Zurisu, How do you go about setting the new cards to appear at the end?* Is that an option in Anki (I use Ankidroid really) or do you change the new cards/day count from 0 to x when you are almost done with reps?

Xanpakuto, What do you mean when you say you "use the knowledge of the hiragana to make structure for the English word so that takes seconds"?

I might give some kind of combination of the above two ideas from you guys--divergent as they may seem--a try.

Xan, do you look in the list on Anki to see the cards that will be due next?*

Thanks.

*My skills are pretty abysmal w/r/t the underworkings of Anki. On Ankidroid, I'd venture about 10-15% of the core audio samples don't play. On Anki on my computer, the replay button doesn't work. I have just kinda dealt with it. Perhaps I'll further pursue these improvements so I'm really getting the benefit from the audio, although it seems like I have about enough as it is.

tashippy Wrote:Zurisu, How do you go about setting the new cards to appear at the end?* Is that an option in Anki (I use Ankidroid really) or do you change the new cards/day count from 0 to x when you are almost done with reps?

Tools->Options->Show new cards after reviews. No idea about AnkiDroid.

I use custom study and/or suspend/unsuspend, though. 0 new cards per day, "Increase today's new card limit" by 5 at a time.

tashippy Wrote:Zurisu, How do you go about setting the new cards to appear at the end?* Is that an option in Anki (I use Ankidroid really) or do you change the new cards/day count from 0 to x when you are almost done with reps?

It IS an option yes (it's in Preferences, change the dropdown from "mix new cards and reviews" to "show new cards after reviews." I actually can't find this option in Ankidroid, so either I'm not looking hard enough, it's not there, or (hopefully not this) it doesn't support that setting at all.

However, I myself do actually change the new card count manually. I leave "mix new cards and reviews" on and keep new cards at 14 (which is an arbitrary number; it's enough to make a good cut into that number 50, but not enough to bloat my reps very much), then when I'm finished with reps, I go and change the number to 24, do those 10 cards, then 34, 44, 50. I like doing it manually because I feel like I have more control over my rep numbers, even though the numbers don't actually change or go down. It's a psychological thing XD But it helps me a lot so I stick with it.

Quote:On Ankidroid, I'd venture about 10-15% of the core audio samples don't play.

It's probably because of how big the deck is--that's a lot of media files. It took my phone WEEKS to download everything from Core2k6k (it would download a certain amount of data, then give me an error, then download a bit more the next time I used it, until finally after many spurts it said that media synched successfully). I'm not sure if your problem is related.

Actually, I'm not really sure if 100% of the files work since I haven't used Ankidroid in a while (but coincidentally I'm going to have to use it a lot this weekend, so perhaps I'll find out )

EDIT: yes they all work fine

tashippy Wrote:[...]
Has anyone used Anki with a focus on collocations?

I generally use Anki for individual vocabulary, but whenever I come across unknown words I try to parse the sentence (focusing on the bunsetsu containing the unknown word). I might use other factors such as audio / kanji composition.

tashippy Wrote:Zurisu, How do you go about setting the new cards to appear at the end?* Is that an option in Anki (I use Ankidroid really) or do you change the new cards/day count from 0 to x when you are almost done with reps?

Xanpakuto, What do you mean when you say you "use the knowledge of the hiragana to make structure for the English word so that takes seconds"?

I might give some kind of combination of the above two ideas from you guys--divergent as they may seem--a try.

Xan, do you look in the list on Anki to see the cards that will be due next?*

Thanks.

*My skills are pretty abysmal w/r/t the underworkings of Anki. On Ankidroid, I'd venture about 10-15% of the core audio samples don't play. On Anki on my computer, the replay button doesn't work. I have just kinda dealt with it. Perhaps I'll further pursue these improvements so I'm really getting the benefit from the audio, although it seems like I have about enough as it is.

I do look at the list in anki and put all words in my notebook, well at least now I do. I used to take my vocab from songs with the same method. But I thought to myself I'll just get these 6000 out of the way fast than take vocabulary from the resources I enjoy, not saying core is boring, not very fun though.

Using hiragana as a structure, hmm how can I explain it. Our brain can't really learn so many things at once, well for me I could learn two things at once. When I look at a kanji, I guess the readings of those that I know, then just finish it off. After you got the hiragana down its really based on the learner. Some people like to break up the sounds and relate it to an English word they know. There was a good video that went over it, I'll look for the link. Pretty much for the word neko. They imagine a cat with a long neck or something like that, for me that doesn't work. Instead.....hmm I can't really answer that sorry it just comes to me. Learning the hiragana takes a bit longer, English word just comes almost immediately.

