Reading Fast and SlowPrint
The speed at which our eyes travel across the printed page has serious (and surprising) implications for the way we make sense of words
By Jessica Love
March 1, 2012
In 1986, an Italian journalist named Carlo Petrini became so outraged by the sight of a fast-food restaurant near Rome’s Spanish Steps that he ended up spawning a movement. The Slow Food movement spread across the West, with adherents championing all things local and artisanal—promoting the humanely raised, sustainably fed, ethically killed, home-cooked chicken, say, over a 10-piece order of McNuggets. These days, it isn’t just food that’s slow. We have proponents of Slow Gardening, Slow Parenting, Slow Travel, Slow Fashion, Slow Media, Slow Art, Slow Money (which urges investment in local food enterprises and organic farms), and Slow Software Development (favoring careful, superior design). Taken as a whole, the movement is a reaction against those aspects of modern Western life that sacrifice quality and connectedness at the altar of convenience and efficiency—which is to say, much of modern Western life.
Should we be surprised that a Slow Reading movement has emerged as well? Though its advocates are a motley crew, refusing to converge on a single leader or even a single name (a Slow Books movement is also in the works), John Miedema’s 2009 book, Slow Reading, is as good a place as any to turn for an official rallying cry. According to Miedema, “Slow reading means exercising choice about how one reads rather than being forced to read as fast as possible.” The movement is not primarily about speed, but about freedom: the freedom, at least on occasion and when the text demands it, to read as though we care about something other than being done reading. In his essay “Reading in a Digital Age” (published in the Scholar in Spring 2010), Sven Birkerts cautions, “The reader who reads without directed concentration, who skims, or even just steps hurriedly across the surface, is missing much of the real point of the work.” This is, I believe, absolutely true. But according to psychologist Victor Nell, the heaviest readers—those who presumably derive the most pleasure in the act—are in fact marked by the variability in their reading speed: they skim a text and hurry along just as much as they linger and contemplate.
So exactly what happens when we read at a gallop? When we read at a trot? It behooves us to understand the reading mind, because the manner in which we read has real consequences.
But first, the mechanics. Reading a sentence, our eyes seem to glide across the page, stopping each long, fluid sweep only at the margin. They seem to glide, but in actuality they hop in a volatile way, landing here and here and sometimes here. Psychologists who study reading call these landings “fixations.” Each time we fixate, we are able to take in approximately four letters to the left of our fixation point and some 15 letters to the right. We’ve trained our perceptual span—the range of text we can effectively discern during a fixation—to be asymmetrical in this way because English words are printed from left to right and it’s more advantageous to see where we’re going than where we’ve been. Were we readers of Hebrew, which is written from right to left, our perceptual spans would take in more letters to the left than to the right.
Thus, with each fixation we glean information from about 20 letters. We can’t necessarily identify all 20 letters: the outermost are relegated to our fuzzy peripheral vision. We may learn only that one letter is capitalized or another contains a curve—information that will help us identify that letter on our next fixation but isn’t going to do the trick on this one. Generally, to decode what’s in front of us, we must fixate every eight letters.
Decoding begins quickly, within 60 milliseconds of landing. The shape of the letter string as a whole—not just its constituent letters—assists us in our endeavor, so odd fonts or cApiTAlizAtiOn patterns slow us down. But before we’ve finished, before we consciously know we’ve read the words fonts and capitalization and have added these to the other words swimming around most prominently in memory, we have already started to plan our next fixation.
During a hop, known as a saccade, we are unable to perceive any letters at all. Should some meddling experimenter briefly alter the text we are reading—change “sense” to “sensibility” and then change it back, for instance—we will never be the wiser. Given that fixations average 200 to 300 milliseconds, and saccades 20 to 30, we spend about 10 percent of our time blind to what we’re reading. But we have learned to “mask” the saccade, to fill in this brief period of blindness with perceptual information from before and after it. We have tricked ourselves into seeing a smooth flow of sentences when we’re really getting a choppy surge of words and half-words. Some quick math suggests that the absolute fastest any of us can read—and actually read each word—is 500 words per minute. This assumes no backtracking, though nearly 15 percent of our saccades are regressions. It also assumes no comprehension beyond word identification. In practice, most of us read about 250 words per minute.
