not unreasonable expectation and I think we we'd seen enough, uh, longitudinal data from other places to be prepared for the fact that those kids who didn't have lots of multiple challenges were probably going to have parallel rates of growth.
00:15:23.960 --> 00:15:37.759
But I really thought, you know, in our sample, we do have kids who have lots of things going on and so we know that those kids, certainly by year three, a lot of them are in different, um, educational provisions.
00:15:38.179 --> 00:15:42.590
They, they have trouble reading, they have difficulties with peers.
00:15:42.919 --> 00:15:49.429
So actually the language that they're exposed to becomes quite different to their peers.
00:15:50.360 --> 00:15:55.190
Um, so I really did think that, uh, we would see gaps getting wiser.
00:15:56.000 --> 00:16:14.779
Um, that's not necessarily the case to the end of, um, primary school, but we've had a little sneak peak at year eight and I would say that, certainly for some things, you can start to see that gap opening, uh, in secondary school.
00:16:15.110 --> 00:16:17.000
And that, that could be lots of things.
00:16:17.000 --> 00:16:26.419
So, um, it could be measurement, it could be those changes in the environment and that it just takes time for you to see the impact that, that has.
00:16:26.960 --> 00:16:33.649
It could also be that when you get to secondary school, the nature of the language that you need to keep pace is changing.
00:16:33.950 --> 00:16:39.620
So when you get to secondary school, we use much more complicated syntax or grammar.
00:16:40.009 --> 00:16:46.580
Words are more abstract and they might have more nuanced interpretations depending on the context.
00:16:46.970 --> 00:16:51.889
And, you know, social relationships are also a lot, a lot trickier at that age.
00:16:51.919 --> 00:17:08.269
So it could be that, um, you know, in primary school, your capacity is kind of enough to keep you going on that track but the nature of what you need to do when you get to secondary school just exceeds that capacity and that's when the gap opens up.
00:17:08.060 --> 00:17:17.720
So it will be really interesting, uh, when we're, when we've got time and headspace to look at that in more detail.
00:17:18.950 --> 00:17:20.420
Mmm, that is incredibly interesting.
00:17:21.380 --> 00:17:27.230
Um, and so what do you think are the sort of practical implications from what you've discovered so far?
00:17:27.230 --> 00:17:39.859
I mean, I'm sure there are, you know, there's a list as long as your arm- but what would you sort of want to highlight for listeners that you think is an interesting y'know action that people can take or lesson that people can take home?
00:17:39.950 --> 00:17:40.730
Yeah.
00:17:41.299 --> 00:18:05.519
So I love these questions because this is where I think"if I ruled the world, what would I do?" So I think one thing that I would really highlight is that school is a good thing that, um, you know, kids do learn, um, school provides a consistent language learning environment and I think that's part of the reason why you get such high rates of stability once kids go to school.
00:18:06.240 --> 00:18:09.569
Um, but clearly closing the gap is difficult.
00:18:10.380 --> 00:18:15.690
Uh, and that really makes me think about what should the purpose of intervention be?
00:18:15.691 --> 00:18:34.470
You know, often when we talk about interventions, certainly at schools, they talk about closing the gap and if you were to go to speech and language therapy, often outcome is measured on some standardised test, which is all about moving you closer to that, uh, typical range or the typical mean.
00:18:35.400 --> 00:18:44.789
Um, so I've started to wonder if, if that's really possible to do and really whether that should be the goal of intervention anyway.
00:18:45.690 --> 00:18:50.130
Um, and I kind of think about it, um, now like other physical characteristics.
00:18:50.131 --> 00:19:01.529
So like, height or weight, which are other skills that we know are genetically influenced, are distributed in the population, um, but they're subject to environmental influences.
00:19:02.339 --> 00:19:08.609
So if, if I, um, well somebody once described me as an"athletic teddy bear", Sue.
00:19:09.549 --> 00:19:13.380
I have a kind of predisposition to be a bit curvy.
00:19:14.460 --> 00:19:36.480
Um, if I wanted to, uh, you know, I can change that through, you know, an intensive environmental modification that involves eating differently and exercising a lot more, and that will have a positive impact- that will move me closer to a typical mean whatever that is and that might have some health benefits as well.
00:19:36.990 --> 00:19:45.150
But once I stop doing that, I just kind of revert to my typical trajectory, right?
00:19:45.599 --> 00:19:58.109
So if you want to make a change, you can't have a short, sharp period of intervention and expect that that will move you onto a completely different track.
