Routes to Reading

Maryanne Wolf, Mirit Barzillai, and Elizabeth Norton
Tufts University
Reading changed the course of intellectual development in our species. Over the span of approximately 5,000 years, we moved from societies in which literacy was extremely rare to a highly technological world in which the ability to read is practically a prerequisite for survival. Although many research paradigms attempt to reveal how the human brain learns to read, few are more interesting and instructive, perhaps, than the methods and questions of cognitive neuroscience. The story of reading's development is a complex tale of equal parts human invention and neural plasticity. The human mind created reading, but that skill could only come about because of the brain's unique capacity to form new circuits. Scientists have long known that reading depends on an intricate set of neural circuits in the brain, but the exact operation of these circuits remains an area of ongoing investigation. Now, a study by cognitive neuroscientists Laurent Cohen, Stanislas Deaene and their colleagues in the March 1st issue of the journal Neuroimage gives us some new insights into the reading brain. Cognitive neuroscienists often break down the study of reading into its components, including processes that are phonological (related to the sounds of language), orthographic (related to the way a language is written t), and semantic (related to meaning). In their seminal work, Cohen, Dehaene and their colleagues concentrated on orthographic processes. In doing so, they have pushed our understanding of what the brain does when it reads anything from the smallest features of letters; oft-repeated letter patterns (such as the "ph" or "ent" in English); to words that vary in length, frequency of usage, position (where the eyes fixate when reading them) and the overall quality, or legibility of presentation. Using imaging methods, Dehaene, Cohen and their colleagues have added to the evidence that the brain of an expert reader taps different "routes," or circuits, for well-known, routinized text, such as a the type you are reading right now, compared with text that is written in a way that is less familiar (For instance, try reading this word q u i c k l y. Or try reading The Canterbury Tales in Old English.) This research suggests that learning to read the different letter patterns in your language is similar to any other task that requires practice. At first, it requires conscious effort and focus on each letter. But then, after a period of practice, the task becomes routine and automated. Your brain is able to read the words without having to process them letter by letter. Strange Words The late, eminent cognitive scientist David Swinney of UCSD described how it is only in the acquisition of routines that later become automatic that we can see processes exposed before they become so smoothly conducted by our unconscious that they are impervious to our investigations. In their most recent study in Neuroimage, however, Cohen, Dehaene, Vinckier, Jobert and Montavont found ways to explore aspects of these larger questions through several inventive and probative methods with adult, expert readers. In essence, the researchers tricked the mature reading brain into revealing what it does when the text to be read is unfamiliar, and can't be automatically perceived and processed. Previous research by their group demonstrated that the brains of expert readers who are looking at typical, routinized passages involve parallel activation of letters and words by neural detectors in an area of the brain often referred to as the visual word form area (VWFA), a region located in the occipital and temporal cortex in the left hemisphere. Although the entire reading circuitry is not yet fully established, it is believed that this pathway (called the ventral system or route), processes words that are familiar at virtually automatic speeds. In the latest study, the researchers sought to determine the limits of the ventral system. They asked 12 adult participants to read words of different lengths that were either intact or degraded (transformed) in one of three ways: they were rotated up to 90 degrees in either direction; extended visually in length, with up to three spaces between letters; or shifted into the far right or left visual field. As one might expect, the more degraded the visual representation of these words, the longer it took to comprehend them. Furthermore, reading time was related to the length of the words only when they were very degraded, suggesting that degraded words were being consciously deciphered. So far, so predictable. Results from functional MRI scans, however, provided far more interesting insights. Although intact or slightly degraded words activated the VWFA and the ventral route, text that was highly degraded activated another area often described as part of the dorsal route. This route has been linked to letter-by-letter processing especially in children who are learning to read. (This is known as serial processing, since each letter is studied in sequential order, (i.e. from left to right in English).) Cohen, Dehaene et al. discuss their findings as further evidence generally supporting a dual-route model of reading in which factors such as development of reading skills, length and degradation of the text, difficulty of the prose, and familiarity with the type of writing influence whether serial or parallel processes are used to read a given word. In other words, the present findings suggest that, in adult readers, when the automatic parallel processes within the ventral system are unable to identify words, the second, serial processing system is invoked. Learning to Read The implications of this study are wide-ranging, providing lessons for reading researchers as well as educators of both developmentally impaired readers and children learning to read. First, in focusing on the distinction between automatic and serial routes of letter and word processing, the study contributes to the growing effort to understand the critical role that automaticity and fluency play in reading. For instance, converging evidence reveals that some children with dyslexia are unable to "switch" over to the left ventral route. The present research extends such work by highlighting the importance of understanding when and how young readers acquire a VWFA capable of automatic processing. Results from this study also suggest ways we can improve reading instruction in the classroom. For instance, educators should place a heightened emphasis on acquiring a repertoire of well-known letter patterns in a language, in addition to the current emphasis on training an awareness of phonemes, the component sounds that make up words, and on training knowledge of letter-sound correspondences necessary for decoding. Most importantly, all of these emphases should be explicitly connected when teachers instruct children who have difficulty reading (see for example our lab's work on the RAVE-O intervention where emphasis is placed on training several linguistic skills (orthographic, phonological, semantic, morphological, and syntactic knowledge) necessary for fluent reading in an integrated, systematic, and fun fashion ). Missing from the above discussion and from the present research study, however, is the less discussed role of semantic knowledge. Numerous studies tie semantic knowledge closely to comprehension, but questions abound about the role of such knowledge during reading acquisition. Where does semantic knowledge fit? How does it affect the rapidity of visual word perception? Does it affect the particular type of system used? Novelist John Updike said that a "good story ends with an open door." A good science story is no different. We look forward to the next contributions, particularly by these researchers.