How big is your Anki deck and how many leeches?

[Silent]

Exactly the point I try to make. You need something to work with. So I started out with flashcards. Now its time to make a shift toward more practical use. This is a gradual process. At one end I try to read real text, educating but still far too intensive to do in meaningful amounts. At the other end I shift focus to sentences instead of simple vocabulary.

[renzhe]

At 4000 cards, you should be trying to read comics: http://www.chinese-forums.com/index.php?/topic/33062-comic-recommendations-and-index-thread/

Then simple books, then proper literature, then classics. Reading is a skill -- just knowing the characters out of context will not get you far, so get practicing

[Silent]

Well, 4000 cards studied does not mean I know them:) A large amount of those 4000 cards are sentences that I've seen only ones or twice. As explained I try to maintain a backlog on these. But you're absolutely right that I should focus more on reading. I think I mentioned in an earlier post that I ordered some books at studychineseculture.com and made the mistake of taking the cheap shipment option. So I'm waiting for the books. They may turn out a bit easy when they arrive. For now I'm mainly focusing on slowchinese.com and in lesser extent chinesestoriesplatform.com for the reading. Also downloaded a fair amount of video material but to get through that is really tough. Even for the once I perceive as relatively easy I disect only a few minutes of conversation in an hours work.

[wedge]

Wow this thread is a real kick in the a$$. I have 863 cards. Each one has a character on the front and on the back has a list of common words that use that character. I try to keep my daily reviews under 150 a day but I guess I really need to step it up. How many cards do you guys review a day?

[renzhe]

I tried to keep it under 150 a day, even with 10,000 memorised cards.

Anything more than that, and I couldn't concentrate well enough, and reviews would take forever.

[WestTexas]

Well right now with my deck of 5300 cards I do around 200 reviews a day, but I've been trying to cram as much vocab in as possible recently so it's a bit high. I'm starting a new job in September and I'm going to cut down to 15 new cards a day. I'm at 30-40 right now.

I didn't start using Anki until I had been studying Chinese for three years, which means I've only been using it for one year total. Personally I found it critical in making the jump from graded readers and kiddie books to newspapers and websites. I don't have the problem others have described of having characters in my Anki deck but not recognizing them. Sometimes when I'm reading and I see a word which is in my Anki deck I won't feel confident about its meaning, but when I look it up to make sure I'm usually right. Actually, I feel that recognizing the words in a natural context is almost always easier than recognizing them in Anki, which has basically no context. I have no 'backlog' - I do every single card that gets scheduled every single day. Well, I miss maybe two days a month, no one's perfect. But I feel that with certain words or characters it's just difficult to get them stick, and before I started using Anki I was kicking myself every time I had to look up a word or character I had already studied. This still happens, but much more rarely, and never with common words. I'm by no means fluent or have great Chinese, but a year ago I could barely get through kiddie books, and now I'm reading the newspaper every day and understanding a lot of it.

BTW, for those of you who don't know, the mathematics may seem unintuitive, but basically the total number of cards you do each day is proportional to the number of new cards you do each day, not the total number of cards in your deck. If you calculate the average number of cards you do each day over a sufficiently long period (IE, basically infinity) you will find that the formula for the number of cards forms an infinite geometric series which has a constant sum. Although I'm not really good at math, I just hazily remember this from college, so feel free to correct me.

[Silent]

I think the number of reviews depends essentially on three variables. Number of new cards, the button you choose after a review and the number of cards you fail. Size of the deck is indeed pretty much irrelevant as after pushing the easy button a couple of times the card is many moons away.

I try not to look at the number of reviews, but at the amount of time spend. For sentences the time spend per item is a multiple from the time spend for ordinary vocabulary. New/unfamiliar vocabulary takes more time than reviewing mature cards. For vocabulary I used to aim at 40 to 60 minutes. Since I completed Old HSK Basic deck I cut back the number of new cards and hope to bring it back to less than 20 minutes in a few weeks. I've ramped up the sentences a bit to about 40 minutes a day. That makes about an hour a day in 3-4 short sessions. The rest of the time I spend on other material. For this I aim at at least half an hour a day. In reality it's usually less as it's too exhaustive. Only when I use an extensive approach such watching a movie with English subs I spend longer, but then how effective is that?