00:19:58.289 --> 00:20:04.289
What you need to do is kind of sustain that environmental input over the longer term.
00:20:04.529 --> 00:20:06.000
Now, obviously that's hard.
00:20:05.520 --> 00:20:23.009
When I think about something like running I think"I hate running because it's hard", but I've found other things that I can do that have the same benefit, um, and, and, and are some things that I can sustain over a longer period of time.
00:20:23.400 --> 00:20:26.339
So when I think about language intervention I think about that.
00:20:26.369 --> 00:20:32.819
I think short-term interventions- they might have an immediate impact, but that impact wouldn't be sustained.
00:20:33.089 --> 00:20:43.680
We have to think about the kinds of, uh, learning opportunities that we can give kids more consistently over a longer period of time and also think about those workarounds.
00:20:44.130 --> 00:21:05.920
And so instead of thinking,"can I change the language trajectory?" I started thinking,"what kinds of language skills do children need to participate more fully in our society?" And so some of the things that we've been picking up on are the importance of language for, um, developing emotional understanding and emotion regulation.
00:21:06.430 --> 00:21:15.789
And we're thinking, you know, we would need to test it, but, you know, we can teach kids language, can we teach them that sort of language?
00:21:15.819 --> 00:21:26.140
And if we did that, would they be able to use it in a way that might attenuate some of the negative secondary consequences of having language disorder?
00:21:26.319 --> 00:21:27.670
So that's how I think about it now.
00:21:29.470 --> 00:21:46.690
I think that's such a great perspective that you bring and I think it chimes with something that I've been thinking about a lot as well, which is, you know, working a bit harder to accept the differences between people and that there is a normal distribution with people closer to the tail end.
00:21:47.200 --> 00:21:52.900
And our goal in life should not be to smoosh everyone as close as possible as we can get to the middle, you know?
00:21:52.930 --> 00:21:53.289
Yes, absolutely.
00:21:54.730 --> 00:21:55.480
Yeah, yeah.
00:21:55.180 --> 00:21:56.470
Well- yes.
00:21:56.109 --> 00:22:16.210
Yeah, I was going to say, it's just figuring out, you know, for those, those kids who do have lots of challenges, you know, we know that's associated with negative outcomes, so, you know, what, what kind of level of language with reduce those negative outcomes, um, and allow them to really meet their potential?
00:22:16.210 --> 00:22:28.930
So I think that's a really open question for the field and one where we need young people with language disorders and their families and the people that work with them all the time to kind of help us think that one through.
00:22:29.470 --> 00:22:31.089
Yeah, absolutely agree.
00:22:32.319 --> 00:22:35.079
Um, so Courtney, this has been fascinating.
00:22:35.500 --> 00:22:49.539
Um, before we, before we finish, um, I think there are probably some early career researchers and maybe some students listening and I wondered if you had any words of wisdom that you wanted to drop before we, before we wrap up?
00:22:50.680 --> 00:23:02.410
Um, so yeah, so I've been thinking about this a lot and I think there's two things that I've really learned from SCALES apart from all the rich things that I've learned about, uh, language disorder.
00:23:02.769 --> 00:23:08.380
And one is, uh, you know, if you really, if it's your passion and you really want to do it- persevere.
00:23:08.710 --> 00:23:17.259
So I wrote two grants, uh, for SCALES that were both rejected before we finally got the third one funded.
00:23:17.759 --> 00:23:29.319
Uh, and almost didn't write the third one because, you know, rejection is hard, I'd been working on it for years, I had a baby in the meantime...I thought, oh, I just, there's no point I don't want to put myself through it again.
00:23:29.950 --> 00:23:30.759
Um, but I did.
00:23:30.789 --> 00:23:38.799
And I'm so glad I did because you know, that time we got, we got lucky and it just yielded this amazing project.
00:23:39.220 --> 00:23:41.500
So stay strong and don't give up.
00:23:42.279 --> 00:23:48.559
And the other thing I would say is to really, um, collaborate with wonderful people.
00:23:48.980 --> 00:24:14.539
I, when I, when I was deciding at that stage of my career, what to do, um, there was a lot of, well, not a lot, but there was some pressure to think about things like an ERC grant, which is, you know, an individual award and I knew that nobody in their right mind would like give me, as an individual, a lot of money to do SCALES because it's just the kind of project you can't do on your own.
00:24:15.470 --> 00:24:28.910
Um, and so I went down a route where we had a team and we had, we had a multidisciplinary team of wonderful collaborators- some of whom are, are more senior, more experienced than me.
