Wednesday, January 03, 2007

Notes on MannX at the U of Kabul and elsewhere in the Gap (A Plan for Victory, Part umpty-three)

Being a geek, I naturally think of the Long War as an information war, and in part as an educational war. If enough people of the Gap get Core-style educations, we win. Really. (Well, probably. At least it would help a whole lot.) But there are several problems in the Textbooks to Iraq solution, and one is the problem of language. Back around 1970, as the sole native speaker of English in my physics classes in Colombia and then Buenos Aires, I was very much aware of this: people in the Gap need to learn some Core language because the information they need is available in Core languages, and often it's in English. Nowadays it just happens that language-teaching software is something I work with, and that -- as a project specifically for the U. of Kabul science programs, but hopefully more generally -- is what this is about. Well, mostly.

Maybe this belongs on the development blog, but it's more of a personal I'm-incompetent-to-judge-this-but-I'm-trying sort of post, so I'll put it here. As a programmer, I've spent a lot of time writing video-control code to teach languages I don't know a whole lot about, mostly Russian; it started in the early-to-mid 90s in Apple frameworks and then C and then C++ with Quicktime, then switched to Java applet form (still with Quicktime, then with JMF), but reviewers said that the installation was time-consuming and confusing even though the program itself was good.

The basic idea is simple: link a video with a transcript and a commentary and a glossary so that, just for instance, you can find a comment in a commentary and then click to play the video segment that this is a comment on. Find a word, and cycle among the segments indexed as containing that word. And so on; any component can point into the others. We are using this for language lessons, but of course it could be used for any video/transcript/commentary/glossary presentation. But only awkwardly. We kept on adding to the program and writing papers(PDF from SWAP 2005) about it, but the installation didn't get easier: the screenshot in that PDF of a computer science lecture with commentary in Pashto for Afghan students is very nice, and I can play it off the web, but you have to do some local installation first. (If you have JMF and a Pashto font, try a simple Colgate-student dialog, JMF. But most people don't, and I don't deny it's a nuisance.)

Finally, in early January 2006, I learned to work with SWF from Javascript so that no installation was required other than Flash -- a vast improvement, except that the download of a big SWF file was tedious and annoying, and nothing happened until the whole file downloaded. But all through the past year FLV files, as in youtube, got more and more interesting, and the Javascript API ditto, as soon as there was a Linux version of Flash 9 I could work with it, and Anssi Piirainen of added a Javascript API for his Flowplayer project which was a copy (at my request) of the mplayer Javascript API, and as soon as I could get enough keyframes into the movie at a (Firefox-tested) KeyframeTest; FLV unit, I could synchronize video with script and a dummy commentary though I didn't bother with the dictionary or ontology.

So now we're working on a much-closer-to-real unit, a Milton Friedman work in progress which is changing a couple of times a day but which synchronizes nicely using FLV transcoded from the Free To Choose videos; we'll build this up for use within Afghanistan (permission granted by Bob Chitester of and then try to go on from there. Why Friedman? Because what he's saying (in 1980, with selections from the 1990 review) is what people in the Gap need to hear, and if they'll learn it from an English lesson (one with subject-matter commentary in one style and linguistic commentary in another) then we are way way ahead.

So will it work? I wish I knew.


Post a Comment

<< Home