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

[querido]

This post is about flashcard-making theory.

It mentions in passing that the "whole sentence method" is just one stopping place on the way to a more general principle.

I hope no one minds that I also posted it elsewhere; I welcome any insight I might be missing. I'm changing my flashcard-making policy, today.

I've been flashcarding everything, characters, words, and sentences, welcoming any redundancy. Then for the sake of consistency I pushed this policy through six volumes, "two years", of my children's textbook series, 3600 cards. While it didn't hurt and I've made great progress, I now *must* decrease the number of cards I'm making (600 per month).

Unless I'm overlooking something, I have two choices:

1) smaller: finally accept the "minimum information" principle, learning all bigger chunks outside of mnemosyne. Characters only, my latest text would yield about 250 cards. Characters and words, it would yield about 400 cards.

2) larger: generalize the "whole sentence method" to the "biggest chunks you're comfortable with method". Volume 7 lesson 1 has six poems of 3 or 4 lines each, and a 1/2 page story (the series is transitioning now from all-poetry to mostly stories). I can now (before studying it) *read* this lesson as real poetry and prose, with the exception of the 19 new characters and the words they form. I could make (in the extreme) two cards (probably utilizing an external player): "read (aloud) 0701", and "listen to 0701 (verify that you could dictate it)". This policy would yield 16 major-assignment cards for the whole volume. Broken a little smaller, into poems and stories, there would be about 100 cards, each carrying the full context provided for their new characters. (What is lost might be some of the assurance that the character would be recognized with no context, or in a different one.)

At the heart of this question is this: a sentence, poem or story has "something" that is more than its bits, and the above decision determines whether that something is learned inside or outside of the flashcard program. As a practical decision, I will probably choose that last option (individual poems and stories), but I haven't really condensed this theory, that your flashcards should gradually test bigger chunks as you learn a language, bringing that "something", the working language, into the flashcard routine.

[anonymoose]

It depends on what your goals are. If you wish to memorise entire poems, as many Chinese schoolchildren have to, then putting large chunks, or even entire poems into the flashcard system has its reason.

However, if you want to increase your ability with the language itself, I think using too large chunks is unnecessary and can even have an adverse effect (waste time).

As you know, there are many ways to say the same thing. The longer a sentence you have, the more possibilities there are for changing words, word order, phrases, etc. in that sentence without significantly altering the meaning. So it becomes an unnecessary burden to memorise long sentences word for word, when there are many alternative and equally valid ways of saying the same thing.

On the other hand, putting individual words into the flashcard system will often neglect the environment in which the word appears, meaning that even if you understand the word, you may still have difficulties using it in practise.

Personally, I find one of the best methods is to test short phrases. The phrases are selected so that frequently collocated words appear together. This has the advantage that you don't have to pointlessly memorise long strings of words that have no necessary relation to one another, yet you will know which words to use in which situation.

The way I do it with Anki is put an entire sentence in Chinese on the obverse side of the flashcard, with the part I want to test in brackets in English, so for example it might look something like this: 当心，有人(under the guise of presenting a gift)忽悠人. Then on the reverse side I would just have the omited part - that is 打着馈赠的幌子. This has the advantage that you will learn 打着...的幌子 as a single unit ready for use next time you need it, and you will also learn 馈赠. Although these two parts don't have a necessary relation with each other, there is still only one way of putting them together, so you won't have to waste extra effort remembering just one arbitrary sequence amongst several posibilities. Alternatively, you could just test the two separately on different flashcards.

[querido]

Thank you.

No, I don't want to literally memorize the poems and stories.

Yes, like you, I want to draw in more supportive context.

But my posts were about a more general policy I see possible.

Unable to settle on a new *system* by just thinking and posting about it, I intend to simply try something and see how it works over time in practice, beginning today. End.

[Ed Log]

do you use spaced repetition method?

[Guest realmayo]

One advantage of what you call "minimum information", ie -- if I understand right -- testing yourself on characters and words in isolation, is that this allows you increase your basic grasp of vocab quickly, with the minimum of time wasted inputting material, & quizzing, etc.

This leaves plenty of time to read texts in Chinese, where this new vocab appears in natural, various, authentic contexts.

I owe so much to using flashcards, I use an SRS programme everyday, but I try to avoid ever spending longer on the flashcards than I do actually reading Chinese texts (eg reading comprehension books, etc).

[animal world]

I use SRS and just the definitions of a word or character. Sometimes, but rarely, i also jot down on the definition side a brief note about something relevant (i.e. the source of the word if it's from a poem or something not run-of-the mill). I use SRS only for the drill, i.e. speed. When i encounter the word again while reading some other text where the definition as learned doesn't fit the situation i do more research. If appropriate, i expand the listed definition of the word in my SRS database.

[querido]

(To Ed Log: Yes, mnemosyne (modified).)

My first try was one extreme: single-characters only. Rejecting that for reasons I gave elsewhere in this forum, one year ago I started a children's textbook series and my flashcarding policy was to *archive every nuance* in these books (including the subtleties of every audio snippet). This turned out to be impractical. But getting a grip on some other (also idealized) system proved to be excessively fascinating.

Trying to keep this post brief...

I said, above:

At the heart of this question is this: a sentence, poem or story has "something" that is more than its bits, and the above decision (about the size of the flashcarded chunks) determines whether that something is learned inside or outside of the flashcard program.

...and I concluded after much agonizing that the "something" is way too big to try to hold in the flashcard program; it's the wrong tool for the job. The right tool is... the brain. It isn't mechanical... some magic has to happen in a realm I can't quite grasp... and it has to be trusted to do this automatically!

Do your flashcards, read and listen, and believe in miracles.

realmayo said:

I owe so much to using flashcards, I use an SRS programme everyday, but I try to avoid ever spending longer on the flashcards than I do actually reading Chinese texts (eg reading comprehension books, etc).

Right.

Conclusion: I'm now doing words only, of one or more characters, as parsed by Wenlin, and my interpretation of the flashcard system as a tool is that it is *my dictionary*. All reading and listening comprehension work, and all verification that I can *write* and *say* what I read and hear is done elsewhere. Also (though this is a different topic), I'm doing front hanzi back pinyin+English only: 2100 words = 2100 cards, presently, and no audio snippets. One compromise: I include "bound forms" (which don't form a word by themselves), noting that in the definition.

My time spent flashcarding has collapsed, and this feels like a system I can move forward with. What a relief.

[xianhua]

I'm doing front hanzi back pinyin+English only

So here we have (Method One):

Front - Hanzi

Back - Pinyin and English

My current method (Method Two) involves:

Front - Hanzi and Pinyin (plus tone in brackets)

Back - English

The first method is ideal with recognising Hanzi, but then when the cards are reversed, the Pinyin is there too, so the task of translating from English to Chinese, in terms of pronounciation anyway, is done for you.

With method two, when the card is reversed, your English to Chinese pronounciation and character recall is tested, but then seeing the Pinyin alongside the Hanzi doesn't test your ability to pronounce the Pinyin for the characters.

In an ideal world, I suppose the Pinyin wouldn't be there at all, but I still need this input. Have I missed some features in Anki, or is this just something to live with? Either that, or I could just be misusing it completely.

[querido]

To xianhua:

First, let me tell you I'm not reversing; I'm using front hanzi back pinyin+English *only*. (This is just a decision I made for myself that I don't necessarily recommend.)

Yes, you're missing a feature.

I think this is what you're asking for:

front hanzi

back pinyin+English

and then flipped:

front English

back hanzi+pinyin

In mnemosyne, this is called a "three sided card". Anki can do this too; check the documentation or ask on their forum. This link explains everything, I think: http://ichi2.net/anki/wiki/ModelProperties

Edited October 25, 2009 at 10:38 PM by querido

[xianhua]

Thanks querido.

I figured out how to do it in Anki (with a little help from their forum).