00:24:29.480 --> 00:24:32.660
Uh, and I've just learned so much from them.
00:24:33.049 --> 00:24:38.029
And then through that project, we bring junior people on and I learned an equal amount from them.
00:24:38.720 --> 00:24:45.049
Uh, and I, I just think the science is much better for having such a strong, rich collaboration.
00:24:45.500 --> 00:24:58.910
So you can have pressure to kind of go, go at your own on your own and get fellowships and all these kinds of things but good collaborative science, I think, reaps huge rewards in the end and in all kinds of different ways.
00:25:01.039 --> 00:25:04.519
Oh, that's such a lovely, such a lovely note to end on.
00:25:04.520 --> 00:25:05.750
I completely agree.
00:25:06.710 --> 00:25:19.039
Um, thank you so much for your time, Courtenay, it's been a huge pleasure to talk to you and to hear more about SCALES, which is a study I think is admired, you know, um, all over the world and quite rightly.
00:25:19.970 --> 00:25:30.829
Um, for anyone who's listening, you'll be able to find out more about Courtenay's work about the SCALES study by following the links that we'll put in the podcast description on Buzzsprout or on your podcast app.
00:25:32.089 --> 00:25:34.519
And thank you so much, Courtenay, and goodbye!
00:25:34.819 --> 00:25:37.700
Thank you, Sue, i's been a huge pleasure to be here.
00:25:37.701 --> 00:25:38.839
I really appreciate it.
00:25:42.160 --> 00:25:45.039
Okay, we did it! I thought that went quite smoothly.
00:00:09.029 --> 00:00:10.679
Oh, it is recording, I see the little figure.
00:00:10.710 --> 00:00:11.130
Okay, great.
00:00:11.131 --> 00:00:14.130
I will do my little spiel and then I'll introduce you.
00:00:13.801 --> 00:00:14.669
Nice, okay.
00:00:16.280 --> 00:00:21.050
Hi, I'm Sue from the Salvesen Mindroom Research Center at the University of Edinburgh.
00:00:21.710 --> 00:00:25.219
Um, I'm recording another PsychologiCALL episode today.
00:00:25.579 --> 00:00:38.600
This is our little podcast, we're aiming to make a bit of evidence-based contribution to the conversations that people are having a lot at the moment about child and adolescent wellbeing and development and learning.
00:00:39.229 --> 00:00:52.369
And today I am very delighted to be talking to Courtenay Norbury, who is a professor at University College, London, and a visiting professor at University of Oslo, and Courtenay is also a speech and language therapist.
00:00:52.789 --> 00:01:00.859
And she's going to be talking to me about the SCALES study, which stands for the Surrey Communication and Language in Education Study.
00:01:01.460 --> 00:01:03.560
So hello, Courtenay, how are you today?
00:01:04.159 --> 00:01:06.920
Hello, Sue, I'm really good, thank you, how are you?
00:01:07.640 --> 00:01:14.180
I am not too shabby at all, thank you very much- I'm wearing a glittery top and there's some snow in my garden from last night.
00:01:15.170 --> 00:01:16.159
Sounds wonderful.
00:01:17.450 --> 00:01:22.549
Um, so tell me what are the kind of headline things that you've been discovering in the SCALES study?
00:01:24.290 --> 00:01:29.510
Okay, so I think one of the main findings is that language is incredibly stable.
00:01:29.930 --> 00:01:52.489
And what this means is that if we were to take a whole year group of children and give them a language test and then line them up in the rank order of the scores that they get on that test and then test them on the same test every year, that rank order pretty much doesn't change from the beginning of school, primary school, to the end of primary school.
00:01:53.719 --> 00:01:59.000
So that's quite striking, but that doesn't mean that children don't change.
00:01:59.269 --> 00:02:06.319
So in the SCALES sample we've demonstrated that actually all children increase their language knowledge over time.
00:02:06.680 --> 00:02:16.580
And the exciting, but also head-scratching bit for me is that the rate of language change seems to be similar across that distribution.
00:02:17.120 --> 00:02:32.810
And what that means is that even those children who had the most severe language deficits when they started school- and many of them have other developmental challenges as well- made as much language progress as their peers who were doing really well when they came into school.
00:02:33.530 --> 00:02:42.199
And this is really, to me, really exciting and in my view demonstrates that universal education is a pretty awesome intervention.