So where do speed-readers fit into all of this? A few empirical studies do suggest that reading speeds far beyond 500 words a minute can be attained, though these studies, for various reasons, are terrible. One found, for instance, that after 16 hours of speed-reading training, participants could read significantly faster than they could beforehand. However, the study didn’t include any control groups, so the participants could have simply familiarized themselves with the experimental procedure from the first test to the second, received easier reading materials the second time around (or materials about a more familiar topic), or even—a palpable fear for this Big Ten researcher—had their baseline reading rates established on the restless Friday afternoon before a football game. Other studies boast impressive improvements but rely on a device called a tachistoscope, which flashes words at participants for briefer and briefer periods of time until—voilà!—they are reading very quickly indeed. But sans tachistoscope (and most of the rooms in my house are, sadly, tachistoscope-free), readers must once again saccade before a new word will appear, thus regaining the reading speeds of mere mortals. Finally, perhaps most disturbingly, some studies never required participants to answer a single question about the information contained in the texts they read, making it impossible to determine whether the speed-readers actually, well, read them.
More than a decade ago, psychologist Keith Rayner, probably the most widely respected expert on what the eyes do during reading, pinpointed a single study of speed-reading as both rigorous and interpretable. In that study, psychologists Marcel Just, Patricia Carpenter, and Michael Masson monitored the eye movements of speed-readers (reading at 600-700 words per minute) and normal readers (250 words per minute). The researchers determined that speed-readers were able to read so quickly because they made fewer fixations than did normal readers: instead of reading every word, the speed-readers sampled the text, reading a few words here and there. When tested on what they’d learned, these speedsters did relatively well on general questions—those they could answer from the fragments they’d read and their preexisting knowledge about the topic. But when tested on the minutiae of the text (the minutiae that had not been sampled, that is), they fell apart. The killer, though—the thing that should keep you from reaching for your wallet, should you feel tempted by a pricey weekend speed-reading seminar—is that the researchers also instructed the normal readers to merely skim the texts at 600-700 words per minute. When the skimmers’ eye movements and comprehension test results were held against the speed-readers’, the match was uncanny.
There may be people whose eyes really can reach 2,500 words per minute, and reputable psychologists just haven’t tracked them down yet. Some people can also twist themselves into a breadbox or remember exactly what they had for lunch on February 16, 1944. There are always folks on the bell curve’s most asymptotic reaches; indeed, I would never have wanted to call John F. Kennedy—perhaps the most famous speed-reader of all—a liar. But when I gallop I am skimming, and so, most likely, are you. We’re letting our eyes move down a line or two every time the ratio of detail to action becomes too lopsided. (And isn’t it always the description that gets to us? Halfway through Marilynne Robinson’s Housekeeping, a friend and I made a game of yelling, “Simile!” every time things were like or as other things—which was often. We’d have given anything to skim, but the medium was, alas, book-on-tape.) Why can’t we remember Falstaff’s first name, or which American city was home to the grown-up Amir in The Kite Runner? It may have been because we simply forgot. But it may have been because we never really read it.
One might assume that given how smart we are and all the nifty tricks we’ve developed to mask saccades and get our perceptual spans to bulge to the right, we’d at least be decent skimmers. One would think we’d be able to skim somewhat selectively. If we want to skim Walden for survival tips, aren’t we able to do so? Unfortunately not. We’re okay at skimming texts for particular words, at treating “Self-Reliance” like a giant word-find puzzle, only if we know exactly what we’re looking for. We can’t get around having to read something to know what it says, and we have to know what it says to know if it is relevant. But because skimming is essentially sampling, we’re just as likely to skip over the relevant as the irrelevant.
The only thing that prevents skimming from being an entirely random pursuit is that some information becomes legitimately available to us before we read: we know where something is on a page without knowing what it says. Therefore, because the first sentence of a paragraph often gives us clues about what will come next, we might safely skip the rest of a paragraph if the first sentence is about pedometers and we seek the ratio of sugar to water most attractive to hummingbirds. Unsurprisingly, this “topic sentence” method appears to be particularly helpful for skimming webpages, where graphics and bullet points corral the eyes wherever content providers or marketers see fit. I have strong, if anecdotal, evidence that this same technique rarely works with freshman comp papers.
