flashcards:"minimum info." vs "bigger chunks"

[chrix]

Yeah, you're right.

At this point I'm not learning characters per se (with the exception of your trad-simp set), only words. Right now I see focusing on writing characters as a waste of time, given the ubiquitousness of computers and the large number of 漢字 I've learnt to write since childhood. If I was to focus on writing characters again, I'd likely revert to using notebook and pencil

[Guest realmayo]

Though I'm not sure about how to properly get it to distinguish between production and recognition cards, I believe you could probably do something by way of tags

I believe anki tags all "forward" cards with the tag "forward", and backwards ones with "backwards", so you can very easily tell Anki to suspend all cards with the "forward" tag.

As for production / recognition, I do almost everything (standalone characters, vocab) both ways. Anki gives me 200 to 300 reps each day, which takes around an hour, sometimes quite a bit more: the reps would be more than halved if I cut out production. I'm not 100% sure what benefit, or how much benefit, I get from being able to produce these, but for the moment at least am continuing this way....

[chrix]

I believe anki tags all "forward" cards with the tag "forward", and backwards ones with "backwards", so you can very easily tell Anki to suspend all cards with the "forward" tag.

Hm, I can't see these tags anywhere. Maybe I haven't set it up correctly? How do I get to run the "reverse" cards as well? (I haven't really felt the need to do so until now for the reasons given above, but from time to time it might be nice to do so)

[animal world]

All the characters i learned during the pre-Mnemosyne era, i know how to write. This is not true for most of the characters i learned since using Mnemosyne. I'm going to set up some routine for this as well and i'll just do the Reverse order in Mnemosyne for my character category. Haven't quite figured out the routine yet.

I've debated whether in the age of computers it is necessary to know how to write hanzi but i concluded that i would like to know this. If nothing else, for my own satisfaction. Even with a SRS program there are subtle patterns in your learning and i don't believe i truly know a character inside out until i'm also able to readily reproduce it exactly as is.

[querido]

To animal world:

I see your two previous posts.

I'm replying with regard to your regret at not having used the "three sided card" option.

While I mentioned that option above, I'm currently using question-hanzi, answer-pinyin+english, *only*. And, I'm also thinking about making flipped copies to bring back writing practice.

I posted to say I think it might be perfectly OK, even ideal, to have the pinyin on the front. Repeating what someone else posted recently: It helps disambiguate in many cases, leaving you facing, and being tested on, only writing. This is relevant to the original topic "minimum information vs. bigger chunks" (I chose "minimum information"): Testing *only* writing might actually be ideal on those grounds.

[animal world]

Thanks, Querido, for your reply. I manually input a few 3-sided cards but didn't like the effect, so i canned that idea.

As to the "writing practice" i'm still thinking about how to proceed. I'll probably do a straightforward reverse, i.e. English WITH pinyin as the question. Or, i'll do this completely manually without Mnemosyne. Then again, i also still keep hoping that Imron's visualization technique might do the trick some day.

[querido]

I'll probably do a straightforward reverse, i.e. English WITH pinyin as the question.

Yes, that's what I meant. And it is probably what I'll do.

[querido]

I have some more arguments or observations about these two approaches and a tentative conclusion.

#1. The "minimum information" (that is, usually, single word cards) approach, which builds something like a personal dictionary, I think requires a feature not normally offered- a limit on the maximum interval. First I'll argue for a limit, then I'll share an opinion about how short it should be and what this implies.

I hope this is finally clear enough:

The algorithm assumes that you haven't seen the item (word, etc) since you last saw its card.

I'll say that another way: It makes no provision for, because it has no way of knowing about, your seeing a one-month-interval item one day ago.

So, the assumption fails for any items you happen to see elsewhere in your studies before their cards are due.

The justification of a promotion depends upon that assumption; for example, you should promote to two months only after proving that you can remember for one month.

But for many common items, you'll never go a month without seeing them, so promoting them breaks the logic of the algorithm.

Because most of the affected items are easy, e.g. 人, we ignore the problem (or don't notice it) because it doesn't appear to do much harm.

But many other items, not common and not easy, are also seen incidentally shortly before seeing their cards, also breaking the algorithm.

A mass of these probably builds for typical students who review outside of the program. And *THAT*, I think, is one reason for the commonly reported observation that the flashcard deck remains mysteriously difficult long after you think, and it is telling you, that you know the items pretty well. I suspect that for *these* cards simply distributing their intervals randomly below some limit would be better than pushing them all past where they should be.

