Language Impairment in the ADHD Context Looking at models of comorbidity, assessment implications, and capitalizing on comparisons between children with SLI and children with ADHD. Presentation Video
Presentation Video  |   November 01, 2014
Language Impairment in the ADHD Context
Author Affiliations & Notes
  • Sean Redmond
    University of Utah
  • Presented at the 24th Annual Research Symposium at the ASHA Convention (November 2014).
    Presented at the 24th Annual Research Symposium at the ASHA Convention (November 2014). ×
  • The Research Symposium is hosted by the American Speech-Language-Hearing Association, and is supported in part by grant R13DC003383 from the National Institute on Deafness and Other Communication Disorders (NIDCD) of the National Institutes of Health (NIH).
    The Research Symposium is hosted by the American Speech-Language-Hearing Association, and is supported in part by grant R13DC003383 from the National Institute on Deafness and Other Communication Disorders (NIDCD) of the National Institutes of Health (NIH).×
Article Information
Research Issues, Methods & Evidence-Based Practice / Attention, Memory & Executive Functions / Clinical Practice Research / Diagnostics, Screening and Assessment Research
Presentation Video   |   November 01, 2014
Language Impairment in the ADHD Context
CREd Library, November 2014, doi:10.1044/cred-pvd-c14003
CREd Library, November 2014, doi:10.1044/cred-pvd-c14003

The following is a transcript of the presentation video, edited for clarity.

