Reading, the Brain, and Teachers
By: Jamie Emerson
Stanislas Dehaene is a passionate and prolific advocate for the popularisation of neuroscientific research. He is the director of the INSERM-CEA Cognitive Neuroimaging Unit and author of a plethora of academic papers, but also writes and speaks to a general audience. His mainstream books focus on topics like the neuroscience of consciousness and mathematics, but most relevant for language teachers is his work on reading and the brain. Dehaene’s book Reading In The Brain: The New Science of How We Read was a key source for my last Think Tank article about reading disabilities (available here) but for this piece, I’d like to summarise a talk given to the World Innovation Summit for Education, or WISE and then draw some connections between this and classroom practice.
“Teachers”, Dehaene argues, “know more about the workings of their cars than they know about the workings of the brain of their children.” That is probably overstating the average teacher’s automotive knowledge, but his point is that neuroscience, cognitive science and psychology should inform teaching more than they do. Stanislas Dehaene truly is a man after the Think Tank’s heart. Or perhaps its brain.
Surrounded by text
Before exploring how reading works in our brains, it’s worth pausing and pondering what reading does in the world. We are surrounded by text to the point that we don’t notice it. For instance, looking up from my laptop and out of the window of the café in which I’m working, I see a dozen examples of writing – car license plates, For Sale and To Let signs, company names and slogans on vans, boards enticing potential customers into cafes and pubs, and details of a church’s services. We take written text for granted to the extent that we forget what it is – a visual representation of spoken language. Spoken language likely predated its written counterpart by hundreds of thousands of years, as explained by John McWhorter in this episode of the Lexicon Valley podcast. Evolution endowed us with the ability to speak, but not the ability to read. That said, nature did give us neuroplasticity, allowing our brains to make use of pre-existing areas and circuits for innovative purposes like reading.
Reading and the brain
Dehaene describes the act of reading as first activating the brain’s visual cortex like any other visual input, before activating the brain’s ‘letterbox’, or the occipital-temporal area near the back of the left hemisphere. What follows is an “explosion of activity” in at least one of two networks in the left hemisphere: one network related to meaning and another related to spoken pronunciation. This activation depends on the extent to which a word is already known to the reader. Both networks are active for novice readers, but as reading skill develops the role of the pronunciation network decreases because more and more words are familiar to the reader.
Dehaene addresses how much of this predates learning to read and how much is caused by it. What predates learning to read is the visual cortex, and the networks for both semantics and speech production. This makes sense, given that the vast majority of babies first see and then speak, well before learning to read. At a neurological level, then, reading is about connecting the existing visual and language systems. The letterbox area which is the basis for much of this connection is what changes in the brain as a person starts to read. In humans before they learn to read and in non-human animals like primates, the region activates in response to the sight of faces and objects. As a person’s reading proficiency increases, the area becomes increasingly activated by the sight of letters and progressively less activated by the sight of other things. At the same time, activity in the right hemisphere increases. Reading therefore seems to displace some aspects of visual recognition from the left hemisphere to the other side of the brain.
In his talk, Dehaene offers this shift in the occipito-temporal area’s specialisation as an explanation for the common phenomenon of ‘mirror writing’ in children. If you’re able, look back over your oldest school journals or those of your children and you may well see words or letters being written as the mirror image of their standard form. I’ve mocked this up in the picture below with letters “c”, “e” and “g”:
Dehaene rebuts the belief that mirror writing is a symptom of dyslexia and instead suggests that it is a normal part of writing and reading development, caused by the previous role of the occipto-temporal area. A face belongs to the same person whatever direction it is viewed from and so the ability to identify a face or any object from different angles is an advantage. This becomes a problem when the occipio-temporal area is repurposed as the letterbox because, unlike other objects, the identity of letters changes when viewed from other angles. A “b” looked at from another angle is “d”, for example. From my perspective as a teacher of English as a foreign language, I wonder if this explains some of the mirror writing I have seen in the handwriting of low-level learners whose L1 orthographic systems differ significantly from English.
Of perhaps greater significance for language teachers, especially those teaching young learners, is Dehaene’s take on the so-called ‘Reading Wars’ which pits phonics against the whole word approach to teach reading. Dehaene states that the whole word approach is based on the misconception that the units of reading for skilled readers are not individual letters but rather entire words. He suggests that neuroimaging studies show that although proficient readers process words much more quickly than novice readers, they still process individual letters. The key difference between the expert and the novice reader is that the expert can decode individual letters simultaneously, while novices do this serially, a letter at a time. This gives the illusion that skilled readers see whole words rather than individual letters.