Limiting the maximum interval is the easiest way I see to mitigate this problem. What to do about any items too annoyingly easy to do every limit*days? Suspend them.

Now the question arises of how long the limit should be. If real reading/listening of the whole piece (from which you took the cards) is begun as soon as possible, and repeated as often as necessary (see below), I suspect the limit could be so short that the effort put into this flashcarding thing might not be worth it. Ten days, taking maybe 20 days to hit that? Mine is presently set at 30 days because I still have 2800 cards, about 1700 of which could be suspended now.

Of course, if your aim really is to memorize a dictionary you'll keep all the cards and have to set the limit higher.

#2 Bigger Chunks

I never necessarily wanted to memorize the bigger but still flashcardable items, such as whole poems or stories, but periodic review to prove that I still understand and can read aloud, etc., everything seems worthwhile. The scheduling of these readings, listenings, etc., can be handled by a flashcard program (the whole item can be on the card, or just its title, assigning an outside lesson). Here, it is not necessarily an application of the memory stretching theory, but simply a scheduling aid. I'm doing this now with the 130+ pieces in my volumes 1-6, representing 1700 items. Reading these aloud with full understanding (at least at the level required by these books) is mysteriously easy, hence this thread and post.

Conclusion, maybe:

Given a choice between not reviewing, to preserve valid conditions for the graduated-interval theory, or reviewing, and rendering the flashcarding to some extent superfluous, the latter seems wiser. When I resume my children's book series at volume 9, I intend to try a "wordlist" method (no flashcards) and scheduled reading/listening. (Although I don't necessarily endorse his system, you could google "Iversen's wordlist method" for more ideas and for *his* reasoning.) In a few months I'll let you know how it worked out.

Please don't tell me I should be studying instead of posting; I'm studying too. :-)

[Guest realmayo]

I'm not too fussed about the fact that I see lots of characters and words in books, real life, etc, between Anki intervals. It's inevitable, and desirable.

If I have a three-month interval for a character, and half-way through that interval I see this character in a textbook or on TV then yes, the next time Anki asks me for it the 'real' interval will only be 1.5 months, despite Anki thinking it's 3 months: if I get it right then Anki will set the next interval at (say) 6 months, instead of 3-4 months. So yes, it's possible that I have an extra 2 months to forget it.

But is this really a problem? The worst-case scenario is that this 6-month interval does indeed prove to be too long, I don't see the character in 'real life' during that period, and the next time Anki asks me for it, I've forgotten it. So what? I fail it. Simple as that. If this forgetting was happening for hundreds of cards, for the reasons described above, then I agree this is a problem. But I don't think it is so severe a problem, and I'd guess that the majority of cards where this problem plays out are ones which I don't remember particularly firmly, and therefore it's no bad thing that I have to fail them.

Also, I'd imagine that for plenty of characters and words, I'll come across them in "real life" fairly regularly. For those cards, Anki becomes less crucial.

And this is the point: for me, Anki is something that makes sure I don't forget stuff. Once cards are a month or two old, then Anki acts as a safety net to make sure I don't forget them.

This presupposes that you are exposed to Chinese outside of Anki, and on a regular basis. But as I say, that's ideal. My 2010 resolution is to never spend less time reading Chinese each day as I spend on Anki.

The best spaced-repetition system for Chinese characters and vocab is reading Chinese texts. The second best is Anki. Because the first one isn't perfect, I rely heavily on Anki (I have about 14,000 cards, average 300+ reps a day) to plug the many holes.

[animal world]

Hmm, it seems i'm not the only one in a chatty mood today:)

I agree with realmayo's great post. The more you see a character/word in real life, the quicker you can zoom through it during your drill. It also gives reassurance that the entry in your SRS is actually being used in the real world. In general, memorization is quick but learning new things is not. Realmayo with 14000 entries in his SRS uses it predominantly as a memorization tool. I have a little over 10000 entries of which 8700+ words/character i know and 1300 entries represent backlog. (Oh, that ubiquitous backlog and not to forget the various files i haven't yet imported to SRS!). So, this year it will still be a year of heavy-duty learning but less than last year and next year it will be easier still. I do read a lot on a wide range of subjects and from different sources. Whenever i come across new words i add them to my Mnemosyne backlog or the other backlog files. I read an article about the Haiti earthquake and then added the Chinese for Haiti and Port-au-Prince to Mnemosyne. I read articles about the arms sales to Taiwan and added lots of words such as cruise missiles. You might say that it's overkill but i just can't help myself and am willing to pay the price for my idiosyncrasies by learning things that are not necessarily a "must."

