That one time a journal article on speech sounds broke the SLP internet

This is the story of when a journal article broke the SLP internet.
 
On accident.
 
Oops.
 
And it was kind of a good thing, and kind of a bad thing, and mostly just really confusing to everyone observing it.
 
So allow me to play out this series of events—trust me, it’s relevant to why someone would ask the question above!

This fall, McLeod and Crowe published this article. Here's a paraphrase of our review of the article:

"This study provides a cross-linguistic review of children’s acquisition of consonants in 27 languages. The authors reviewed 64 studies from 31 countries to inform practicing SLPs’ expectations of children’s speech sound development across languages… SLPs can use this information and the overall patterns of speech sound development to decide what is “typical” across languages and to inform their clinical decisions.”

Sounds great for SLPs working with non-English-speaking children, right? What a great cross-linguistic resource! And when we sent our review to the authors (that’s part of our internal audit process) they said Yep, yep, looks good! And we published it.
 
Then all hell broke loose. Not because of our review. But because the authors released infographics describing their data. Here’s the infographic set that put U.S.-based SLPs in a tizzy. 
 
Now, here’s what SLPs thought this was:

• Oh, wow, “new” norms
• Oh crap, this is going to explode my caseload!!?!

 
So! We have some things to talk about here. I’ll try to group them into question themes, below:
​

Are these new norms?

​
​No, man! They’re not! McLeod and Crowe’s norms data wasn’t new OR different from what’s been published for the last several decades. Instead, this article is basically a summary of previous developmental norms studies. Think about it like the authors grouped previous data together, and combined it all so you could view an average across studies. First, they asked—at what age do 90% of children produce the sound correctly? They grabbed that data from our field's best studies, and averaged it. The infographics represent the average age, across studies, at which 90% of kids produce the sound correctly (or 90% of productions are correct; see article for details). Cool? It may look weird because it's displayed with categories instead of bar graphs or some other version you've been used to seeing. But, frankly, if you’re going to use any study at all for age of acquisition, it should be this one! Because it's aggregated data.

Why are these norms so shocking to everyone?

​Well, first you have to ask the question, “What norms are SLPs using now? And how are they interpreting them?” 

Google “Speech Sound Development Chart” or “Speech Norms”. See the massive range of what SLPs have been looking at? See how different they look at face value?
 
And that’s the part of the problem, here: often times, SLPs will look at a norms chart and interpret the ages to mean “age at which a kid should qualify for therapy”. Don't do that! All the norms charts look a little different because they’re based on different studies with different methods of interpreting and displaying the data. Some show bar graphs with the range of mastery, from 50% to 90%. Some charts show the average age of mastery instead. Sometimes "mastery" is defined differently. Some charts collapse across word positions. Or clusters. Or isolated words vs conversation. The list goes on, and on, and on…
 
But, basically, the point is this: When you look across multiple reliable sources for age of consonant acquisition and mastery, they’re not very different at all. Sources like this and this align well with these “new” (not new!) norms charts.

Guys, the details matter. Other bloggers and scientists have talked about this before. You can’t read all these charts the same way! And when you are reading the charts, make sure you're interpreting them correctly! And, whatever you do, do not interpret bars like this to mean that it’s “normal” to not acquire the sounds until age 8. That’s not what that says. Students qualify for therapy well before the tail-end of that bar graph. 

(Further, qualification decisions just can’t be based on norms charts alone because too many other factors are at play (e.g. patterns of errors). For a deeper explanation on qualification and speech sound disorders, see #week31 in the Clinical Research for SLPs Facebook group).

---

Are these norms accurate?

​Yes. I know that's not necessarily what you wanted to hear. But, yes.
​

So what does that mean for my caseload? What am I supposed to do?!

​Yeah. That’s the yucky part, right!?
 
Because the chart itself isn’t really the point, is it? It’s not that the majority of SLPs don’t know how to read these charts. I think the majority of SLPs do know how to read these charts just fine. The problem is that they’re concerned with the general take-home message, and that is:
 
Nearly all speech sounds are typically acquired by age 6; many by age 4.
 
And when we start thinking about what this means for our caseloads, it basically means that we should have a LOT more kids in speech therapy in the late preschool and kindergarten years. It means that waiting to treat things like “r” until after 1st grade isn’t well-aligned with the evidence on when speech sounds are mastered.
 
And that stinks. Because we want to be doing what’s best for these kids. But we're facing barriers (workload, caseload, silly rules our SPED administrators put in place...). So this chart doesn’t feel good to SLPs, at all. This makes us feel:
 
Annoyed
Sad
Frustrated
Angry
 
And that’s the part that non-SLPs just couldn’t see. That's the part that "broke" the SLP internet.

