Humans Can't Read: Misdefining of the Mystifying act of Human Reading

This article is dedicated to Patty Warman (and Steven Pinker)

From the point of view of computer programming, the means by which human children learn to understand and speak their mother tongue is dauntingly complex. And, they learn it without having carefully to recall the complex cognitive and muscular details of how they do it. The same is true for sign language for deaf children of fluently signing parents. Why is reading different? This article presents several mutual answers to that question.

There seem to me three basic parts to the learning and teaching of reading/writing by and to humans: The Mechanical, The Social, and The Developmental. I shall briefly discuss these three in turn. But, before I do, let me introduce the subject more generally.

Children are not the ‘glorified hammers’ that we call ‘computers’. Moreover, the qualities with which human infants pay attention to the things that are 'out there' in front of them is quite unlike how developmentally normal adults often feel that they themselves are so 'able' at certain things. This adult ‘able-ness’ is the ‘deliberateness’ which so easily becomes the tendency to over-analyze and overly-micro-manage.

Adults are often too good at this kind of ‘deliberateness’. It was adults, not children, who invented the proverbially dreaded 'foreign language classes' by which a merely academic knowledge of language was forced to act as the means by which a new language was acquired. If children were restricted to this means for acquiring a first language, then the ‘nature of speech’, as it would then be learned, would be as ‘analyze-ably’ artificial as what Steven Pinker and Charles Darwin thought of print:

[Humans have] an instinctive tendency to speak, [but no human] has an instinctive tendency to write. More than a century ago, Charles Darwin got it right: Language is a human instinct, but written language is not. Language is found in all societies, present and past. All languages are intricately complicated. Although languages change, they do not improve: English is no more complex than the languages of stone age tribes; Modern English is not an advance over Old English. All healthy children master their language without lessons or corrections. When children are thrown together without a usable language, they invent one of their own. Compare all this with writing. Writing systems have been invented a small number of times in history. They originated only in a few complex civilizations, and they started off crude and slowly improved over the millennia. Until recently, most children never learned to read or write; even with today’s universal education, many children struggle and fail. A group of children is no more likely to invent an alphabet than it is to invent the internal combustion engine. Children are wired for sound, but print is an optional accessory that must be painstakingly bolted on.

—--from the Forward to Diane McGuinnness' book, Why Our Children Can't Read

Did Pinker actually say 'bolted on'? Does he think a person who can read is accurately likened to a cyborg? Does he think that any task the brain has learned to proficiently perform is not a mere extension-and-refinement of the brain's own natural abilities?

A bicycle is an artifact which a group of children is not likely to invent purely on their own. But, this does not prevent a human easily learning to ride a bike if he is allowed to approach that learning task in his own time, and in any unpressured way in which he mechanically sees fit.

In simple fact, the skill of riding a bike is merely an extension of the skill of walking. Consider the scooter: a bike with a very low side-to-side pivot point by way of the smallness of the wheels. Such a bike allows a novice more easily to learn to ride than does a bike with wheels ten times as big-around. This is simply because that bigger-wheeled, ‘normal’ bike is proportionately more difficult to learn to ride for a beginner who has not yet correctly imagined how to ride a bike.

A tight control of the handlebars is essential to riding a bike in rough terrain. And, the proportion of locomotion is more-or-less equal to the ease with which a novice rider may keep upright. But, neither the handlebars nor the speed-and-manner of locomotion is essential to the irreducible task of riding-without-falling over. In fact, absent any sense-and-ability of that irreducible task, a person effectively seeking balance in rough terrain only by controlling the pivot of the front wheel (by the handlebars) is sure to fall over. In other words, if the rider sits more-or-less passively on the bike while seeking to keep upright primarily by control of the handlebars, then the rider is treating the bike as if it were a trike or a car with a steering wheel.

Phonics is like the handlebars, and travel and means-of-locomotion is like the comprehension. But, without a basic efficient use of the brain in the visual recognition task, and this in efficient coordination with the level of comprehension ability, the person is left to learn to get around the text like a tourist learning to get around a foreign city with a tall cardboard box affixed around him blocking his view of anything but the ground around his feet.