Xanpakuto Wrote:Here's how I brute force.
Each line looks like this

Politics - せいじ 政治
Airplane - ひこうき 飛行機

I look at the kanji and think of the hiragana for it, not the English meaning. After I memorize the hiragana for it I then memorize the English word. I use the knowledge of the hiragana to make structure for the English word so that takes seconds.

For order it's pretty simple. Study one word, to to the second, before the first word falls out study it again. Keep going repeating the hard ones, then repeat the easy ones maybe after ten words. Then anki all of them! Takes me about an hr to have all of these pre studied and in anki.


Edit: I also want to start studying production, still finding the most efficient way to do it. But I do know ill be using my expo board more.

I like the way you learn your vocab, I'm going to try learning vocab the way you do it. I feel like my prestudying/brute forcing the cards is lacking. It seems more time spent prestudying cards in the beginning means less time spent in the long run trying to relearn it..

I love my expo board

tashippy Wrote:

killua Wrote:I think you got it wrong. I started a little less than two weeks ago and I'm gonna do the last ~300 in a couple of days

from http://forum.koohii.com/showthread.php?p...#pid195014

zurisu Wrote:But I'm adding 50 cards/day right now

from http://forum.koohii.com/showthread.php?p...#pid194974

Sheesh, people here always seem to find a way to make me feel like I'm not doing enough myself. Not that that's a bad thing though-- it motivated me to do 40 extra today, at least.

Also, I totally forgot I posted here... A lot has changed in that time. If anyone cares, I stopped doing production cards. That was 2 hot 2 handle. I'm not even using them for core anymore either. I switched the template... I kept failing old cards with words I really shouldn't be failing.

I sort of realized that it's one thing to create a vivid picture of a definition in your mind upon first seeing a word, and it's quite another to be able to evoke that picture every time you lazily glance at a question prompt.

And I can vouch for zurisu's statistics. I'm doing 40 new / day with about an 80% average for young and mature (85% on a good day), and the daily cards seem to hover around 500.

edit: also, if you guys are having trouble with big review stacks, I highly recommend the metronome method. Check the thread by Splattered.

Haych Wrote:

tashippy Wrote:

killua Wrote:I think you got it wrong. I started a little less than two weeks ago and I'm gonna do the last ~300 in a couple of days

from http://forum.koohii.com/showthread.php?p...#pid195014

zurisu Wrote:But I'm adding 50 cards/day right now

from http://forum.koohii.com/showthread.php?p...#pid194974

Sheesh, people here always seem to find a way to make me feel like I'm not doing enough myself. Not that that's a bad thing though-- it motivated me to do 40 extra today, at least.

Also, I totally forgot I posted here... A lot has changed in that time. If anyone cares, I stopped doing production cards. That was 2 hot 2 handle. I'm not even using them for core anymore either. I switched the template... I kept failing old cards with words I really shouldn't be failing.

I sort of realized that it's one thing to create a vivid picture of a definition in your mind upon first seeing a word, and it's quite another to be able to evoke that picture every time you lazily glance at a question prompt.

And I can vouch for zurisu's statistics. I'm doing 40 new / day with about an 80% average for young and mature (85% on a good day), and the daily cards seem to hover around 500

edit: also, if you guys are having trouble with big review stacks, I highly recommend the metronome method. Check the thread by Splattered.

Oh my if I had 80% I would literally cry. If I ever go below 95% retention rate I'm going to bring down the pace, not that it ever happen before of course ^^

I dunno man, I think that attitude will catch up with you eventually. Forgetting is only natural, you can't beat yourself up over it. There's some stuff you can do.. cramming.. altering the intervals.. but sometimes you just gotta be happy with what you can remember. I see a lot of material in a day, and I can't slow down to deal with the hard stuff. I've done that before and found it gets me nowhere. I'm pretty much sunk the moment I start worrying about what I've forgotten. My pace slows to a crawl.

in context. learnig words become almost effortless after a while. effortless is an exaggeration. I mean easier. you really shouldn't killl yourself over retention rate sand whatnot because it's just going to get easier and easier to learn/remember/notice words as you learn more so... just hold on to that thought. just because you have something in anki and you scoared the perfect score doesn't mean you know the word anyway. it's happened to be before. I'll have it in my deck and i have no problem with it but then one day i'll hear that word and then it doesn't register with me with the sounds or i won't be able to catch the word... then when i type it up/find out and then i go oh that word. anki doesn't relaly prove anything other than the fact that you understand it within the context of the card which probably also doesn't involve audio so there's a chance the word will still fly over you when you hear it in audio..

