Language at the Speed of Sight
Language at the Speed of Sight: How We Read, Why So Many Can’t, and What Can Be Done About It
by Mark Seidenberg
A detailed introduction to the psychology of reading from cognitive, computational and neuroscience perspectives
Mark Seidenberg is a cognitive neuroscientist, now retired, but still advocating for improving reading instruction using insights from research. Language at the Speed of Sight is a detailed introduction to how we read, accompanied by an analysis of why reading instruction fails to take account of the research, and how that might be changed.
What it’s about
The author discusses reading from a range of perspectives including writing systems, cognitive psychology, computational modelling, neuroscience, and education.
After a brief introductory first chapter, he begins, in Chapter 2, with a discussion of the differences between reading and speech:
“Reading and speech share a common linguistic core, but the modality differences mean that some kinds of information are easily communicated in one but little or not at all in the other.” (p. 20)
Chapter 3 is a fascinating outline of theories about the origin of writing. Seidenberg’s observations of how writing systems are optimised to reflect the structure of the languages that they represent were particularly interesting. He points out that a full alphabetic system of representation could be considered superior to other systems because it’s a universal system in the sense that every language could be written using an alphabet of manageable size. But actually:
“The fact that every language can be written in an alphabet does not mean that doing so will be the most efficient solution. The goal is not to use the writing system with the broadest applicability; rather, it is to use one that is well-matched to the properties of the spoken language.” (p. 53)
For example, as Japanese has relatively few syllables, it can be written efficiently with a syllable-based system. In contrast, Hebrew can be efficiently represented with a consonantal alphabet because that reflects the way that consonantal roots are the basis of families of words.
Chapter 4 explores reading from a visual perspective. When we read, our eyes don’t move smoothly across the page. Instead, they move in a series of small jumps called saccades. This chapter explains how the limit of our perceptual span (the amount of information that we can take in in one fixation) makes it impossible to ‘speed read’ with effective comprehension. The author points out that “the amount that we can see on a fixation seems pretty well matched to how rapidly we can make sense of it” (p. 68).
Chapter 5 examines the statistical properties of written language. Seidenberg explains how, when we begin to read, every word is a unique pattern, but as we encounter large numbers of words, we pick up on regular patterns about which letters are likely to occur together and in what syllable positions they occur. As he humorously puts it:
“We gain orthographic expertise by reading. In the course of gathering all that spelling data, a person can also enjoy some books.” (p. 92)
Chapters 6 and 7 forms the core of the book, examining how reading develops, and the phonological (sound), semantic (meaning) and orthographic (spelling) systems that become connected during that process.
Dyslexia is the focus of Chapter 8, which discusses how dyslexia is defined, and how people with dyslexia can be identified. The author examines the causes of reading impairments, and also argues that there is little evidence to support the notion of dyslexia as a ‘desirable difficulty’ that leads to advantages in other domains.
Chapter 9, titled ‘Brain bases of reading’, is brief introduction to the main areas of the brain that are linked to the different components of reading that were introduced in previous chapters. Seidenberg describes how by evidence from acquired and developmental reading disorders, as well as data from typical readers, has shaped our knowledge of reading in the brain.
Chapters 10, 11 and 12 are perhaps of less interest to readers outside the US, as they describe reading standards in the US, problems with teacher training, and possible solutions. Nevertheless, the central point is relevant beyond the US: basic linguistic concepts and evidence from reading research should be incorporated into teacher training so that teachers know how to help children learn to read!
The best bits
Seidenberg’s academic research includes computational modelling of learning to read using neural networks, so he is well-placed to explain how these work. These networks are described in relatively simple terms, and the author describes how modelling connections between semantics, orthography and phonology can reproduce patterns of behaviour that we see in readers. Perhaps most interesting are the attempts to simulate reading difficulties caused by a phonological impairment by using ‘noisier’ input to the model – and the resulting slower learning and difficulty ‘sounding out’ nonwords within the model.
The less good
I have reviewed the first edition paperback, published in 2018. Since then, there has been an explosion of interest in the ‘science of reading’, to the extent that Seidenberg’s analysis of the state of reading instruction in the US is somewhat out of date. I notice that there is a new edition due for publication in 2026, and look forward to reading the author’s thoughts on the swing towards explicit teaching of many aspects of reading. Seidenberg’s recent blog post indicates that it will be a thought-provoking and possibly controversial read!
Who is it for?
Although written for the layperson, it’s not an ‘easy read’ – you’ll need to engage your brain to take it in. Mark Seidenberg has researched reading for several decades and he goes into a fair amount of depth. However, it’s well worth the effort. Although the content focuses on the processes underlying reading and reading impairments, rather than on pedagogy, I’d still recommend this to literacy teachers because a solid understanding of what’s involved in reading will help you to understand what needs to be taught and why some students struggle to read.
However, if you want a more direct focus on what reading research can tell us about how to teach, Christopher Such’s The Art and Science of Teaching Primary Reading (reviewed here) is more reader-friendly and practical.