But I can still use a “wait and see” approach for speech sound disorders, right?

​Well, I mean. <sigh> Look, here’s the thing:
 
To-date, many of us have used a “wait and see” approach with speech sound disorders. That’s how we keep our caseloads under control, right? And, of course, those silly scientists who say that “r” should be treated in kindergarten are completely out of touch with reality, right? Right?! Eh—maybe a little. Maybe a little in that it’s easy for them to forget the workload barriers SLPs face. So when publications “come at you” with statements advocating SLPs do something—like getting these kids in speech therapy sooner—it can be annoying because we feel like we have little control over the issue. And then suddenly we have a workload discussion on our hands, and who do we turn to for help with that? (For real: for those of you wanting to help SLPs, we could really use a “Workload Fairy”. Somebody start that business!)
 
Because, seriously​. Nobody’s treating “r” in Kindergarten without first making massive changes to how SLPs work in the schools. (And in the meantime, the private practice SLPs just take these kids. From parents who are able to pay. Which—thank goodness for the private practice SLPs—but what we’re looking for here is speech therapy even for families who CAN’T afford to send their child outside of school.)
 
And that’s why this issue is really frustrating for school-based SLPs! Because we feel like we have zero power to fix it, and we’ve been treating speech sound disorders with a “wait and see” approach for so long that the concept of treating early just feels really foreign.

So, Meredith—how do you feel about the fact that neither your team nor any of the scientists involved in the early interpretation and dissemination of this study accurately predicted how SLPs would feel about it?

​Now, for me, personally, this is the part that’s most fascinating. And disappointing. And sad, frustrating, etc.
 
First, my team is usually really good about interpreting research exactly the way an SLP would. Not picking out the parts that scientists care about, but picking out the parts that SLPs care about. But none of us caught that when we read this study. Probably because these aren’t new speech sound norms, and showing “new speech sound norms” wasn’t the point of the study, at all. So it just didn't cross our minds. (See quote block in first paragraph for what the "point" of the study was, or read here.)
 
But! That doesn't matter. What matters is how and if it impacts clinical practice. So I am disappointed that we didn’t catch this immediately with our review.
 
But watching how this all played out needs to be a lesson for all of us. Things I learned:

1. Nobody sees science unless it’s shared. And shared the right way. With things like infographics. So BRAVO! to the scientists who did that, truly. They should be super proud that their study went viral.
2. Dissemination needs to be done by people who deeply understand the audience (the audience of this study was SLPs). Otherwise you get clinicians saying things like, “That study is wrong!” Why? Because you have to be able to understand how your audience will interpret what you say. So there’s a lesson here for scientists and people who care about science communication: your perspective isn't enough. It's your audience's perspective that counts. And you have to spend a lot ​of time with your audience to get that right.
3. Dissemination is ideally done by people with an audience. Now, here’s the part that’s really funny to me—those original infographics the scientists created? They’re quickly being replaced by copycat images, disseminated by people who know either, a) where to put these things so SLPs can actually find them, or b) already have a massive audience of SLPs (think—people who create products, like therapy materials, for SLPs). Don't believe me? Go Google “New Speech Sound Norms” or any other term you’d use to try to find those original infographics created. Can’t find them, can ya? But if you dig deep enough, you'll start to find other people’s images based off of the originals. (Or go search where you like! Pinterest? Instagram? Yep—a whole bunch of recreations of the originals, in just a few months’ time!) So what’s the lesson here? 1) Make sure to think about the location for dissemination. Will people be able to find it? 2) If you're a scientist, consider partnering with people who understand marketing. I’m a big advocate for more of our scientists partnering with people who have large SLP audiences. That’s a win–win for everyone.
4. Dissemination needs to be done with bias in mind. Careful analysis of bias. So that we don’t screw it up! For example:

• Our scientists may be biased by what’s “best”—as in, what minimizes adverse impact, what the best treatment case scenario is for our clients, regardless of what’s realistic (because they don’t operate in clinical reality on a daily basis). So, what's realistic, given our workload burdens as school-based SLPs? What's realistic, given that most SLPs know it’s far easier to treat tough sounds like “r” in slightly older kids because they are so much more self-aware? Sometimes it feels like our scientists are working under the assumption that tough sounds like “r” treated in Kindergarten is equivalent (equivalent in treatment intensity required to get success) to “r” treated later, when clinicians know that’s not the case. And, it's not that our scientists are wrong about sounds needing to be treated earlier. They're right. But they're often unaware of their biases and difference in perspective, and thus don't do a very good job of either communicating what needs done, or planning treatment studies in a way that directly addresses SLPs' concerns.
• Our clinicians may be biased by what’s “easiest”—as in, requires them to jump through the fewest hurdles to get success. Clinicians also tend to be biased toward the way things have always been. Because it's a pain in the rear end to consider doing something differently! And it’s hard to imagine doing things entirely differently if you’ve found success (or perceived success) in your current path. Clinicians are also biased by the fact that we see more kids with disorders on a daily basis than typically-developing kids. So our *gut* for what's normal tends to be a little low.