What I do is just learn them in groups. The more the better. The Optimized Kore deck does this job well. Sometimes, I do stories too if I am in the mood. Then, go in the wild and read read read.

Haych Wrote:I dunno man, I think that attitude will catch up with you eventually. Forgetting is only natural, you can't beat yourself up over it. There's some stuff you can do.. cramming.. altering the intervals.. but sometimes you just gotta be happy with what you can remember. I see a lot of material in a day, and I can't slow down to deal with the hard stuff. I've done that before and found it gets me nowhere. I'm pretty much sunk the moment I start worrying about what I've forgotten. My pace slows to a crawl.

I couldn't agree more. I felt the same way too in RTK. It was really stressful! What I do now is if I forgot a card, I just press hard, study it once more and move on. Consume my native materials, have fun and just enjoy the journey. If I keep on punishing myself just because I forgot a card, I don't think I will last long. I will eventually hate studying Japanese and quit.

bimspramirez Wrote:I felt the same way too in RTK. It was really stressful! What I do now is if I forgot a card, I just press hard, study it once more and move on. Consume my native materials, have fun and just enjoy the journey..

Oh I see, when you forget a card, you just hit the button that says you didn't forget it.

I really don't see the value in lying to the SRS. It's trying to get an accurate assessment of what's in your head at that moment so that it knows what to show you the next day and on into the future. You hitting the "hard" button on cards you've forgotten is just going to throw it off.

If you're having a problem with missed cards taking too long to get back up to speed, you could monkey with the intervals for re-learn cards so that they get back to a longer interval earlier. That way it doesn't feel like you're wasting so much time on cards you just needed your memory jogged slightly for.

As well, a better strategy might also be to get over your fear of failing cards. On the surface you're talking a big game about how you don't really care, but you appear to care enough to lie to a computer program that nobody else is looking at but you and is designed to help you based on your feedback.

erlog Wrote:Oh I see, when you forget a card, you just hit the button that says you didn't forget it.

Yeah, I do that on the first review. For the succeeding, I fail it. LOL.

I haven't tried changing the intervals. I'll look into it, thanks!

Xanpakuto Wrote:If I ever go below 95% retention rate I'm going to bring down the pace, not that it ever happen before of course ^^

If you really happen to have 95% retention in Anki then your scheduling is screwed. You see, Anki comes with default Forgetting Index = 10. This means that the scheduling aims for the at most 90% retention rate. Anything below, and from the algorithm's point of view, the intervals are to long, anything above - the intervals are to short. In other words, 95% retention rate means that you unnecessary do reviews of cards that you remember. I would prefer to learn new cards in that time but that's me.

I guess I'm using anki wrong, wouldn't care less though as its benefitting me and its enjoyable. My easy interval for new cards is 1 day. I'm not at my computer right now so the percent thing, is 150? Maybe that's the cause for inaccurate reading. Once I fail a card it goes right back to pretty much new with its one day interval.

If I make the littlest error, either mess up in hiragana, English word, take longer than 10 seconds I hit fail.

Inny Jan Wrote:95% retention rate means that you unnecessary do reviews of cards that you remember.

I think Xanpakuto's relationship to those numbers is exceptional because s/he studies the vocab before it comes up in Anki. Most everyone else seems to first encounter the material when the respective card comes up.

Xanpakuto Wrote:I guess I'm using anki wrong, wouldn't care less though as its benefitting me and its enjoyable. My easy interval for new cards is 1 day. I'm not at my computer right now so the percent thing, is 150? Maybe that's the cause for inaccurate reading. Once I fail a card it goes right back to pretty much new with its one day interval.

Starting ease=150 would explain it. The default is 250. You'd be buried under reviews if you didn't have such a great recall rate.

Quote:If I make the littlest error, either mess up in hiragana, English word, take longer than 10 seconds I hit fail.

English "word"?

Erm sorry, it was worded odd. Pretty much if I make the smallest mistake on the card I fail it.

-Take longer than 10 seconds
-Mistake on the reading
-Mistake on the (vocab word), lets say I'm studying 猫 thinking it was a dog, but 1-2 seconds later I know I'm wrong without looking at the card at say it's a cat. <--- I will mark the card wrong.

I appreciate your perfectionism, and had my own bout with it myself a couple years back... but the point of anki is that it helps you put a huge amount of information in your head with optimal efficiency.

