A Future Language Learning Platform

May 31, 2013

Here are a bunch of popular language learning websites:

Anki
LingQ
Memrise
Duolingo
Readlang (OK, not as popular yet but I had to throw it in there :-)

Notice a common thread? They all re-implement their own spaced repetition system and all except for LingQ¹ don’t share their data². This means if I’ve invested time with one tool, there’s a resistence to switching to a new one. And the experience I have with a new tool will be inferior until it learns about me.

Imagine if all these tools backed onto an open learning platform which stores your progress on each word or concept. You could encounter a new word on Duolingo one day, test yourself on it with flashcards in Anki the next, then encounter the same word in an iPhone game the next day, and all these events would contribute to the same user profile. New apps would hook into this data to provide you with a personalised experience from the start. Wouldn’t that be awesome?

Of course, getting all the existing sites on board with this is a tough proposition. But if this existed and got a bit of traction, users would demand that new language learning tools or games use it.

I’m not sure exactly what it’ll look like but it should provide an open standard API to:

Store a list of your unknown vocabulary
Record your encounters with that vocabulary, both passive and active
Use word frequency lists to prioritise useful words
Use a spaced repetition to provide a stream of word to learn, along with related data like example contexts and hints

For bonus points it could incorporate grammatical rules, understand conjugation of verbs, provide good dictionary definitions, translations, etc…

I predict that one day an open platform like this will exist, and an ecosystem of connected language learning tools will emerge.

LingQ does provide an API allowing read/write access to the words and the spaced repetition data, which is good, but doesn't seem to have attracted much of an ecosystem as far as I see. Largely due to the paywall for users.
I suppose you could look inside Anki's local sqlite database, but if you're developing for the web, that doesn't count.

</small> (PS. As for how these ideas may affect Readlang, I’m not exactly sure yet, but stay tuned!)