IE 11 Not Supported

For optimal browsing, we recommend Chrome, Firefox or Safari browsers.

Can AI-Powered Tech Help Educators Eliminate Dyslexia?

Rather than letting learners cope with the lifelong struggles associated with dyslexia, one technology company is using an artificial intelligence-enabled tool that corrects problems and sets them on a course toward proficiency.

BoyReadingLaptop
CCO 1.0
K-12 school districts and families across the nation are testing an artificial intelligence-powered tool that, according to its developer, “dissolves” language processing problems rooted in dyslexia.

Dr. Coral Hoh, CEO and founder of Dysolve, said some of the results from University of Delaware’s ongoing clinical research from 16 school districts in seven states will be released this winter, along with testimonials from college students who used an earlier version of the tool to correct their dyslexia when they were in middle school. Her goal is to help schools nationwide replace existing dyslexia diagnostic and coping and corrective methods with better technologies.

“Language works very fast. This is why we need computing,” Hoh said. “[Existing methods] still leave people struggling with it for the rest of their lives.”

Dr_Coral_headshot.jpg
Dr. Coral Hoh, founder and CEO of Dysolve
Image courtesy of Dysolve
The Dysolve website notes dyslexia as an “unexpected reading problem” where the student’s intelligence, vision, hearing, school and home environments, literacy training and motivation do not play a role in their inability to comprehend written words.

By trade, Hoh is a linguist. She says the 40-plus years of research on the spectrum of dyslexia has been useful in determining root causes, like brain disruptions where the student is trying to process too much information in the small working memory space.
This occurs when language processing should be moved onto the subconscious to allow people to perform language tasks without having to “think” about them. This then frees up working memory space for attention, memory and learning.

Hoh, who is based in upstate New York, said the inspiration for the tool dates back 30 years, when she tried to help a friend who struggled with reading. She used her own evaluation methods to pinpoint that comprehending vowels was her friend’s biggest obstacle, “but I didn’t know how to correct that problem.”

She would build on that experience, later teaming up with mathematicians and software engineers who shared her passion for decoding natural language. The earliest version of the Dysolve software was released in 2017. Five years later, they partnered with the University of Delaware to conduct more comprehensive research on the cloud-based tool. The clinical trials in Mississippi, Kansas, Illinois, Ohio, North Carolina, Wisconsin and Louisiana are expected to continue until at least the end of the 2023-2024 academic year, though some findings will be released before then.

Her body of work in linguistics and past research on dyslexia, including the body of knowledge on phonics, laid the foundation for the tool, she explained. Its artificial intelligence engine quickly assesses the user’s capabilities and, in real time, creates a personalized game for the user that hones in on their reading deficiencies and corrects them in an engaging manner. A typical student diagnosed with dyslexia would use the tool about 15 minutes a day, three to five days a week.

Dysolve generates evaluation and learning “cells” that target the individual user’s problem, according to the company website. Those cells then retrain the user’s brain to process the aspects of language that were causing the reading deficiencies while also continually monitoring progress.

“It’s autonomous AI all the way through the evaluations and corrections without any human support or instructions,” Hoh said.
Dysolve - Jellyfish game.png
Dysolve's dyslexia diagnosis and correction technology involves interactive features that help to pinpoint where students are struggling with reading.
Image courtesy of Dysolve
A short demo of a personalized game on the website cartoon shows jellyfish floating in an underwater lost city scene. The game evaluates the child's ability to break the spoken word, "keeps," into its sound segments. The child hits the up arrow key on their laptop for each sound segment, sending a jellyfish up to the surface. If the child cannot tell the correct number of sounds in a spoken word, Hoh said, then spelling and reading would be difficult for an alphabetic writing system like English. She explained that a soothing water theme is used for these games because language processing difficulties can make children feel anxious.

Each game, Hoh added, asks users to complete one task. This allows Dysolve AI to isolate the language process targeted for the purpose of assessments and corrections.

Hoh said Dysolve’s success will be measured by the same school assessments that apply to all students, whether exams, standardized testing or report cards. Because the tool is still being tested, schools are still invited to use Dysolve for free. Individuals can sign up for paid subscription-based services, Hoh said, adding that one of her marketing pitches is its ability to handle large volumes of students and free up school staff.

“Whether the school is big or small,” she said, “we can accommodate everyone.”

Editor's note: Clarification was added to the section describing the relationship between language processing and working memory.
Aaron Gifford is a former staff writer for the Center for Digital Education.