14 ms·

Without claiming myself an expert in research on reading, and not a "speed reader" myself, I am probably able to deliver something of a lecture on this topic. D

Without claiming myself an expert in research on reading, and not a "speed reader" myself, I am probably able to deliver something of a lecture on this topic. Disregarding my lack of credentials, here's something you might find interesting in an unstructured manner. I apologize in advance for lack of sources; you can question me on anything I write, and I assure you there are sources but I won't bother digging. Generally, proponents of speed reading don't know their stuff. Speed reading is based on the notion that the brain is underused in comprehending text; training would supposedly untap this potential. Don't count on it. Some say that if you read faster, you force your brain to focus, and so comprehension goes up. This works to an extent, noticeable, sure, but far less than hyped. And it is not a technique problem: it is a discipline problem. If you're generally focused at what you do, you're probably among the faster readers. So most people read at about 250-400wpm, and this is related to the speed of the "inner voice," aka "subvocalization" (or, better, the phonological pathway/loop). So the idea is that if you reduce, or eliminate the inner voice, you can read faster. Programs that focus on these will claim 1000+ wpm. There is some truth to this, but these programs aren't all it takes. Now when your eye sees text, it the brain takes time to parse it and register meaning. Conscious meaning, IIRC, happens after a 350ms mark, which means it is biologically impossible to read word by word, line by line, beyond the speed of the inner voice. This means that if you want to read faster, you're reading in blocks of text . What does this mean? It means that when a normal reader sees a word, you're seeing a sentence or paragraph. Peripheral vision? Not necessarily. Peripheral vision doesn't have enough acuity to support parsing at a level required for detailed reading. IOW, most of the peripheral vision talk by these programs is BS. The cases where PV is useful is when you have good familiarity with the text contours. How do you train this? The hard way: build your vocabulary, and read a lot. Reading is a "self-bootstrapped process," if you will. This is, AFAICT, the ONLY fool-proof, non bullshit method. FYI, I recall two speed readers, who both read very fast but in different ways. First one reads blocks of text in a glance. He would see a paragraph and understand it in the time you read a sentence, in a kind of real-time, bulk-parsing process. The second one would read text in a "serial" fashion without regard to the meaning, into some kind of buffer, then think about it after he's done. The catch: both are geniuses. The bottleneck in deciphering meaning from text is not at the text->parse stage (this is what SR programs focus on). It is from the parse->meaning stage. IOW, brain power. Fortunately you can train this, but probably not with some book or some program. This is pretty thoughtstream. If you have other questions I'm happy to discuss.

rguzman 19y ago
Thank you, this was informative. I'll believe you about the sources. Who are the two geniuses? I've heard the block of text at a time thing before. In fact, I even think most people who I know that read significantly faster than me do so this way. The strange part is that when asked, people have told me that they have a hard time deciding whether they read a word a time or in larger blocks of text. My only question would be: can this skill be learned/trained? If so, how does one go about doing that? I'm a bit ambivalent about the bottleneck, at least personally. I guess it depends on what kind of reading you do most of the time, but I'm inclined that I read more materials that are written in simple text that was meant to be consumed easily and that in those cases the bottleneck is in the text->parse stage. When reading technical material (which I also spend a fair amount of time doing) the bottleneck is definitely on the parse->meaning stage and that process probably takes several orders of magnitude more time, but speeding up the text->parse stage would be worth its while still. I say this because the process is a little bit different when reading technical material. You get through the text and you think about it for a long time, do problems, discuss, perform thought experiments, etc. But everything in the list seems to be a different stage than the parse->meaning stage. It is more of a meaning->true understanding/intuition stage. Something like first you read and you need to comprehend what you have been told at a "language" level. Then you go and understand the concepts more abstractly and at different levels. However, it is difficult to draw the line between those two stages.
- whacked_new 19y ago
  These geniuses were an accomplished scientist, and a high school friend (who will probably become an accomplished scientist). It shouldn't be hard to find these kinds of people among "smart" folks though, simply because they absorb lots of information on a regular basis. If we arbitrarily set the upper limit of normal reading to be at 800wpm (very optimistic and unlikely), there are people who still vastly exceed this speed. It's not just skimming: they will learn the stuff just as anyone else would, in an almost magical process. You're right about the bottleneck, which makes my previous post's discussion not so good. There are multiple bottlenecks, and improving one area will not guarantee a result like those programs promise. If it's technical material, obviously, if you were Feynman you'd be able to quickly read hard paper without feeling uncomfortable (hence my reasoning about brain power). If it's Bearstein Bears, then, yes, you can fly over pages with total comprehension. As you said, here the speed of parsing text would make a big difference. This is related to the "block of text" thing, which can be trained. It's not "reading in blocks" per se, but the training of fast visual pattern recognition. In this sense, it is similar to the pattern recognition the Rain Man used to count 270 or so matches in one glance. The most direct strategy then, is simply twofold. One, increase your vocabulary. Not just so you can answer test questions: you really need to know these words cold. Unfamiliar words take longer to retrieve from memory. Obviously, Charles Dickens would take longer to decode than Douglas Adams. (Hitchhiker is interesting though, since there are many invented terms; maintaining speed on this requires a relevant, but different, skill, and shouldn't be confused with that involved in vocabulary building). Two, read more, and ideally, when you read non-essential material (daily news perhaps), consciously push yourself (this takes discipline). The point of this is not for speed of parsing words. It's for recognizing the shapes of lumps of words. To illustrate: if you see "the Jedi reader reads fantastically quickly" a thousand times, you'll begin to parse that as a combined shape, in a single fixation. You will slow down, though, if you see "fantastically quickly, the Jedi reader reads." On the most part these two sentences are semantically equivalent, but the latter will take longer to parse, simply because it takes a different shape. But if you read that many times, you'll start chunking it like the other sentence. So reading a lot is simply to build up your arsenal of "familiar phrase shapes." Again, this is the most foolproof way of training yourself. There are other things that may bring small improvements. But this is basically the heart of the matter. There are people who can parse unfamiliar word forms very quickly. The "bottleneck" here, as I understand it, is at the pipeline between CPU and RAM, and cache size. Genetics plays a big role here, so I usually don't bother.