As to the original topic of this thread, I've changed from Minimalism (just the minimum answer/description) to Baroque (ornate details). This became annoying as Mnemosyne would perpetually expand and contract like an accordion during the drill. Yeah, i know i can use the program at full-screen but i don't like that. So, then i took a lot of this extra data out and much of this is now sitting on a backlog file.

One thing i now have begun to add for entries that are not run-of-the-mill is the source of the information (article in X newspaper, poem, a friend saying this to me, etc). This will validate the information. If the source is a poem i'll note whether the expression is still in use today. Before, i would sometimes stare at entries and wonder where the heck i got it from and whether the entry was trustworthy. Right now, i'm spending a lot of time reading poetry with a friend. Many of the words, meanings and cultural background i learn from these exercises are precious to me, so i add them to Mnemosyne. Yet, i don't want to end up talking like a Tang poem and run the risk of Chinese strangers ever asking me, "Hey, Lady, in what millennium were you actually born?"

Besides memorization, Mnemosyne also serves the important function of being a repository for all things Chinese to me (within reason, of course, and not counted the blasted files i haven't imported yet). I would say that at least 98% of the entries in Mnemosyne i learn actively. When i come across an entry such as 太子港 (Port-au-Prince) i really don't want to see it day after day until i know it inside out. If i see such entries every month or so, i'll probably learn them over time, too. If i ever need it for anything, i'll know where to find it!

Querido, how do you suspend an entry in Mnemosyne? The only thing i can think of is to create a new category, call it something like Suspend and reclassify the easy stuff to it. I'd rather not create more categories than i already have.

[Guest realmayo]

One thing i now have begun to add for entries that are not run-of-the-mill is the source of the information (article in X newspaper, poem, a friend saying this to me, etc). This will validate the information.

Recently I've started doing exactly the same thing! I don't know if it has any benefit but I hope it helps prevent SRS being just an empty, abstract exercise. And I think it reflects real life function: if I am talking to someone and trying to remember a newish word I might well remind myself that this was the word I saw in the newspaper article about frogs .. which then leads to the recall of the word I need.

[querido]

Thank you for your nice replies.

To animal world: I meant "suspend" generally. In mnemosyne I guess I would export to a "suspended" deck. Mnemosyne version 2 will probably have a suspend, as anki does. I haven't suspended anything yet, as that is the harder point to justify, but my argument above for a maximum interval limit is sound I think. (By the way, the pleco guy agrees with this, having already thought about it himself, and an optional configurable max limit might be coming to plecodict. When he mentioned it in passing, I took the opportunity to argue for it as well as I could, there in the pleco forum, since I can't modify pleco as I can mnemosyne/anki.)

To realmayo: I can accept everything good in your post yet still have something else important to say, but it is too long-winded and is of too little interest and I can't prove it. I'll let my previous post stand as the best I can do.

[imron]

If real reading/listening of the whole piece (from which you took the cards) is begun as soon as possible, and repeated as often as necessary (see below), I suspect the limit could be so short that the effort put into this flashcarding thing might not be worth it.

This is kind of the conclusion that I have come to - or rather it is what I have found works for me. I prefer to re-read things (or read new things with similar vocab) than put things in an SRS. I really like the theory behind SRSs, and from time to time I do make stabs at flashcarding, but I just find that in practice, reading or re-reading something also gives a similar effect. It might not be optimally spaced, but it's also more interesting than doing flashcard reps.

[querido]

Thank you imron.

[chrix]

I'm of two minds regarding the issue raised by imron. I didn't use SRS until after joining the forum, so I have studied stuff the traditional way all my life (the 填鴨 method, and of course by reading and looking up things on the fly).

So in that method, reinforcement would just happen by reading other texts (I could never read texts several times, unless needed for work, this would be too boring for me) and coming across words I had read before. For many words I actually remember where I read them, so in a way each word would have its own little history for me.

However, the problem with the on the fly method is that sometimes in a conversation I would remember there being a certain word but just not be able to come up with it. Flashcarding has changed this immensely, not only because it has made the remembering process easier, but also because it creates some kind of record for you to look up if you run into this kind of situation again.

