I've suffered a lot with anki burnout, and the advice for that I see the most is "delete cards that are too easy and cards that are boring." I've been thinking about that lately. I have a lot of cards...let's use 初めて here...that are due for review again 1 year from now or longer. But that word is in the core2k. It's a very common word. And if I'm exposing myself to native materials everyday or at least once a week, I should see 初めて again before the year is up. That means it's not really due for review again. By that logic, native material should be enough for me to retain 初めて forever, and I no longer need to review it in anki. (I suppose there's the fear that I'll stop looking at native materials and forget 初めて without anki, but there's really no reality in which I no longer read manga but still do anki reviews.)

So wouldn't it make sense to do a search for cards with intervals a year or longer between reviews, and delete those?

If you're a crazy polygot with a 50k deck and there's low-frequency words in there you won't find in native materials and therefore have to review in anki or you won't see them, yeah, don't delete those. But for us meager manga-readers with less than 2,000 cards, why do we need anki for long-term review of really common words?

I think you could certainly delete a word like that without worrying too much about it. The real question, however, is how much is deleting it going to help lower your current number of reviews. Unless you're really deep into your deck, I can't imagine you having enough cards in your deck that appear with that kind of infrequency to make a huge difference (even if you delete 365 such cards, you're only saving yourself one review a day). Deleting such cards is probably a good start, in other words, but if you're really burnt out, you might want to be a little more aggressive, especially if you're confident that the word is one that's going to come up frequently in native material. Anki, for me at least, is just to get started/familiar with a word, not to master it or "learn" it long-term.

Maybe this is the Anki addition-addict in me talking but wouldn't it be more effort to dig through anki and pick which cards are REALLY worthy of deletion than to click Easy once every few years and say "YES DAMNIT I KNOW WHAT A 子供 IS" and then move on with the next half second of your life?

systematically...

jessem Wrote:So wouldn't it make sense to do a search for cards with intervals a year or longer between reviews, and delete those?

Not really, no. It only makes sense to delete easy cards before you start reviewing them, not after they're mature. The interval is not what makes cards easy. Words that aren't easy end up with long intervals too. Besides, you're not saving all that much time, by deleting cards that will never come up more than once. If you delete 500 such cards, you're at most saving half an hour of review time in total (it shouldn't take more than 3-4 seconds to review a word that you know well).

Easy cards are not what causes people to burn out. Having a lot of cards that are too difficult does. Those are the ones you should be suspending (both manually, and by having a low leech threshold).

By systematically deleting cards with long intervals, you basically create a system that only keeps cards that are too difficult (those are the cards that never mature to a long interval), and deletes everything that you should be keeping. Instead, a good system to implement is one that does the opposite: deletes cards that are:
1. too easy from the start
2. cards that you recognize as too difficult either on first encounter or at any time later on
3. cards that have 5 or more fails on them.

I use Ankidroid for reviews (the charts aren't as nice as the desktop version...), but I'm sure these practices can be performed as easily on the other versions:
1.) If it's a card you'd hit 'easy' on, suspend/delete it (I prefer to suspend cards... because).
2.) If it's a card that you consistently get wrong and it aggravates you, suspend/delete it (being aggravated or angry won't do you any good for remembering that card you failed ten times; I suggest suspending these so that you can look at them again when you have more context that you can use to remember the card)

Going back through Core6k ('cause I quit reviewing and forgot a bunch of things), following these guidelines (especially the first one) has kept me from having multiple hundreds of reviews a day, since I remember more cards than I forgot. I'll admit I'm a bit stubborn with the second one though: I hate admitting that I just can't remember a word; thankfully, I don't encounter those kind of cards too often anymore.

Anki is just a tool. It's not actually your brain, and so I think the advice of deleting cards is mostly okay. For me, though, I always felt funny about deleting cards. So I did something a little bit different.

I aggressively split my card decks up, and I just let some of the older easier decks lapse if I'm taking a break for a while or spending more of my study time on harder and newer material. I didn't delete them because I want to keep a record of my studying, and it's not like they're taking up a crazy amount of space or anything.

I have found myself going back and catching up on those lapsed easy decks when I find myself with more free time. Plane flights/vacations are good for burning through a bunch of Anki cards, and some Sundays you just want to put a movie on the TV and roll through flashcards without thinking too much.