Some people fail to see that reading, and learning to read, can be as natural, in-process, as the learning of a spoken language that occurs outside a 'foreign language class'. Reading, and understanding speech, both make use of the brain’s own natural abilities, and these by extension, just like bike-riding is an extension of the essential balancing abilities of walking. Every ‘artificial’ outward ability, from tying your shoes, riding a bike, driving a car, or manufacturing-and-programming a Harrier jet from the raw earth to the machine shop and the keyboard, makes use of core natural abilities. It’s just that some abilities are far more readily attractive to a given human learner at a given point in her development.

In fact, many people learn to read without the hassle which has become the hallmark of modern, micro-managed, protocol-centric pedagogy. This is not to say that all such hassle is always avoidable. After all, we live in an imperfect world, and, even in our most healthy state, we have, at best, an imperfect cognitive function. But, since all learned abilities are just extensions of core natural abilities, no learned ability need be thought of, much less taught, as strictly artificial or protocol-bound. There are always two, opposite, rational ways to go in helping treat a given problem; but those two ways must be balanced against each other: neither way is alien to the other. But they can be forced to be too separated.

Only to the extent that some of the brain's core abilities are brought into conflict with one another do the efforts to perform the ‘task’ become a frustration. By way of analogy, in the martial arts one can tense one’s arm, and one can speedily thrust one’s arm forward; but, to execute a dynamically efficient punch, one cannot be simply tense and simply fluid at the same time.

Now, on to the three parts of learning and teaching of reading/writing to humans.


Many persons recall having experienced getting lost and then trying to recognize familiar landmarks in order to get un-lost. If you are out traveling and you want to get home, but you feel lost in an anxious way, then you may start to use extra effort to try to recognize the landmarks around you. You may try to make them be what you want them to be: landmarks with which you are, in fact, familiar.

But, if you fail to recognize landmarks that you do know (or mix them up in some way), or if you mistakenly think that some of them around you are the very ones you are hoping to see, then when you find yourself mistaken and still lost, you may remain anxious, and so continue to make mistakes. Or, you may become even more anxious, and so you try even harder to make things look familiar, with an increasing sense of failure and disorientation when your greater efforts result in greater failure. And, if you finally do manage to get un-lost, yet all of the mistaken effort you spent had no positive relation to your final success.

So, the effort of mistaken applications at performance does not contribute to a final success at the object. This applies to any activity or skill. Many amateur martial arts novices tense up their whole arm during the entire executing of a punch, either without knowing it, or with the mistaken feeling that it contributes to the force of the impact. A brute-force effort actually expends far more energy than need be spent, and does so at the expense to which the core-abilities are effectively sensitive to every one of the different sources of data, not just to the source of the protocol.

When a person feels pressured to try to recognize, or to learn to recognize, a particular collection of visual information (whether a letter, a sequence of letters, an animal, a landmark, etc.), he tends to try too hard in all sorts of ways that fundamentally cut down on the delicate and complex cognitive coordination of the visual recognition system. This is why imposed 'whole-word' reading instruction has ‘generally’ poorer results that simply imposing on children to learn graphemes: for the latter there is so much less visual information to learn to recognize, and the functional units are all very small.

But, of course, simply recognizing graphemes does not equal reading. And, the brain is made to use as much of its wide range of resources as efficiently as possible, so that, given task of visual recognition, the brain seeks to accomplish that task not only by the use of as many parts of itself as it can bring to bear, but as little of each of those parts as possible.

The visual recognition problem comes when, whether by external intent or accident, or even by deliberate scientific purpose of the person, the brain is forced to drastically limit the number of kinds of its parts which it is capable of bringing to bear on the recognition task. When that happens, the brain is forced to try to make a too-narrow set of its applicable parts do the job which it could far more efficiently do otherwise, through the positive compounding effect which its different parts have on each other. The brain is not a simple, linear-processing machine, such as a modern computer, the latter being invented by the brain as an external enhancement of only some of the brain’s natural, and naturally inter-informative, abilities.

But, unlike a computing machine, no one normally sees anything in a strictly ‘brute conscious’ way (though there are various levels at which people normally try to learn to recognize new visual information). Rather, they see it in a ‘lucid’ way. This ‘lucid’ way of seeing is made possible by the naturally inter-informative parts-and-functions of the brain. So, the more ‘brute conscious’ the means of accomplishing a visual recognition task, the less normal, and less efficient, the brain is caused to function to accomplish that task. Like trying to stop your own heart from beating, such ‘brute conscious’ visual tasks are far too narrow to be worth what the living organism cannot help but function toward: living life normally, that is, with as full a set of its abilities as it has ready access to.

