Tuesday, November 04, 2014

Is unlearning harder than unschooling?

Patrick Desmet
This blog has developed something of a split personality. The bulk of the readership (well over 10,000 visitors on a typical day) is here to read about the unfolding geopolitical tragicomedy, and the various and assorted stages of collapse which are to follow. A much smaller group (under 10% so far) is also interested in my effort to single-handedly solve a certain well-defined problem with the English language. And an even smaller group, numbering in the hundreds, is actually participating and contributing to this effort. Now, the geopolitical tragicomedy is certainly fun to watch, sort of like watching icebergs capsize, but since there is not a lot any of us can do about it, it's something of a time-waster; whereas fixing an actual problem, and making many people's lives better as a result, seems like a worthwhile pursuit—at least it does to me.

The problem with the English language is basically this: close to half of the population in English-speaking countries has a lot of trouble grokking the weird old English spelling system in all its multifaceted glory. They go under the unkind label of “functionally illiterate.” Most people aren't even aware that they have a problem, beyond a strange sense of unease, and have no idea what the source of it might be. The rest are in denial. Being a language nerd, I am perfectly aware of what the source of the problem is, and we will get to it in a moment. The obvious symptoms of this problem are, basically, three:

1. Most people stumble through life afraid to speak any of the words that they only know from their reading, fearful that they will make fools of themselves by mispronouncing them in public.

2. Most people stumble through life afraid to write down words which they know perfectly well from hearing them repeatedly in context, but which they haven't encountered in their reading, for fear of misspelling them.

3. A small minority of people is at ease with both written and spoken English, but, having had to spend a large part of their lives being indoctrinated inside educational institutions, these people lack the ability to think independently, and simply recycle their acquired linguistic clichés while watching helplessly as the surrounding culture becomes ever more primitive, vulgar and banal.

And here is the cause of all this confusion: English doesn't sound the way it's written, and it isn't written the way it sounds, making written English just too damned difficult for most people. Yes, there are lots patterns, and they apply to quite a lot of the words, but since it is often hard to tell which pattern applies to which word the only way to learn written English is to memorize both the sound and the spelling of each word, separately. And while the ability to speak is innate, and rests on the ability to easily remember the sound of words, as strings of syllables, consisting of phonemes, the ability to remember the spellings of words has nothing to do with language per se and is more akin to being able to memorize long, arbitrary strings of digits—a savant-type ability that many perfectly normal people happen to lack.

Add to that the annoying feature of English spelling to sometimes spell the same word differently depending on its context (there are only some 2000 such nuisance words, but the damage this causes is disproportionate to the relatively small number of cases). Take the word that sounds like “tu.” It is spelled in three different ways: “to,” “too,” and “two.” You might think that this helps distinguish meanings, but it doesn't, because “to” stands for two completely different things: a preposition “to you” and an infinitive marker “to go.” Likewise, “too” stands for two completely different things: in excess, as in “too much,” and likewise, as in “me too.” If such distinctions were important, there would have to be two more spellings of “tu”: perhaps “tew” and “tue”? But in fact such distinctions do not exist in the living, spoken language, but only in the crufty, outdated English spelling.

In general, without piling on such artificial complications, reading is by no means a savant-type skill, and researchers in the field of psycholinguistics have experimentally determined the two factors that are key to learning how to read quickly and easily. The first factor is called phonological awareness: the ability to recognize and reproduce the individual sounds that make up a language. For instance, the “er” in the word “ermine” is a vowel—same one as “ur” in “fur” and “ir” in “fir”. The “g” in “beige” is the same consonant sound as the “s” in “Asia” and “z” in “azure.” Such things are immediately obvious to someone who has phonological awareness. The fact that the same sound can be written using different letters is at best an annoyance.

The second key factor to being able to learn how to read quickly and easily is called the alphabetic principle. The vast majority of written languages in the world use an alphabet, which evolved from just one basic insight: you can take a picture of a thing, and make it stand for just its initial sound. Then you can endlessly reuse a small set of pictures, which is easy to learn, and string them together to make arbitrary words and sentences. So, Ancient Egyptians took a stylized picture of a bull, “alp,” and used it for the sound “ah.” As it got passed along through the ages, it morphed from a stylized picture to an abstract symbol, which lost its original meaning and was given arbitrary names such as the Greek “alpha” or the Canadian “eh.”

Likewise, our ‘B’, before it got spun 90° clockwise, was a picture of a house, “bet” with two arches. It then morphed into the Greek “beta” and the English “bee.”

