[Katsuo]

nadiatims Wrote:not bad not bad.
But you already knew those words

Er, yes. In fact I learned そっくり a few years ago from the Apple advert I linked to above.

Quote:Also they took you more than 4-5 seconds each right?

Only a couple of seconds each to think of, but that's because (a) I use mnemonics a fair bit and as result often come up with them quickly, and (b) they were relatively easy ones.

zachandhobbes Wrote:I have a slightly tangential question.

When people refer to the "X000 vocab will cover 95% of convos"...

Does this statistic account for different forms of verbs?

行く
行きます
行って
etc etc

The poster above answered the question already but I just wanted to point out that percentages are very deceptive and kind of a dangerous game to play. I remember when I started out I thought, "ah sweet 2000 words make up 80% of spoken language" but what I didn't realise was that meant I don't understand 1/5 words being said. All of a sudden 80% went from sounding like understanding a LOT to actually not understanding that much at all. Even when reading novels, knowing 99% of the vocab means you're actually looking up 2 words per page roughly!

It's kind of nice in the beginning though that you do get such excellent mileage in terms of R.O.I of time spent learning vocab, though just bear in mind it's slightly misleading.

Thanks to both replies, they were informative.

At Mezbup, I'd rather not understand 1/5 of words being said, than not understand 4/5.

(At the point where I am right now, I pretty much understand like... 1/10th)

Katsuo Wrote:

nadiatims Wrote:not bad not bad.
But you already knew those words

Er, yes. In fact I learned そっくり a few years ago from the Apple advert I linked to above.

Quote:Also they took you more than 4-5 seconds each right?

Only a couple of seconds each to think of, but that's because (a) I use mnemonics a fair bit and as result often come up with them quickly, and (b) they were relatively easy ones.

a. a couple of seconds? like 2/3 seconds? I could see making the the sound connection to sock relatively quickly, but the well formed story in your post surely took longer no?
b. Well I just chose some random gitaigo. But you admit some difficult words will take longer then?

regardless of how long a mnemonic takes to form, do you think it reduces the number of reviews required to take an item to maturity or the point of non-review? How many reviews on average do you think you review each item?

There's a lot of question marks in this post.ごめむ

[Katsuo]

nadiatims Wrote:a. a couple of seconds? like 2/3 seconds? I could see making the the sound connection to sock relatively quickly, but the well formed story in your post surely took longer no?

For そっくり the thought "match up like a pair of socks" came in a couple of seconds, but adding the laundry idea for clarification took a few more.

Quote:b. Well I just chose some random gitaigo. But you admit some difficult words will take longer then?

Yes. As a test I just tried a few other gitaigo. Sometimes mnemonics do come to mind straight away, but other times not. E.g. ごっそり (all, completely, often associated with theft of everything) took about 30 seconds for a good hook.

Quote:regardless of how long a mnemonic takes to form, do you think it reduces the number of reviews required to take an item to maturity or the point of non-review?

It makes a big difference to me for non-kanji words; without any kind of mnemonic I usually just forget them.

nadiatims Wrote:...do you think it reduces the number of reviews required to take an item to maturity or the point of non-review?

I agree with Katsuo. I am definitely much more likely to forget words in the early stages if I don't create a mnemonic for them. It might take a while to become accustomed to creating mnemonics, but it gets easier as you go along. I think the extra time it takes to create the mnemonic is well worth it and saves time in the end.

I've also used mnemonics a lot for remembering my Japanese students' names, e.g., ゆいな (picture her putting *You in a* box); ゆか (playing Ukulele); わたる (A Prussian soldier on his horse charging into battle at *Waterloo*. I give almost all (~800) of them a mnemonic and am able to remember most of them (with help from Anki). (Admittedly, there are times when a mnemonic eludes me and I force their name into my head through brute force.)

In the hopes of making it easier for students to memorize words, I sometimes demonstrate the power of mnemonics for them. They usually comment on how cool it is, but go on in their usual way of bashing words into their heads by moribund repetition. It's a little like a person using a rock to dig a hole even though there's a shovel available. (I guess I just need to make the shovel seem more attractive.)

I agree totally with both of you on the power of mnemonics for remembering something in the short term, next hour, next day etc. However I find their necessity drops away once you realise that remembering something in the short term is not actually necessary for a long term task like language study. It's just extra time spent reviewing and extra time spent devising mnemonics. I strongly recommend people increase their minimal intervals to at least 2 weeks. What I found when I started doing that, is that there are many words, usually 20-30% that I still remember when I review them after that first long interval with no 'work' required. And the words I have forgotten I tend to at least remember adding, so it's not totally forgotten anyway. These trace memories, however faint they may be, are important because they show that even though you don't recall an item completely, it is still in your memory. You remember the word or you remember adding it, you just don't recall the meaning. When it comes time to relearn the meaning, it will stick better because you know you've encountered the item before. It's not totally fresh. You're building upon previous knowledge. By waiting about 2-3 weeks before re-reviewing things in this way, I usually find I am able to learn all items in a list after 1-4 reviews per item spaced over 2weeks to 2 months. Reviewing a word or relearning a word's meaning in this way takes almost no time (couple of seconds?). So the time savings on a per word basis should be obvious.

semperanimus Wrote:I will.
trying to find an appropriate list atm.

1. Use subs2srs to get a giant number of sentences and use MorphMan and Anki's wonderful conditional fields feature to make vocab cards from the i+1 sentences using the base form of the only word you don't know.
2. ?
3. Profit

but surely by the time you get half-way through that deck, the i in your i+1 has changed to like i+10 or 15... so this seems a bit short-sighted

tokyostyle Wrote:MorphMan kept putting off sentences that were extremely trivial for me because it thought they were i+N when in reality they were i+0 or i+1.

My best guess was that it just didn't have enough data, but by the time I have given it enough data I would be well beyond needing it.

Depending on how much work you do inside Anki you may well have a lot of knowledge that isn't represented in known/mature cards, but it incorrectly thinking a card is harder than it is shouldn't be a big problem as you'll get to them eventually- the only real problem it could cause is if you run out of i+1 cards and even after an update you still don't have any new i+1 cards (though this is solved by just making your deck bigger or in the worst case, going through harder cards and just suspending liberally).

tokyostyle Wrote:It also didn't help that I had started doing MCD cards before I tried MorphMan 2 so my results were even worse. All of my MCD cards are by nature i+0 or i+1 so at that point it was much easier to just suspend the outliers by hand since they are so few and far between.

Yea, I could see it not being very useful for MCD decks.

tokyostyle Wrote:I can see how MorphMan 2 would be awesome if I could somehow train it with my knowledge and then import a bunch of unknown data, but I have neither a way to train it or a bunch of pre-made decks I want to learn.

If you have a decent amount of knowledge from sources outside Anki I'd highly recommend shoving a bunch of stuff you can read into a text file and creating a database from it then merging that into your known/mature dbs.