Consider the cognitive task of recognizing the face of a human stranger in terms of its being a human face rather than a pixilated treetop-like abstraction. This is a vastly complex task. But, even if we had ready access to a strictly brute-conscious kind of knowledge of that task, we could not put that knowledge to natural use anywhere near as easily as we already otherwise do by simply and naturally performing that task. This is because such knowledge would require so much more brain glucose to apply than what we already use by accessing that knowledge in the minimal-and-subconscious way that we already do naturally access it in performing that task.

So, human life is not centrally about what too many robotics and computer engineers of the past century had liked to think it was. And, contrary to so many pro-phonics reading experts of the protocol-centric pedagogy of the middle of that century, human reading is fundamentally more than simple visual-analysis protocols and input-sound conversion. In fact, the task of normal human-agent reading is fundamentally more energy-efficient, economic, and per-agent-variable, than what any 'seeing machine' has ever been made to do.

When too many of a brain’s parts-and-functions are forced to ‘sit on their hands’ contrary to the aim-and-purpose of the person, or contrary to the performance expectations of the teacher, then that person becomes frustrated. But, if the brain is allowed to do its natural job, it develops depths of connection to the task which, quite literally, are a kind of romance with that task.

This romance is the love of reading, in whatever way a person desires to read. Even if, for scientific reasons, one of those ways is a ‘brute conscious’ way, the point is that that person’s reading efforts are his or her own efforts, effectively pre-coordinated with his or her own expectations and purposes, not those which, by being externally imposed, happen to be in conflict with that person’s effective, if not internally elucidated, knowledge of how to accomplish the task. If robotics engineers took over toddlers’ natural efforts to learn to walk, there would be a crisis of bipedal learning. In other words, we do not teach reading, we teach human beings.

Now, look at the following set of three little shapes and tell me what you see (the HTML coding does not allow the 'lesser-than' sign to appear here, so I have substituted a back- and forward-slash to indicate where the 'lesser-than' sign should be).


Do they merely look like themselves? Or, do you tend to make them appear to be more than they are? Are they a pair of glasses on a nose? Or, maybe they’re a universal sign for “bicycle-and-rider”? Or, maybe a bird eating a seed? The fact that you can so easily see such a simple, static visual shape like that in different and meaningful ways shows that the human visual system is not keyed to the visual trivia constituting the shapes themselves, but to any and every part of your brain that is brought to bear. In fact, the way you get to know a new person (visually, or in general), and thus to get an ever-deepening sense of them, is, essentially, also the means by which you get to know a simple, abstract shape in terms of any meaning you assign to it.

In some circles, reading is expected to be so unnatural that it is, indeed, ‘bolted on’ by an unnatural means, thus fulfilling its own expectations. But, imagine the following hypothetical set of conditions: 1) infants and adults each have a graphics screen of such form, and attached above their faces in such a way, that each person can see not only all data on everyone else's screen, but on their own screen; 2) each screen is hooked up to the corresponding person's brain so as to allow him to produce static line graphics at will; 3) no one has hands, mouths, or hearing; 4) all adults can already read, and can already ‘write’ on their own screens).

You will notice that, in this hypothetical set of conditions, infants can immediately identify the line graphics produced by another person as being directly associated with that person and with that person's behavior. You will, I hope, also notice that this hypothetical set of conditions presents various automatic feedback systems related to reading and writing---feedback systems that, in the real world, are comparable to those available to hearing infants in regard to speech.

In such a hypothetical world, infants would learn to read as easily as hearing infants today learn to understand and produce voice signing (speech). If you doubt this assertion, then let me mention two commonly known facts. One, hearing infants in the real world do not learn hand sign language if that signing is not used by the adults around them and in relation to them. Two (despite One), all infants learn hand signing as naturally as voice signing (speech) if the adults use it in relation to them---even though hand signing was 'invented', rather than 'evolved over millennia’, and even though only speech is supposed by Pinker and Darwin to be the 'biological imperative'.


