Who else goes both ways with SRS?

[jiacheng]

Personally, I study in three directions:

characters -> pinyin & meaning (reading comprehension)

pinyin/pronunciation -> definition & writing (listening comprehension/dictation)

definition -> pinyin (expression)

I think that all three are necessary skills and you can get significant benefit by studying them all. If I were to eliminate any of them, it would be the first, but I am reluctant to do it because it is probably the easiest and least time consuming test.

[Gymnosopher]

Having recently been doing this myself I was wondering what sort of long term results/experiences those of you posting before have found?

For me, I have been heavily leaning on Anki for both production (writing hanzi) and recognition whilst studying Chinese full time and whilst it is tough at times I think I'm really reaping the rewards in a matter of just months. In fact, the hardest point was probably catching up on the backlog when I switched to writing as well and doubled my card count, alongside swallowing large amounts of vocab from the lower course level I skipped but don't want to miss out.

After having made the noob mistake of first not caring for tones and characters before studying more seriously I felt it necessary to be harsh on myself to ingrain these aspects. And although I've heard of people doing 3 sided 汉字-拼音-English cards in 6 directions I've stuck with just 2 sides. Reason being that if I have a character card I need to be able to say the character properly (i.e. get the pinyin right) as well as know what it means - which I suppose it like Chinese reading a character for the word and meaning too. Reverse is much the same, an English idea into Chinese needs me get the character and pinyin. While I have audio play when the answer shows I don't have any listening to writing cards as with individual characters/words out of context I can imagine this being incredibly hard.

In reality this is me writing out the character and putting a tone mark above it (I found that if I wrote full pinyin I remembered that more than the character) and then inserting using my Chinese IME - which drills home the pronunciation (along with the nuances of similar pinyin/sounds) whilst also makes me more harsh on myself if I got it wrong and no benefit of the doubt "oh I did (mean to) say that really". If doing reps on my iPod it becomes writing out the character on my leg etc. which actually lets me see the character surprisingly well (to the point of rubbing out out and restarting if I slip up :z).

The reason I moved to cards both ways was due to the frustration of having characters I wanted to learn and doing it the old fashioned way of writing out hundreds of characters dozens of times only to forget them by the time it came around to wanting to write them. In fact, it was more like completely forgetting the vocabulary I though had been registered and needing to go back to scratch. So in that respect the SRS is helping me practice them in an organised way to help remember.

Also, I hear a lot of people online raving about Skritter. I tried the tutorial and found it to be like a paid version of Anki for the luxury of using a mouse to write characters (unless you're using a tablet), so if so many people find SRSing writing then why not just do it within Anki and by hand? Sure, there are less 'hints' in the real world but it's the world I'm going to be spending a lot of my time in...

Finally, re some comments right at the start, I don't see why it would necessarily be beneficial to separate reading and writing apart other than for reasons of time constraints. I think it's just part of the many versus single deck debate - having a single deck makes it harder but more authentic in that you don't know what to expect and easier to keep on top of when not having words cards left right and centre.

I'm still undecided on sentence cards (something I'm yet to experiment too far with aside from some phrases) though have a feeling doing these both ways will be too much of a hassle and I plan to just use them to prompt which character/word I'm specifically looking for as am beginning to find it difficult to guess what card is due, be it 特别/特/尤其/非常/尤 etc and writing them all out doesn't help know which contexts they would go in. Of course this is only a problem of production as I don't mind understanding them all to be synonymous.

[xuexiansheng]

@Gymnospoher- I'm a recent Anki convert. I have been making it a central part of my study, since it FORCES me to keep on top of chinese everyday. (Don't want to fall behind!) I did the recognition only for when I was study for the HSK4. And had a similar experience to post #2

I now think going to recognition-only was a mistake. It let me push one aspect of learning to an absurd extreme. I should have simply slowed down. This year I'm having to go back and catch up in production anyway, mainly so far via full-sentence production cards, which seem to be working well.

This feeling, of being overwhelmed by the flashcard deck, should probably be taken as a warning to slow down and consolidate.

