“language learning”
13 posts under this tag.
大人 = big people = adults
小人 = little people = kids (this and above like in toki pona)
写真 = reality copy = photography
売買= sell buy = trade
靴下 = shoes under = socks
台訶= pedestal talk = speech
赤ちゃん= little red (one) = baby
西日 = west sun = setting sun
姉妹 = older sister, younger sister = sisters (older/younger sister are basic concepts!)
One of the coolest things about an idea-sign language, which motley Japanese at times is, is that it encourages making new words by combining simpler ones. It does this as a necessity (there are only so many signs you can remember), by making of words stable roots (idea-signs tend to be more stable than letter bundles—for one thing they don’t reflect pronunciation changes), and by not allowing for sound loan words (“Bon weekend!”), where meaning is lost in grafting a word from one meaning net into another.
Here a couple of interesting, basic examples:
火山 = fire mountain = volcano
下女 = down woman = maid
電話 = electricity talk = telephone
出口 = out mouth = exit
入口 = in mouth = entrance
Just think of the responsibility, the challenge, the opportunity. One third of the population is still young enough to be natural born digital citizens (see Classmate PC and the OLPC XO laptop), to easily master an international language (whichever one), to be taught about doubt (“Just think of the tragedy of teaching children not to doubt…”), to receive the best education we can give them…
Remember that character in Neal Stephenson’s Diamond Age, catatonic at page 169 at discovering a quarter million Chinese girls thrust to his care? Well, look around and realize we’ve been given a ship of 1.8 billion souls. Just think of the opportunity.
(Statistic according to the U.S. Census Bureau, international)
Starting an artificial language has been a recurrent dream of mine. As a subscriber to the Sapir-Whorf hypothesis (that the shape of our language is the shape of our thought), a believer in ending Babel through an auxlang, a pathological formist, and an admirer of the grace, elegance, and pleasure to be found in conlangs such as Esperanto and toki pona, I believe the enterprise worth a lifetime, worth my lifetime.
But of course, given my extremist bent, I want to start an artificial language that subsumes all languages. A language to make languages, like in John Varley’s beautiful Persistence of Vision. An extensible language to gobble up and be enriched by the thoughts and feelings of as many souls as the universe will ever have. A perfectly regular language that can be learned in a week but never mastered. The creation of a self-conscious, language-obsessed culture but learnable by the illiterate. A language so abstract and basic, it can be embodied inside any symbolic system, be it based on sounds, graphics, gestures, raised dots, or farts; be it English, Maori, or Farsi. A language of infinite expressibility, synthetic and analytic, vague and precise, formal and casual, exquisite and coarse. A language that will outlast the stars.
The key, I think, lies in internal flexibility. The ideal is to do for language what the Hindu-Arabic numeral system did for numbers. Not only will there be no arbitrary, capricious limits to word creation, it will be a language of pure word creation, able to convey books in a word, lifetimes in a sentence. It will be a language complete in itself yet always growing.
After years of frenzying about it late at night, the language finally got its first name, despite it not yet having a transliteration, let alone any words. It’s self-referentially called, among infinite names, the-language-this-word-belongs-to.

Here’s a (controversial) idea for a language test inspired by the famous Turing test for artificial intelligenceWP:
a native speaker of language X engages in conversation with two other parties, one a native speaker of language X and the other a student of language X as a foreign language; if the judge cannot reliably tell which is which, then (and only then) can the student be said to speak language X.
The test could be easily constrained to test for more specific capabilities: one could test for written command of language X by only permitting written communications, test only for accent by limiting “communication” to the spoken repeating of the judge’s written sentences, and so on.
It is simply stated but almost a “thought test”WP—it could be done, but there would be a myriad practical complications and scaling would be a bitch. What’s important about it, though, is that it is a valid test to demand of (foreign) language learning: passing it should at least be its hypothetical goal.
The problem is that ridiculously few people would pass it if it where applied today. And because it seems impossibly difficult most people turn away, dismiss the test as wrong or irrelevant, and sink their heads in the sand (“what shouldn’t be, can’t be right”). Which only highlights the current sorry state of language education. It is NOT asking too much. It is not asking for exceptional performance—it doesn’t ask of you to be a Nobel-prize, a literati, or a rapper. It’s merely demanding average, pick-a-guy-from-the-street native-speaker capabilities. Why isn’t that a valid goal to ask of language education?
You could say that most people don’t need native-speaker level to start benefiting from a foreign language and that’s entirely true. But it is just as true that not reaching it is a serious, frustrating, even painful hurdle to communication. A hurdle that will plague ever more people the more the world shrinks. Some of the world’s smartest people can’t get their r’s right hard as they try. And we mock them for it. (Soon, we will be the mocked ones for not getting our intonations right.)
Well looked, Turing level is perhaps even a modest goal. We all possess it already in the language we are born into and we all contained within us the same language potentiality at birth. So it should be perfectly achievable and shouldn’t take nearly as much time as starting from zero.
Yes, I know. We are nowhere near knowing how to reach such a level efficiently. It’s too hard and too long a goal—currently. But we should at least strive for it. (And be honest with students on what the status quo of our language technology is: no more “Learn to speak Chinese in 21 days!”—for now.) Languages are some of the most complex and powerful artifacts we have created. It’s only to be expected that their learning is one of the most complex and difficult challenges we face.
But it is also one of our most rewarding (and valuable) experiences. I want to commoditize it.
Chances are we are on the brink of Turing level language translationELZR. Why aren’t we even close to practical Turing level language learning? I’d still want it.
I have always envied Alexander the Great, because he had Aristotle as a personal tutor. In those days, Aristotle knew pretty much everything there was to know. Even better, Aristotle understood the mind of Alexander. He understood which topics interested Alexander, what Alexander knew and did not know, and what kinds of explanations Alexander preferred. Aristotle had been a student of Plato, and he was himself a great teacher. We know from his writings that he was full of examples, explanations, arguments, and stories. Through Aristotle, Alexander had the knowledge of the world at his command.
With that, Danny HillisW, E introduces his idea for Aristotle, an AI tutor that will move in a smarter web he calls the knowledge web. I find his dream somewhat unconvincing, somewhat pedantically unrealistic and somewhat suspicious of oversimplification. (Even though he considers it but a steppingstone towards Neal Stephenson’s Young Lady’s Illustrated PrimerWP, ELZR, which I love.) It is from the eminent responses to his essay where there’s gold.
As much as I truly hate domain hoarding when I’m out there looking for a spiffy domain to my latest webapp, I confess compulsive domain buying is one of my guilty pleasures1. I’m hoarding, I know, but perhaps my scale will redeem me. Those bastards—you know who you are—who hoard (“park”) thousands of domains, financing the whole murky enterprise by filling their spoils with semantically-related ads disguised as directories… well, may they be strangled to a slow, painful death by his noodly appendage.
My two most recent acquisitions are ThisWorldIsTooDark.com and Nellodee.com.
The first domain is a phrase that has haunted me since I first read it at a local exposition2 (thanks to Andrea for telling me about it) of the work of Cultural-Revolution China’s Li ZhenshengWP. A photoreporter of the main newspaper in China’s far Northeast during China’s Mao mire, Li kept negatives of his work against orders and they may be the best remaining record of the horror. Andrew Stuttaford wrote a harrowing review of Li’s Red-Color News SoldierAM and he didn’t escape the phrase either:
More typical, and more tragic, was Wu Bingyuan, a technician accused of counterrevolutionary activities (a pamphlet). Li recalls that when Wu heard his sentence, death, “he looked into the sky and murmured, “this world is too dark”; then he closed his eyes and never in this life reopened them.” The photographs show Wu being paraded through the streets of the city. Later, shackled and bound, he’s pictured at his place of execution. His eyes are still shut. We see him kneeling, back turned to the firing squad. His eyes are still shut. The final image is of Wu’s corpse. His eyes are still shut.