Thank you for giving me the opportunity to talk to you about some work that I've been doing looking at various ways of capitalizing on the comparisons between children with specific language impairment and children with ADHD.
I'm going to first start with the acknowledgments and recognize the support that I've been getting from NIDCD and the support from consultants and my project manager, Andrea Ash. As well as the community contacts and a team of research assistants that have over the years helped me work out what I've collected.
The ADHD Context
When I was coming up for a title for the talk, I thought I was quite clever with this. Because the notion of an ADHD context can actually be placed out in multiple ways. And so what I'm going to talk through here is the issue of the ADHD context as a co-occurring condition that does occur with language impairments more frequently than we might expect it to.
But then that leads into the question: how to interpret these co-occurrence rates in the context of the possibility that we might have measurement error working its way through how we identify the co-occurrence rate.
And then the next question will be, what happens when a child has both an ADHD and a language impairment? And that moves us into the realm of different models of comorbidity. And whether or not ADHD plus language impairment represents a subtype of language impairments. Or is it the case that the combination of ADHD plus language impairment would make your language impairment worse? Or your symptoms with ADHD more severe?
And then one thing I started to play out here was this idea that the group of children with ADHD might represent a better control group than the traditional comparisons that we have between children with SLI and typically developing peers and I'll talk ourselves through why I think that's the case.
And then, finally, I'll talk about the possibility that ADHD might be the best thing that could happen to a child with language impairment. Because that could increase the likelihood that they'll receive services.
ADHD as a Common Co-Occurring Disorder
The first topic is taking a quick survey of what's available on the issue of whether or not ADHD and language impairments co-occur. First we have to kind of lay out on the table what we're talking about with ADHD and SLI.
I've been playing around with different ways to set this up. But it helps to think about ADHD and SLI in the larger context. So one way of talking about this would be to point out something that's quite obvious, that everyone knows what ADHD is. Everyone has had their personal lives affected by someone with ADHD. Much more so than the idea that SLI is affecting multiple people. And so we can characterize ADHD as a very well established clinical brand.
Whereas SLI is sort of marginalized in these discussions. I've offered the standard definitions of both. In both of them, there's this notion that there is an unexpected difficulty relative to other areas. These are two conditions that seem to be comparably matched in terms of prevalence rates from epidemiological reports. In both areas, they represent the most common version or common type of the larger class of disorders that they belong to.
One thing that's very interesting about SLI relative to ADHD is that ADHD is a condition that is monitored by the Centers for Disease Control. We get information on disconnects between the prevalence rate based on epidemiological reports relative to how many kids are actually being diagnosed with ADHD. And there is a discrepancy. So many more kids are being diagnosed with ADHD than the epidemiological reports would suggest. But we are completely flying blind when it comes to SLI. It is not tracked by any entity. The Centers for Disease Control, the U.S. Department of Education, or ASHA. But based on epidemiological and longitudinal reports, it's pretty clear that SLI is being wildly underdiagnosed.
There is a male to female ratio well established in ADHD. It's slanted towards males. And in reports of SLI, we're finding out that things are much more balanced.
Both conditions have a variety of different tools available to clinicians with standardized and non-standardized procedures.
ADHD is recognized as a highly comorbid condition, as co-occurring with just about everything else. And what we know about specific language impairment is that it does frequently co-occur with reading disability, but there's some mixed signals about other areas that may or may not be compromised.
In both areas, heterogeneity has encouraged different subtyping initiatives. And in both cases, the subtyping initiatives haven't shown developmental stability, but they seem to be clinically useful to think about ADHD without hyperactivity, ADHD without attention problems, and then the combined profile. And similarly in SLI, we talk about expressive and receptive deficits. In the ADHD world, it's interesting because there is a long-standing interest in establishing markers for differential diagnosis. And we'll take a look at some of the things that have been done there. In contrast, it seems like the SLI world is working in the different direction. So we start off with a constrained definition of who could fit the profile of SLI. And then we try to spread it out into areas outside of language.
And then, finally, it's clear that ADHD is a well-resourced disability with lots of financial and research dollars. A very large, critical mass of investigators working on it. And the SLI world is wildly under-resourced relative to the prevalence rate of the condition.
Prevalence and Co-Occurrence Rates
If you were to try and synthesize what's available out there on ADHD and language impairments in terms of comorbidity, you can muster citations to support the claim that these things are frequently co-recurring. But you can also identify studies where there is much less of a prevalence rate of co-occurrence between the two conditions. And in some studies, the co-occurrence rate gets very close to the population estimates.
I'm going to lay out a figure here in which we're going to try to take a look at prevalence rates and co-occurrence rates from zero percent to 100 percent. And so to place things into reference, we'll start off with what's the official prevalence rate of ADHD based on the APA and meta-analyses of epidemiological reports. And then we'll contrast that with the diagnostic rate for both sexes that came out in 2013. And then we can disaggregate that value and identify that, the Centers for Disease Control reports that 15 percent of males school-age population have been diagnosed with ADHD. We also know, again, by virtue of the data collected by the Centers for Disease Control, that their diagnostic rate of ADHD actually varies quite considerably across different areas. So it's, the western states including Utah have some of the lowest rates in the country. And the southern and eastern rates have much higher reported values.
Then we bring in language impairment and/or specific language impairment from the Ottawa and the Iowa epidemiological reports. And one thing that you can note right away is that there's a discrepancy in the recency of the information that's available to us in these two disorders. So we have a 15 year old data that we're all using to provide support for how prevalent specific language impairment is.
As I mentioned before, nobody seems to be tracking this in a systematic way. And if you take a look at the caseloads or surveys that ASHA has collected over the years. ASHA's actually done a better job of tracking how many kids with ADHD are receiving serves from speech language pathologists. But we don't have any ability to identify kids who have primary language impairment or specific language impairment and how prevalent or how prominent they are in speech pathologist caseloads.
Then what I've done is I've tried to identify sort of the highest and the lowest values that are out there in terms of ADHD plus language impairment co-occurrence. So that almost covers the entire range with values around 3 percent to over 90 percent. I want you to think about where might be see the distribution of other reports? Should it lean up more towards the high end or the low end? So that there should be some kind of pattern that comes up when we take a broader sample of reports.
What happens when you do that is you find a value, you find a report that almost lands on every possible value that you could think of. So we've got reports that suggest these things are happening at more or less a 50 percent rate, and other reports that suggest it's much lower than that and other reports that suggest it's much higher than that. So, basically, this is a very generous body of research that can support any theory you want about the nature of language impairment and ADHD links.
Potential Contributors to Cross Signals
So there's some contributors maybe to these cross signals that are variable in the literature. One is the well attested phenomenon of the of the Berkson's bias. So that children who are receiving clinical services are going to have more severe symptoms and likely to have multiple conditions. And that you get a distorted view unless you step out and look at things from an epidemiological perspective. But as you can see, there's quite a bit of overlap between those other ranges as well. It does seem to matter whether or not you're asking the question, how many kids with language impairments seem to have symptoms that are consistent with ADHD versus how many kids with ADHD seem to have symptoms of language impairment. But, again, there's still quite a bit of overlap.
We could also take a look at whether older reports, where our diagnostic schemes were still in development versus our more sensitive, hopefully, and more psychometrically intact or developed protocols. Again, there's still quite a bit of overlap.
And it also seem to matter somewhat whether or not you're controlling for nonverbal IQ when you're talking about language impairment and ADHD. Where we get lower values and those studies that consider the impact of low nonverbal IQ in both language and ADHD.
But there's another issue here, and that is these co-occurrence rates estimates are meaningless if we can't trust the tools that we're using to identify language impairments as being capable of identifying the language impairment and not misidentifying ADHD and vice versa. And in both clinical arenas, we've got a wide variety of choices. And in both areas, we certainly have measures that are capable of differentiating a typical status from atypical status.
But that's a different question from identifying different kinds of atypical status. And so some of our measures are really good at identifying kids who are not functioning in the normal range, but end up being very poor when you're trying to use them for differential identification.
Socioemotional Behavioral Assessment