Now I have:

Forward:

Question: Hanzi

Answer: English and Pinyin plus tone in brackets

Reverse:

Question: English

Answer: Hanzi and Pinyin plus tone in brackets

[Guest realmayo]

Or consider:

1 Hanzi in nice biggish comfortable font

2 English (and pinyin if necessary ... or half the pinyin)

3 Hanzi + pinyin + English + other stuff if needed to help you remember or understand

Forward

Q: 1

A: 1 + 3

Reverse

Q: 2

A: 1 + 3

[animal world]

and then flipped:

front English

back hanzi+pinyin

In mnemosyne, this is called a "three sided card"

I wish i had seen this remark as i started off with Mnemosyne, NOT when i'm 8500 entries into it!!! Although it might not have helped me then either.

Just played a bit with the Mnemosyne database edit card function by right-clicking. It seems you can add sound or image but there's no capability to turn an existing card into three-sided-ness. Don't believe you have that capability either when you're importing files but only as you add individual cards manually. I do the latter only rarely.

So, now i'm considering a gutsy move: to eliminate the pinyin altogether for cards that i know already very, very well. This, presumably, would entail lots of manual edits and i'm just not up to it. Maybe let sleeping dogs lie.

[chrix]

I haven't used mnemosyne before, but it's really easy to set up with anki. Why not try and import your cards into anki and set it up the way you like.

BTW, question for anki users, do you know of a way where you can flip back and forth while reviewing your cards? So one would be the forward type, the next the reverse.

For instance, you'd get

1. QUESTION: Hanzi

ANSWER: Pinyin, English

and in the next turn it would be

2. QUESTION: Pinyin

ANSWER: Hanzi, English

I had one deck that worked like this, but they kinda cheated and doubled the number of facts just with the Hanzi and English fields reversed.

[renzhe]

EDIT: I misunderstood your question. Nevermind

When I tried Anki, it insisted on making me write characters out before testing whether I knew what it meant, with the explanation that this is the way you must learn, because it's better, and the only way around it was a hack of some sort. I didn't like that, I don't think it works for me, and I don't think that it's the fastest way to learn to read Chinese.

In fact, I don't bother much with active (production) cards at all.

[chrix]

what do you mean, it forced you to write out characters?

[renzhe]

Well, not write, but to produce them in one way or the other.

Given the meaning, produce the character, see if it was correct. You're still grading yourself, like with everything else.

[chrix]

you mean, like enter the character? They no longer have it set up like that, I think. "Don't ask me to type in the answer" is the standard setting now.

[renzhe]

There was nothing to type.

Anki asked me what "good" means, and I had to answer "好". (in my head).

What I wanted to do is for Anki to ask me "好" and the answer should be "good". (in my head).

I believe that the second approach is faster in terms of getting you reading real materials (and expanding your vocabulary). But a few years ago, the software insisted on doing it the other way around, with long arguments why it must be done that way. I remember that it took me half a day to figure out how to turn it off when I was setting it up for my gf.

Nowadays, I have more than 10,000 recognition cards, and only 2000 production cards (for the 2000 most common characters).

[chrix]

I see, so that's what they call "reverse" cards now. 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, or eventually split up your deck.

I mostly do recognition anyway, the other way round has too many problems...

[renzhe]

Well, one thing to keep in mind is that "production" and "recognition" are two different processes, which take different amounts of time.

So you must have two cards for each fact, because your scheduling will be different for them. You might be very good at recognising 察, but very bad at writing it. Flipping them back and forth assumes that it takes the same effort to go both ways, but it doesn't.

Actually, learning production before recognition is not a bad way to approach learning of facts in general, but I'm absolutely unconvinced that this is the right way to approach 1) language learning in general, and 2) Chinese characters in particular. At the very least, you need many things carefully organised for this to make sense (the order in which you learn characters, etc.) I personally find it easier to get through the most common vocabulary, get loads of exposure over a few years that cements most of the important stuff in my mind, and then see what needs to be polished.