Test Learning Language Learning in Two Minutes


If new here, see laptop v. education or faith v. miracles.

There is fear that OLPC deploys the cart before testing the horse. There are questions about teachers and localized educational content. Critics say learning learning is untested theory.. this idea that curious, creative, communicative kids will somehow construct learning troubles doubters. These are reasonable concerns. So can we actually test "learning learning"? Even push toward the edge of teacherless kids constructing localized educational content? I, Duke Crawford, believe we can.

Nepal olpc art
Limbu script on OLPC XO

If we narrow focus to language learning, we can test the theory in practice. How? Millions of the laptops ship soon and in many languages: Arabic, Yoruba, Hausa, Kinyarwanda, Kiswahili, French, Portuguese, Spanish. Kids may use these machines to learn some English, too. So you may dare describe these laptops as a ''multilingual'' education project.

As these kids interact, some may want to grow bilingual, some even multilingual. How will the children learn language? Some may want instruction. Others may want construction. For these, let's provide a simple tool, maybe even with no explanation, then test the results..

Let the kids try learning language learning. Let them wixi.

Wixi? Wixi is simply a wiki for language learners. Wixi lets language learners try new things, as they emerge. Like, for starters, maybe getting text twext. Twext? Twin text formatted to give language learners more comprehensible input. The basic twext translation interface is so simple even a child can use it.

On a wiki platform, twext translation errors will tend to be corrected, you know, the wiki way. So using wixi and twext, kids who know more language might help teach kids who know less.

Kids choosing material that kids find interesting may cause content to be instructive. For example, even without teachers, kids know lots about popular music. Might popular melodies and lyrics teach language? Even without teachers, services combining popular music, images and lyric texts are today proving to boost literacy in India.

Nepal olpc art
A little OLPC wixipedia?

Children are, by nature, creative language learners. With simple free wixi software running on communicative machines, kids may find new ways to help even know-it-all-adults to learn more language. Not impossible? Worth a try?

Wixi software development ain't rocket science and there's plenty of room to grow. A demo is done and a multilingual version will be working within weeks. With minimal support from the OLPC team we can deliver an XO integrated wixi activity for children to try starting this summer. Your cost today? Two minutes.

If you want to test an actual case of "learning learning", why not take a little action? With very little risk now, we may soon learn a lot.

Related Entries


twext is a great tool. This is what the OLPC needs, a killer application that shows what the XO can be used for. Scratch (scratch.mit.edu) is another one that lets children crate stories, animations and games without programming.

"If we narrow focus to language learning, we can test the theory in practice. How? Millions of the laptops ship soon and in many languages..."

That would make sense if the cost were $0.00 - however, no smart governement would invest hundreds of millions of dollars to finance such experiment.

Pilot Projects are the obvious and smart answer. But we already know what Negroponte thinks of Pilot projects....

Wixi looks like a great piece of knowledgeware for the OLPC machine, the Intel machine, the machine (probably) coming soon from China or your everyday laptop. Let's make it available to those who need it.

@jason.. glad you like it and thanks for letting me know

@troy.. wixi is free: not restricted to olpc, but could be a useful tool (with luck "killer app") to test "learning learning"..

cost of wixi ain't anywhere near $$ millions.. baby steps more like $1 or $2K.. $10K could deliver many useful features worth testing..

cost of wixi could be approach $0 if free software developers participate.. you might assume cs student ie linguistics might wanna play.. or that a cs student or class might risk some time for potential reward of a growing user base.. so far, ixne

@jack http://twext.cc/go/18#.twx looks for a data design to make wixidata platform independent, database independent, open as possible

It is great to see more and more learning applications with constructionist concepts.

Our generation who have been educated in traditional instructional style still think too often in terms of topics like a tool for language, another tool for math etc.
This might not be necessary anymore or it might even be opposed to the natural way of learning. You did not learn to walk because you conciously decided it. You learned it while pursuing another goal like wanting to go to places beyond your privious reach. This creates high motivation. The applications should include such motivation. E.g. kids might learn foreign language while investigating something highly interesting that is only available in that language. In this situation the application should come in and support and facilitate learning.


the original poster DID SAY it would take "millions of laptops" (with the associated hundreds of millions of dollars to buy them) to test his "theory" - I will quote him now:


"we can test the theory in practice. How? Millions of the laptops ship soon and in many languages.


However, if I'm mistaken, then he can test his theory independent of the OLPC Project. Perhaps he is even doing that now...I don't know. What do you think?

"millions of laptops" (with the associated hundreds of millions of dollars to buy them) to test his "theory"

But this could be any laptop.