That leads me into the next point here which is out of all the choices that we have in front of us for identifying ADHD symptoms in children with language impairments, which ones might be a better choice over others? And then, conversely, which language measures might be better choices for measuring language abilities in kids with ADHD?
Part of this background in thinking about how to identify behavioral markers for differential diagnosis that both language impairment, specific language impairment, ADHD have undergone conceptual revisions over the years. One revision that's influences this discussion about language impairment and ADHD is that the notion of ADHD has evolved across the different DSM schemes. And so in DSM-III, we had something called situational ADHD where you could have ADHD if your symptoms existed in either the academic or nonacademic context. So you could have ADHD that functions from 8 to 3, Monday to Friday, September to June, right. In DSM-IV, that was identified as a problem and then the criteria was moved. That at least some of the ADHD symptoms needed exist in multiple settings. And for children, that means school and home settings. And then the DSM-V, there's a much more deliberate attempt to identify, actually move away from a situational ADHD kind of concept and notice that children with specific learning disorders may appear to have ADHD but really don't. And so if the limitations for children with specific learning disorders do not impair outside of academic context, then you cannot have ADHD is how it's currently formulated.
There's a lot of enthusiasm for coming up with a behavioral measure. A situation where children do something that would allow us to identify performance that would place them at risk for ADHD. And in this context, these measures called executive function tasks have been developed. But I'm going to suggest that these executive function tasks actually represent very poor choices for taking up this issue of identifying overlap between ADHD and language impairments.
One of the main reasons for this is that it's recognized that the executive function tasks, tests, and measures haven't lived up to their potential yet in terms of being able to consistently identify children with ADHD. And so roughly about half the children with an ADHD sample will perform within the clinical range on any given executive function task. Leaving the other half to perform within normal limits.
This obviously compromises our ability to use them as the stand in measure for ADHD. This doesn't stop people from developing various home grown versions of executive function tasks. That's a problem for bringing up these clinical symptoms, because those particular procedures have unknown levels of reliability and validity. For those standardized executive function tasks that are available, their test/retest reliabilities are not adequate. And in the particular context of identifying language impairments, an older report by Riley who's one of the authors for these executive function tasks, they reported an 81 percent positive false rate when using executive function tasks relative to parent and teacher-rated symptoms of ADHD. So children who had ADHD and a language impairment versus kids who just had a language impairment, the kids with language impairments were coming up positive on the ADHD measures, the executive function measures.
There's another body or type or class of behavioral measure of attention referred to as continuous performance tasks. And continuous performance tasks have the interesting possibility of being very good at differentiating typical status from atypical status. But as I mentioned, it seems to be the case that virtually any disorder of childhood will display weaknesses in continuous performance measures. And so this is a direct quote, a list from this book and notice the variety of conditions: Mental retardation, seizure disorders, maltreatment. And then there was a study that used general medical referrals. So basically kids that were in the waiting rooms of the pediatricians were given a continuous performance assessment, and they all came up scoring within the clinical range.
It is true that children with specific language impairment have also shown weaknesses on these measures. In fact, it would be really weird if they didn't relative to the variety of conditions that these tools are able to pick up on. And so it would be, as stated by the authors of this volume that to rely on these performance measures as the primary diagnostic tool would yield unacceptably high measures of false positive rates. A lot of these continuous performance measures use numbers and letters as their stimuli. And you could consider that to be problematic for kids with language and/or reading problems.
There is a continuous performance measure that does not use numbers or letters that tests the variables of attention. But its specificity rate is 22 percent. So I'm going to suggest we not go there.
The gold standard in this area in terms of clinical practice seems to be the use of standardized rating scales. And they're often collected from parents and teachers.
I'm going to suggest in the context of looking for ADHD symptoms and children with language impairments that we move towards preferring parent ratings over teacher ratings. And there's a couple reasons for that. One is that when parent ratings are positive for ADHD status, 90 percent of the time they agree with the teacher assignment of ADHD status. And people have argued that that suggests that they are sufficient for the diagnosis of ADHD. But overall, if you are to look at the levels of agreement between parent and teacher ADHD ratings, these have been consistently modest in the literature, less than .5 levels of the concordance. And teacher ratings also have been shown to not be concordant with observational measures of ADHD symptoms. And some of the work that I did a long, long time ago, looking at teacher ratings of children with specific language impairment from kindergarten to first grade, to second grade with Mabel Rice. What we found was that the teachers across different grades couldn't agree on which kids were presenting the ADHD symptoms in the clinical range. So a child with SLI might be characterized or considered by their teacher as having a lot of problems in the areas of in attention, hyperactivity, impulsivity. But when the child moves into the next grade, they're suddenly performing within normal limits. So that's a potential problem.
It's also the case that heritability estimates that are based on parent ratings are higher than those based on teachers. And that the teacher ratings and the variance that's captured by the teacher ratings overlaps with the variance that's captured by parental ratings. So parental rating is like a subset of these attempts to move towards genetic measurements or genetic factors in play for ADHD.
There's also direct evidence that if you get teachers standardized rating scales to identify ADHD symptoms in children with language impairments that they would basically display bias against children with language impairments. So in a study that was done by Charach et al. where they looked at kids with language impairments, and children with reading disabilities, and children with intellectual disabilities. It was the group of kids with language impairments where the concordance between the teacher ratings and a blinded psychiatric interview was the most discordant. So there was twice as many false positives as true positives in the group of kids with language impairments. And that didn't happen in the other clinical groups.
So I'm going to suggest that parent ratings are going to give us the clearest signal on what's going on with kids with language impairments in the areas that are tapped into by the disorder ADHD. But I don't think we're ready yet. One thing we also have to do is take a look at the scales themselves for the content and see if there's possibility that there might be confounds there. And so one thing I did a while ago was take a look at some of the more commonly used pediatric psychiatric scales and noticed that many of them included language and academic items. So I don't think it's a stretch to suggest that if you include items like won't talk, has speech problems, or poor schoolwork, that you're also going to be tapping into the dimensions of language impairment.
In the world of ADHD, it's not uncommon to take standardized rating scales when the purpose is for differential diagnosis and make adjustments based on overlapping symptoms. So what we did was we took the child behavior checklist and the Conners scales which represent the two most commonly used pediatric, psychiatric instruments. And we removed the language and academic items that these items contained, and that's the list right there: we took out won't talk, speech problems, doesn't seem to understand what is being said to him, et cetera.
What we did was we took those items out of the scales and so one thing that could happen when you do that is you basically compromise the diagnostic integrity of the tool to differentiate ADHD from typical status. But we didn't find that at all. What happens when you do this is it actually increases their capability to differentiate language impairment from ADHD. And it didn't compromise the ability to differentiate kids with ADHD from atypically developing peers.
Our ROC curves with the adjusted scales were in that low to high 90 range which is quite good. And one thing that when you start to collect a behavioral rating scale on different clinical groups is that you start to realize that it's a matter of scale. So that if you see a difference between kids, which has been reported in the literature, between children with specific language impairment and typically developing kids, and there's significant group differences, the question is whether or not that level of difference is the same as you would see in a group of kids with a psychiatric condition. And it's clearly very different. That the symptoms you would see in kids with ADHD are much, much, much more severe relative to typically developing controls than you would see when you compare kids with SLI and kids with typical development.
Psycholinguistic Assessment
So now I've suggested that what we want to do is take a look at parent rating scales when we're looking at the co-occurrence rate between ADHD and language impairment. And then I'm going to move now towards taking a consideration of our own available, possible language measures if we're looking at language abilities in kids with ADHD.
What I'm going to suggest off the bat is that we avoid vocabulary. Things that are called verbal IQ and the pragmatic indices that are available to us because these things have problems in addressing this question. A very helpful review by Spaulding, Plante, and Farinella, they went through the (at the time) basic choices we have in terms of standardized tests for language assessments. And what they noticed was that the items that are measuring vocabulary have the weakest levels of sensitivity and specificity. And so that's why we shouldn't do vocabulary.