Another point that Dehaene stresses is the amount of “self-teaching” that a child does when they learn to read. The underlying framework of the L1, its phonemes, grammar and lots of its vocabulary is acquired before a child learns to read, and so reading is a matter of the child learning to match graphemes to phonemes.
My one-year-old daughter, for example, will be intimately familiar with the song “The Wheels on the Bus” long before she comes to try to decode its lyrics written on a page or screen. When she’s ready, her task will be to perceive the correspondences between the symbols on the page and the sounds and meanings she already knows. She’ll need the explicit guidance of a teacher to work out the correspondences. However, Dehaene stresses, once these visual-auditory-semantic connections are made and strengthened in the brain, the child can read independently and thus start to access the meaning of words more directly, bypassing the auditory route and making direct visual-semantic connections.
Takeaways for teachers
What should foreign or second language teachers make of this? I’m generally cautious about applying insights from L1 research to L2 pedagogy, because of the many differences between acquiring a first language and learning another (see Meisel, 2011 for a detailed introduction). However, I think in the case of reading we can expect more similarities, because as noted earlier, unlike speaking and listening, reading and writing must be consciously learned, starting with direct, deliberate instruction. L1 and L2 reading are therefore closer in character than other aspects of language. Moreover, the neuroscientific evidence suggests that reading in an L2 uses most of the same areas and pathways as reading in an L1, albeit with additional activation of areas of the brain associated with cognitive control (Kupar et al. 2021). 1
1 Cognitive control is the brain’s ability to keep goals foremost and adapt behaviour and concentration to achieving them. It gives us flexibility and achievement.
In short, Dehaene describes the neuroscientific picture of reading as one of connecting existing systems and then strengthening these connections. The role of the existing systems should not be forgotten by language teachers. Remember that in our L1, we learn to read by using the language we already know and after a few years we progress to learning new language by reading. Coverage is the relevant technical term here, meaning the amount of a vocabulary in a text that we are already familiar with. The level of coverage required for good comprehension is high, with some estimates at 98% (Hu and Nation, 2000). Teachers need to choose texts carefully, using tools like Text Inspector or Vocab Kitchen to profile a text’s vocabulary and then editing accordingly. If working from a textbook with no choice in the matter, teachers can pre-text vocabulary prior to reading.
Similarly, the importance of strengthening connections is hard to understate. This requires practice and teachers or self-directed students need to make this repetitious and sometimes difficult task as tolerable as possible. Selecting or asking students to select interesting topics, working to build interest in textbook topics, and designing meaningful purposes for reading with information-gap tasks are all plausible methods for this. Encouraging extensive reading is another.
In his excellent, evidence-based and accessible guide to language learning, Paul Nation advises that extensive reading, where students read what they want for pleasure and lots of it, should make up about a quarter of course-time. Two thirds of this extensive reading time, he argues, should feature some unfamiliar words; “no more than two unfamiliar words in every 100 running words (about 1 unknown word or less every five lines).” At lower proficiency levels, this requires graded texts, because authentic texts with such controlled vocabulary are rare and primarily aimed at young children and thus uninteresting to older learners. According to Nation, the remaining third of extensive reading time should be for reading “very easy material in order to develop reading fluency.” In total, students should be reading, at minimum, 1-2 hours per week, or 20 minutes per day (“do lots of easy reading every week” Nation suggests). The two insights from Dehaene’s talk unite here: the large proportion of already known vocabulary means both that pre-existing linguistic knowledge is employed and reading large amounts is tolerable. Having students select topics based on their own interests aims to move reading beyond tolerance and into the realm of enjoyment.
While it is reassuring that extensive reading is supported by neuroscience, a more basic but perhaps more profound insight for teachers to take from Dehaene’s presentation is this: a spiky profile, where a learner excels in some skills but is weak in others, is the rule, not the exception. Remember, virtually all literate people, at one time in their lives, had sophisticated speaking and listening skills and a good grasp of the systems of their L1 with little to no ability to read.
Ultimately, although accessible, Dehaene’s explanation of the brain processes is not simple, but it might be worth studying. They give teachers another lens to use to view their students’ development, and to help understand some of their problems. As Paul Nation’s guide shows, integrating evidence from neuroscience into an approach to studying, teaching, and developing language programmes is possible.
Jamie Emerson (MA, DELTA) has taught, designed, and managed English courses since 2012 in the UK, Europe, South America, and Asia. He has written for a variety of academic and trade publications and spoken at numerous conferences. He works for Advance HE, a member-led charity for the Higher Education sector.