This historical process of starting with a picture to represent a speech sound to eventually arrive at an abstract symbol is just a bit of bootstrapping; you can just as easily start with a set of abstract symbols, and many creators of alphabets (quite a few alphabets were created by just one person) did just that. But the principle of each one is that one symbol represents one speech sound, and vice versa. This is the alphabetic principle. Languages that follow the alphabetic principle are easy to learn to read: all you have to do is make the sound that corresponds to each symbol, form the sounds into syllables, the syllables into words, and the words into phrases in a process that is intuitive and automatic.

The hilarious thing about English is that it is almost as if someone set out to create a writing system designed to thwart one's ability to gain phonological awareness or to use the alphabetic principle. Children are taught that “A is for apple” even though it is perfectly obvious that: “A is for acorn.” “Æ is for æpple” would have made more sense, but then the Dutch printers who were the first to typeset English text (because the English were a bit backward at the time) were stingy about enlarging their assortment of Latin characters. (On the other hand, they were paid by the letter, and so they doubled as many consonants as they could get away with.)

If a primary school teacher happens to be savvy about “phonics,” the children are taught that the letter “C” stands for either “kuh” or “suh”—failing to mention that it also stands for “shuh” (acacia) or “chuh” (cappuccino) or nothing at all (Leicester)... unlike, say the letter “S,” which also stands for “suh,” and “zuh,” as well as “shuh” (sure) and “zhuh” (Asia)... unlike, again, the letter “T” which often stands for “tuh,” but also “shuh” (nation), “chuh” (nature) or nothing at all (castle). Just these three letters give you a network diagram that looks like this:

And the kids are supposed to be able to make sense of this? At what point are children allowed to learn a set of symbols that unambiguously represent the speech sounds of English? That's the most hilarious thing of all: never. So much for phonological awareness. As far as the alphabetic principle, sounding out English words one letter at a time doesn't result in anything that sounds remotely like spoken English. So much for the alphabetic principle.

Now, it is still possible to learn to read by brute force—without phonological awareness or the alphabetic principle—by memorizing each written word as if it were a sort of poor man's Ancient Egyptian hieroglyphic (composed of an abbreviated set of Latin characters) and memorizing how each word sounds on top of that. And that, unfortunately, is what most people end up doing. Such exercises in rote memorization are only possible when the mind is young and malleable, and at a certain point before too long the learning process comes to a halt. And so we have literate adults—literate by virtue of the fact that over many years of effort they have succeeded in memorizing both the spellings and the sounds of some reasonably large number of words. But this doesn't mean that they have ever achieved phonological awareness, or have even heard of the alphabetic principle.

Having recognized the problem, I went ahead and came up with a solution. My idea is to give parents a tool that allows them to quickly make their kids able to read a version of written English that obeys the alphabetic principle and is designed to impart phonological awareness, so that their kids can then go on learning largely on their own. This is what I mean by “unschooling”: eliminating the need for kids to go through eight years of regimented instruction just to gain a limited ability to read their native language. I want English to behave like many other, well-behaved languages, where teaching kids to read basically means teaching them the alphabet.

Since I wanted this tool to be useful to the largest possible audience, I did some research into what visual, graphic characteristics are likely to pose the least problem for dyslexics, the visually impaired and those with limited manual dexterity. Reusing the Latin alphabet was out of the question, because that would only cause confusion. After some experimentation I came up with a minimal set of symbols:

Then I mapped this set of symbols, and their variants, to the phonemes of English, and used them to create an alternative English orthography that obeyed the alphabetic principle, unambiguously associating one symbol with one sound, and vice versa. The variants are simple: consonants are tall and skinny, vowels are short and fat; there is a horizontal bar that denotes “openness” (“w” is more open than “m”) and a diagonal stroke that denotes “voice” (“v” has voice, “f” doesn't). And that's pretty much all there is to learn.

And I gave the whole thing a catchy name: Unspell.

My first attempt to unleash Unspell upon an unsuspecting world consisted of a little wallet card, which contains all that anyone needs to memorize in order to start reading “unspelled” English:

To my amazement, this was all the priming needed for a few families to teach Unspell to their kids, including some dyslexic kids on whom the teachers had given up, and who had remained virtually unschooled. Pretty soon one of these dyslexic kids was using Unspell to write birthday cards to his parents, to everyone's joy unconfined.

My second gambit involved offering to the world a lavishly illustrated booklet that taught Unspell in twelve short lessons and easy exercises using pictures of animals:

The lessons look like this:

...and the exercises like this:

 It's too early to tell how well this second phase of the experiment is going as far as the kids are concerned, but one thing has become clear already: whereas this sort of unschooling does not seem to be hard to achieve for the kids, there is a problem with the adults who are supposed to be unschooling them.