It allowed me to cram a bunch of characters and meanings into my head for the HSK test, which worked out great since 90% of it is based on recognition only. (i.e. doesn't matter how good your 听力 is if you can't read a) 太吵 b) 灯不亮了 c)关不上们 d)不起作用了。。。等等. Only the last page of 10 questions did you have to produce sentences on your own. And you could still look at the characters you had used in the rest of the test!) Now I'm building my own decks (if you haven't downloaded the Chinese Input Plug-in, GET IT!) based on older textbooks. I'm forcing myself to do recognition and recall at the same time and it is definitely helping with my character production. I just need to find more reasons to be writing characters, you can only do exercises on your own for so long.....! (How would you translate

?) I look forward to seeing if it will help me over 1 year, since it has been a great help the last 4months.

[realmayo]

An update, for what it's worth. A couple of years ago I stopped doing any production at all: I had started studying Chinese full time for the first time and was adding lots of vocab every day, and the time spent on Anki was getting excessive, so I switched to recognition only. Recently I've found that my recognition of characters either on their own or in a new piece of vocab has got a lot worse, so I've restarted production, but only for individual characters (ie pinyin + english + other details if needed => write the character).

[drungood]

I've only been doing this for a short time so I can't speak of long-term results, but so far it is working quite well:

1) I have a Hanzi single-character deck which is 100% production (English word to Chinese character) and I hand write each character.

2) I have a vocab/sentence deck which is also 100% production, but instead of handwriting each character, I type each character in Wubi. If you haven't heard of Wubi, it's a typing method that forces you to recall what each character looks like, unlike Pinyin IMEs. I've only been using Wubi for 2 weeks but I can already type a bit faster than I can write. As I continue using it I expect to get quite a bit faster.

This deck has Pinyin+Audio on the front, and Hanzi+English meaning on the back.

From my experience, if you can produce vocab/sentences, you will also be able to recognize them immediately whenever you see them. So there is no point in doing recognition separate from production. If you don't care about production at all, you can solely do recognition. But don't expect to magically be able to produce by practicing recognition (this only works the other way around).

The apparent weakness in this method is that I am never drilling English --> Chinese. Yet, in my experience, whenever I want to say something in Chinese, I can remember what it is I want to say automatically by drilling Chinese --> English cards.

[imron]

The apparent weakness in this method is that I am never drilling English --> Chinese.

I don't see why this is a weakness. For years I have only drilled Chinese -> Chinese and find it far superior to either English -> Chinese or Chinese -> English.

[drungood]

Yes, Chinese --> Chinese is what I will do once I am capable, but I've only been studying Chinese a few months so it's not possible until I build my vocabulary a bit further.

[WestTexas]

80% of my cards are Chinese characters-〉English meaning or Chinese characters -〉pr. The other 20% are basically English->Pinyin, and consists of (A)words I wanted to say in conversation but couldn't remember the Chinese for (B)words that seemed useful when I put their Ch->English card into the deck C)English->Chinese sentences that force me to use some specific type of grammar.

I don't do any writing when I study Anki. I don't see the point of memorizing how to write every character you learn unless you are planning to go to school here, but that's another argument. For me it is sufficient that I can type a given word. If I know what the English->Pinyin is, and can recognize that word when it is written in Hanzi, then I can type it. I have not had any problems with this so far.

[Don_Horhe]

I also recently switched to recognition only, after going both ways for almost 3 years, as I was spending way too much time. I still write out by hand the words/phrases I'm testing myself on, though, as it's important for me. I think drungood has a point in saying that production includes recognition, so I might give production only a try as well.

[Gymnosopher]

My update a few months on backs up my own opinion that production is important, at least at the early stage - though it came through some painful mistakes of my own.

I too was trying to get to grips with the HSK 4 word list and imported it directly into my deck. This was terrible as some terms I'd never come across and out of context I just wasn't getting their meaning - so struggling through them and hoping they would get suspended! So it became impossible to produce these and with a large number of new cards I began to stop taking new production cards. Then if I didn't have time I started just doing my recognition cards thinking 'at least' I'd get those done as I found production takes about 3 times as long - eventually putting production on hold altogether, telling myself I'd come back and catch up later.

And the result? Well I passed my HSK exam - mostly recognition as said above, but found in my exams (studying Chinese in the mainland atm.) production was verrry rusty. I also agree with the idea that getting the production rounds out your knowledge and helps your remember in general. Having finally got my break I managed to get back on top of things and the production has made a big difference.