If your goal is to get perfect scores playing the anki game then by all means continue, but if your interest is in memorizing thousands and thousands of things in an efficient manner, lying to anki/perfectionist grading is only slowing you down.

I appreciate that it feels good though.
I'm hard core man!

Xanpakuto Wrote:I guess I'm using anki wrong, wouldn't care less though as its benefitting me and its enjoyable. My easy interval for new cards is 1 day. I'm not at my computer right now so the percent thing, is 150? Maybe that's the cause for inaccurate reading. Once I fail a card it goes right back to pretty much new with its one day interval.

If I make the littlest error, either mess up in hiragana, English word, take longer than 10 seconds I hit fail.

Mistaking the reading or meaning definitely merits beginning a new lapse.

What you're doing sounds fine to me.
I also agree that it is most efficient to initially learn words through "effortful processing".

From what I can tell, our pass-criteria is similar:
if recall is immediate I press easy, if it takes some effort I press good, if I'm uncertain I got it right and got it right press hard.

Trust your own intuition. Learning is an individual sport.

'Easy' increases the "ease" of a card and multiplies the normal interval by 130%.
'Normal' brings "ease" closer to average, 250%. modifier at 100%.
'Hard' decreases "ease" and has a modifier of 70% (?).
'Again' begins a new lapse

"Ease" determines the interval, the modifier changes that interval.
Intervals range from 1.80x to ~1.96x to ~2.01x, iirc
(Assuming 220%, 250%, and 280% ease)

EDIT after looking through the manual "1.80x to ~1.96x to ~2.01x" is wrong. The "ease" already determines the interval. Re-calculation is not necessary.
Also, the modifier affects the final interval. (View anki manual for more details).

Aspiring Wrote:Trust your own intuition. Learning is an individual sport.

this is definitely true. however there are people that study how we learn, and we ought consider their opinions as well (like for instance the people who created the program we're using)
the supermemo grading suggestions are:
5 - perfect response
4 - correct response after a hesitation
3 - correct response recalled with serious difficulty
2 - incorrect response; where the correct one seemed easy to recall
1 - incorrect response; the correct one remembered
0 - complete blackout.
it's based on a 6-button program (which would be nice) but you get the idea...again these are people concerned with optimization and efficiency.

Aspiring Wrote:'Hard' decreases "ease" and has a modifier of 70% (?).

this is not true. hard actually increases your interval by 1.1x. the lack of a 'soft fail' option in anki is something that has been bemoaned for years. I find that a pragmatic if imperfect way to get around the problem is to nerf your fail button a little.

It always seemed masochistic to me to start at 0, 10min, 1day after failing a card that had been mature for 6 months. So I have it set at 20%... then if it comes back failed again after that it becomes a new card again like the old fail button would.

Aspiring Wrote:Ease" determines the interval, the modifier changes that interval.
Intervals range from 1.80x to ~1.96x to ~2.01x, iirc
(Assuming 220%, 250%, and 280% ease)

This is not the complete algorithm... as there is also a 'momentum' factor built in depending on your history of successes/failures with the individual card.

http://www.supermemo.com/english/ol/sm2.htm

After a bit of searching...

Quote:Starting ease controls the easiness that cards start out with. It is set when you see a card for the first time. It defaults to 250%, meaning that once you’ve finished learning a card, answering "Good" on subsequent reviews will increase the delay by approximately 2.5x (eg if the last delay was 10 days, the next delay would be 25 days). Based upon how you rate the card in subsequent reviews, the easiness may increase or decrease from what it starts out as.

Quote:Interval modifier allows you to apply a multiplication factor to the intervals Anki generates. At its default of 100% it does nothing; if you set it to 80% for example, intervals will be generated at 80% of their normal size (so a 10 day interval would become 8 days). You can thus use the multiplier to make Anki present cards more or less frequently than it would otherwise, trading study time for retention or vice versa.

Quote:For moderately difficult material, the average user should find they remember approximately 90% of mature cards that come up for review. You can find out your own performance by opening the graphs/statistics for a deck and looking at the Answer Buttons graph - mature retention is the correct% on the right side of the graph. If you haven’t been studying long you may not have any mature cards yet. As performance with new cards and younger cards can vary considerably, it’s a good idea to wait until you have a reasonable amount of mature reviews before you start drawing conclusions about your retention rate.

On the SuperMemo website, they suggest that you can find an appropriate multiplier for a desired retention rate. Their formula boils down to:

log(desired retention%) / log(current retention%)

Imagine we have a current retention rate of 85% and we want to increase it to 90%. We’d calculate the modifier as:

log(90%) / log(85%) = 0.65

If you plug the resulting 65% into the interval modifier, you should find over time that your retention moves closer to your desired retention.