So what now?

At this point, I think the next step that matters most is understanding what questions still linger in SLPs’ minds, and using our science to address those. I would predict it’d be things like:

• Can I successfully treat tough sounds like “r” in Kindergarten? Or am I going to try, and fail, and waste a bunch of time?
• What would the impact be on my caseload of bringing kids into speech therapy earlier? Could I make that work? If I'm in the schools, is this more likely to be an RTI thing? Regardless of what we call it: who out there has made it work, and how? I want to see what that looks like.
• Say I choose to disregard this data and keep treating a little later. What is the risk of delaying treatment of speech sound disorders until kids are more mature (e.g. 2nd/3rd grade)? (Hint: Literacy is a huge one. Search #week9 here.)

 
Insert some comments, below, about what questions or concerns you have! Because, frankly, our science doesn’t matter much if it can’t be implemented. And successful implementation is where our focus needs to be now.
​
​

An addendum...

[Updated 1/24/2019] I was chatting with Marie Ireland, and she reminded me of something else that should be said as we're on this topic! Don’t forget that your place of employment may impact how you use information like this in the assessment process. In the U.S., for example:

• School-based SLPs provide services to children with disabilities who otherwise can't make progress in the classroom. 
• Private-practice SLPs provide services to children who may benefit from services.

 
Thus, a school-based SLP can’t use these norms alone to make eligibility decisions. Here’s what Marie had to say:
​

"When words like caseload and schools are used, we all need to remember that eligibility for special education—including speech services—has specific steps that have to be considered.

I worry that when we jump to discussions of when to start serving kids by age or sound error we miss the big picture—the big difference between clinical services and educational identification as a child with a disability who qualifies for speech in schools.

Now of course, if you are not in schools then IDEA doesn’t apply, but since caseload keeps getting mentioned we really should think about decision making requirements in schools. The most basic way to think about special education eligibility and services under IDEA is that to qualify for services you must:

1) have an impairment

2) that impairment results in an educational impact

3) the child requires specially designed instruction to make progress
 
Before we can ever determine the need for treatment in schools, we have to be sure we have data to answer all three questions for eligibility.

The article Evaluation and Eligibility for Speech-Language Services in Schools gives an overview of the regulatory requirements of IDEA and examples of some state differences. Important data like intelligibility ratings, stimulability probes, and any other state required data are needed to answer the IDEA questions.  We should be sure SLPs and decision making teams have data from teachers and parents looking at the educational impact caused by the articulation errors before we begin to discuss services in schools.

We need to remember that IDEA services in schools, including speech, are for students who are disabled. Giving services to a student who doesn’t meet all three parts of the federal definition of special education is inappropriate, results in overidentification, and is a violation of the child’s civil rights."   

~ Marie Ireland, SLP Specialist at the Virginia Department of Education
... read here and here to learn more about Marie's work.

Another addendum... because we just can't stop.

[Updated 2/27/2019] Two of my favorite colleagues, Sarah Bevier and Lisa Kathman from SLPToolkit, have a podcast called "True Confessions". Sarah and Lisa bring such clarity to the realities of being a school-based SLP, and incorporating evidence-based practices into that environment. With invited guest Dr. Kelly Farquharson, a top expert in speech sound disorders, the result was a wonderful episode, titled "We've been interpreting the articulation norms all wrong."
​
Dr. Tiffany Hogan, of "See Hear Speak" podcast also has a fantastic episode on this topic! See episode #3, "Speech norms, eligibility for speech treatment, and advocacy".

And finally, four new journal articles on speech sound disorders and qualification decisions were published this week as well! The clinical tips contained within are fantastic:

• Speech Sound Disorders in Schools: Who Qualifies? (intro to the forum)
• Standardized Tests and the Diagnosis of Speech Sound Disorders
• Using Developmental Norms for Speech Sounds as a Means of Determining Treatment Eligibility in Schools
• It Might Not Be “Just Artic”: The Case for the Single Sound Error
• Eligibility and Speech Sound Disorders: Assessment of Social Impact

Between this blog, the podcast, and these four journal articles, you'll be set with understanding this topic!

[Updated 8/8/2020]

• A new paper out on this topic! On only U.S. children. Infographics to accompany it, here. 

Written by Meredith Harold, PhD, CCC-SLP