But pragmatics seems to be an area that would be a natural fit for ADHD. And one reason for that is you can basically redefine ADHD symptoms into a pragmatic deficit. And so that was done by Camarata, Hughes and Ruhl. And there's also, you can even go further than that. You can take the DSM manual and redefine every psychiatric condition as a pragmatic impairment. And that was actually done in a book called "Language and Psychiatry". Where there's a section for each area in the DSM, it's a borderline personality disorder, ADHD, psychotic disorders, schizophrenia. Would all, potentially, if you have a very open view of what pragmatics is, anything that creates intrapersonal distress could be defined as a pragmatic deficit. And that's not a good thing when you're try to find out how often things co-occur. In some of the data that we've collected, we've played around with taking a look at how pragmatic measures load or don't load into different areas. Bruce Tomblin did a similar analysis in his book-length treatment of the Iowa epidemiological sample. And when you do that, what you find out is that pragmatic measures do not load onto a common factor with other language measures. Which, by itself, isn't a bad thing because we think of pragmatics as being potentially a separate domain from semantic syntax phonology, morphology. But what it does load with is other psychiatric scales. So the children's communication checklist, the pragmatic problems subscale. It loads with anxiety, ADHD, and other syndrome scales that you can measure with the child behavior checklist or the Conners. It's also the case that when interventions have been directed at improving pragmatic and social cognitive skills, they don't seem to have an impact on semantic and syntactic skills. It's suggesting that these are separate domains, so we should probably not go there for measuring language impairment.
Instead, I'm going to suggest that those emerging markers of specific language impairment, non-repetition, sentence recall, and grammatical tense marking indices representing good choices. In part, because they're being used in genetic work to look at the underlying mechanisms for language impairment. It's also the case that these measures have been really good for differentiating typical status from atypical status.
And so the next question is, how good are they at differentiating between different clinical conditions? I'm presenting box plots so you can see the distributions within these groups. You can also, when I do this, you can also take a look to see if there's like a hidden subgroup within the ADHD sample. So these are clinical samples. So these are children receiving services for language impairment, children receiving services for ADHD, and a sample of typically developing controls.
This is what happens when you take a look at tense marking. There is no difference between the group of kids with ADHD and the typically developing controls. They're basically functioning at ceiling levels of performance. These are 7 to 8 year olds that we're looking at here. And notice that within the SLI group, there's huge variability. It basically covers a range of possible scores with the subgroup of kids who are performing the poorest with language impairments in the SLI group.
This is what happens when you take a look at nonword repetition. There's a little guy there in the typically developing group that's not like the others. Breaking out and lining up more with the distribution for kids with language impairments.
This is sentence recall. Now we've got a kid in the SLI group that's breaking away, good for him, and lining up exactly with the normal range. But it does look like sentence recall is a measure, perhaps, of verbal short term memory or grammatical processing is a good indicator of language impairment status.
In this particular study, we also have narratives. And there's more overlap between the groups in narratives, but it is clearly differentiating. And I included narratives in that study because I thought that if there was an area of language where kids with ADHD would potentially show weaknesses, it would be narratives. Because of the planning and organizational demands that are involved there. It would make perfect sense to expect kids with ADHD to show weaknesses. And they didn't.
Then after that report came out, there was a nice replication of that study in a Dutch sample of children. In Parigger's report, she also included measures of executive functioning and asked the question, whether or not executive functioning would line up or predict some of the weaknesses in language abilities? And what they found basically replicated what we found. That nonword repetition, sentence recall, tense marking and narratives were not associated with ADHD status. And that there was no significant correlation between those measures and executive function. And there was a prediction that had been out there that the language impairments in children with ADHD might be linked to their executive function problems, and, in particular, we would see a link between executive function problems and pragmatic difficulties. And that was also not found in this Dutch sample of kids.
It's possible that there might be communication markers for ADHD that are unique to ADHD. One area that hasn't been explored in the specific language impairment research, I don't think we typically talk about vocal problems as a big primary symptom of children with SLI. But that does seem to be something that's associated with ADHD. And the idea being that it's linked to their hyperfunction in general, including vocal hyperfunction. And so those kids have poor voice qualities relative to typically developing controls. And I've also seen some differences in conversational samples between kids with ADHD and SLI in the number of utterance formulation problems. And so in the world of SALT where you put everything in parentheses and that turns into a maze, when you do that, it looks like kids with ADHD are doing a lot more mazing, false starts, um's and oh's and repairs relative to kids with SLI.
ADHD + LI Comorbidity
This next slice is going to be taking a look at what happens when you use parent ratings and those clinical markers of SLI and you ask a different question. You ask, in those kids who have ADHD and language impairment, is it the case that their symptoms in either area are worse relative to the kids who present with a more pure clinical profile?
Comorbidity is an interesting topic in and of itself. And when you think of all the different ways two disorders might interact with each other becomes quite intricate. And so you might have a situation where as I mentioned there might be additive or interactive effects where the symptoms are worse and the comorbid group relative to the non-comorbid or pure group.
And we have addressed this issue in the literature on kids with specific language impairment or language impairments in other areas. And so people have looked at what happens when you have a language impairment and weak nonverbal abilities. These kids are referred to in the literature these days as nonspecific language impairment. But in earlier eras, we would have referred to those kids as having borderline intellectual disability. It's been shown that children with the nonspecific language impairment perform worse than children with SLI across various language metrics. And that their growth is actually been shown to be slower. And it's also the case that children who have additional nonverbal limitations are at greater risk for academic and, in particular, behavior problems.
Another comorbidity that's given a lot of treatment in our literature has been the comorbidity between language impairment and reading disability. And studies that have looked at the language abilities of kids who also have reading disability, there is no difference in terms of severity or profiles for kids who have language impairment with or without a reading disability.
And so those are two possible scenarios that could happen when we take a look at what's going on with ADHD and language impairment.
There's a third possibility. And that is, you can have subtractive effects when you have co-occurring disorders. And that can happen if you have a protective mechanism that's associated with one disorder that's brought in by virtue of being co-occurring with another disorder to offset the risks that they're associated with the other disorder.
I have two samples that I've been looking at in the research projects. And so one as I mentioned is the clinical sample where I went out and identified kids by virtue of receiving services. And then we also have access to a community sample that's being collected over time by virtue of the fact that we're going out into the schools in Salt Lake City and screening kids for language impairments, and then bringing those kids into the lab who fail the screening and identifying those who have language impairments.
What happens when you do that is you identify kids who have language impairments that are not receiving services. And if we interpret what's going on with those epidemiological reports, that's actually the more common profile for specific language impairment. It's more likely for children not to be receiving services than to be receiving services. So we have some of those kids in the sample as well.
What I have laid out here was actually just accepted for LSHSS, is we have 19 kids with SLI, 19 kids who have ADHD plus language impairment in a control group. And I have nonword repetition, sentence recall, and then the different probes from the TEGI (testing grammatic impairment) for measuring different aspects of command of the tense feature. And if you go over to the column on contrast, what you see is that the ADHD plus language impaired group is not significantly different from the kids with specific language impairment. So it doesn't seem to be the case that there's an additive or interactive effect on language abilities when a child also has clinically identified ADHD.
And, in fact, if you go into the details here, in each case, the kids with ADHD plus language impairment are doing better than the children with specific language impairment. So not only is it not showing up as an additive or interactive effect, it's providing what seems to be a modest subtractive effect.
Here is some box plots. And, again, I'm doing this deliberately to put out for your inspection whether or not there's a potential subgroup of kids with ADHD plus language impairment that are breaking out of these box plots and coming up as outliers. The outliers are in the SLI group. Now that was nonword repetition. This is sentence recall. Again, there's a little guy there in the normal, in the typically developing control group that's having difficulties. And then here is the TEGI screening score which is a composite of the present tense and past tense measures. So the lowest performers in the sample are the kids in the SLI group.
