A framework for speed reading

Link post

Chances are you’ve heard about speed reading. A magical technique that allows you to glance at a page of text and immediately extract all of its meaning. But after digging deeper and trying the techniques yourself, you realized that it does not seem to work for you. You may even feel guilty about not being able to read fast enough. After all, if there is a way to read 10,000 words per minute, are you not wasting a lot of time? I was in a similar position myself not too long ago.

Eventually, I decided to build my own framework for reading. I collected some bits and pieces from various sources, added my own insights to it—and voilà! You can read the result here.

The good news is: Once you understand some basic facts and have the right terminology, the mystery around it fades away. What is left are a few practical and useful techniques that help you optimize your reading.

The basics

A fixation point is a point that your eyes fixate on. Every time you look at something intently, you are placing your eyes on an imaginary point. It could be on the lamp on your desk, on the nose of your conversation partner or on the title of a book lying in front of you.

The fixation area is the area around a fixation point that you can still see and extract information from—without directly focusing on it. In other words, it is everything that you can see without moving your eyes. It can be approximated by a round shape. This should not be confused with peripheral vision, which refers to the outmost parts of your entire field of view. The fixation area only refers to the periphery of a single fixation point.

A saccade is a single atomic eye movement, going from one fixation point to another. Of course, the fixation area is also changed through that movement. We tend to believe that our eyes move continuously, when in fact, they make short and discrete movements.

Now, let’s apply these basic concepts to reading.

Of course, it is impossible to read a word or sentence without ever having seen it [1]. Therefore, to read something in full, each word must have appeared in a fixation area at least once. Since each additional fixation point takes time, we look for a method that minimizes them while still covering the entire text. If we also assume that we read from left to right and leave no gaps, there aren’t many options left.

We end up with the following standard reading method:

  • Read from left to right, placing 2-3 fixation points per line

  • Don’t go back in the text (avoid regression)

  • Place the fixation points in a way that exploits the fixation areas as much as possible

  • In particular, don’t place them on the outer edges of the text, but indent them a few centimeters (for example, on the third and third-to-last word)

Standard reading method

This concludes our take on the foundation of reading. All other reading methods are based on this standard method.

It’s not about speed

Numbers are tempting, especially if they are indicators of intelligence. We all know that those who read are intelligent. So those who read very quickly must be very intelligent, right? Unfortunately, numbers can also be misguiding.

After all, reading something quickly does not imply that you understood the material or that you will retain it. Reading speed is a nominal value—it is useless without putting in into proper context. If you gave somebody a boring book to read, he will claim to have read it in no time. His reading speed will be through the roof. But it is doubtful that it has brought him much use.

Instead of optimizing reading speed, we should optimize reading itself, i.e. the extraction of meaning from text. Sometimes this means to read faster and sometimes to read slower. The main way to accomplish that is to remove the mechanical barriers that keep you from reading as fast as you could. This is what proper reading should be about and this is also the goal of the method outlined above. Once you’ve mastered it, the mechanics of reading will step into the background and your focus will shift to extracting and processing meaning [2].

Thinking speed: A natural limit

Your reading speed is upper bounded by your thinking speed. You can see reading as thinking by proxy, whereby you reproduce an author’s thoughts in your mind. But if reading is merely a form of thinking, then it must be bounded by the same limits as thinking.

What are the limits of thinking speed? This is hard to determine, but I suggest to do an experiment: How fast can somebody talk for him still to be intelligible? Auctioneers talk at about 250 wpm. If you listen to a podcast at 3-4x the normal speed, you will still understand most of it. So, thinking speed is a multiple of normal talking speed (at about 150 wpm), but it is still within one order of magnitude.

There is no magical sauce that allows you to circumvent this barrier. People who claim to read faster than that are skimming the text rather than reading all of it (see fragmented reading below).

Regardless of your actual thinking speed, the ultimate goal is to get your reading speed as close to it as possible. Because that is all you can do anyway. Good reading is about removing barriers that keep you from reading at the speed of thought. Once you understand that, you understand speed reading.

Voice and visual reading

While learning and experimenting with reading, one insight was particularly shocking to me: There are two separate and very distinct ways how to read. Both of them are useful in different contextes.

The first one, voice reading, refers to a mode of reading where you pronounce words out loud in your head. It is as though you are both creating and parsing a conversation that takes place only in your mind. This is how most people read.

The second one is visual reading, where the meaning is directly extracted from the visual perception of written text. The words are not pronounced, but directly understood.

It is worth noting that the mechanics of reading as outlined above stay the same for both modes. The only thing that changes is the method you use to turn words into meaning. This is also why shifting between these two modes is so subtle.

Subvocalization is the internal speech typically produced when reading. You can characterize voice and visual reading by the degree to which they use subvocalization. Research suggests that it is impossible to shut it down entirely, but it can be significantly reduced, which allows you to do more visual and less voice reading. This ultimately results in a higher reading speed.

