Where the Trouble Begins
There is more to dyslexia than meets the eye. One researcher believes the problem starts with sound

Photo by Steve Prue
Reading is a complex skill: you move your eyes across a page, sound out words, recognize visual patterns, retain information, and build sentence structure. There are lots of places where things can go wrong. And for people with dyslexia, who have no trouble comprehending spoken language, what goes wrong is their ability to decode words in print.
“There is something fundamentally different about the way their brains handle language that permits spoken communication but makes written communication challenging,” says Tyler Perrachione, an assistant professor of speech, language, and hearing sciences. The difference, he believes, may be in the way their brains process variability, or variations in word sounds.
Researchers have identified two hypothetical—and historically competing—models to explain how the typical brain processes variability. According to the episodic model, the brain recognizes speech by comparing it to specific words it has heard before.
Watch: Tyler Perrachione, an assistant professor of speech, language, and hearing sciences, explains how communication works in the brain. Video by Hesse Costa, WBUR
According to the abstractionist model, the brain strips away variability to access the underlying, or abstract, sounds of the words. When listening to speech, the brain also develops a catalog of information about the speaker’s voice; as the voice becomes more familiar, the speaker’s words become simpler to understand. If two people with different accents speak the word phone, for example, the listener’s brain removes the accents to focus on the phonology, or the fact that phone is constructed of the phonemes (speech sounds) f-ō-n. It is here that Perrachione thinks the trouble begins for individuals with dyslexia.
When shifting from listening to reading, an individual must match word sounds to letters, recognizing that the f sound in the word phone is the same as in the words floor, finish, and physical. Because letters can combine in many ways to represent speech sounds, learning to read relies on having strong abstract representations of these sounds in the brain.
While researchers habitually debate which model the brain uses to process speech, Perrachione believes that the typical brain employs both: the abstractionist model enables it to recognize words more efficiently, while it can fall back on the episodic model to recognize words with more effort. In his research, Perrachione aims to discover whether individuals with dyslexia rely on one of these models at the expense of the other.
“When children with dyslexia are reading words and making decisions about sound, they’re bringing their language network online less than children without dyslexia.” —Tyler Perrachione
In collaboration with colleagues at Massachusetts Institute of Technology and Massachusetts General Hospital, Perrachione used functional magnetic resonance imaging (fMRI) to measure the brain metabolism of first- and second-grade children with and without dyslexia. He instructed the children to read a series of words to themselves and press a button when two consecutive words began with the same sound. Perrachione found that children with typical reading skills showed more metabolic activity in areas of the brain that are related to reading and language. They also showed more activity in areas related to hearing because they heard the words in their heads when reading silently.
In the same study, children with dyslexia activated smaller areas of their brains. “When children with dyslexia are reading words and making decisions about sound, they’re bringing their language network online less than children without dyslexia,” Perrachione says. They’re not activating the sound areas because they have difficulty connecting the phonemes to the printed words; for instance, they have more difficulty recognizing that the word phone comprises the sounds f-ō-n.
Typical readers use both methods of processing language, Perrachione says, which allows them both to recognize new words and to process familiar speech efficiently. They compare only unfamiliar speech to their catalog of stored words. Children with dyslexia seem to compare every word to their catalog, which requires their brains to exert more effort.
“If they don’t have the abstract phoneme representations of speech sounds to serve as an intermediary, then it’s harder to get to the word,” Perrachione says. “It’s being blocked by not having a good way to translate between print and speech.” Perrachione’s goal is to devise training strategies to help children with dyslexia perform this translation process, enabling them to read more quickly and efficiently.