Can we move beyond an instrumental approach to reading?
The Science of Reading: Information, Media, and Mind in Modern America by Adrian Johns. University of Chicago Press, 2023. 504 pp, $32.50
Fifteen years ago, when my first child was born, I did what all modern college-educated mothers do: I hit the books to research the most expert-endorsed methods of childrearing. But in all the reading about nap schedules and attachment theory the question that weighed on me the most was still a few years from being relevant: How was I to teach this child to read? In my search for an answer, I read through contemporary polemics, education research journals, ancient pedagogical treatises, and scores of century-old primers. Getting reading right felt like the most significant of my educational undertakings.
I was not alone in my anxiety. Few issues provoke as much ineffective educational policymaking as concerns about imparting literacy to young Americans. As University of Chicago historian Adrian Johns explores in his new book, The Science of Reading: Information, Media, and Mind in Modern America, the urgency of these questions dates to the turn of the twentieth century and has had consequences far from the reading classroom. And full disclosure: My graduate school transcript includes an incomplete for a class I took with Johns shortly before I had the aforementioned first child and dropped out of grad school without writing the final paper.
Johns is best known for The History of the Book, a study of the aftermath of the printing press. In it he takes issue with previous, more technologically determinist accounts. In this new volume, he again narrates a scientific history of unexpected applications and unanticipated consequences that carry us far beyond the page.
As Johns recounts, the science of reading originated in the 1880s discipline of psychophysics, which aimed to “provide psychology with the kind of rigorous, precise, factual foundations that the physical sciences enjoyed.” Perhaps unsurprisingly then, when psychophysicists turned their investigations to reading, they focused their attention on factors amenable to quantification: motion and speed. However quantifiable in theory, measuring speed and eye movement across the page required the development of new scientific instruments before much useful measurement could be done. These instruments included, in particular, the eye-movement camera and the tachistoscope—a device for precisely timing the exposures of visual stimuli.
It is by tracing the refinement and use of these instruments across the following century that Johns argues that this minor scientific subfield has had an outsized effect on subsequent history. His list of the offspring of the science of reading includes not only the obviously text-related fields—reading instruction pedagogy, library science, speed reading fads—but also less obvious phenomena. The latter include airplane cockpit design (improved by the use of the eye-motion cameras developed originally to track readers’ path across the page), Marshall McLuhan’s media theory (shaped in part by his visit to the lab of a prominent scientist of reading), and enduring conflicts over the deployment of scientific expertise in policymaking. Indeed, Johns argues that the furor provoked by the 1955 book Why Johnny Can’t Read set the template for subsequent battles, from climate change to vaccine safety.
Johns is also carefully attentive to the demand as well as the supply side of the science of reading, and there was plenty of demand. With the assistance of the new technologies of steam printing and stereotyping, the presence of print in our everyday environment increased on a staggering scale, involving not just books, magazines, and newspapers, but billboards, advertisements, memoranda, and other documents of the modernizing management-intensive professional environment. As has happened repeatedly in the history of written language, the amount of text seemed to be growing far beyond the ability of anyone to keep up, resulting in a sort of textual FOMO. “One must learn to read in self defense,” wrote James McKeen Cattell, the first professor of psychology in the U.S., one of the founding scientists of reading, and also the editor responsible for establishing Science as the premier venue for public science publishing. And self-defense required not just learning to read but learning to read in a new way.
Prior to the print explosion, reading for common people was, as historian Robert Darnton has written, “often a matter of recognizing something already known rather than a process of acquiring new knowledge.” You learned to read with the Lord’s Prayer, the Psalms, maybe a catechism. Reading was a sacred activity or a social one, and not infrequently both. Written words were still so tightly connected to their oral articulation that Ivan Illich argued medieval manuscripts are better thought of as “scores” than “texts.” Somewhere in the fourteenth to sixteenth centuries this began to change, but well into the nineteenth century reading instruction remained anchored to earlier reading practices. Aimed at elocution, it was expressive, oral, and social.
Perhaps that sort of reading was all well and good when your regular daily exposure to print was limited to a Bible, The Pilgrim’s Progress, and a shared newspaper. But by the 1880s it was no longer adequate to meet the demands of modern reading. It was too slow and too focused on the enunciation of individual sounds and words. Modern reading had to happen “at the speed of thought” so that the meaning of a text could be grasped as a whole. The type of reading that the scientists of reading privileged was silent, speedy, and extractive—the very type of reading they perceived themselves to practice.
And so, with the help of their new instruments and early efforts at standardized assessments of reading comprehension, scientists of reading diagnosed the first American reading “crisis.” Eugenicists (there were several among the scientists of reading) worried that proficient readers, assaulted with more text than was healthy, were in danger of reading too much, weakening the eyes and mental hygiene of the race. Others did not read enough because they could not. Scientists created a new category for the “functionally illiterate”—the large portion of the population who could slowly, laboriously, and haltingly decode some written text, but were not competent enough to enjoy reading or make it a daily habit. Johns sums up the anxiety that came to a head around the turn of the century: “How could this immense social machine—reliant as it was on sophisticated systems of engineering, management, finance, commerce, and politics—survive and flourish, if its people could not read the instructions?”