There may be concerns about whether visual reading really exists. I want to argue that not only does it exist, but it is the primary way in which we navigate daily life. When you enter an office and see a chair, you don’t say to yourself “This is a chair”, but recognize it as such and move on. If you see a sign in traffic, you don’t say to yourself “I have the right of way”, but you simply know that and then act on it. This means that you have been doing plenty of visual reading all along (just not with books)!

The problem with visual reading is that is not how we have been conditioned to read. Words—unlike traffic signs—are typically spoken out loud, so we tend to do the same with written words. This results in subvocalization. Fortunately, this conditioning can be untrained to a large extent.

A little caveat: Even when you are doing visual reading, it still may be that you hear a small voice in the background. But the words are all scrambled up and only part of the text is read. Subvocalization is so deeply conditioned in ourselves that it is quite hard to get rid of it entirely. However, as long as you are not focused on these voices as your source of meaning, it should be fine.

According to most speed reading literature, voice reading is an inferior mode of reading. Also, subvocalization needs to be suppressed at all costs. But this is a fatal misconception. It is a knee-jerk reaction to think that the faster method is automatically the better. In reality, there are many situations in which voice reading is the favorable method—or is even indispensable. I will outline some of them here.

The clearest example is poetry: It is a form of art whose expression of beauty is fundamentally dependent on the way that the verses are pronounced. You cannot speed read (here: visual read) poetry. It was only when I understood the distinction between voice reading and visual reading that I understood the appeal of poems. Until then, entering this world was impossible for me.

For humans, the meaning of text is intertwined with how it sounds. Powerful messages sound right, in that they produce exactly the right echo in the reader’s mind. This is why the choice of words matters so much. When you read visually, your focus is on the visual image of the words rather than the sound. This means that you miss the deep emotional effect it produces. For that to happen, you have to read slowly—to allow the sound of the words penetrate your soul.

If you are a writer and you want to assess the quality of a text you’ve written, you will be forced to use voice reading. It is the only way to truly feel a piece of text. It is only then that you know whether the text really creates the impression that you want it to have. However, if your goal is to simply recall the content of a longer text you’ve written, then visual reading is the better option.

If you edit a piece of text and use visual-reading, you won’t notice that a word is missing since you have trained yourself to look over grammatical constructs, as they don’t add much to the meaning of the text.

Comprehension is another issue. If you voice read a text, you typically pronounce every single word. Also, it is slower, giving your mind more time to process the information. Also, as stated before, meaning is intertwined with sound, so pronouncing words should produce a deeper meaning. Pronouncing the words also may aid in memory, as we typically retain better what we have heard (rather than merely understood).

So, voice reading is not inferior, but different. And consequently, it is used for different purposes.

How do you choose which one to use? A good rule of thumb is[3]: Visual reading is about information, voice reading about emotion. In visual reading, you are focused on the sheer informational value of the text. In voice reading, you are focused on how the text feels. If you only ever do visual reading, you deprive yourself of a lot of meaning and enjoyment for the sake of efficiency.

Full and fragmented reading

There is no need to read the entire text to understand it. This is another major insight that helped me improve my reading. This also motivates another distinction: full reading and fragmented reading.

Full reading is reading the entire text, word by word, without skipping or jumping over any parts of it. Most often, this is done using voice reading and by pronouncing every word. But visual reading is also possible. This is the appropriate mode for all situations that require full comprehension: technical documents, dense textbooks, problem descriptions to name a few.

The standard method that is described above is a form of full reading, as we required each word to appear once in a fixation area.

Fragmented reading takes a different approach: instead of reading the full text, you only read fragments of it. For this to be useful, these fragments have to be chosen strategically. One approach is to only read the first sentence of each paragraph. Another approach is to focus on the central words within a text, i.e. those that provide the most meaning (or bang for the buck). This means skipping grammatical constructs such as “a” and “the” entirely.

In terms of the terminology we introduced earlier, fragmented reading means to place fewer fixation points on the text than would be necessary for full reading. But as long as they are well placed, this is not a problem.

You may ask: If you don’t read the entire text, how are you supposed to understand it? I bet that if I gave you a collection of words, you would be able to derive the story behind them. Try it yourself: “mom”, “video games”, “education”, “son”, “autism”. Since we already have a lot of background information on each fragment and know how they are related, there is only one line of interpretation that makes sense.

The key here is to acknowledge that you are puzzling together pieces. This means that, to some degree, you are simply guessing what the text is about. Depending on the text and the amount of background information you have, this can work very well. But it is certainly a psychological barrier to be overcome, as we have been conditioned to read everything in full.

The more background information you have, the easier it is to attain higher speeds. In fiction in particular, there is a set of common tropes that get repeated across books. If you know them, you will understand what the book is about without reading all the details.

There is a trade-off between speed and accuracy[4]. If you want higher speed, you need to be prepared give up accuracy.