I want to do something at thisworldistoodark.com that honors Wu’s memory but I still don’t know what. What I do know is that the phrase is forever carved into my memory.
The other domain, nellodee.com, is thankfully from the opposite end of human possibilites. Nellodee is the full version of Nell, the name of the protagonist of Neal Stephenson’s excellent Diamond AgeAM, a toddler from the future slums that chances on a state-of-the-art learning machine. This book-machine, the Young Lady’s Illustrated Primer, the book-within-the-book, is every self-learner’s wet dream: endlessly interactive, infinitely patient, all-knowing, self-adapting, story-driven, fractal (the basic outline of the book’s story is presented at the very beginning, from then on you advance the story by zooming in on any particular fragment of it, the fragment develops into a full-fledged story, and on it goes). It has left me so deeply impressed that I have to do my share to bring it eventually to life. Toki Pona seems like the perfect subject to try my clumsy hand at the Primer concept with a simple web-app—it’s a small, simple, and enjoyable subject, and I’m already sort of an expert in it. We’ll see.
So why am I telling you all this? To assuage my conscience. You see, perhaps I dawdle for years before actually implementing any of the above ideas and so I’ve configured both ThisWorldIsTooDark.com and Nellodee.com to redirect here, to this very post, in the meantime. If you are doing (really doing, not pie-in-the-sky woulda doing) something really cool, are missing a good domain, and either of those two would be a great choice for your project, I’d be glad to give them to you. Gratis. Full-ownership. With my best wishes.
Is a translator who doesn’t believe linguistic intercommunication problems should or could be remedied as ridiculous as a doctor who doesn’t believe that diseases should or could be remedied? Or put another way, is a translator who believes that linguistic intercommunication problems can or should only be palliated as ridiculous as a doctor who believes diseases can or should only be palliated, not cured?
I frankly don’t know. But those who know me can see on which side I’m leaning. The thought came to me tonight and am still grappling with what it would mean.
This is from Douglas Crockford’s Survey of Javascript (never program JS without your Crockford!). I thought it quirky at first, surprisingly helpful later. (Emphases added.)
The && operator is commonly called logical and. It can also be called guard. If the first operand is false, null, undefined, ”” (the empty string), or the number 0 then it returns the first operand. Otherwise, it returns the second operand. This provides a convenient way to write a null-check:
var value = p && p.name; /* The name value will
only be retrieved from p if p has a value, avoiding an error. */
The || operator is commonly called logical or. It can also be called default. If the first operand is false, null, undefined, ”” (the empty string), or the number 0, then it returns the second operand. Otherwise, it returns the first operand. This provides a convenient way to specify default values:
value = v || 10; /* Use the value of v, but if v
doesn't have a value, use 10 instead. */
Short-circuit logical operators are a well-known, simple idiom in several languages, but they can sometimes be confusing to read, specially when nested. What I want to point out here is that next time you have to go through code that uses them, try reading them as guard or default, as the case may be. You’ll grokEE them immediately, trust me.
Isn’t it striking, the power of names?
Oh! In Tokyo, the New Trend Is ‘Media Immersion Pods’, a New York Times article from a while ago on Tokyo’s media youth, is important, very important. This is me, this is my generation.
And, really, what’s so wrong with getting lost on the Internet; watching soccer or baseball on satellite television; devouring Us Weekly or Time Asia; and organizing solo marathons of Tim Burton or Kurosawa movies? The craving for media sprees runs deep, and, like so many Internet-era developments, Gran Cyber Cafés seem to answer an almost carnal need for uninterrupted access to pixels and screens and Web sites and instant-messaging and iTunes. And when that need is satisfied, you can always return to life in the city, at least for a while.
And this is it. Screw Chinese, screw German or French (both of which I already studied for a year), I’m off to learn Japanese.
|