I have already observed that kids learn Unspell much faster than the adults. It seems that the ability to read spelled English can be a hindrance to learning Unspell. My guess is that it has to do with the two missing ingredients: phonological awareness and knowledge of the alphabetic principles. The kids quite naturally accept the fact that each letter makes a specific sound and then merrily chew their way through arbitrary text sounding out each letter. The adults, on the other hand, only know how to do one thing: sound out entire words—from the set they have memorized in their youth—by looking up how they sound in memory. In order to become proficient in Unspell, they have to unlearn the trick of looking up whole words in memory and learn the new trick of sounding out phonemes.

And it turns out that few of them are willing to do so. A few parents might consider learning something new for the sake of teaching it to their kids, but most would rather not. Some good enough sports to suffer the indignity of having their own kids race ahead of them, but many are not. And if that's how it is with the parents, the situation with the grandparents is often even worse: not only are most of them unwilling to entertain the notion that there is a better way to teach kids to read English (after all, it was good enough for them!) but they no longer have the flexibility of mind to acquire radically new skills even if they tried. They just want to read to their grandkids.

This may seem like an insurmountable problem, but in practice it is very simple: every single word in an unspelled piece of text is subtitled in spelled English; like this:

And so the grandparents can read along in the spelled version, the grandkids can follow along in the unspelled version, and everyone is happy.

Think of Unspell as the training wheels on a bicycle. Training wheels are very useful: they let a kid get from point A to point B without falling down. But adults don't go bolting training wheels onto their own bicycles in order to teach kids how to ride.

In the case of spelled English the training wheel metaphor breaks down because here the need for training wheels is terrain-dependent: once you learn a given text (stretch of road) you may decide to take the training wheels off, only to wipe out on the very next stretch when you hit a stretch of visually unfamiliar words. And so the need for Unspell never goes away completely; it will still be needed for rare words, words of foreign origin, place names, personal names and other oddities that abound in English text.

But the need for adults to become proficient in Unspell in order to teach it can be dispensed with quite easily. Consider it done.

The most recently updated versions of Unspeller have been updated so that every single piece of unspelled text is fully subtitled in spelled English. This should make it possible for English-speaking adults with scarcely a 4th grade education to teach Unspell to children—without having to become proficient at reading it themselves—just like the many swimming coaches who lack the ability to float and who demonstrate the breast stroke and the crawl with one foot on the bottom of the swimming pool. Don't worry, all that matters is that the kids learn to swim—or to read.


FiveGunsWest said...

My 6 year old thinks your system is much easier!

Bogatyr said...

I'm curious why you don't just use the International Phonetic Alphabet, which is already well-established and widely used...?

Dmitry Orlov said...

Ruben -

Tell us more. How far along is your 6-year-old with reading Unspell?

Bogatyr -

Here are the top reasons why IPA is a bad choice:

1. It's basically Latin, so it would interfere with spelled English, causing one's spelling to decline as one learns Unspell.

2. It's basically a research tool invented by Linguists, so it's ugly as sin.

3. It is at the wrong level of detail. Unspell uses meta-phonemes, which are dialect-neutral. IPA just captures speech sounds of a specific dialect.

4. There are no examples of extended passages of text written in IPA because it's relatively unreadable.

5. The various upside-down-and-backwards letter forms are a nightmare for dyslexics and special needs students.

And so on and so forth.

Rhisiart Gwilym said...

Brilliant! This seventy-four-year-old mind loves it, is revelling in it already, and will be proselytising it forthwith to friends home-schooling their children.

And all this written in the diabolical orthography of standard English spelling. Wish I'd had Unspell when I was five or six. Mind you, I was always one of the fortunate few who had the knack naturally, to the point where I can still do sub-editing.

Is Cyrillic easier...?

Unknown said...

Is there a font available such that I could type in unspell?

Dmitry Orlov said...

That's what this project needs--people to spread the word.

Cyrilic is easier. But Russian is hard, beyond the way it's written, which is quite straightforward.

Dmitry Orlov said...

There is an entire font family, to be released shortly.

Marc L Bernstein said...

Your unspeller is beautiful. I do have at least a couple of questions. The 1st question is easy -

(1) Have you passed your system around so that linguists have had a look at it? For example, Noam Chomsky might be quite amused. Sure he's 85 years old, but what the heck?