We looked at correlations across these different scales. And we found a positive correlation between DSM ADHD symptoms as measured by the child behavior checklist and sentence recall and third person present. Now what that means is because the DSM scales are oriented so that higher scores mean more severe symptoms and lower scores mean more normal performance. If we have a positive correlation between language and ADHD symptoms, that means that kids with a more severe ADHD symptoms are showing relatively better language abilities than the kids with less severe ADHD symptoms. And this is based on a combined sample of the SLI and the ADHD, LI groups.
And then I can update some work that's going on looking at narratives. So these are, again, this time we have the ADHD group added in as another comparison. And what's going on here is, again, that the ADHD and SLI groups are very similar. And notice what happens here, though, is that the ADHD group ends up being a better control. So the group mean for the ADHD group is 100, the group mean for the typically developing group is around 110. And that's actually a very common aspect of our literature. We have consistently compared kids with specific language impairment to not a typically developing group, but an advantaged group. So it's not uncommon for our control samples to have IQs that are higher than the means based on the standardized tests. And that's happening here in the narratives where the kids with ADHD are actually more normal than the kids with in the typically developing group.
If we step out a little bit and take a look, these are Z scores at the relative severity of symptoms across these different language areas. It looks like we have the biggest differentiation going on with tense marking in both the SLI group and the ADHD post language impaired group. And that actually as you move into narratives, the symptoms become less severe which is kind of interesting because you'd expect as you bring in more complexity into the task, that that would sort of compound the deficit. But what could also be happening is that as the task becomes more complicated, children have more compensatory resources to draw upon for accomplishing those tasks. And that it's really when you measure in with a laser-focus at these areas of weaknesses that you're able to find the strongest signals for language impairment.
So in summary, what we found here was very clear phenotypic boundaries. If we take a look at parent ratings and when we're asking the question of behavior. When children have ADHD, that's often associated with a much more pervasive behavioral problem profile than it is with kids with SLI. The behavioral profiles of kids with SLI tend to demonstrate weaknesses in attention and weaknesses in peer relations. Whereas kids with ADHD tend to have a much more pervasive profile of weaknesses across multiple areas. The presence or absence of ADHD had very little impact on the clinical markers of specific language impairment. Executive function markers are not correlating with language impairment symptoms. And then there might be a differential marker if we take a closer look, perhaps, at voice symptoms in kids with ADHD versus SLI.
And these core language areas, there's no evidence of a double deficit or ADHD plus language impairment subtype. And, in fact, having ADHD might represent a buffer against risk of poor outcomes. Maybe what's going on here, I don't have any data to speak to this. This is all speculation at this point. But maybe what's going on here is that if a child has ADHD symptoms and language impairment, that's going to encourage an earlier referral and/or more intensive intervention. And that's going to create this sort of subgroup of kids with stronger language skills.
ADHD as a Test for Non-Linguistic Theories of SLI
So as I mentioned in that slide on narratives, it looked like the kids with ADHD were actually functioning as a better control group than the typically developing kids who, by comparison, were actually performing at above average levels: Almost the standard deviation above according to the norming distributions of that scale. And I think that opens up an opportunity to start thinking of kids with ADHD, who I have just demonstrated can present without language impairment, might represent our preferred control group going forward when we take a look at the nonverbal abilities with kids with SLI. And this was actually quite nicely done by a study by Janis Cardy and colleagues. Where they took the task, the impaired temporal processing and the generalized slow processing accounts and collected the information processing measures that would be used to evaluate those accounts. And what they found was that in kids with ADHD who did not have any symptoms of language impairment, they were actually performing poorer on those information processing measure than the kids with SLI, who are supposed to have their problems with SLI because of their poor performance on these particular measures. And so this is a very nice demonstration of the student to test the necessity and sufficiency of various nonlinguistic accounts for the problems of kids with specific language impairment.
And this area is low-hanging fruit. There's all kinds of things that people can do. Because kids with ADHD have had weaknesses reported in nonverbal IQ, dichotic listening, central auditory processing, implicit learning, procedural learning, working memory, serial reaction time, fine motor skills, and gross motor skills. This sounds familiar because this is also the list that gets invoked when we talked about specific language impairment as the soft signs that might be contributing to their difficulties. And so we can take a look at whether or not any of these have anything to do with language symptoms by bringing in comparison groups of kids with ADHD.
So as I mentioned, we really can't test sufficiency or necessity when we stripped our consideration to comparing kids with SLI to just their typically developing peers. It's only through these cross etiology comparisons that we can start to tease out whether or not something is representing a causal factor. There's lots of other reasons why it's better to compare between clinical groups than to compare a clinical group to a typically developing group. And that is this low average presentation that sometimes gets introduced in the discussion of SLI is a problem for SLI. It's a problem for every clinical group there is. That's true for kids who have stuttering. It's true for kids with various psychiatric problems. It's just a general observation of relatively low performance. It's also the case that if you match, if you compare kids with SLI to kids with ADHD, you're going to get more similar demographic profiles. So you're going to get more similar levels of comparison with regards to another educational level. A single parent households and parental stress. Because both groups are dealing with the fallout of having a child with a disability.
The children, themselves, have similar experiences of academic failure, stigmatization, and peer rejection. All the good and bad things that come with receiving clinical services. And I think there's a very clear difference between when you use typically developing controls, they're often agreeing to participate in your study because they're curious. Or maybe they have a sense of civil responsibility. As opposed to a family that comes to your study because they desperately want more information about their child's strengths and weaknesses and the families are often in crisis because of that.
And I think we need to really be cautious about getting really too excited about nonlinguistic theories of SLI until we've taken that step of testing them against this group of kids who have lots of overlap in nonlinguistic areas to kids with specific language impairment.
ADHD as Contributing Factor to Educational/Health Disparities
And finally, that thing that I mentioned at the beginning that maybe having ADHD is the best thing that could happen to a child with a language impairment. I'll impact that here right now. As I mentioned earlier, and this is something that for some reason people are just waking up to, that the majority of cases of specific language impairment never receive services. And, in particular, it looks like if you're a girl with specific language impairment, you're out of luck. Boys are much more likely to receive services, even with the same diagnosis and even with the same level of language impairment. There's also a very clear overrepresentation of males and referral for language assessments. And this is unfortunate because it looks like women with positive histories of language impairment are actually at more risk for negative outcomes than their male counterparts, and, in particular, sexual assault is elevated in women with a history of language impairment.
In a recent community sample of ADHD where they were taking a look at how many -- it's an epidemiological study, and they're asking how many kids with ADHD also presented with language impairment. They collected this information that about half the kids with ADHD who had a language impairment had access to speech services. But in the non ADHD group, only 16 percent had. So that's, that's the basis for this idea that if you have ADHD plus language impairment, the prognosis looks to be better than if you don't have ADHD.
I'm going to wrap things up here with a quote from one of our participants. And this is a 9 year old guy with a standard score on the CELF in the mid 70s. He's also marking tense about 75 percent of the time. And his nonverbal IQ is 112. And so for that, I think for that reason, the family's encountering some frustration with getting attention drawn to his language problems. "So because of his learning disability with the reading, the language has always been the back burner, you know what I mean. They figure, well, if he would learn to read or if he could learn, if we could help him learn to read or learn to help with this and this and this, then the language thing would probably go away to a point. So it seems like it was never that huge to anyone but us. And they're always like it might be his hearing. And we're always like, it's been checked 100 times. And so it's almost as if specific language impairment does not compute in clinical services. It can't possibly happen. It has to be due to something else. So let's send him out for another round of hearing assessments. Or let's focus on the primary academic sequelae, the secondary effects of the language impairment. And maybe the language issues would naturally adjust themselves as a consequence."
Questions and Discussion