rguzman 19y ago

Thank you, this was informative. I'll believe you about the sources. Who are the two geniuses? I've heard the block of text at a time thing before. In fact, I even think most people who I know that read significantly faster than me do so this way. The strange part is that when asked, people have told me that they have a hard time deciding whether they read a word a time or in larger blocks of text. My only question would be: can this skill be learned/trained? If so, how does one go about doing that? I'm a bit ambivalent about the bottleneck, at least personally. I guess it depends on what kind of reading you do most of the time, but I'm inclined that I read more materials that are written in simple text that was meant to be consumed easily and that in those cases the bottleneck is in the text->parse stage. When reading technical material (which I also spend a fair amount of time doing) the bottleneck is definitely on the parse->meaning stage and that process probably takes several orders of magnitude more time, but speeding up the text->parse stage would be worth its while still. I say this because the process is a little bit different when reading technical material. You get through the text and you think about it for a long time, do problems, discuss, perform thought experiments, etc. But everything in the list seems to be a different stage than the parse->meaning stage. It is more of a meaning->true understanding/intuition stage. Something like first you read and you need to comprehend what you have been told at a "language" level. Then you go and understand the concepts more abstractly and at different levels. However, it is difficult to draw the line between those two stages.

whacked_new 19y ago
These geniuses were an accomplished scientist, and a high school friend (who will probably become an accomplished scientist). It shouldn't be hard to find these kinds of people among "smart" folks though, simply because they absorb lots of information on a regular basis. If we arbitrarily set the upper limit of normal reading to be at 800wpm (very optimistic and unlikely), there are people who still vastly exceed this speed. It's not just skimming: they will learn the stuff just as anyone else would, in an almost magical process. You're right about the bottleneck, which makes my previous post's discussion not so good. There are multiple bottlenecks, and improving one area will not guarantee a result like those programs promise. If it's technical material, obviously, if you were Feynman you'd be able to quickly read hard paper without feeling uncomfortable (hence my reasoning about brain power). If it's Bearstein Bears, then, yes, you can fly over pages with total comprehension. As you said, here the speed of parsing text would make a big difference. This is related to the "block of text" thing, which can be trained. It's not "reading in blocks" per se, but the training of fast visual pattern recognition. In this sense, it is similar to the pattern recognition the Rain Man used to count 270 or so matches in one glance. The most direct strategy then, is simply twofold. One, increase your vocabulary. Not just so you can answer test questions: you really need to know these words cold. Unfamiliar words take longer to retrieve from memory. Obviously, Charles Dickens would take longer to decode than Douglas Adams. (Hitchhiker is interesting though, since there are many invented terms; maintaining speed on this requires a relevant, but different, skill, and shouldn't be confused with that involved in vocabulary building). Two, read more, and ideally, when you read non-essential material (daily news perhaps), consciously push yourself (this takes discipline). The point of this is not for speed of parsing words. It's for recognizing the shapes of lumps of words. To illustrate: if you see "the Jedi reader reads fantastically quickly" a thousand times, you'll begin to parse that as a combined shape, in a single fixation. You will slow down, though, if you see "fantastically quickly, the Jedi reader reads." On the most part these two sentences are semantically equivalent, but the latter will take longer to parse, simply because it takes a different shape. But if you read that many times, you'll start chunking it like the other sentence. So reading a lot is simply to build up your arsenal of "familiar phrase shapes." Again, this is the most foolproof way of training yourself. There are other things that may bring small improvements. But this is basically the heart of the matter. There are people who can parse unfamiliar word forms very quickly. The "bottleneck" here, as I understand it, is at the pipeline between CPU and RAM, and cache size. Genetics plays a big role here, so I usually don't bother.