The transition between full and fragmented reading is fluid. You can always decide to use fewer or more fixation points, leading to a more fragmented or a fuller type of reading. The fact that we don’t have two strictly distinct types of reading is a testament to the flexibility of this framework. Also, it is often unclear whether you are reading in full at a very high speed or already doing fragmented reading. In practice, though, these distinctions are not important as long as you manage to adapt your reading mode to the material at hand.

A lot of speed reading literature glorifies fragmented reading as the one-true-way of reading. But to make this clear: In many cases, fragmented reading simply is completely inappropriate. Not reading a math problem in full is a terrible idea. You wouldn’t want to go to a doctor who reads your clinical history in a fragmented way. Nor would you do that with a dense technical document of a device you are trying to repair. Whenever a text is packed with information or if every bit of information is crucial, full reading should be used instead.

At this point, we can use our terminology to state what most books on speed reading actually mean by this term. It is very blurry because it is used in many different ways, without ever making the underlying distinctions clear. Sometimes it refers to fast reading in general, sometimes to intelligent skimming and sometimes to visual reading as opposed to voice reading. Using the above terminology, we can clear this up and hopefully never have to use that dreadful term again. The definition of most books most likely is this: speed reading = visual reading + fragmented reading + maximize fixation area. This is certainly faster than voice reading + full reading (or: slow reading), but, as we have seen, there are many situations in which it is not applicable!

How to read? It depends

If you adhere to a rigid system, your performance will suffer. Trying to read poetry using visual reading or applying fragmented reading on a doctor’s note will lead you nowhere. You have to read flexibly and be prepared to switch from one mode to the other instantly. In particular, you have to acknowledge that different reading tasks require different reading modes.

There are many factors that play a role: If you are sleepy, it is unlikely that you will be able to process information quickly—so, choose voice reading instead. If you are interested in a single piece of information, but are looking at a very long document, then full reading would be a waste of time and you should switch to (very) fragmented reading. If you want to deeply understand a quote, you should read it using full reading + voice reading.

The mechanics of reading should also be kept flexible. Say that you chose a method where you use two fixation points for each line (both slightly indented). What do you do on the last line of a paragraph that may only contain three words (the rest of the line is blank). Do you force yourself to place another fixation point on the blank space, just to enforce your method? Or do you break the rules and make a vertical saccade to the beginning of the next paragraph? The answer is: First of all, don’t use unnecessary fixation points only to follow some rule. Second, it depends on the text you are reading. There is no single valid way to transition from one paragraph to the next. But as of now, you should have the tools to figure this out for yourself.[5]

The optimal strategy will always be some mixture of all techniques. In many scenarios, a good strategy is the following: First skim through the text to get a rough idea of its content, identify the parts that interest you the most and then read them in full. This is a combination of fragmented reading and full reading.

Choosing the right reading mode is a matter of experience. After experimenting with different ways to read (and remember, they are all valid ways of reading), you should grow more and more confident in your ability to choose the right one for the job.

Conclusion

Since most of us are reading all the time, trying to improve it is a worthwhile investment. The problem is that there is a bewildering amount of material on this topic—and a lot of it is of dubious quality. Here, I tried to present a realistic and down-to-earth approach, as well as introduce some key distinctions that help to think about this topic.

This approach can be summarized like this: There is no one-size-fits-all solution. There is no magic trick that allows super-human reading speed. And that’s fine. After all, all we really need is to get rid of the barriers that keep us from reading at human-level speed!

We have only touched on the mechanical aspect of reading. There are more advanced questions such as choosing between different strategies to read a book. But with this framework, I am giving you the trunk of the tree. It is up to you to fill up the branches.

Why does reading have to be so complicated? Why can’t we simply read naturally? Unfortunately, our brains were not designed to read. And since this ability is not built-in, we need to use what we have to work around that. The concepts presented here make reading slightly more complicated, but they also help remove the barriers that keep you from proper reading. Which, in turn, makes reading easier! And after gaining some experience with them, they will feel natural, too.

The first time I stumbled upon speed reading was in high school. But the materials I studied contained too much mumbo-jumbo for my taste. Also, trying their techniques rapidly reduced my rate of comprehension, which was very discouraging. I put the topic aside and came back to it only many years later. After some reflection, I finally understood the key ideas behind good reading and started applying them. This was the first time that I was truly satisfied with how I read.

I hope the ideas in this guide are as useful to you as they were to me.


  1. ↩︎

    I am stating such obvious facts because many books I have read on this topic don’t.

  2. ↩︎

    This is analogous to the vim or emacs editors. Like proper reading, it takes some effort to learn them. But eventually, they help you to get rid of mechanical tasks (such as moving the cursor) that interrupt the writing process. Once you are are proficient in them, they allow you to type at the speed of thought.

  3. ↩︎

    This may also be a gross oversimplification since voice reading is helpful for full comprehension, which is clearly about information.

  4. ↩︎

    I intentionally refer to accuracy and not comprehension, as it is possible to comprehend something that is inaccurate.

  5. ↩︎

    I am aware that this is very technical, but in my experience, it is these small technical details that can ruin one’s reading experience.