Say, the XO, Classmate (if Intel sells them), a Chinese or Indian laptop, successor projects.

Whichever way you take it, supplying children in the developing world with technology will cost money. Currently, the OLPC is the cheapest option. And teh best developed one in terms of hardware and educational content.

If you think that modern technology is not cost effective in developing education, then you might have a tough job convincing those in the developing world. Many in higher administration did study in the West, and are experienced in using modern technology.


@troy.. fyi i'm the fool original poster 8p.. yes, original post framed wixi as mass-olpc dependent but in reality wixi is a free web-based app, already started at http://sf.net/projects/twexter ..

design goal of wixi data is platform and database *independence*.. so ideally machine architectures don't restrict web-connected human language learners..

http://twext.cc/go/io rather naively reaches for a SOAP design so interacting web-based services, ie chunking, machine translation, formatting and other services can easily connect/disconnect as per individual user needs..

when http://twext.cc/go/18 tool is delivered in the next few weeks, theoretically anyone (using firefox for now) will be able to test and develop the wixi language learning interface.. so to be clear, the proposed test is neither exclusive to the sugar platform nor the XO..

still, due to b.) wixi's simple open design and a.) OLPC's desire to implement localized educational content in a constructivist context, original post seeks support to develop a native sugar xulrunner/javascript activity that hundreds of kids in brazil, uruguay, nepal using the XO can test now..

as you point out, original post assumes shipment of a zillion machines, which is not necessary to conduct a little test of wixi.. thanks for helping to clarify the intent

Thanks, Duke, for the clatification.

I want to point out that

a. I do NOT consider you a fool (no reason whatsoever)

b. I have nothing against your project.

My point is that the XO machine needs to conduct pilot tests in order to determine what works and what doesn't work. THEN, interested parties will be in a position to place orders. It would be really crazy to buy thousands, perhaps millions of machines without a single test. That's why nobody has placed an order...

"You did not learn to walk because you consciously decided it. You learned it while pursuing another goal like wanting to go to places beyond your previous reach. This creates high motivation."

- ~ - ~ - ~ - ~ - ~ - ~ - ~ - ~

thanks roland.. one assumption of wixi on the XO is, connected and communicating like never before possible, kids will be uniquely motivated to learn each other's language.. not to win brownie points for whoever, but rather to have more fun playing with each other.. maybe fiddling w/ scratch?

so you put your finger on the trigger.. motivation/intent is key.. so is "affective filter", (are we having fun yet? not bored or stressed but *engaged*?) that's why the wixi approach aims not top down but bottom up..

an aside, if the XO UI can toggle between languages, and maybe even include a little twext text, that may provide another immediate and practical context that can induce language learning.. for those motivated..

duke, could you, please, explain bottom up and top down in this context a little bit more?

@roland: re: top down v. bottom up..

"top-down" approach, in language instruction, often uses methods that worked just fine for the language teacher, but aren't productive for most learners.. here in mexico kids study english for *years* but can't speak.. they want english, but associate the process with boredom, anxiety resulting in frustration.. many end up learning that they can't learn..

above "sage on the stage" teacher role has "guide by side" alternatives, such as Blaine Ray's TPRS or http://storyasking.com method, where teachers *involve* the students in an interactive learning process.. students/teachers develop stories together.. it's a playful way..

again, due to motivation, it doesn't work for all students, but better serves the majority.. learners end up able to comprehend and even speak in the target language.. thus prepared for meaningful interaction with native speakers, learners can then "scaffold" toward more language acquisition..

twext speculates a more "bottoms up" approach: if motivated and provided with twext between *interesting* text (ie popular parts of popular songs), then learners get meaningful context (when needed) that can provide repetitive comprehensible input.. it's all about repeated comprehension of specific language usage in variable contexts..

repeated comprehension in variable contexts creates foundation for the most effective teacher: meaningful interaction with native speakers.. http://sk.com.br/sk-krash.html

no claims here that twext is the only way, a better way, or even that it works at all.. but it is simple, flexible, intuitive and worth a try..

update: article said multilingual twexter would work months ago.. sorry, that hasn't happened yet.. el hombre propone pero dios dispone.. if you're a computer science student, please make killer proposal for http://twext.cc/go/google#summer_of_code

above twext.cc/go/ links are now simply twext.cc/ links (above twext.cc/go/ links lead to blank pages, so simply remove "/go/" from URL)

if visiting from ycombinator, please overlook bathwater, seek baby.. current PHP plugin system reaches wide.. lisp/arc appear to be, from my naive perspective, powerful and focused ways to get text twext.. possibly a scalable opportunity, not only to get rich, but to create wealth: