Iversen’s Method, Vocabulary, Japanese, Kanji, and Anki

 Foreign Languages, Japanese Language  4 Responses »
Jul 272009
 

After seeing this post by Vosmiura on the RtK forums, I’m going to try Iversen’s method of learning lists of vocabulary before entering them into Anki.

I’m not usually obsessed with getting particularly high pass rates or having high long-term retention percentages (mine is already around 96%), but I have been noticing lately that there are certain words that just don’t seem to stick, no matter how often I see them, and it’s bugging the crap out of me.

If you jump down a few posts later on, Vosmiura provides graphical evidence of how his retention rates improved in Anki over a 47-day period. It improved for short, medium, and long-term retention, so that’s not too shabby.

The basic gist of Iversen’s method is simple. I’m paraphrasing from his post here. (Scroll down about halfway down to the big post.)

How to Make Word Lists Work

Take a list of 5-7 words in foreign language X you want to learn, which have corresponding meanings in your native language Y.

Write the words in foreign language X in a column on a piece of paper in one color of ink. Then learn all of the meanings in your native language Y, and only write them down in the next column when you know all of them and can write them without hesitation.

So if you go down your list of 5-7 words, and you keep missing one, don’t write down the translations for any of them yet. If you have to struggle to remember one word in your native language Y, don’t write anything down yet. Keep going at it until you can. If you have to look stuff up, then look stuff up.

Once you can remember everything, then write down all of the translations in your Y language in a different color ink.

Now go and cover up the original words in the foreign language X column. Based only on the words you see in your Y language column, use the 3rd column to reconstruct the X column the same way you had to construct the Y column. That is, you can’t write anything down until you can write everything down correctly.

So when you’re done, your sheet looks something like this:

X language --> Y language --> X language

With one column for each.

Now comes the tricky bit: applying it to Japanese, which has kana and kanji for a lot of words. If a word has no kanji, you’re fine. It’s just English and kana. Not a problem. But kanji will complicate matters, as they always do.

Vosmiura’s approach is to break it down like this:

Kanji --- Kana --- English

He covers the kana and English columns while looking at the kanji. That way he makes sure he has the meaning and the reading correct.

He also varies the way he tests the list. If the list has words in the order a-b-c-d-e-f-g, he doesn’t always test in the order abcdefg. He often tests gfedcba, or acfedgb, or any other random order.

I think it’s a good idea to avoid getting the cde words lost in the middle.

Remember, we’re good at remembering firsts and lasts, but horrible at remembering stuff in the middle.

I’m going to try messing with the order a little to fit my models better, and see how it works. It may work, it may fail spectacularly.

So I’m going to try setting it up like this for now:

Kana --- Kanji --- English --- Kana --- Kanji

That way, I get Iversen’s X-Y-X pattern, and I get my production needs met. Although in this case I guess it’s more of a X-X’-Y-X-X’ method.

Monolingual types will probably froth at the mouth a bit, but I’m not a monolingual zealot. Whatever gets my error rate down is cool with me.

I am becoming more and more “theory agnostic” and am just using whatever works best for me.

Oh, and the mountains are still gorgeous.

Links:

This blog is protected by Dave\'s Spam Karma 2: 3159 Spams eaten and counting...