Try, some time, having your live conversations with house guests without any exchange of sensory data except what you can get from print (no watching your guests’ hands or eyes, just seeing the static visual result of their having just finished writing the words they would otherwise simply speak). You probably will quickly begin to feel how utterly inconvenient and one-dimensional such a means of conversation is, since your guests will be right there available to simply talk to and see, just as a normal hearing-and-seeing human has the right to expect as normal. There's a complex set of reasons for why most children do not learn to read without a minimum of overt, deliberate instruction. You can find all of those reasons in that 'live'-but-inconvenient manner of printed conversation.

But, you may find those reasons on your own only by pondering and re-pondering the question for maybe days or months on end. Because, a one-time experiment requires a lot of pondering and sorting out and extrapolating. A baby, on the other hand, in the total process of finally becoming a basically proficient language agent (understanding and producing language), feels quite free to consider as much---or as little---linguistic data as she pleases from her social environment.

The now-popular Rosetta Stone language learning software replicates some of that environment, partly by making it possible to interact with a poor-but-effective virtual language-user in whatever language you’re hoping to learn. Another part makes use of the common-sense principle that the more language-relevant data you are immersed in while the language is being produced, the less overt 'thinking' you have to do to become proficient in the language. In other words, you don’t have to stare at a blank wall trying to remember what you’d heard and what it had seemed to mean.

In fact, since natural language is based on usage rather than on protocols, natural language is structured as a complex of nested recursions which lend itself to being learned by usage-oriented (i.e., telic) agents. The form-manipulation machines which today popularly are called ‘computers’ are not telic agents any more than they are used by telic agents; and, in fact, such ‘computers’ do not actually know even a single thing, no matter how much ‘language’ data is stored on, and retrieved by, them. A pocket calculator has no more sense of quantities and their relations than a speaker system in an auditorium has the least idea of the audible words it is faithfully reproducing. So, the forms shown on your computer’s monitor (or, if you’re reading this from a printout) have no bearing on what your computer ‘knows’. Merely nominal ‘knowledge’ is not knowledge at all.

And, without actual knowledge, all that’s left is mindless protocol. Protocol must be bolted on, but only to the extent that actual knowledge is prevented, or otherwise not brought to bear.

The nature of the data for all forms and mediums of signing are cognitively equivalent. What is different are the ways in which they each are most easily and thus commonly used. As implied in the above hypothetical world, static-graphic, permanent signing in a prosthetic medium is most naturally indirect and passive, so that to make it otherwise would require (other) prosthetics of a very high-tech form.

So, when a person is reading silently, you cannot see the relationship that is going on between the reader and the text. Reading is, thus, both a non-social language-activity and a hidden language-activity. Moreover, static-graphic, prosthetic notation has many purely practical comparative and more immediate disadvantages to direct forms of language. Here's a simple picture to give you a clue to what that means:

You can't ask for more juice in your bottle if you are unable not only to hold and manipulate a crayon upon a surface, but to keep your mom in the room and looking at this action while you write out your request. Using your vocal chords is far less costly in terms of energy, and in the cost-to-dimensionality, and cost-to-speed, ratios. And, you can hear your mom talking from across the room even if you haven’t yet the ability to turn your head to see her, or from in another space (room/womb) if there is a wall between you and her.

So, writing is a more intellectually abstract activity than is speaking with someone directly. This is why most people never become prolific writers, even though it is technically and logically possible for anyone who can write to produce a lot of good writing. When you write a letter to grandma in the evening (even if it is by email), you're not interacting with anyone in that activity (unless, of course, you happen to be writing it physically or virtually alongside someone). So, writing most easily and usually is a lone activity. And, so is reading.

In the case of either hand or voice signing, the person is, of course, automatically identified by the infant with that person's signs. But, for reading to be social, it has to be made specifically social by adding the social dimension to it, either by reading aloud to someone, or by talking about what you read; it is not social in itself, it is one or more degrees of social separation. Infants do not become language users by way of TV or radio programs that do not respond to their needs.