The 2nd question involves a case of a vowel that within your system gets only 1 symbol, but actually seems to be a particular case of 1 basic sound followed by another. I'll try to explain what I mean below.

I would love to use your symbols but I can't find them on the typewriter. I'll fumble along with the ordinary English alphabet.

The "i" sound as in "bite" or "buy" seems to me to be the same as :

the "ah" sound as in "got" or "la" (think of do re mi fa so la ti do) followed quickly by the "ee" sound as in "bee" or "he".

So for me, your "i" is almost (but not quite) the same as "ah-ee" .

The variable here is in how long you hold the "ah" sound.

So "i" is a special case of "ah-ee" in which the "ah" sound is maintained for only a small fraction of a second.

Does this make sense to you?

Marc L Bernstein said...

A brief addendum to my last comment -

The "A" sound as in "bait" or "bay" is again 2 separate sounds stuck together, in this case the "eh" sound as in "get" or "met" followed by the "ee" sound as in "bee" or "gee".

The "eh" sound is typically maintained for only a small fraction of a second.

Dmitry Orlov said...

1. Chomsky once referred to English spelling as "close to optimal for a language such as English" for being.able to distinguish things like right, write and wright. Since that is as far from optimal as I can imagine, we have very little common ground.

2. The term you are looking for is "diphthong," which is a glide from one vowel to another. In English, the dividing line between vowels and most diphthongs is fuzzy and.varies with accent, so it's safest to treat them all as distinct phonemes, teach them by example, and say nothing about their composition.

Marc L Bernstein said...

If Chomsky actually said that the English language spelling is close to optimal, he is laughably wrong. From my limited experience with languages other than English, I have found that Spanish runs far closer to Chomsky's characterization.

Thanks for the heads up on diphthongs. I looked the term up on Wikipedia and became suitably informed. It's not so bad to figure things out yourself on occasion, even if you're "reinventing the wheel" sometimes.

Silent Otto said...

Gday Dmitri , looks like something written on the wall of a pyramid ....
I wonder if ancient egypt was full of cheez- doodle munchers who loved nothing better than their weekly fix of collapse- porn iceberg capsize watching ...like me ?
Cheers mate
You are doing a good thing with Unspell

Raymond Duckling said...

Dear Dmitri,

Thanks for the remainder for not getting lost in the collapse porn.

Regarding project Unspell... I think it may have some potential, but ultimately it will be a matter of reaching a critical mass of users.

So, I hate to be *that* guy, but... as a non-native speaker, for whose descendants English spelling will most likely fade away like a dream you cannot recall upon waking up... how can I help? Or is it the right call to leave English speakers to fight their own battles?

Helix said...

OK, Dmitri, in many of your posts, you emphasise the value of cultural continuity, but in your unspell implementation, you purposefully break with our traditional alphabet, which is virtrually a lynch-pin of our civilization. What gives?

While I can see why the International Phonetic Alphabet is a poor choice, I cannot help but think that modified or embellished versions of at least some of our current alphabetic characters would work quite well and make it much easier to transition to such an alphabet.

Dmitry Orlov said...

Kutamun -

Unspell is designed to specifically NOT look like ANYTHING, to avoid confusion.

Raymond -

Critical mass is unimportant. Unspell will work fine for 1 or more users; since everything is either delivered electronically or via print-on-demand, quantities don't matter. The project is fully funded and far enough along that it can easily become self-sustaining regardless of the size of user base.

If you think English spelling will fade away, you are just plain wrong. English is the language the entire world uses to communicate, in spite of all of its warts.

Helix -

A bunch of scribbles do not a culture make. Unspell is specifically designed to not look like any sort of Latin to avoid confusion. The last thing the world needs is two different ways to write English using the same alphabet.

English-speakers may be fond of the Latin alphabet, but for the the vast majority of the world's population that is not the case. Since English is MOSTLY used by people who grew up speaking some other language, acceptance of Unspell in the English-speaking world may not even matter too much.

I do care about cultural continuity—with languages other than English especially. If English is simplified and relegated to its function as an international utility/communication tool, then other languages will be given more room to flourish.

Unknown said...

In an effort to help adults learn, perhaps a game is in order. Have you considered something akin to Scrabble, but Unspelled? The element of fun and challenge could be more appealing to adults, and it would still be an effective means of teaching children. It would also facilitate introducing the language to people who are outside your circle of direct readership.

Helix said...

Dmitri, I respectfully disagree. The current alphabet is by now strongly embedded in both our psyches and in our cultural tradition. IMO, making use of whatever useful parts of it there are would vastly improve the chances of transitioning to a more sensible system. Not doing so risks the project following the same trajectory as Esperanto, an elegant solution that never comes into widespread use.