00:02:43.340 --> 00:02:58.460
But the downside, the challenge is that that means that any gaps that were there when they started school really persist all the way through, and in our sample it's about a two year gap between the most able and the ones that have the most challenges.
00:03:00.580 --> 00:03:06.490
Well, that's, that is fascinating because it definitely goes against I think what I would predict.
00:03:07.360 --> 00:03:18.759
Um, but before we dig into that result in a bit more detail perhaps we should just turn back time a little bit and you could tell us a bit about what motivated you to set up the SCALES study.
00:03:19.539 --> 00:03:27.340
You have to turn back time quite a lot, Sue[laughs] cause I've been working on SCALES for a decade now, can you believe that?
00:03:28.210 --> 00:03:48.250
Right, so when I started in this field as a speech and language therapist, a million years ago, we talked about specific language impairment and as a clinician I always found that really puzzling because when you see a child who has, has language impairment, it's rarely the only thing that kids, those kids find challenging.
00:03:49.030 --> 00:04:03.460
Um, and as I started my PhD and started to learn more about developmental processes it seemed to me quite unlikely that language would be the only affected system when a child was not developing as we might expect them to.
00:04:03.131 --> 00:04:12.849
So, but I had to be kind of wary of the fact that, you know, most of the kids that I would see were coming through a clinic.
00:04:13.300 --> 00:04:25.540
And so what we call co-morbidity or these co-occurring challenges that kids have, um, might be elevated because the more things you've got going on, the more likely you are to get to a clinic.
00:04:26.050 --> 00:04:34.000
So what I thought we really needed was to look at, um, language profiles in the general population.
00:04:34.870 --> 00:04:37.420
Um, and so that's, that's how SCALES was born.
00:04:38.170 --> 00:04:52.029
Uh, and also we didn't, we didn't have any UK prevalence data about language disorder and the existing prevalence data, um, been collected by the very wonderful Bruce Tomblin, but more than 20 years ago and in Iowa.
00:04:52.569 --> 00:05:11.860
So I really thought that it would be useful to know how many children were starting school in the UK with lower than expected levels of language, what other developmental challenges they had at the time, and then how those things play out over time and what impact that has on other aspects of development.
00:05:14.680 --> 00:05:18.610
And so, so tell us a bit about the method that you're using in SCALES.
00:05:18.620 --> 00:05:23.500
This is a longitudinal study- you're following up the same children over time, is that right?
00:05:23.529 --> 00:05:28.540
Yep, it is a, it is a longitudinal study- very longitudinal study.
00:05:29.339 --> 00:05:35.949
So what we did is we, um, we have very great partnership, uh, in Surrey county council.
00:05:35.951 --> 00:05:41.470
So they were very willing to, uh, let us do this and, and really helped us do it.
00:05:41.471 --> 00:05:43.240
We wouldn't have been able to do it without them.
00:05:44.009 --> 00:05:58.790
And what we did was we, we tried to get every state maintained school in Surrey that had a reception class to take part, and just over 60% of all of those schools did- so we had about 180 schools taking part in all.
00:05:59.149 --> 00:05:59.839
That's amazing.
00:06:00.529 --> 00:06:11.839
[laughs] What we did was we bought reception class teachers out of teaching for a day and made them sit down and do some online questionnaires for every child in the class for us.
00:06:11.870 --> 00:06:14.480
So, bless them, they are amazing people.
00:06:14.899 --> 00:06:28.730
And in these questionnaires, we asked about, uh, the child's language and communication, their behaviour...um, and then we have the early years foundation stage profile which is, um, early learning attainment goals.
00:06:29.810 --> 00:06:42.290
So we started this in, it was children who started a mainstream reception classroom in 2011 and we had screening data for over 7,000 children.
00:06:43.220 --> 00:06:49.279
Um, and then we took, uh, a subsample of that big group and we followed them up over time.
00:06:49.281 --> 00:06:51.980
So we started off with about 600 children.
00:06:52.339 --> 00:07:04.699
We saw them in year one, year three, year six and we tried to see them last year when they were in year eight, uh, but COVID then decided to make that particularly challenging for us.
00:07:05.089 --> 00:07:07.639
So we were only able to see about half of them.
00:07:08.209 --> 00:07:19.819
Uh, but we do have, um, data on their language, their reading, their problem-solving skills, their social skills, um, and more recently aspects of their mental health.
00:07:19.821 --> 00:07:22.459
So it's a really rich data set.
00:07:24.589 --> 00:07:32.420
And so for those follow- ups your- this isn't just teacher report now, this is more face-to-face work as well with the kids.