Question: You mentioned that there were a range of studies about comorbidity and prevalence, and that there's a very wide range of what's been shown. You also mentioned that the prevalence of ADHD varies by geographical region. Is there any relationship between where the studies were conducted and the prevalence that they found in the comorbidity?

It's certainly another factor that you'd have to consider. And it does make it very difficult to synthesize that information because that's another potential dimension. That's actually the diagnostic rate, not necessarily a prevalence rate, being reported in those states. As opposed to the epidemiological reports where it's a more deliberate community sampling. So, the one way to interpret that is there might be an overidentification of kids with ADHD and that, for whatever reason, varies across different states. The other thing that's important is that all of these conditions, we believe, are influenced by genetic and environmental effects. And if that's true, then there should be geographical variability based on the kinds of exposures kids have relative to different industries and the gene pools and the different areas. And so that's not necessarily a problem. It would actually be a problem if we defined SLI as some kind of cut off, and then we worked the demographics of the community around so that 5 percent of that population was always SLI regardless of any reference to a common metric.

Question: In the more recent article you mentioned that just got accepted, in your ADHD sample did you control at all for medication? I was wondering if there's any relationship with the length of time they've been on medication prior to you seeing them in the lab.

That's actually a big omission, I apologize. So the convention in the ADHD world is to always test kids on your experimental measure when they're off of medication. And so we asked parents to suspend medication for 24 hours before they come in. What we've been doing is the children come in for eligibility testing and then experimental testing. And during the eligibility testing, they're on medication because that would be a reasonable accommodation kind of argument. But when they come in for the language testing, they're off of their medication. And it's quite a trip collecting language measures on kids with ADHD. Because you have to deliberately choose not to enter into a power struggle, and let the kids do whatever they want as long as they're answering the questions. And so we've got videos of kids that are sitting on the floor, kids that are pointing with their toes, kids that are tearing things up and making confetti while they're answering questions. These kids get lots of breaks. They run a little circuit around the corridor. And they also, when they complete a task, we let them roll a die. And if their lucky number comes up, they get a prize which turns out to be a really good way of capitalizing on their risk for gambling addiction that's associated with ADHD. So in contrast, if you were to watch a video of a child with SLI who is sitting almost in an ideal fashion with his hands folded and missing every single item on the CELF. And knowing that he's missing every single item on the CELF and does not engage in any of those off task behaviors that we're seeing with the kids with ADHD. And I think the trick is to not just get hung up on some kind of like we need quiet hands and you need to pay attention to me and that sort of stuff that you can get trapped into as a clinician. Because we don't care, we just want to collect their language and let them go back. Catch and release. We don't have any interest in fixing their problem, we just want them to tell us what they can do with language.
As far as a relationship with the length of time they have been on medication, notice there's a multitude of medications that have different delivery systems and different titrations. The only way you could get at that is if you were entering into the process where children are just being diagnosed with ADHD before they're put on medication. That would be the window of opportunity to look at what happens when you're on medication versus off medication. And there was a report by Rosemary Tannock that suggested that in a small group of kids that were in that situation, that their narratives improved when they were on medication. But you're right. I mean, there could be residual effects when the kids came in that made their language performance a little better than it should have been.