If you're studying effectively newer material should help reinforce older material naturally. It used to take me like an hour to do 225 cards. Those same cards now take 20 minutes because I've gotten a lot better at Japanese. So a backlog of easy material you've let lapse can look a bit daunting at first, but as long as you've kept moving forward you'll find it pretty easy to burn through when you find yourself with the time to do so.

So I guess my advice is, "Just keep swimming." Keep moving forward, and don't let yourself be bogged down by old material if it has outlived its usefulness. If you don't want to delete it then just re-organize your cards so you can pay attention to the stuff you want to pay attention to. Decks, tags, filtered decks, and subdecks are all great features to make use of in Anki.

Why are people recommending to delete hard cards? Those are the ones you need to keep. Those are the ones you will have problems with. It's not like you can just delete the card and then it will be deleted from the language or something. If you added it, it is most likely worth remembering. Unless you want a pointless deck full of cards that you mark easy every day.

Delete cards as you do your deck ( just press delete button ) Making that efforts to search for the cards and assessing which ones are common or easy is waste of time. Another option is start a completely new deck if the deck that unsalvegable. I've done that before because I was forced to because I didn't anki sync and it was fine.

kameden Wrote:Why are people recommending to delete hard cards? Those are the ones you need to keep. Those are the ones you will have problems with. It's not like you can just delete the card and then it will be deleted from the language or something. If you added it, it is most likely worth remembering. Unless you want a pointless deck full of cards that you mark easy every day.

I believe the recommendation is to suspend cards that are needlessly difficult - the 5% that you keep having trouble with over and over. The difficult words get easier over time, so go back to them eventually once you've exhausted the low hanging fruit. That way you can cover more ground with the moderately difficult cards that you need help remembering, but you don't end up getting demoralized failing the same cards every time. So the recommendation is more of a goldielocks situation where it's not too difficult that you get demoralized and not too easy that you're not learning anything, but just right.

jessem Wrote:I've suffered a lot with anki burnout, and the advice for that I see the most is "delete cards that are too easy and cards that are boring." I've been thinking about that lately. I have a lot of cards...let's use 始めて here...that are due for review again 1 year from now or longer. But that word is in the core2k. It's a very common word. And if I'm exposing myself to native materials everyday or at least once a week, I should see 始めて again before the year is up. That means it's not really due for review again. By that logic, native material should be enough for me to retain 始めて forever, and I no longer need to review it in anki. (I suppose there's the fear that I'll stop looking at native materials and forget 始めて without anki, but there's really no reality in which I no longer read manga but still do anki reviews.)

Do you mean 初めて? 始めて is the テ形 of 始める.

I'd get rid of all everyday words like that if you're a comfortable reader/listener. If you use the language even a little, it's unlikely you'll forget anything in the top 1000-2000 if you already know the words well.

I think that reality you mention may exist. If you end up busy with school/work and socialising and you have 500 reviews, then the manga goes out of the window.

yogert909 Wrote:I believe the recommendation is to suspend cards that are needlessly difficult - the 5% that you keep having trouble with over and over.

It's not necessarily 5%, could be a lot higher, as well. Your low hanging fruit analogy is spot on. But, usually, low hanging fruit are a lot less than 95%. Can even be less than 50%.

When adding vocab cards, my criteria for "low hanging fruit" (for Kango at least) is that I have to either know the word (from audio immersion), or know at least one of the Kanji, including its reading, well, for it to be a keeper. If it's a word I never heard, made up of Kanji readings I don't know yet, it's an automatic suspend because it's too hard. Needless to say, I suspend quite a few vocab cards.

When adding sentences, the criteria for low hanging fruit is that it shouldn't have more than two words I don't automatically recognize. And that's because I have the audio in the question, in sentence decks. If I didn't have the audio to help me, then the criteria would be that it has to be an n+1 sentence (I have to be able to read and understand all the words except one).

I didn't use these criteria as a beginner, learning Japanese (and I haven't used SRS for any other languages in the past). But I sure as hell intend to use them for future languages I learn, from the start. Knowing what I know now, if the only deck available was difficult enough that these criteria would exclude 90% of the cards, then that's exactly what I would do: I would read the cards, one by one, as they come up, and suspend 90% of them. It's not an easy thing to do, admitting that I'm too stupid for 2/3 or more of a deck others have bragged about completing in one go, but it's exactly what I should've done with the Core sentence deck, for instance, when I first started reviewing it as a beginner. Instead, I struggled through sentences that were way above my level, and they delayed my progress significantly.

jimeux Wrote:Do you mean 初めて? 始めて is the テ形 of 始める.

