I grew up in a tiny New York City apartment, packed in alongside our four cats and my father’s immense personal library of some 3000 books. My father designed books for a living, and he revered them. His books were everywhere in the apartment, covering every possible surface in the house, the radiators and toilet tanks included. To my father, these books were objects of art: beautiful to hold, beautiful to look at, and beautiful to read.
Though my father’s outsized romance with books didn’t entirely rub off on me, he did instill in me an appreciation for the book as a technological invention, a remarkable piece of engineering whose importance is arguably like none other ever devised. And yet, given that at its core reading is nothing more than a tool, engineered around a set of compromises and constraints, it’s far from perfect.
Unfortunately, the system of reading we inherited from the ancient scribes —the method of reading you are most likely using right now — has been fundamentally shaped by engineering constraints that were relevant in centuries past, but no longer appropriate in our information age. When books were scarce, and few people could read, the fact that some inherent flaw in the design of reading may have hindered reading was not much of a concern. But today, in an era of computers —where it is possible to instantly download virtually any book ever published and read it on a device we carry in our pockets— what limits our reading is the capacity of the brain to absorb the available content. The problem of our millennium is that we simply can’t seem to get the information into our minds fast enough to satisfy our needs.
What may have seemed like a good design idea for reading a millennium ago, may not be such a good idea today. Take for example the shape and spacing of letters. In the centuries before the printing press was invented, it took scribes an inordinate amount of time to pen a book. Therefore, the need to work quickly likely influenced the design of our letter shapes, as these were formed to make their drawing by hand as fluid and rapid as possible. Other design constraints we inherited from the past, no longer relevant today, relate to the high cost of materials. Parchment was expensive. So letters were designed to compress and cram the symbols on a page — so much so that spacing and even punctuation came to be omitted and had to be introduced in the 12th Century to facilitate reading.
While cramming symbols tightly together may have seemed like a brilliant way to save on parchment, scientists are now beginning to understand that this design decision runs afoul of the way the brain processes visual information. It was demonstrated in the 1970’s by Herman Bouma of the Netherlands that when similarly shaped objects (such as letters) are clustered tightly together, this clustering interferes with the brain’s ability to discern the elements of the cluster, a phenomenon known as “crowding.”
While scientists do not understand why crowding occurs, its effects are easy to see. Consider the following string of letters: Dwzrh k wbp. Can you make out the letter “k” while looking directly at the letter “D”? Most people can. But now see what happens when the “k” becomes crowded: Dwzrhkwbp. Most people find it impossible to make out this letter once spacing becomes tight. This phenomenon limits the number of letters we are able to perceive at a glance, and it was shown in 2007 by Denis Pelli of NYU and his colleagues that crowding fundamentally limits our speed of reading.
Recognizing that it may now be time to reconsider the traditional design of reading, a disjointed band of modern-day Guttenbergs — comprised of scientists and technologists from around the world — are beginning to search for innovative engineering solutions aimed at making reading more efficient and effective for more people. Their ultimate goal is to re-invent reading so that we are no longer limited by anachronistic design constraints imposed by the quill and pen, and instead devise a new form of reading limited only by the capabilities of the human brain. Our team at the Laboratory for Visual Learning — a collaboration of scientists from UMass Boston, MIT, and Harvard, that includes Marc Pomplun, Chen Chen, and me, is honored to play a role in this research.
The reason I became interested in this field is because I have dyslexia, and reading was always a challenge. Reading for me was so painful and non-productive, in fact, that for a period of about 20 years I almost stopped reading altogether. But then, by chance, I discovered that when I used the small screen of a smartphone to read my scientific papers required for work, I was able to read with much greater facility and ease. Intrigued by this personal discovery, with funding from the National Science Foundation, we first tested this effect in 2010 in a small sample of college students, and found the technique had promise. Then, in a comprehensive study of over 100 high school students with dyslexia done in 2013, using techniques that included eye tracking, we were able to confirm that the shortened line formats produced a benefit for many who otherwise struggled with reading.
We weren’t the only ones to notice such effects. Though smartphones hadn’t yet been invented when they did their original research in the late 1980’s, Gadi Geiger and Gerome Lettvin at MIT, as well as Keith Rayner and his colleagues at UMass Amherst, discovered instances of individuals whose struggles with reading were diminished by restricting the span of text processed during reading. And later, when the use of computers, smartphones, and e-readers became commonplace, more people began to notice this short-line effect, including many non-scientists like David Knight who was inspired to create a web process to help people shorten lines, he called “Friendly Type,” based on this effect. Other beneficial format modifications were also discovered. For example, Marco Zorzi and his colleagues in Italy and France showed in 2012 that when letter spacing is increased to reduce crowding, children with dyslexia read more effectively.
Using the redesigned text on a smartphone, I was able to break my fast and begin to read books that had previously bedeviled me. And while this helped me rediscover the joys of reading, it also served to arouse my curiosity as a scientist: Why would something as simple as shortening the lines of text suddenly make it easier for me to read? Our theory, based on our eye tracking study, is that short lines help some people who otherwise struggle because they serve to guide attention during reading, and promote forward tracking in the text.
Shortening the lines, though simple, is not the only way to guide attention in reading. A clever web application called Beeline Reader, developed by Nick Lum, a lawyer from San Francisco, may accomplish something similar using colors to guide the reader’s attention forward along the line. Beeline does this by washing each line of text in a color gradient, to create text that looks a bit like a tie-dyed tee-shirt. The colors of adjacent words gradually bleed from one color to the next, to perhaps help guide attention forward. Though research has not yet been carried out to confirm this, Lum reports that many users find this technique beneficial in reading.