So time constraints are still an issue, and I would eventually want to move to just production of individual characters but that seems quite a way off - not sure how you'd do one English word for one character either, I've been thinking it'd be more like '别' giving me 特x 区x x的 to trigger. Also still not got any sentence's for recognition or production but am ever considering it due to grammar issues - just kept feeling the vocab shortage block.

Now to get over my current issue - last 4 weeks of reps were in iPod Anki when travelling and now back desktop Anki wouldn't accept it, corrupting my iPod database and giving me an imaginary backlog of 2500 cards! The ups and downs of technology...

[laurenth]

To Gymnosopher: I once had the same syncing problem with Anki. As the most recent version of my deck was on my (Android) phone, I just copied the Ankidroid file to my desktop and deleted the old desktop version. I hope this helps.

[Gymnosopher]

I've given you a green as it would have been helpful but alas it's too late!

I did the 'check database' in the desktop version and it had no issues and synced online, then I tried to sync Anki mobile where I was going to wait for the 'keep local or remote?' and keep the anki mobile version to copy back. However it didn't ask and just started to download and over-write my deck... I shut down the application hoping that the download would be first and then the deck replaced but it was too late, seems like the overwriting and download happen simultaneously, so me closing the app had corrupted the database and it wouldn't open. I then copied the file to my desktop via 'download to itunes' etc but the anki file was 0kb and opening with anki gave the same errors as the phone - about not being an anki file or being corrupted and 'no table found' or something.

Anyway, no choice but to go back through the backlog and slog away until the daily card count is manageable again. Aside from losing data about card ease and other stats the only real issue (other than wasting time) is that cards I'm getting right now are being shot off into the future - three to six months seems to be the norm. I'm gonna have a lot of trouble in 3 months time it would seem! The algorithm is out of sync with reality - the opposite of people wanting to put Anki on pause for their break... The other option would be to reschedule all due cards as new and deal with the issue now but don't think I could go through with an even bigger task! Guess I just need to study hard now to make my memory of those cards catch up and minimise long term damage :\

[realmayo]

Are you sure there are no backups on your phone? I only ask because I know that the desktop version produces loads of backups automatically.

[Gymnosopher]

Desktop version has loads of backups but those are pretty big files, I guess the phone version doesn't do this because of memory size issues? The iPod is jailbroken so I did run a search for .anki files but those it found were all like 0kb or 5kb I think so pretty empty - infact I think those anki files were made when I did the export of the deck after the database wiped/corrupted, I guess it makes sense that the files in the app version would be stored in a different version?

Running a search online the Anki docs seem to show up info about backups for desktop version and mention that if you sync online this is a backup too - however that file was changed already. So seems online doesn't backup and nor does the phone..

Edit: It does seem you can open the database in the app under library/appsupport/database, which is an SQL file, straight into anki desktop - though this just seems to be the file after the last sync for me.

[Pendragon]

I'm not sure if I'm adding something new, but after lots of experimentation I ended up with these decks:

- character --> type pinyin (using numbers for tones)

- write character (e.g. Q: "右bian1, 左bian1, 身bian1, 旁bian1", A: "write 边")

- CN word --> EN meaning

- EN meaning --> CN word (e.g. Q: "他的[...]真是太好了; memory", A: "记忆力")

I use Tatoeba and Nciku sentences for this

Also I found it's most efficient to add words only if you have a text or audio file (could be chinesepod for example) where this word is used. That way reviewing your usual reading and audio files helps reinforce the vocab your learning through Anki. The alternative is to download an existing Anki list and work through it at random, but I found that's much harder.

[Ludens]

Are there any people here working with SRS/anki decks with a large number (say, more than 10.000) of production (English > Chinese) cards?

I've been going both ways (EN>CH; CH>EN) for a year and a half now, and I'm starting to think this isn't sustainable in the long run. My deck has a few thousand production cards now, and already I'm struggling with keeping apart synonyms. Some strategies to overcome this include:

• providing example sentences
  
• adding comments (on the differences between synonyms)
  
• providing the (near-)synonyms
  
• providing pinyin
  

Of these strategies, I've only used the first two, and while it's still (somewhat) manageable at 4000-5000 production cards, I'm sure it won't be at 10.000-20.000.