00:07:32.810 --> 00:07:47.329
Oh yeah it is, um, so we had a whole army of wonderful research assistants who would go into schools and they would assess children directly on standardised tests of language and cognitive ability.
00:07:47.689 --> 00:08:02.689
We had child questionnaires- they tell us about their experiences of school, we did experimental tasks on their recognition of emotion or their executive control skills, so a whole range of things.
00:08:03.680 --> 00:08:06.019
Um, yeah, very rich dataset.
00:08:06.620 --> 00:08:08.839
Yeah, it's amazing, it's completely amazing.
00:08:09.620 --> 00:08:17.480
Um, so I can imagine that the, I mean, the options, the things to explore in that are absolutely, you know, there are so many.
00:08:17.480 --> 00:08:26.930
But if we think about this, this, um, finding that you flagged at the beginning about sort of stability of language profile across development, right?
00:08:27.230 --> 00:08:37.460
Could you tell us a bit about how you analysed that stability, what, what kind of techniques you were using to, to chart that change or lack of change?
00:08:38.299 --> 00:08:48.950
It's, it's very complicated and I am eternally grateful that Andrew Pickles is a collaborator on the branch because he is possibly the world's greatest statistician.
00:08:49.820 --> 00:09:05.909
Um, so one of the reasons that it's complicated is because, uh, you know, we have the smaller sample of, of 500 or so kids that we've seen all the way through, but we want to use that sample to say something about the bigger population that they came from.
00:09:06.659 --> 00:09:11.370
Um, and there are statistical techniques that you can use to do this.
00:09:12.330 --> 00:09:18.299
Um, so one of the ways we started off was to use weights, and that takes account of our study design.
00:09:18.301 --> 00:09:31.019
So for example, in our study, um, teachers rate boys as having more language difficulties than girls, but we recruit, we brought into our smaller sample, equal numbers of boys and girls.
00:09:31.500 --> 00:09:40.769
So you can use a weight to kind of say, well, the boy data should count for more because there are more boys with this profile in the population.
00:09:41.429 --> 00:09:41.700
Right, right, right.
00:09:41.940 --> 00:09:58.559
Um, but this becomes trickier to use when you've got longitudinal data because as you go along you not only have to account for the initial study design but you also have to account for why people drop out of the study, because people always do drop out of the study.
00:09:59.460 --> 00:10:20.309
Um, and so that's, that's where Andrew has come in recently and he's been looking at some really cutting edge techniques that will look at variables that you have for the entire population, so our screening questionnaire, and then use those to predict what your missing data would look like.
00:10:21.269 --> 00:10:26.850
Um, and that allows you to just, uh, make use of the whole sample that you started off with.
00:10:27.629 --> 00:10:30.539
Um, so that's a really tricky statistical one.
00:10:30.840 --> 00:10:40.559
But I think one of the things that, um, has really become apparent to me over time is is the issues and challenges that we have with measurement.
00:10:41.009 --> 00:10:53.070
So, um, so for example, we have a study where, um, we want to know, um, how language influences attention skills and vice versa.
00:10:53.549 --> 00:11:01.799
So for language, we have a standardised test- it's very reliable, it's the same test, the same items at every point in time.
00:11:02.370 --> 00:11:10.529
But the attention measure is a teacher questionnaire, um, which kind of intrinsically has less reliability.
00:11:10.740 --> 00:11:13.620
And also you have different teachers at different time points.
00:11:14.309 --> 00:11:32.879
So you have a much higher stability estimate in language than you do in attention and, because of that, you've got more room to explain variation and attention because there is more variation than you do to explain any variation in language.
00:11:33.210 --> 00:11:36.360
So that, that has made interpretation really difficult.
00:11:37.139 --> 00:11:48.509
Uh, and it, it has made me think a lot harder about how we measure the things that we're interested in and what other factors can influence that measurement.
00:11:50.159 --> 00:11:51.779
That's such an important point.
00:11:52.299 --> 00:12:13.779
And of course even you weren't using a teacher questionnaire, you know, I've found this with studies, um, just where we have quite a wide age range of children involved, it's very hard to find an experimental task based measure of attention that is easy enough for say a five year old and hard enough for a, for a 10 year old, right?
00:12:14.169 --> 00:12:15.820
That is the same measure.
00:12:16.029 --> 00:12:17.860
You know, it's, it's just incredibly difficult.
00:12:18.419 --> 00:12:19.230
Absolutely.