Another very popular method researchers have used to minimize the costs incurred in tracking is to use a computer to flash words sequentially, one after the next, in the same spot on a computer screen. Called rapid serial visual presentation (RSVP), this method was first introduced by Molly Potter and her colleagues at MIT in 1976. Using this method, Potter’s team showed that the brain can extract useful information from text or other visual scenes in as little as 15 milliseconds, compared with the 200 milliseconds it takes the eye to move from word to word.
In principle, RSVP should be able to produce a ten-fold increase in reading speed. But, a number of researchers, including Gordon Legge at the University of Minnesota, one of the foremost pioneers in this field, found that the gains attained using RSVP were much more modest. While large gains were possible when the words being flashed were random and unrelated, when they were sequenced to form sentences, and comprehension was required, only modest gains could be achieved. It was thought that one reason for this limitation is that it takes time for the brain to process and interpret the words and sentences, even when RSVP eliminates the need to move the gaze and track the line.
Reading was invented as a way to save speech on paper, and so when reading speeds are compressed using RSVP, comprehension becomes limited by the brain’s ability to rapidly process speech. Normal speech is a bit slower than the speeds at which most people read — about 150-200 words per minute (wpm). Remarkably, however, a team of researchers led by Laurianne Vagharchakian in France in 2012 found that when speech was compressed using a computer, and played at dizzying speeds —about 600 wpm— people could still comprehend the speech quite well. But when speeds were increased even more, comprehension abruptly collapsed. When they examined how the brain responded to this ultra-fast speech using MRI imaging, they found that the higher-level language processing regions became saturated with information when the speed barrier was crossed. It was as if the brain had run out of space to temporarily store the incoming stream of information, overwhelming the limited speed of the language processing areas. The result was that, whether or not the hyper-accelerated information stream was auditory (compressed speech) or visual (RSVP reading) comprehension abruptly failed.
Accelerated reading thus appears to be fundamentally limited by a bottleneck in the speed of higher-level language processing. If so, how can we ever hope to devise a technology that will allow us to keep pace with the ever-increasing demands for reading? To address this question, our laboratory is investigating two possible technological solutions: one aims to increase the throughput of the brain’s reading buffers by changing their capacity for information processing, while the other seeks to activate alternate channels for reading that will allow information to be processed in parallel, and thereby increase the capacity of the language processing able to be performed during reading.
The brain is said to be plastic, meaning that it is possible to change its abilities. Zvia Breznitz and colleagues at the University of Haifa demonstrated that when people are forced to practice reading using a process they call “reading acceleration program” (RAP), people can be taught to roughly double their reading speed, without compromising comprehension. RAP uses a computer to enforce reading at a fixed rate, by gobbling up text like in a game of Pac-Man. People read at accelerated speeds that are gradually increased over time. This practice revises structures of the brain used for reading, so that people can maintain an accelerated reading pace months later, even without the use of a computer to drive their reading.
While RAP shows considerable promise, we feel there is much more that can be gained by freeing the design of reading from its vestiges in the pen and quill, to rethink the process from scratch. Consider that we process language, first and foremost, through speech. And yet, in the traditional design of reading we are forced to read using our eyes. Even though the brain already includes a fully developed auditory pathway for language, the traditional design for reading makes little use of the auditory processing capabilities of the brain (except, perhaps when reading aloud, or subvocalizing our words). While the visual pathways are being strained to capacity by reading, the auditory network for language remains relatively under-utilized. This then suggests the possibility that the auditory network can be used in conjunction with visual reading to create parallel pathways for reading in the brain that can be used to accelerate processing.
In our laboratory, we are investigating a potential re-design for reading that intends to break through the brain’s speed barrier for comprehension. We intend to do this by building on the neurological circuits for auditory processing, and use these in parallel with reading. Here, people read using a highly accelerated visual presentation of the text (e.g., using RSVP or RAP), while at the same time they listen to a highly compressed auditory rendering of the same text, using compressed text-to-speech. Forcibly accelerating the rate of language processing in both the visual and auditory domains simultaneously, the brain can process the information in parallel. Our theory is that the high-speed language inputs from each of these perceptual streams will reinforce the other, to effectively increase the buffering capacity of the brain, and allow people to read with comprehension at faster speeds than is possible with either technique alone.
Preliminary indications from a controlled experiment, conducted in our laboratory at UMass Boston, have been encouraging. Here, we examined new methods for reading, wherein 40 college students, both with and without dyslexia, read using a smartphone driven by Voice Dream reader, using RAP concurrently augmented with compressed text-to-speech (along with other visual modifications to address crowding and attention). Though our results are still preliminary, what we are finding is that this new method of reading is far more effective (in terms of comprehension and speed) than any method attempted so far, whether people have dyslexia or not. Importantly, our early indications suggest that the least effective method of reading may be the one society has been clinging to for centuries: reading on paper.
What our research suggests is that the current methods we use for reading — based on ancient engineering constraints no longer relevant in today’s society — make only limited use of the brain’s capacity for information processing. As technology continues to change how text is delivered and consumed, all of us will soon be able to make use of this research — carried out by the world’s community of modern-day Guttenbergs —to push back the brain’s speed barrier, and read with greater speed and efficiency. And while in the past such an evolution in reading may have taken thousands of generations to take hold, with the explosive growth in computer technology, it is very likely such changes will become available for all to use in less than a single generation.