I should probably discuss this on the other thread on SRS strategies, but I do think that principally speaking learning words in context will be more efficient than learning them out of context. But I'm still learning the HSK vocab and the basic chengyu because being exposed to primary vocab out of context will help you later when you do encounter them in context.

As far as characters go, I have decks for these because of the nature of the Chinese writing system - reading texts alone will not necessarily reinforce the readings of characters, and for an error-free pronunciation you'd have to be absolutely clear about the exact tones....

[animal world]

Qerido, my apologies for not having exactly addressed your post as real mayo expressed my opinion better than i could have done myself. I already warned that i was in a chatty mood.

My goal is not to turn Mnemosyne into a dictionary but it's becoming more a personal document. I could also see that in a few years time Mnemosyne will be of much less value to me. But now that i'm still actively learning new characters and words, Mnemosyne is a tremendous aid to prevent chaos in my mind and prevent forgetting things i learned a while back but have not had much occasion to use since.

Once you become more proficient in a language you gradually develop your own vocabulary. This is particularly true if you have the opportunity to spend a lot of time in an environment where that language is used habitually. To compare this to English, words such as "cool, dude, the F word and its derivatives, etc, etc, etc" are not in my vocabulary. I know them but don't use them. Dirty words such as "credit default swaps" are, unfortunately, very much in my vocabulary. I've never acquired a liking for American sports and there are football and baseball terms i don't know or understand. Something like that occasionally comes up early in a conversation with strangers. They hear my ignorance on this subject, note my accent and then immediately assume that i'm an immigrant "fresh off the boat." Quite amusing! In any case, I see myself slowly acquiring my own vocabulary in Chinese that includes the "must know" and everyday words but also terms that reflect my interests.

Recently I've started doing exactly the same thing! I don't know if it has any benefit but I hope it helps prevent SRS being just an empty, abstract exercise. And I think it reflects real life function: if I am talking to someone and trying to remember a newish word I might well remind myself that this was the word I saw in the newspaper article about frogs .. which then leads to the recall of the word I need.

Exactly! There's another benefit. If the interval becomes long, you might have forgotten this word/expression and wonder whether it's legit or still in use. The source notation gives context and is somewhat of a substitute for double-checking everything you learn with a native speaker. Last month i added the idiom 引吭高歌 (it means singing joyfully in a loud voice) to Mnemosyne. My note says that i got it from a Xinhua article about snow swans that overwinter at the coast of Shandong. These swans happen to have a special meaning for me after having seen documentaries about them and i hope to watch them with my own eyes one day. Every time i come across this entry and the note i experience a pleasant sensation as i visualize these swans.

[querido]

People wanting to learn... figuring out how to make that happen... a forum in which to chat about it...

It's all beautiful.

[animal world]

Querido, you appear to visit the M forum regularly. Any word on when version 2 will come out? And whether it might address the following:

- The "The Text you entered was not found" message always comes twice.

- Ability to refine one's search in the deck

- Ability to bold or color certain text.

Thanks!

[querido]

He isn't giving a date. Also, he has said he's quite busy (they're having a baby this month, or something).

I do check for V2 news almost every day.

I am playing with the V2 code, but it isn't ready to use yet.

Because I wanted the more powerful deck browser now, with the multiple tags giving the ability to search on and grab subsets and suspend/reschedule/etc., I am a little sad to say that I switched to anki about six weeks ago. I expect those features in V2 and could switch back because I like his code better!

[animal world]

I'm sorry to hear both about the delay and you switching to Anki. How do you like Anki so far and did you have to start all over again with ranking your cards?

I used Anki only for about 4-6 weeks when i downloaded the 20K HSK sentences and then i got very bored with the sentences and quit Anki altogether. During that time i tried to create a different category; from my test with 6 entries only 3 made it into Anki and i don't believe as a separate category. Don't know why. I didn't feel like learning the mechanics of using Anki as i was happy with Mnemosyne. One feature i liked was that you could learn ahead without it messing up anything.

You just mentioned another feature i'd like: rescheduling. Assuming that means an ability to tell the program that you won't be using it for x days and it will rearrange the scheduling of the cards. In other words, you can go on vacation and won't be greeted by 2 million entries scheduled for your review when you come back. Does Anki have that?

During my one visit to the forum i also read something about a baby. That shouldn't be a reason to delay version 2 though. Some people just don't have their priorities straight!

Once you know Mnemosyne version 2 is out, please post this information here somewhere. Thanks!