But it is doubtful that instructors could skim those even if they wanted to. Difficulty slows readers down, and awkward wording is about as difficult as it gets. (After a quick but dangerous trip to my computer’s thesaurus, that last clause became obdurate articulating is nigh on as recalcitrant as it annexes.) Once a passage begarnishes itself with odd or obsolete usages and syntactic constructions, we have to work harder to make the text coherent enough for us to move on. Even the most difficult words and constructions get easier with repeated exposure, however. Just as we can, over time, become accustomed to our bartender’s thick Irish brogue, we can adjust to difficult texts by changing our expectations about what we’ll encounter. The first time we read a sentence like The boy handed the candy bar drew a picture, it seems odd. But after reading a sentence like The boy driven to school drew a picture, the original isn’t quite as hard to get. Ordinarily, we’d assume that the boy had handed the candy bar to someone else. But because driven clues us in to the sentence’s reduced relative clause (in which the who was is dropped from The boy who was driven), we are able to interpret handed in the correct way. We have, in short, learned how to parse the sentence.
Not everything can be learned or, rather, unlearned, of course. Some innocuous words in the English language may be particularly resistant to rehabilitation—corner, relish, and delicious, for example. These words all look like they contain morphemes, the smallest units of meaning in a language, but they do not. They appear to contain, respectively, -er, re-, and de- (which do contain semantic information: even if you have never heard the word regoogle, you can take a stab at what it means). When we first learn to read, at least some psychologists believe, we memorize these morphological letter patterns as stand-alone chunks that can be combined with root words and other morphemes to form words. Then, when we come across a word that contains one of these patterns, we attempt to deconstruct it into its basic components. This process works for words like runner and redo and deconstruct. For words like corner, relish, and delicious, however, it’s a deconstruction that must be undeconstructed every time.
Slowing down isn’t necessarily a bad thing; the speed with which we read something is linked to our ability to remember it later. Recent research by psychologist Daniel Oppenheimer and his colleagues suggests that reading information presented in a font that is trickier to decipher—and therefore takes longer to read—leads to better memory of the message. Older studies have also found that increased illegibility leads to better metacomprehension, or grasp of our own comprehension: our best guesses about how well we’ll remember t_xts wh_re so_e of t_e _ett_rs a_e m_s_ing are more accurate than our guesses about how well we’ll remember texts where no letters are missing at all. Illegibility forces us to consider words more carefully (or, in extreme cases, to generate them ourselves), leading to a richer, deeper memory of them. Still, before you send that text message without your glasses, remember that it must read somewhat fluently or people will (a) not bother reading it or (b) hate you.
Reading quickly does have its advantages. For one, it seems to ward off mind-wandering, that enemy of even the staunchest Moby-Dick fan. I know from my own reading experiments that the surest way to get 20-second response times to simple yes-no questions like “Is bird a real word in English?” is to present the words too slowly. My participants will have begun pondering their weekend plans or, worse, fallen asleep.
Another benefit of a nice steady clip is that more information can stay prominently in memory at once. Think about it: when we have to remember an address or a telephone number, we recite it quickly to ourselves until we can write it down. If we recite it slowly, we lose it. A speaker of Mandarin Chinese can remember longer telephone numbers because the words for the digits one through nine are all quickly pronounced and monosyllabic—no long thr-eeeees or unwieldy se-vens. The ability to keep different pieces of information simultaneously active in memory allows us to make connections, to bind individual words and phrases together into something coherent. Read………too………slowly………and………things………stop………making………sense.
It is not enough, of course, to speak simply of speed when characterizing the way we read. We can read slowly because we’re working hard to parse a sentence, or simply because we’re savoring it. We can skim because we’re impatient to finish or because we need to know what happens next. Though attitude is trickier for psychologists to quantify than reading speed, we do know a few things. For one, we appear to read more engaging passages in a qualitatively different manner than less engaging ones. Some of my own research with psychologist Gail McKoon suggests that readers are likelier to make a correct inference about a pronoun’s referent (e.g., to realize that she refers to Mary) when the pronoun appears in a slightly longer, more engaging passage rather than a shorter, less engaging one. If a passage is too short to engage readers (the short passages in our experiments were just three sentences long), then pronouns—usually understood so effortlessly—are often not matched with their referents: she isn’t Mary per se, just someone.
Psychologists Richard Gerrig, David Allbritton, David Rapp, and others have demonstrated that we incorporate what we want to happen into our understanding of what is happening. Egocentrics that we are, we let our preferences about plot outcomes (e.g., that the bomb not detonate, that the bad guy take it in the balls) drive our expectations. Thus, when the bomb explodes, or the bad guy does not, in fact, take it in the balls, we’re disappointed, but we also stumble a bit—as measured by a slowdown in reading time. Really? we think, though probably not in these words; I could’ve sworn he’d be a soprano by now. And when asked even moments later to verify what just happened, we’re slower to accept those outcomes we didn’t want in the first place.
The more readers report being “transported” in a text (as measured by responses to questions about how well they can imagine the events in the story taking place, or how distracting they find their physical surroundings), the likelier that lessons learned in the story realm will speak to the beliefs they carry with them in the real world. After reading The Diary of a Young Girl, for instance, an engaged reader transported to wartime Holland might be likelier to support an individual’s right to freedom of religion. A transported reader is a swayable reader.
Can an e-book transport us the way a physical book can? Can a website? The sense that they cannot is what gives the Slow Reading movement its urgency. Most psychologists assume that people read entire blocks of text on a screen in much the same way as on paper—at least once factors such as eyestrain, scrolling time, and page refresh rates are accounted for. Psychologists have good reason to want this to be the case. Much of what we know about reading in general comes from reading on computers, where we have more control over the text a reader comes across at any given moment. If reading on a computer, an e-book, or an iPad involves different mental processes from reading a physical book, or if readers adopt drastically different strategies, some long-held psychological assumptions would need upending.
But the real problem with screen reading, as Nicholas Carr has explored in his book The Shallows: What the Internet Is Doing to Our Brains, is that the device gives us quite a bit more than a screen with some text on it. Even on Amazon’s Kindle, the most book-like of the e-readers, readers still have immediate access to word definitions. This convenience alone has increased the number of instances in which I have interrupted my reading midsentence by approximately 600 percent. On the Kindle Fire, you are also susceptible to Carr’s ultimate “interruption system”: the Internet.
Most psychologists agree that cognitive resources are a zero-sum game. Unless reading is automatic—and, at least beyond the level of word recognition, it absolutely is not—it is bound to suffer from such interruptions. But earlier times weren’t free of distractions, and deep, sustained reading has never been a popular pastime. For most of the written word’s 5,500 years on this planet, such reading has been the province of monks and scholars and the occasional Duchess of Newcastle. Furthermore, the shallow (think “lazy”) processing of language is not limited to reading. In some very early psycholinguistic studies—well before the Internet became a cultural force—people were simply stopped and asked questions such as, “How many animals did Moses bring on the ark?” Listeners relied so heavily on their expectations about what they were being asked that they never noticed the incorrect information contained in the questions. More recently, studies have shown that without proper punctuation or prosody to help us along (prosody is the singsongedness that makes a question sound different from a statement and distinguishes proGRESS from PROGress), we make an astonishing number of comprehension errors—such as attaching a single word to both the object position of one verb and the subject position of another—without ever bothering to correct them. We’re lazy because we’re smart enough to have figured out that laziness is usually sufficient, and much like well-fed lions, we laze with confidence.
Nonetheless, the capacity for deeper comprehension—among listeners and readers—has survived for centuries and will weather the digital naysaying as well. Our brains and the brains of our children will wire themselves as they please, perhaps favoring more efficient skimming over concentrated parsing, or perhaps not. Either way, nothing can be done in a decade or two that cannot, with some dedication, be undone. Any concerns we have are less akin to CO2 emissions than to light pollution: unplug and the problem will take care of itself. It’s an unlikely solution, but not, at the level of the individual, an unthinkable one. Nothing but a bit of friction is preventing today’s teenagers, even those seemingly incapable of reading anything longer than the back of a cereal box, from being, 20 or 40 or 60 years from now, squarely in thrall of Thoreau.
All readers are created equal—more or less. Our idiosyncratic jerks and wobbles nonetheless form a familiar pattern of movements across the page. We all slow down some when the going gets tough (or when the burgers come with relish), and speed up again when the tough gets predictable. That’s just the mind we’ve been given. What’s ours, as Slow Reading reminds us, is the freedom to sample one chapter and linger over the next, to read aloud or to flip shamelessly for dirty words, to allow ourselves to hope something will happen and to be moved by what does. So a question remains: How, then, to best love a book? If we love books at a gallop, we are almost certain to forget much of the scenery we encounter. If we love at a trot, we will never feel the wind against our face. The only solution, it seems, is to love again and again, with different, deliberate speeds.
Jessica Love is a contributing editor of the SCHOLAR. She holds a doctorate in cognitive psychology and edits Kellogg Insight at Northwestern University.