If the problem was a vast and ever-growing jungle of text that somehow had to be read and used, the solution seemed to be to get everyone reading more, faster. Corporate America and the educational apparatus of the state turned to the science of reading for practical strategies, demanding for its insights to be (often too quickly) operationalized into marketable classroom techniques and technologies.
Alas, insights from the science of reading have thus far not translated well to the classroom, as a recent report on the latest skirmishes in the reading wars shows. A 2020 report from the Department of Education found that 54% of American adults cannot read at a sixth-grade level. Despite over a century of scientific investigation into the act of reading, we seem no closer to achieving its goal of implementing “functional” mass literacy via mass education. And with the death of the bipartisan education reform movement and the post-pandemic culture-war turn in public schools, one wonders if we are giving up on trying to teach all Americans to read because it has proved too hard.
Far easier, it turns out, is teaching computers to “read,” particularly when you start with the instrumental reading privileged by the science of reading. As Johns documents, the “experiences, machines, and theories” of the science of reading in the first half of the twentieth century were easily “transcrib[ed]…into a new idiom of cognition” by the new related disciplines of cognitive, computer, and information sciences in the second half. Practitioners of these disciplines saw the concerns of the scientists of reading as consonant with their own concerns about “information-processing” and human creativity. Cognitive scientists looked back to the work of the early scientist of reading Edmund Burke Huey, considered the first to “characterize . . . reading as an information-processing activity.” Text became “information,” and readers’ speed and accuracy were now tested as a measure of “the human being as an information channel.”
Even if the terms have now changed, the problem that incited the science of reading remains. Indeed, the overproduction of “information” has only accelerated. Computers have made it still faster and cheaper to produce (digital) text. Now that AI can generate acceptable “content” with hardly any human effort at all, it is about to get much, much worse. The only thing that will possibly be able to keep up with it all are reading machines. Despite feeling a bit like a technological hair of the dog, as a tool for getting some sort of handle on a vast amount of information now in existence, reading machines are undeniably useful—I used chatGPT, along with other sources, to try to get a better understanding of roughly how machine reading actually works in preparation for this review. People of a certain age can remember how early Google Search felt a bit like a superpower, and it’s easy to see the appeal of an AI-powered smart search engine trained on more text than you or I could read in a lifetime.
But The Science of Reading reminds us that the type of reading we have “taught” machines to do is just one historically situated practice of reading, and that there are alternatives. Now may be the time to revive an understanding of reading that is formative and not merely instrumental, expressive and not merely extractive. The rising popularity of audiobooks suggests such a revival may already be underway. Now may be the time for a reorientation from information to language, starting with a recognition that not only is the average person now exposed to more written language than ever before but they also produce more written language than ever before. Once reading instruction was not taught in isolation but as one piece of a larger process of linguistic training. In its most systematized form, the Jesuit Ratio Studiorum, instructors explicitly taught several types of reading, but also directed the whole educational process toward the development, not of the student’s extractive reading abilities but of the individual’s own creative powers for communication and self-expression.
Johns sees in both the original scientists of reading and their cognitive and computer science followers a similar shared concern for “the autonomous, creative, exploratory properties of human nature.” This is an emphasis I also remember appreciating in Johns’s teaching. Alan Kay’s visionary Dynabook, which directly shaped the development of the personal computer, was conceived as an educational device that would have “a renaissance man effect”—Kay’s mockup of the Dynabook displayed the first page of Irving Stone’s novel, The Agony and the Ecstasy—but these hopes have not been fulfilled. Today when asked about the effects of personal computers on education, Kay analogizes to the effect of putting a piano in every classroom: “If there is no other context, you will get a ‘chopsticks’ culture, and maybe even a pop culture. And this is pretty much what is happening. In other words, ‘the music is not in the piano.’” Whatever progress we may have made technologically, we seem at best to be slowly and agonizingly recovering the pedagogical knowledge of earlier generations.
Technological determinists, both those who think AI brings salvation and those who think it brings our doom, together serve one comforting function: They relieve us of the responsibility of agency and, with it, the responsibility of education. The hope now seems to be that, in addition to reading the ocean of information we have created, reading machines will do a better job than us of solving the human literacy problem, too; Bill Gates recently predicted that within eighteen months, AI tutors will be teaching children to read.
Having now taught five children to read, I wonder how—and even if—he could possibly believe this. Perhaps a machine may one day be able to teach a child to read like a machine (though I have my doubts), but certainly only a person can teach a child to read like a person.
Sara Nardo lives with her husband, five children, and a dairy cow in Memphis, TN. She maintains an old-school homeschool blog, The Gothic Homeschool.