00:12:19.230 --> 00:12:43.740
And we hit that wall with our work on emotion recognition because not only do you have to have something that basically means the same thing over time, but we have huge variation in the language and cognitive skills within the population, so you have to have something that, at age five, everyone can do because if you've got floor effects that also affects the interpretation.
00:12:44.610 --> 00:12:45.570
Um, so yeah.
00:12:45.571 --> 00:12:53.490
We, we have done, we, we've tried to be upfront about that and, uh, you know, journal editors have varying degrees of sympathy with that.
00:12:53.490 --> 00:12:58.049
But it's, it's been quite an eye-opener.
00:12:58.051 --> 00:13:07.350
And I think the other thing is that, you know, a lot of our experimental measures are designed to show group differences, not individual differences.
00:13:07.679 --> 00:13:17.669
And so when you then try to, uh, take an experimental measure and relate it to one of your measures of individual differences, doesn't always work.
00:13:18.269 --> 00:13:19.799
Um, and so that's quite tricky too.
00:13:20.669 --> 00:13:22.590
Uh, and that's just in the psychological field.
00:13:22.591 --> 00:13:27.059
Once you get onto education assessment there's many more challenges with that.
00:13:26.850 --> 00:13:31.799
So, measurement has become increasingly important to me and that's a surprise.
00:13:33.720 --> 00:13:35.220
Yeah, no, but I think it is.
00:13:35.220 --> 00:13:52.440
And I think this is one of the reasons why people maybe sometimes get impatient with how slowly research moves, but quite often you come up against new discoveries and questions, but before you can address them, you need new measures or, you know, more broadly validated measures.
00:13:52.470 --> 00:14:00.149
So one of the reasons why we are so, so slow, I'm afraid people out there who were impatient with research, we're doing our best!
00:14:01.860 --> 00:14:02.669
Absolutely.
00:14:04.220 --> 00:14:17.340
So I could talk to you about measurement for ages, but let's go back to the, to the original kind of headline finding that, um, children are staying broadly in the same rank order over time.
00:14:17.370 --> 00:14:23.639
But also, I guess we could think of it in terms of how far away they're standing from their neighbors, right?
00:14:23.640 --> 00:14:26.129
Those spaces are about the same as well.
00:14:26.879 --> 00:14:54.980
So this is, I find this very counter-intuitive because I think my assumptions about language and development is that if you are, um, you know, at a kind of language disadvantage early in your life with a small vocabulary perhaps, or, um, something like that, that's going to compound and actually increase the gap between you and your peers, you know, and you're going to fall further behind.
00:14:55.940 --> 00:14:58.340
Was that, was that what you expected as well?
00:14:58.341 --> 00:15:00.210
Am I being very naive and ignorant with that expectation?
00:15:02.120 --> 00:15:23.600
No no, um, well I think, I think that is a not unreasonable expectation and I think we we'd seen enough, uh, longitudinal data from other places to be prepared for the fact that those kids who didn't have lots of multiple challenges were probably going to have parallel rates of growth.
00:15:23.960 --> 00:15:37.759
But I really thought, you know, in our sample, we do have kids who have lots of things going on and so we know that those kids, certainly by year three, a lot of them are in different, um, educational provisions.
00:15:38.179 --> 00:15:42.590
They, they have trouble reading, they have difficulties with peers.
00:15:42.919 --> 00:15:49.429
So actually the language that they're exposed to becomes quite different to their peers.
00:15:50.360 --> 00:15:55.190
Um, so I really did think that, uh, we would see gaps getting wiser.
00:15:56.000 --> 00:16:14.779
Um, that's not necessarily the case to the end of, um, primary school, but we've had a little sneak peak at year eight and I would say that, certainly for some things, you can start to see that gap opening, uh, in secondary school.
00:16:15.110 --> 00:16:17.000
And that, that could be lots of things.
00:16:17.000 --> 00:16:26.419
So, um, it could be measurement, it could be those changes in the environment and that it just takes time for you to see the impact that, that has.
00:16:26.960 --> 00:16:33.649
It could also be that when you get to secondary school, the nature of the language that you need to keep pace is changing.
00:16:33.950 --> 00:16:39.620
So when you get to secondary school, we use much more complicated syntax or grammar.
00:16:40.009 --> 00:16:46.580
Words are more abstract and they might have more nuanced interpretations depending on the context.
00:16:46.970 --> 00:16:51.889
And, you know, social relationships are also a lot, a lot trickier at that age.