Question: I was wondering about your thoughts on your use of the narrative testing to see if you could differentiate the ADHD from the SLI group. And, like you, I found it very interesting that there wasn't that expected difference. But I noticed on your table of diagnostic accuracy that it seems like you're getting the best sensitivity and specificity for differentiating SLI and ADHD with the sentence repetition task. And it feels like that might be that optimal balance between the varying language specific, like grammar formulation things and the sort of short term working memory kind of considerations that you might think come in with ADHD.

The ADHD population is one that has been roughly characterized as being weak with working memory. Some reports suggest that also extends into verbal working memory, but it clearly is there in visual working memory. And maybe sentence recall isn't verbal working memory, right. I don't, I don't think we really know what sentence recall is. It's kind of like a hot dog. So there's some meat in there, but there's also some fillers and tongues and lips and things. I don't know what is actually going on with sentence recall. Except that it's really good at identifying kids with language impairments. And maybe we don't need to know why, it just does. But, you're right. It is interesting that that was coming out strong. And in that sample, the kids with ADHD were actually a little bit better than the control group when it came to sentence recall. I mean, it could be spurious. It could just be a weird scoop of kids that we kid that time.
And that's the reason why in the follow up project that I have going on, we've used sentence recall as our community screening procedure. It's very quick. It's easy to train undergrads and grads to do it. You can go out and invade the schools, gym. We've been collecting it under real world circumstances, so there's all kinds of distractions that come about when you're doing it in the gym and other people moving around, and it seems to be working pretty well at identifying kids with language impairments.

Question: When we're looking at assessments, sometimes a big concern of clinicians is the impact of a highly structured testing environment on the scores. Do you see or are you noticing trends where a highly structured environment boosting performance in one domain versus another for any of your populations?

Oh, I think that's an empirical question, and you'd have to go out and actually test that. I think the more things you measure, the more possibilities of discrepancies you produce.
One of the things that's happened with that community sampling is we also get kids who have good language skills, but have ADHD or autism or behavioral disorders as virtue of the mass screening. And so we haven't been able to really look at that deliberately. But it does seem to be the case that language sort of flows independently of those other variables. And that's not always appreciated unless you bring in the clinical comparison group. But the presence of discrepancies may just be, may be a common attribute for kids with atypical development.

