It’s not a stretch to say that teaching children how to read is the most important thing we ask elementary schools to do. The old axiom in education is that children spend the first three years learning how to read, and the next nine years reading to learn. If schools fail in this most important task, students don’t stand a chance.
So, if teaching kids how to read is so important – if it’s the most important thing schools do – why are we so bad at it? Is it the schools? Is it the kids? Both?
I spent nearly 20 years as a journalist writing about the American public school system, and this fundamental question dogged me, no matter where I was working. In districts large and small, rural and urban, from New England to the Sun Belt, nobody had a good answer as to why so many kids couldn’t read.
It all crystallized for me in 1999, in a small-but-growing Florida school district north of Tampa, when I took on one of the most bewitching stories I ever chronicled as a reporter: How do you teach a dyslexic child to read?
Unfortunately, the answer to that question was: Most of the time, you don’t.
That seemed to be the future for a kindergartener named Chelsea, who was struggling to read. Her family was pushing the school to act with urgency because they knew what was at stake. Twenty years earlier, Chelsea’s father had drifted through the same school district, functionally illiterate, until he finally dropped out to sell shoes at the mall.
Teaching a child to read in the late 1990s was a lot like it is today: a high-touch, labor intensive process that relies on a host of hit-or-miss strategies, many of which lack any basis in research. As a reporter who visited a lot of different schools over the course of two decades, it always seemed to me that every campus (and even every teacher) had their own way of doing it. In some schools, the fuzzy concept of “Whole Language” was the prevailing theory, while others were heavy on phonics. Some blended the two. To an outsider, it really did look like a giant roll of the dice.
At the time I started following Chelsea’s progress, there were only a handful of educators and researchers who believed that reading was a highly specialized skill and that teaching it effectively required specialized training. Most colleges of education didn’t offer teachers-in-training instruction or guidance in how the brain learns to read because the science was so new. The prevailing theory at the time was that kids would “naturally” learn if schools submersed them in words and literature.
Right about that same time the hard science around reading and literacy was beginning to take shape. Human genome researchers had linked four chromosomes to dyslexia, thereby giving weight to something teachers long suspected but could never prove: Illiteracy can be hereditary. That certainly seemed to be the case with Chelsea–her family suffered from at least two generations of illiteracy.
Meanwhile, researchers at Yale University were using Magnetic Resonance Imaging machines to peer inside the brains of readers and dyslexics. When they compared the images, they saw striking differences in how brains processed the written language. At the same time, the National Institutes of Health had convened a National Reading Panel, which laid the scientific foundation for effective reading instruction that was rooted in phonemic awareness.
For reading and literacy advocates these breakthroughs were illuminating. Scientists were revealing evidence that specific types of reading instruction could, in effect, “rewire” the brains of dyslexic children and that literacy was absolutely obtainable for them and other struggling readers.
I, too, at the time was optimistic that the momentum and promise of these scientific discoveries would lead to new approaches and techniques related to reading instruction. The science seemed to be saying that illiteracy was, at best, the result of poor resource allocation in schools or, at worst, a policy decision. Surely, the new science would shift the landscape.
Unfortunately, by and large, the brain science and discoveries of these early reading researchers have had little to no impact on how schools teach children to read.
Why? National Public Radio’s Emily Hanford revisited the literacy issue in 2019. Her resulting documentary, “Hard Words,” put it bluntly: “Most teachers nationwide are not being taught reading science in their teacher preparation programs because many deans and faculty in colleges of education either don’t know the science or dismiss it.”
As a result, literacy rates among students haven’t budged since I followed Chelsea through her first two years of school. According to the white paper, Brain Science and Reading Instruction: The 2013 national assessment of 12th grade students’ achievement in reading was actually lower than it was in 1992. “Clearly, we have yet to find the ‘magic bullet,’ if there is one, for improving student reading achievement,” the white paper’s authors noted.
The brain science on reading might not be a “magic bullet,” but it’s clear educators have largely chosen to ignore it. Either that, or they don’t know what to make of it.
Today, I am no longer a reporter. But from my perch at The Learning Agency and our sister organization, the non-profit Learning Agency Lab, I’ve glimpsed a new generation of scientific breakthroughs that are embracing the discoveries of learning science. And while far too many teacher training programs continue to snub the science, entrepreneurs, researchers and learning scientists are using that knowledge to advance Artificial Intelligence tools, computer-assisted games, and algorithms that offer an amazing opportunity to advance the field’s ability to identify and intervene with dyslexics earlier, and with more efficiency.
One of the most interesting efforts uses eye movement recognition software to detect dyslexia in children with 89.8% accuracy. Researchers at the University of Jyväskylä in Finland funneled thousands of recordings of children’s eye movements into a database and then wrote a machine learning program to analyze them. The computer quickly learned which eye movements were “abnormal” when compared to those of strong readers.
“Tracking [eye movements] during reading is especially fruitful in the case of dyslexics, as it has been proven that readers with dyslexia have different eye movements than normal readers,” the researchers wrote. “Dyslexics display more and longer fixations, shorter saccades, and overall more irregular eye movement. Knowledge of this phenomenon serves as a valuable starting point in building a tool to separate normal readers from dyslexics.”
Researchers at Indiana University took a similar, yet different, approach. Instead of using machine learning to predict dyslexia from eye movement data, IU scientists used Optical Character Recognition software to analyze childrens’ handwriting samples.
The link between messy handwriting and dyslexia isn’t new. What’s new in this instance was the creation of a computer program capable of examining the handwriting of thousands of students and predicting – with 77% accuracy – which ones were problem readers. This potential tool is particularly exciting because researchers have noted that most teachers are “not aware of the handwriting difficulties” of their dyslexic students.
MIT and Florida State University researchers are developing AI systems capable of “listening” to speech recordings of early kindergartners with the intent of predicting which children will struggle with reading in the future. It’s hoped that they will be able to automate the speech screening analyses into a tool that is usable in all schools without expert administration.
Why am I optimistic about these new technologies when so little has changed on the reading front over the last 20 years? Primarily it’s because these tools are in development outside of the education system and they may not require specialized training or advanced degrees to implement in classrooms. Given that, we may not be stymied by the teacher colleges that Hanford noted in her NPR reporting, and that I also saw over the years.
Secondly, these tools are diagnostic in nature, meaning they are built to identify problems that we don’t yet see, or at least before they become insurmountable. And they’ll be data-driven. We’ve seen in other aspects of education that once a problem is identified, publicized, and defined with objective data, it becomes hard for schools and policy wonks to ignore.
Maybe then, schools will be forced to recognize, and implement, the science of how the brain learns to read.
– Kent Fischer