I'd get rid of all everyday words like that if you're a comfortable reader/listener. If you use the language even a little, it's unlikely you'll forget anything in the top 1000-2000 if you already know the words well.

I think that reality you mention may exist. If you end up busy with school/work and socializing and you have 500 reviews, then the manga goes out of the window.

Your post doesn't add up. If a deck has thousands of words that someone is comfortable reading, then how exactly do 500 reviews take up so much time that there's none left for manga?

Someone who's comfortable reading a word should take at most 3 seconds to look at that word, and press space/space or space/4. 500 reviews made up of words like that should take 1500 seconds, which is 24 minutes. Plenty of time left over for manga.

Again: cards that are too easy are not the ones causing people to be overwhelmed by reviews. Cards that are too easy consume an insignificant fraction of the total time people spend reviewing. They don't matter, and they're not worth spending time on to try and get rid of them. The cards that cause the problems are the cards that are too hard.

kameden Wrote:Why are people recommending to delete hard cards? Those are the ones you need to keep. Those are the ones you will have problems with. It's not like you can just delete the card and then it will be deleted from the language or something. If you added it, it is most likely worth remembering. Unless you want a pointless deck full of cards that you mark easy every day.

Maybe you need to learn the word at some point, but it doesn't have to be now. If it's demotivating and taking up more time than its worth, it can be safely deleted with the knowledge that some day it will be encountered again in a different context and will be learnt anyway.
If you're adding new words every day and get to mark them all as easy something is clearly going very well.

Lot of good discussion and ideas here, thanks everyone! I feel like I have a better understanding of choosing anki cards as a whole now. If it's a ridiculous card like 私 I'll just delete it instead of hitting Easy next time I see it, but otherwise, I guess I'll let my anki cards do as they do.

A lot of food for thought in this thread for all those writing Anki-replacements...!

kameden Wrote:Why are people recommending to delete hard cards? Those are the ones you need to keep. Those are the ones you will have problems with. It's not like you can just delete the card and then it will be deleted from the language or something. If you added it, it is most likely worth remembering. Unless you want a pointless deck full of cards that you mark easy every day.

I'm recommending that people SUSPEND hard cards, not delete them. And I'm talking about pre-made decks, in which the user isn't actually the one who added the cards.

If it's a deck you made yourself, then, obviously, the recommendation would be to not add hard cards to begin with. Not to add them and then suspend them.

And the reason for these recommendations is this: Language learning isn't hard. It takes a lot of time, but it doesn't have to be frustrating, or challenging, or willpower draining, or however else you want to describe "hard". It should be a smooth, enjoyable process. Same as all learning, really. Staying within your comfort zone will not only make you enjoy yourself, but it will also make you a more effective learner. The notion that pushing people out of their comfort zone will somehow make them better at a task is a myth.

Stansfield123 Wrote:press space/space or space/4

As an aside, I find using the numpad keys much more comfortable.

I've written about this before but it bears repeating that according to the supermemo theory page, peak efficiency is in the 70-80% accuracy range. Of course there are other considerations than efficiency. But if you are going for volume, it makes sense to add cards that your probability of answering correctly falls in that range and weed out cards that are either too easy or too difficult. Personally I tend to get demoralized if my accuracy gets below 80% so I shoot for the 80-85% even though it's slightly suboptimal.

yogert909 Wrote:I've written about this before but it bears repeating that according to the supermemo theory page, peak efficiency is in the 70-80% accuracy range. Of course there are other considerations than efficiency. But if you are going for volume, it makes sense to add cards that your probability of answering correctly falls in that range and weed out cards that are either too easy or too difficult. Personally I tend to get demoralized if my accuracy gets below 80% so I shoot for the 80-85% even though it's slightly suboptimal.

My stats are 84% for all my reviews (of all decks), and 85% for the past month. This includes both learning and reviewing though. For reviewing alone (young and mature cards), I'm at 89%.

Personally, that second stat of 89% is what I pay more attention to. That's the one I aim to keep at around 90%. I don't care where the learning stat is, because I don't view the learning feature in Anki as part of the SRS process. I look at it as a tool to help go over new material, not as the first step in SRS-ing it.

In fact I wish Anki kept a separate count of the time it takes to review vs. learn cards, as well, so that if I decide to take my time when learning new materials (by switching to a browser, and looking stuff up), it wouldn't drive my review times up.

edit: Actually, come to think of it, Anki DOES keep separate time for each category of reviews. That's how it knows to create that color coded "review time" graph. It just doesn't actually output those numbers anywhere. Maybe there's a plugin for that, I'll go look.

anotherjohn Wrote:

Stansfield123 Wrote:press space/space or space/4

As an aside, I find using the numpad keys much more comfortable.