Reading, as we know it, is a lone and rather abstract activity. And, the static-prosthetic medium (such as pen and paper, or paper and ink marks called letters) has no natural appeal as a medium of language: it just sits there. In fact, give a baby a book who has never seen one before and he will treat it for what it is: a prosthetic to play with, as something which has no special social utility; he will not see to use it in the fully open, truly ideationally rich way which is allowed by the range-and-readiness of vocal forms. In other words, print has no apparent socio-linguistic dimension: It's no better that a set of TV screens each of which displays a different, unmoving, abstract shape. To the normal infant, even the ‘countdown’ on Sesame Street has more appeal.

So, while printed language can serve well as a record for an isolated person, the real point even of printed language is social: Robinson Crusoe wants to experience knowing that others know the things he writes, and he wants even to experience knowing that they know that he knows that they know. So, while print is, in the end, usage-dynamic, the perception of the form of print normally is not associated with anything more important that the arbitrariness of its forms: it is void of meaning. Compare this to the product called a ‘drawing’ of some familiar object, which is clearly representational, though it uses the same medium as print. This is why the practice of drawing is so easily acquired from seeing the product or its production: the product is so near to the core natural abilities, so that one need not have ‘gone to university’ to appreciate the point of the product; No Assembly Required.


There are different kinds of language agents. A machine agent is fundamentally different from a human agent---the latter being both fundamentally more complex and primarily superior. The reason a human can make, program, and then use, a computing machine---with its fundamentally more simple internal operating structure---is because a human already has the ability to think in whatever limited, simple-protocol, manner he wishes.

But, this ability to think in a subset of the full human set of thinking abilities becomes a debility when the person‘s cognitive functional autonomy regarding the input structure is overridden or otherwise interfered with. Today, this interference is committed usually with the best of intentions, by an assembly-line\programming method of teaching. So, within an imposed and micro-managed 'learning' context, children whose socio-linguistic development is 'lagging behind' that of their average age-mates need relatively more unambiguous terminology, within such a context, in order to keep from being unwittingly forced into acquiring robotic habits of approaching print.

Hence the supposed indispensability of Diane McGuinness’ ‘correct’ version of terminology for teaching phonics (namely, 'letters represent sounds', not 'letters make sounds'). But, it is these very children whom politically pressured teachers then find must be induced the more to submit to the 'programming'. Such an approach to teaching human beings is like standing on the Earth while thinking there’s a genuine need to make an entire separate planet from scratch in order to grow some plants: you get so caught up in figuring out how to make a planet from scratch that you fail to recognize what you're standing on.

When an entire nation of children are forced by law into this one-sided, dictatorial, 'educational' relationship, the conveniently bureaucratic solution to children's increasing 'learning failures' is to do lots of statistical, third-person research to figure out how to improve the programming, and, or, to load children down with even more of these menial mental chores within a given span of time and beginning at earlier and earlier ages. One ironic result is the impression, on the part of many such 'educated' persons, that getting a 'basic education' outside of the factory model educational infrastructure is akin to making a Saturn V rocket from scratch with never anything more in the way of tools than, say, a bowie knife and some pretty beads.

But, learning is not necessarily the same as external performance (i.e. ‘task time’). This is because performance data is commonly processed by the brain even after one's body has ceased the performance. This is how a child learns to walk within far less 'task time' than might be expected by a hypothetical robotics engineer who has had no exposure to toddlers.

But, it is considered normal by pediatricians that some children spend a lot more 'task time' learning to walk than do others, and that, either way, some attain basic proficiency in walking at only twice the age of others. This variability is all the more necessary---not less---for any supposedly ‘artificial’ skill such as reading, programming VCR’s, and flying Harrier jets. In fact, some people who cannot manage to pass their pilot exam on a flight simulator nevertheless can pass it in a real airplane.

So, it’s only when the brain’s natural abilities are misdirected by what (often merely) happens to be a disharmonious instruction-or-prosthetics that learning something becomes an unpleasant kind of struggle. What is needed is the 'Goldilocks zone' of mutually complimentary coordination between all applicable parts-and-functions that are available to a particular brain at a particular stage in its development. And, boys and girls are not uselessly redundant of each other, either in terms of their respective gender's positive qualities, or in terms of the developmental details between boys and girls. In other words, there is a statistically significant difference between boys and girls in terms of at what age, and in what manner, they are best taught to read. The problems for the future of their literacy arise when they are held down onto someone's idea of a perfect, artificial reproduction of humanity’s natural educational 'planet'.