One important thing to note however is that the tradeoff between time spent studying and retention is not linear: we can see here that to increase our retention by 5 percentage points, we’d have to study 35% more frequently. If the material you are learning is very important then it may be worth the extra effort - that’s something you’ll need to decide for yourself. If you’re simply worried that you’re forgetting too much, you may find investing more time into the initial learning stage and/or making mnemonics gives you more gain for less effort

From the following links:
http://ankisrs.net/docs/manual.html#reviews
http://ankisrs.net/docs/manual.html#reviewing
http://ankisrs.net/docs/manual.html#studying
http://ankisrs.net/docs/manual.html#what...s-anki-use?

I re-quote myself (the manual describes this better in the 'studying' section):

Quote:'Easy' increases the "ease" of a card and multiplies the normal interval by 130%.
'Normal' brings "ease" closer to average, 250%. modifier at 100%.
'Hard' decreases "ease" and has a modifier of 70% (?).
'Again' begins a new lapse

As dtcamero mentioned, previous answers affect the next interval. This is because previous answers alter the ease of a card.

Apparently, the "algorithm" in anki is a bit more user-friendly and customizable (e.g. 'Ease' reflects the subsequent interval(s); you can customize the steps; modifiers can be changed manually)

"Ease:
The approximate amount the interval will grow when you answer a review card with the "Good" button." manual.html#card-info

Avg. ease is a better indicator of performance than retention rate. Take notice of both ease and retention, see how they reflect against each other

Aspiring Wrote:After a bit of searching...

Imagine we have a current retention rate of 85% and we want to increase it to 90%. We’d calculate the modifier as:

log(90%) / log(85%) = 0.65

If you plug the resulting 65% into the interval modifier, you should find over time that your retention moves closer to your desired retention.

the tradeoff between time spent studying and retention is not linear:

As dtcamero mentioned, previous answers affect the next interval. This is because previous answers alter the ease of a card.

Avg. ease is a better indicator of performance than retention rate. Take notice of both ease and retention, see how they reflect against each other

Thanks for finding these quotes. It's about time I start thinking more in this way and studying that manual. I tend to think in a "yeah yeah just give me the cards and make it easy" way sometimes. But I think learning these things help understand how learning takes place, which is invaluable knowledge.

dtcamero Wrote:

Aspiring Wrote:Trust your own intuition. Learning is an individual sport.

this is definitely true. however...

I was going to respond to this as well but I'm glad you got there first.
It's important to be aware of individual learning styles. I think it's also valuable to communicate with others and see how they learn. People often feel short-changed if they meet a speaker of their target language who wants to speak the learner's L1--in other words, they both want to get practice in the other's language, but I think you can gain insight into the learning process even if you speak in your L1 and watch them learn.

Aspiring Wrote:If you haven’t been studying long you may not have any mature cards yet. As performance with new cards and younger cards can vary considerably, it’s a good idea to wait until you have a reasonable amount of mature reviews before you start drawing conclusions about your retention rate.

I might use that 'hard' button too much when learning new words to change the pace.

dtcamero Wrote:...nerf your fail button a little.

I need this especially when I dip back into my old RTK deck.

tashippy Wrote:Thanks for finding these quotes.

No problem

I must note that these quotes were taken directly from the anki manual, and that I deserve no credit for the work put into it by anki developers.

*cough* quote=ankimanual *cough*

This has been an interesting read and I've learned a lot about some of the inner workings of anki that I never bothered to elucidate. However..

Aspiring Wrote:log(desired retention%) / log(current retention%)

This formula seems a bit dubious. It accurately predicts that you would need infinite review time if you wanted to bring your retention from anything to 100%, and if you didn't care whether you remembered a single item, you could bump the interval modifier up to anything you want and never do another review again.

But its telling me I'd need 4 times more studying (modifier to 0.25) to go from 80-90%... Seriously? I think that's a little crazy... The log means its a continuous exponential decay, but there's got to be a linear region in there... I think if I did twice as much reviewing , I could easily cut my failed cards in half, and get to 90% or more... of course, I'm not going to do that. It would just be discouraging and time-consuming...

edit: nevermind, i was looking at 80-95%.. 80-90 is more like 0.44, which is close to my estimate of double time for halved failures.. I still stand by the next statement tho

So yeah, edit the modifier if you feel like it, but you're probably better off using common sense.