Means you're probably right handed. Check out this add-on for righties instead: https://ankiweb.net/shared/info/2090822731

I like xbox 360 controller + xpadder and mapping the 4 main buttons to the 4 answers

[RawToast]

Stansfield123 Wrote:

kameden Wrote:Why are people recommending to delete hard cards? Those are the ones you need to keep. Those are the ones you will have problems with. It's not like you can just delete the card and then it will be deleted from the language or something. If you added it, it is most likely worth remembering. Unless you want a pointless deck full of cards that you mark easy every day.

I'm recommending that people SUSPEND hard cards, not delete them. And I'm talking about pre-made decks, in which the user isn't actually the one who added the cards.

I saw a few mentions of deleting cards on this thread, but that should be kept to duplicates/unneeded cards like 私, これ, etc that the user has covered before and really does not need to cover again when importing a new deck.

I would recommend Stansfield's approach, as I've found it to improve my own Anki experience. I've stated this all before, but it never hurts to repeat the point:

When you're suspending cards your thought process should be "I'll come back to this later," this is because the card is simply too hard right now. To carry on failing the card will rarely be fruitful and ends up wasting time that should be spent doing something more productive (reading, listening, talking, simply adding more new cards, etc). Sometime later you can enable those cards and try them again and you will find that some of them stick -- this tends to occur because you have gained some exposure to the item from native media, lessons, or from other cards (where it was not the point tested).

This point also goes for RTK items (for those doing recognition). Let the kanji that you struggle with get suspended, you will see them in vocabulary or grammar reviews and then you can re-enable them. I found specifically adding vocab items for them helped me remorse the RTK items.

Stansfield123: thanks for the tip, a very handy addon
----------

This is my Anki graph for today's reviews - looks like I spend about 30 mins/day reviewing mature cards (about 1/3 of the total), and it does indeed feel like wasted time, though I suspect this is in part due to setting the default ease too low early on.

As I can now read with reasonable fluency I'm beginning to wonder if there's a case to be made for never reviewing (very) mature cards at all. The initial goal of being able to read Japanese was rather broad. A more specific goal would be something like, "To be able to read stuff that interests me for an average of X amount of time per day without looking up more than Y number of words", with the added assumption that one is actually doing this in addition to flashcard reviews. SRS is then just a useful way to bootstrap new words.

In addition:

- The purpose of the exercise is to read Japanese, not to pass flashcard reviews in perpetuity.

- Mature cards are no guarantee of recognising a word in the wild, as unfamiliar context can be a strong distractor. Whether I recognise a word for the first time in the wild seems to depend to a significant degree on the context in which I encounter it, regardless of whether the card containing it was full-sentence or single-word. In fact, some cards seem to become increasingly context dependent as they mature past a certain point, particularly cards from pre-made decks that were lacking external context from the outset.

- While it is true that reviews taper off with time, this is a very slow process in practice for large decks.

- Recognising a word in the wild seems to produce much stronger reinforcement that a flashcard review.

- I tend to make new cards for words I fail to recognise in the wild regardless of whether I've already got a card for it anyway.

- etc.

Though on reflection, the main reason for the rant is because nearly all of my mature cards are from the core6k/10k, which I loathe. Maybe I'll just stop reviewing those

anotherjohn Wrote:...looks like I spend about 30 mins/day reviewing mature cards (about 1/3 of the total), and it does indeed feel like wasted time, though I suspect this is in part due to setting the default ease too low early on.

- While it is true that reviews taper off with time, this is a very slow process in practice for large decks.

You could use the reschedule function to increase the intervals of the cards you feel are too easy. I've done this before for sentences that I knew going into it most were going to be easy, but I still wanted to study them infrequently. The beauty is that you can push the cards out to a year or whatever and then when they come back around and you find a difficult one and fail it, it's back to a low interval so you can study it more thoroughly while you rarely ever see the easy cards.

Edit: I just looked at your anki stats and you're getting over 98% accuracy even on new cards. That doesn't seem very efficient. Perhaps you should suspend a lot of them or push them out to 5 year intervals. Even the new cards maybe you should start them with several month intervals so you're not reviewing things you mostly know already every day. And holy crap, you have 78,000 cards in your collection! That's more than most comprehensive dictionaries.