00:16:51.919 --> 00:17:08.269
So it could be that, um, you know, in primary school, your capacity is kind of enough to keep you going on that track but the nature of what you need to do when you get to secondary school just exceeds that capacity and that's when the gap opens up.
00:17:08.060 --> 00:17:17.720
So it will be really interesting, uh, when we're, when we've got time and headspace to look at that in more detail.
00:17:18.950 --> 00:17:20.420
Mmm, that is incredibly interesting.
00:17:21.380 --> 00:17:27.230
Um, and so what do you think are the sort of practical implications from what you've discovered so far?
00:17:27.230 --> 00:17:39.859
I mean, I'm sure there are, you know, there's a list as long as your arm- but what would you sort of want to highlight for listeners that you think is an interesting y'know action that people can take or lesson that people can take home?
00:17:39.950 --> 00:17:40.730
Yeah.
00:17:41.299 --> 00:18:05.519
So I love these questions because this is where I think"if I ruled the world, what would I do?" So I think one thing that I would really highlight is that school is a good thing that, um, you know, kids do learn, um, school provides a consistent language learning environment and I think that's part of the reason why you get such high rates of stability once kids go to school.
00:18:06.240 --> 00:18:09.569
Um, but clearly closing the gap is difficult.
00:18:10.380 --> 00:18:15.690
Uh, and that really makes me think about what should the purpose of intervention be?
00:18:15.691 --> 00:18:34.470
You know, often when we talk about interventions, certainly at schools, they talk about closing the gap and if you were to go to speech and language therapy, often outcome is measured on some standardised test, which is all about moving you closer to that, uh, typical range or the typical mean.
00:18:35.400 --> 00:18:44.789
Um, so I've started to wonder if, if that's really possible to do and really whether that should be the goal of intervention anyway.
00:18:45.690 --> 00:18:50.130
Um, and I kind of think about it, um, now like other physical characteristics.
00:18:50.131 --> 00:19:01.529
So like, height or weight, which are other skills that we know are genetically influenced, are distributed in the population, um, but they're subject to environmental influences.
00:19:02.339 --> 00:19:08.609
So if, if I, um, well somebody once described me as an"athletic teddy bear", Sue.
00:19:09.549 --> 00:19:13.380
I have a kind of predisposition to be a bit curvy.
00:19:14.460 --> 00:19:36.480
Um, if I wanted to, uh, you know, I can change that through, you know, an intensive environmental modification that involves eating differently and exercising a lot more, and that will have a positive impact- that will move me closer to a typical mean whatever that is and that might have some health benefits as well.
00:19:36.990 --> 00:19:45.150
But once I stop doing that, I just kind of revert to my typical trajectory, right?
00:19:45.599 --> 00:19:58.109
So if you want to make a change, you can't have a short, sharp period of intervention and expect that that will move you onto a completely different track.
00:19:58.289 --> 00:20:04.289
What you need to do is kind of sustain that environmental input over the longer term.
00:20:04.529 --> 00:20:06.000
Now, obviously that's hard.
00:20:05.520 --> 00:20:23.009
When I think about something like running I think"I hate running because it's hard", but I've found other things that I can do that have the same benefit, um, and, and, and are some things that I can sustain over a longer period of time.
00:20:23.400 --> 00:20:26.339
So when I think about language intervention I think about that.
00:20:26.369 --> 00:20:32.819
I think short-term interventions- they might have an immediate impact, but that impact wouldn't be sustained.
00:20:33.089 --> 00:20:43.680
We have to think about the kinds of, uh, learning opportunities that we can give kids more consistently over a longer period of time and also think about those workarounds.
00:20:44.130 --> 00:21:05.920
And so instead of thinking,"can I change the language trajectory?" I started thinking,"what kinds of language skills do children need to participate more fully in our society?" And so some of the things that we've been picking up on are the importance of language for, um, developing emotional understanding and emotion regulation.
00:21:06.430 --> 00:21:15.789
And we're thinking, you know, we would need to test it, but, you know, we can teach kids language, can we teach them that sort of language?
00:21:15.819 --> 00:21:26.140
And if we did that, would they be able to use it in a way that might attenuate some of the negative secondary consequences of having language disorder?
00:21:26.319 --> 00:21:27.670
So that's how I think about it now.
00:21:29.470 --> 00:21:46.690
I think that's such a great perspective that you bring and I think it chimes with something that I've been thinking about a lot as well, which is, you know, working a bit harder to accept the differences between people and that there is a normal distribution with people closer to the tail end.
