Learning Languages

Here's how I learn a language: start with the orthography, pick up as much grammar and vocabulary as possible in two weeks, try to read a newspaper article, fail to understand much except perhaps a single sentence and a few other words, then lose interest as a result. It took me years to properly learn how the Semantic Web works, so I'm not always impatient, but I do normally expect to learn things fast. And when I don't learn a language to an intermediate level in two weeks, I find it hard to compel myself to continue.

So I've been thinking about compelling ways to learn a language—or fast ways to learn, or both; because speed comes from immersion, and immersion either comes from great perseverence or from being compelled. If I were forced to learn a language, one that I wasn't interested in if there is such a thing, I hope that I'd be able to persevere in learning it; but as it happens I only learn languages by choice.

The most exciting learning system I've used so far was actually rather accidental. It's the Kalusa method: a language which is constructed solely by a corpus which is anonymously contributed by interested parties and then voted on by the same. Anyone can make new words and new grammar, and anyone can rate existing sentences. Each new "entry" in the corpus is provided in the form of a line of Kalusa and the equivalent line in English, and then the words in the corpus user interface are hyperlinked to searches for those words.

In this way, it's very easy to look up how a particular Kalusa word has been used so far, and to find translations for English words. Two strongly related obvious weaknesses with this method are that i) idioms don't work so well, and ii) the English translations are very often far from literal so it's difficult to work out the structure. To counteract that I'd force there to be a literal translation nugget, and also give the opportunity for user notes. In any case, the general principle is the interesting thing: for some reason, working out the web of the words is very compelling indeed. It's learning and discovery in its rawest form. But coining new words and idioms and coming up with interesting new grammars is at least half of the fun.

It might not rescale to an existing language, but I can more or less envision a way in which it could be done. Sentences would be ranked by the abilities of the person entering them, either beginner (mistakes expected), intermediate (mistakes possible), or fluent (mistakes rare); there'd also be a feedback method allowing people to flag a sentence as being wrong and to suggest an alternative to it. This should encourage all kinds of folk to contribute. The main guideline would be to only illustrate one concept, that is one word or grammatical construct, per sentence. In other words, keep the sentences fairly child-like and unrealistic, for the sake of demonstration. I think that this is important when learning a language, because you're trying to find patterns and memorise huge quantities of new data. Keeping it simple and yet fluid at the same time is how children learn languages: so many modern language books are dedicated to teaching people how nouns are constructed and how adverbs work in a particular language. Kids don't know that; they aren't taught that until much later. It's interesting, but it shouldn't be the central method of teaching.

I'm also a very hands on learner, so I like to learn things in practical highly mnemonicistic ways rather than just reading and reading and reading. Reading is important, and in fact I'm only really thinking of reading comprehension when I mean learning a language, but I think that the linear aspect makes it hard to learn. Pictures, designs, motifs, situations, connections, art—all of this can be used in interesting ways to form more lasting impressions of words. It's harder to remember a word that's been used in a book than one which is printed in big letters on a sign outside your house, or one that's been said by a comedian on TV, or something like that. I don't remember genders of nouns very well if I'm just told to remember them, but if I underline the words with red and blue, or imagine the referent of a feminine noun as wearing a bow and skirt, suddenly it becomes that much easier.

So I've been designing some flashcards that kinda illustrate what I mean, by using as many quirks as possible to really make the words and phrases memorable. It's actually pretty hard to design things to be memorable, and it's hard to rate how memorable they are to since the more time you spend attempting to make something memorable, the more likely you are to remember it anyway just because you've spent a long time on it. You need independent feedback.

One technique that people use is to put Post It notes on things around their house with the names of things in the language that they're learning. I think that's a good idea. I've also wondered about putting posters of language things up in places where people spend lots of time: in the kitchen, above the TV, by the bed. Even if people don't focus on them especially, you get a kind of subliminal learning from them. Of course, this method doesn't necessarily make it more compelling, it's a kind of forced learning by locationeering, but it might help when you're already finding other methods compelling.

As for making it compelling when one gets to the intermediate stage, obviously communication with other speakers is the way to go. On the writing front, maintaining a journal could be a good idea; but what about a wiki or something a little more interactive? If you're learning a rare or dying language, then getting involved with efforts to revive it and make it more widely known are a good thing: and there'll usually be some other people who are avid about doing the same, so it can be a good community thing. On the other hand, I've heard that people can get quite competetive about it too.

So in summary, I'm just thinking about slightly more innovative ways of learning a language: by using more naturalistic methods such as you'd use to teach a child; by using lexicons and interactive methods with other adults such as wikis; and by making linear learning techniques less linear and more mnemonic by using various motifs and other ideas.