Question: What you're showing, with the population with a combination of language impairment and ADHD faring better seems quite counter intuitive. And there are certainly many explanations which you've gone through, but one that we kind of tip-toed up to is whether or not some of the pharmaceutical effects given to kids with ADHD may be useful, suitable, or helpful for kids with language impairment without ADHD? Is there any discussion of that?

So in the investigations of the impacts of these medications on children with ADHD, the concept that it could improve like their academic performance, for example, has been examined and it doesn't. So the primary purpose or function of these medications is to get the kids to chill out. And it's the hyperactivity element that is more affected by the medications than the inattention component which is the component that you'd expect to be more tightly linked to language impairments or reading difficulties.
But you're right, the possibility of pharmaceutical interventions for language impairments is absent. We haven't really explored that at all. But there's reasons to be conscious about that for all the reasons that. That's the main reason why we would talk about ADHD as a controversial diagnosis is because of the treatment component that we're concerned about. That's why the public is skittish about ADHD is because of the interventions that are used.

Question: I was wondering about the age group in which the data is collected, because I do a lot of differential diagnostic testing and what I always ask about is the kids who are the middle and high school level kids who have that combination of attention deficit and language issues that are manifested in the longer utterances, the more complex utterances.

Well, so to sort of clarify, as I was speeding through my slides, I didn't point out that so these are kids in the 7 and 9 year age range. And I deliberately selected that age range because that's the most common age of diagnosis of ADHD. And it's also when you take a look at the language development of kids with specific language impairment, that's also a period where they look like they're kind of catching up to their peers. People refer to it as a illusory recovery. And then as you move forward, they start to fall behind their peers again. And so that's like a, that's a window where diagnostic confusion could be heightened. And there's an interesting adjustment in the DSM where the age of onset was set at 7 and DSM4. Now it's been raised up to age 12. Which does introduce the possibility that a child could start off as SLI and then sort of develop or mature into ADHD. I mean, that possibility is there. And it would be important to be able to have metrics that track both ADHD and language over a more extended age range.

Question: Have you had an opportunity to look at qualitative differences? For example, I'm trying to get out what causes somebody to make a mistake for sentence repetition. What kind of a mistake it is. Is it a grammatical one? Is it a vocabulary one? Can you get attentional errors versus language errors?

Yes. Well, no. I haven't looked at it. But I agree with you that there's multiple ways to be wrong. And in the project that I'm doing with Geralyn Timler, we're digging around in the narratives in a little more detail. And so she's looking at micro and macro structures of narratives to see with the idea that ADHD would be more associates with the macro kinds of errors. The organizational pieces. And that maybe the language impairment would be linked to the connections across sentences, the discourse markers there. But that is, that's a rough slog to go through. I'm not doing it. But I agree. I wasn't expecting to see this as a subtractive phenomenon. But I think it does link into that observation that who ends up getting services from speech language pathologists? What puts you in the front of the queue? It is a Y chromosome and an attitude problem, apparently. And if we want to have caseloads that consist of individuals with Y chromosomes and attitude problems exclusively, then we're going about it the right way.

Question: Regarding the idea of preferential access to services by kids with ADHD and language impairment versus the kids with SLI, we've got a problem of these kids not drawing enough attention to themselves. Likely not being enough of a problem in a classroom. But even though the effects of this may not be visible today in this class right now, the long term effects are clearly something that teachers would want to address. How do you think we can change systems? How can we change the culture? Just if you have any thoughts.

Here's my manifesto. The world according to me. We're at a very clear disadvantage by the fact that we have no mechanism for tracking how many kids with language impairment are out there. We can't articulate whether the prevalence is getting higher or lower or how it varies across regions. And we can't rely on the Centers for Disease Control. And the Department of Education reports lump speech language of all different kinds together. So that's essentially worthless information for this issue. And so we need to, perhaps, institute our own census of what's going on out in the clinical world.
The irony of all this, of course, is that the reason why kids with specific language impairment don't get service is because they have a specific language impairment. And when I interact with clinicians and you check in. I learn very quickly, you never approach clinicians and ask them, can you refer your kids with SLI to me? Because they'll respond back and say I don't have any kids with SLI on my caseload. It's not how clinicians are parsing these things. I basically ask now, give me any kid who doesn't have an intellectual disability and I'll sort them out on my end of things. And so there's a clear disconnect between our research base and how that's translating into the clinical enterprise.
And some clinicians will tell you they don't believe that SLI exists. And that's a scary thought. I think part of what's going on there is that in our research discussions, I think it's important to look at potential nonlinguistic weaknesses in these kids. But if you start to inventory these things, then it starts to look like that there really isn't a thing called SLI. And if there's not a thing called SLI, then I don't need to go looking for it. And the consequence of that dynamic is what we see in front of us where there's 80 percent of these kids not receiving services.
I don't think the solution is for the research community to embrace the terminological confusion that's going on in the clinic, right. So if we tried to, if we try to adjust ourselves to move around the terms that are being used in the clinical enterprise, we have a choice of amongst a dozen different terms to go after. And as Mabel pointed out in her introduction, you cannot identify causal mechanisms when you have these convoluted discussions of: what is ADHD in a language disorder versus just a language disorder kinds of issues. SLI has a real scientific value in allowing us to identify what are the necessarily components of a language impairment.