00:21:47.200 --> 00:21:52.900
And our goal in life should not be to smoosh everyone as close as possible as we can get to the middle, you know?
00:21:52.930 --> 00:21:53.289
Yes, absolutely.
00:21:54.730 --> 00:21:55.480
Yeah, yeah.
00:21:55.180 --> 00:21:56.470
Well- yes.
00:21:56.109 --> 00:22:16.210
Yeah, I was going to say, it's just figuring out, you know, for those, those kids who do have lots of challenges, you know, we know that's associated with negative outcomes, so, you know, what, what kind of level of language with reduce those negative outcomes, um, and allow them to really meet their potential?
00:22:16.210 --> 00:22:28.930
So I think that's a really open question for the field and one where we need young people with language disorders and their families and the people that work with them all the time to kind of help us think that one through.
00:22:29.470 --> 00:22:31.089
Yeah, absolutely agree.
00:22:32.319 --> 00:22:35.079
Um, so Courtney, this has been fascinating.
00:22:35.500 --> 00:22:49.539
Um, before we, before we finish, um, I think there are probably some early career researchers and maybe some students listening and I wondered if you had any words of wisdom that you wanted to drop before we, before we wrap up?
00:22:50.680 --> 00:23:02.410
Um, so yeah, so I've been thinking about this a lot and I think there's two things that I've really learned from SCALES apart from all the rich things that I've learned about, uh, language disorder.
00:23:02.769 --> 00:23:08.380
And one is, uh, you know, if you really, if it's your passion and you really want to do it- persevere.
00:23:08.710 --> 00:23:17.259
So I wrote two grants, uh, for SCALES that were both rejected before we finally got the third one funded.
00:23:17.759 --> 00:23:29.319
Uh, and almost didn't write the third one because, you know, rejection is hard, I'd been working on it for years, I had a baby in the meantime...I thought, oh, I just, there's no point I don't want to put myself through it again.
00:23:29.950 --> 00:23:30.759
Um, but I did.
00:23:30.789 --> 00:23:38.799
And I'm so glad I did because you know, that time we got, we got lucky and it just yielded this amazing project.
00:23:39.220 --> 00:23:41.500
So stay strong and don't give up.
00:23:42.279 --> 00:23:48.559
And the other thing I would say is to really, um, collaborate with wonderful people.
00:23:48.980 --> 00:24:14.539
I, when I, when I was deciding at that stage of my career, what to do, um, there was a lot of, well, not a lot, but there was some pressure to think about things like an ERC grant, which is, you know, an individual award and I knew that nobody in their right mind would like give me, as an individual, a lot of money to do SCALES because it's just the kind of project you can't do on your own.
00:24:15.470 --> 00:24:28.910
Um, and so I went down a route where we had a team and we had, we had a multidisciplinary team of wonderful collaborators- some of whom are, are more senior, more experienced than me.
00:24:29.480 --> 00:24:32.660
Uh, and I've just learned so much from them.
00:24:33.049 --> 00:24:38.029
And then through that project, we bring junior people on and I learned an equal amount from them.
00:24:38.720 --> 00:24:45.049
Uh, and I, I just think the science is much better for having such a strong, rich collaboration.
00:24:45.500 --> 00:24:58.910
So you can have pressure to kind of go, go at your own on your own and get fellowships and all these kinds of things but good collaborative science, I think, reaps huge rewards in the end and in all kinds of different ways.
00:25:01.039 --> 00:25:04.519
Oh, that's such a lovely, such a lovely note to end on.
00:25:04.520 --> 00:25:05.750
I completely agree.
00:25:06.710 --> 00:25:19.039
Um, thank you so much for your time, Courtenay, it's been a huge pleasure to talk to you and to hear more about SCALES, which is a study I think is admired, you know, um, all over the world and quite rightly.
00:25:19.970 --> 00:25:30.829
Um, for anyone who's listening, you'll be able to find out more about Courtenay's work about the SCALES study by following the links that we'll put in the podcast description on Buzzsprout or on your podcast app.
00:25:32.089 --> 00:25:34.519
And thank you so much, Courtenay, and goodbye!
00:25:34.819 --> 00:25:37.700
Thank you, Sue, i's been a huge pleasure to be here.
00:25:37.701 --> 00:25:38.839
I really appreciate it.
00:25:42.160 --> 00:25:45.039
Okay, we did it! I thought that went quite smoothly.