Providing the synonyms could help out, but as more and more (near-)synonyms are learned, answering an SRS card would become more like solving a puzzle for words with more than say 5 synonyms. Both adding new facts and answering existing production cards would become more and more time consuming.

To me, the obvious advantages of production cards are practising my handwriting skills and being able to recall (and thus use) words instead of just being able to recognise them. But are these worth the costs: the huge amounts of time spent adding, editing and answering these production cards?

I would love to hear (more) experiences of people who stopped doing production cards and instead went recognition-only. Do you feel that words only learned from Chinese to English are learned less thoroughly? Do you notice any difference in real-world use of the language? How do you keep up your handwriting skills, if at all? And above all else, was it worth it, or do you regret going recognition-only?

[icebear]

I would love to hear (more) experiences of people who stopped doing production cards and instead went recognition-only. Do you feel that words only learned from Chinese to English are learned less thoroughly? Do you notice any difference in real-world use of the language? How do you keep up your handwriting skills, if at all? And above all else, was it worth it, or do you regret going recognition-only?

I went the other way (from single side recognition to all sides on Skritter) and I see a lot of benefit for me at my level (just over 2000 characters, just under 4000 words). That said, I have noticed that reading and watching TV is becoming a lot more important... I think that should only continue. No one learns their native language via flashcards indefinitely - eventually children transition to just learning in context and looking up rare unknown words when needed. If you are getting to a point where your flashcards are that unmanageable you might want to think about switching to that "natural" form of SRS. The same would work for your handwriting, if you make a point of daily journaling (on various topics).

[feihong]

I have both types of cards, but since I can already recognize more than 3000 characters I don't bother creating any new English to Chinese cards. I'm also not afraid to delete cards that I think are not useful any more. My Anki deck is around 5000 and I have a much smaller Pleco deck that's in the hundreds.

[Ludens]

I went the other way (from single side recognition to all sides on Skritter) and I see a lot of benefit for me at my level (just over 2000 characters, just under 4000 words).

Icebear, what are the main benefits you have noticed so far? How are you dealing with synonyms for the production cards?

I have both types of cards, but since I can already recognize more than 3000 characters I don't bother creating any new English to Chinese cards.

Feihong, do you feel that you know the words you do have production cards for better? In writing and conversation, do you experience a negative impact for words you never studied as production cards (i.e., can you mostly use the words you have studied as recognition-only)?

I´m on the verge of deleting my production cards, but am still a bit afraid it would lead to active knowledge becoming passive.

[icebear]

Icebear, what are the main benefits you have noticed so far? How are you dealing with synonyms for the production cards?

I only did recognition for quite awhile and then switched last fall to Skritter, where I have cards going 4 ways (char-pinyin, char-tone, English-hanzi writing, hanzi-English). I haven't noticed an overwhelming burden from the switch, I think because I reviewed a lot of well-known initially since I had zero writing practice, and Skritter seems to be agressive about scheduling cards into the future (which I think is the way to go).

As far as tangible benefits, it seems like I'm much more likely to recall a random word in conversation than before - I find myself surprised sometimes at the words I'm coming up with, as does my tutor, who sometimes is a bit baffled how I would know X obscure word (which is usually business/economics related from a news article I've read). Also, the obvious benefit that I can write a couple thousand characters more or less correctly.

Regarding synonyms, they are tricky sometimes, but I've been lucky that most characters I've added so far have fairly well distinguished English definitions. The batch that do overlap I end up failing enough that I really memorize the difference in their English wording - which is to say, I may not know exactly the best way to use them in Chinese, but I think that burden should fall more on a tutor correcting my speaking rather than me spending infinitely more time on the flashcards. I'm only at 4000 words or 2000 characters, so as that balloons I [hope] to shift away from Skritter and increasingly towards productive activities to act as a natural SRS, as well as shifting a majority of my free time reading (news and novels) to Chinese, which sounds like a great push from what I've read from imron. I plan on ramping that up significantly when I move back to China this summer, some because they'll be easier there, but mostly because it provides a nice clean breakpoint where I can establish new habits.

They (flashcards) are a means to an end, and even when I tend not to be extremely harsh with them they seem to be doing a reasonable job at that. Still, I think rounding that corner towards 'natural SRS' is important and hope I'm ready for it in the not too distant future. It sounds like you may be nearing that point also (5000 items - is this 5000 words or 2500 words?).