BTW, by this, I'm not trying to denigrate your efforts. I have been toying with similar ideas for quite some time now, although I have admittedly not made anywhere nearly as much progress as you have. Good luck with this project, and keep us posted. Maybe we can use this as some kind of secret code!

kayr said...

"And so we have literate adults—literate by virtue of the fact that over many years of effort they have succeeded in memorizing both the spellings and the sounds of some reasonably large number of words. But this doesn't mean that they have ever achieved phonological awareness, or have even heard of the alphabetic principle."

Thanks Dimitri, this is so my problem. I had real difficulty in school with this very thing. English, spelling especially, was really difficult for me and now I know why. I have your book and after a first run through I realized that I wasn't sure that I was "hearing" the separate sounds. I have a friend who is also a language nerd and I was going to chat with her about the sounds to make sure I was getting them right. Is there a plan for an audio file of the English phonemes?

Dmitry Orlov said...

Thinner Buddha -

There is a game directly built into Unspell. Take unspoiled text and remove the bar, the stroke, and the height/width distinctions. Then try to decode what it says. For an extra level of difficulty, get rid of spaces between words.

Another game: the 13 symbols of Unspell, with the two orientations of — and | combined as one, fit onto the 12 facets of a dodecahedron, a Platonic solid. Toss it in the air and try to figure out what it's saying. Could be used for divination as well.

Helix -

Transitioning is not in the cards. This is a PARALLEL system, and, as such, it cannot impinge on the existing one in any way. So, for instance, if Unspell were to reuse the Latin alphabet, and provide alternative spellings, so that, e.g., "queue" would be spelled "kyu" and "bureaucracy" would become "byurokrasi" then 1. everyone would vomit and 2. it would be dead on arrival.

Kayr -

I should really put a page together so that people can hear the shapes of Unspell. It seems so basic, yet somehow lots of English-speakers have never learned to distinguish the speech sounds they make. Their perception of speech has become synesthetic with seeing written words; synesthesia being a sort of neurological disorder where, e.g., one sees colors when hearing musical notes.


This is a very valuable work, and I am sure Chomsky would applaud it actually. I can see a lot of applications for it, including teaching people about phonemes and preparing them for studying foreign languages.

I am, however, irrationally attached to letters. I enjoy seeing their various shapes, as if each one has its own character or personality. A word or phrase expressed in graphical letters is like a dance to me, much like a combination of Yoga postures together make a phrase or an expression of some sort. There is definitely different energy in different sequences of postures, and I feel the same about letters, so I don't want to lose this particular "illusion".

But writing is a relative thing, it is a secondary level "detachment" from reality so to speak (where the spoken word is the first), so it is not innate to our brain, and it would be interesting to find out the long-term effects of this method on creativity, memory, etc. At the very least, it is a great exercise for the brain.

Phonono said...

Basically, you are doing what several civil servants were ordered to do by the Korean sovereign Sejeong the Great, almost 600 years ago (1446), with the creation of the Korean syllabary Hangeul.
And it looks similar, to boot.

Rob said...

Just received my copy of Unspeller yesterday. I figured you would be on to something if I could learn in less than an hour and am not far off having just worked through the book. Potentially a work of genius. Some feedback:

A copy of your "wallet" sounds translation sheet somewhere in the book would be handy for when finished the exercises and learning to read proper.

As well as the "wallet" size reference sheet you could put all the animals on a single page with unspelled spelling for reference/ checking sounds.

The vowels are harder to learn than the consonants which came pretty easy. Double the exercises on the vowels would help. Like in spelled English I find myself working out the consonants and filling in the vowels by intuition.

There are a few accent minor accent issues chapter can be "chapta" or "chapterr".

Will get another copy for my daughter to learn next.

Dmitry Orlov said...

Trevor -

Thank you for the feedback and the "potentially a work of genius." I resemble this remark :)

I will be sure to include a quick start guide with all the animals and each symbol explained somewhere near the back cover. Luckily, with print-on-demand technology such changes are quick to make.

I tried to provide as many exercises for the vowels as I could but ran out of animals. I suppose I could ask for more animals...

About chapta vs. chapter, remember that Unspell is phonological rather than phonetic, and requires more or less decoding depending on accent. So, it is "chapta two" or "chapterr two", but it's always "chapterr eight". The "r" sound is always lurking in the background, but it isn't always audible.

Please let me know how it goes with your daughter.