[Sarevok]

I recently passed the 18000 mark - that's the total number of cards I accumulated since I started using Pleco roughly 4,5 years ago (which was in my third year of studying Chinese, so it doesn't include cards for basic vocab like 爸爸妈妈 etc.). But while that number might look quite impressive, I am fully aware, that it doesn't represent my actual vocabulary size as many of those cards have never been reviewed - right now 3200 cards fall into the "uncategorized" category (I just one-click created them at some point and never got around to properly categorize them, let alone actually reviewing them), I estimate the total number cards-which-never-made-it-into-active-repetition-cycle to be around twice that number...

During that time, I experienced a few crashes and had return to a backup file several months old. At one point (about a year ago) I even deleted all score-files on my own and thus sort of started anew - with the same cards, only picking certain categories for review with a hope to eventually work through them all (but I have slowly come to realize, that the probability of achieving that goal is not very high). At times, I didn't do my daily repetitions for weeks or even months - that led to huge backlogs, which forced me to remove some categories from my active repetitions. All in all, it's a huge mess, even though I spent so much time painstakingly categorizing my cards and all...

I've been thinking about it a lot recently and the the idea of getting rid of everything and starting anew also crossed my mind. Reading imron's posts might have actually put some weight on that side of the scales. As much as I love Pleco and it's flashcards system, I might switch to Anki and do sentence only flashcards in the future - Anki is probably better suited for this task and it has the added benefit of having both desktop and mobile version (and their syncing) available right now (or once I make the switch from WM to Android). Tough call, seeing the amount of time and work put into reviews and developing my current category system...

Anyways, thanks for this great thread!

[imron]

When is the vocabulary enough to shift to practical use? How to bridge the gap between plain vocabulary and practical use?

I agree with Renzhe to try starting out with simpler content like in comics, and/or graded readers (not sure if this is what you ordered from studychineseculture.com).

The other thing to do is just take a paragraph (or sentence) of native content and just pretend it's a giant word list rather than an article (this is not really much different than just choosing random cards from an HSK list or some other deck that you haven't put together yourself).

For example, taking the first sentence from a random article from Sina's tech news. We have:

据多位新浪微博网友反映，百度今日下午从16:00左右开始出现大面积访问故障，具体表现为：百度首页无法访问，或登录百度后无法进行任何搜索，均提示系统出错请稍后访问

Which breaks down into the following word list:

据

多

位

新浪

微博

网友

反映

百度

今日

下午

从

16:00

左右

开始

出现

大

面积

访问

故障，

具体

表现

为：

百度

首页

无法

访问，

或

登录

百度

后

无法

进行

任何

搜索，

均

提示

系统

出错

请

稍后

访问.

So simply go through and add as many of those new words as you feel comfortable learning in a day to your SRS/flashcard program. Repeat again for the next day with an article from a similar source/topic (this way you'll get repeated vocab). At the beginning you may only be able to deal with one or two sentences, but gradually you'll be able to increase to paragraphs and finally whole articles. I found that newspapers are better for this than novels because they often have more direct sentences that try to express or summarise a given point in a few short sentences and often avoid more descriptive terms and phrases that you find in literature.

[feihong]

Just checked my Anki deck: 5864 cards, 0 leeches.

I started this deck roughly two years ago, but after studying Chinese for a number of years, so it doesn't include many common characters and words. I'm adding cards at a more conservative rate now, so I only have about 40 reviews per day.

[realmayo]

I have 13,000, no leeches: if I find I'm getting a card wrong annoyingly often then I'll suspend it and, if I think it's important I'll make an effort to relearn it, if not then I'll leave it for the time being. Have found it very useful overy few months or so to gather up the cards I've suspended and look at them together, because often there will be two that I'm confusing for each other, or several where I'm mixing up characters with the same components, but I'm not aware of these connections at the time I'm actually doing my daily reviews.

Would suggest never being shy to suspend a card that seems particularly bothersome: sometimes there are cards which I always seem to have trouble with when I see them in Anki, but don't find problematic when I see them in a book or mag.

I like Renzhe's "jump start" descriptor: I used Anki the same way, to get to maybe about 1500+ characters and a similar number of words, in this admittedly rather artificial way. After which point I could read much more extensively, and Anki became more completely a not-forgetting tool, into which I shove new vocab and characters once I've come across and learned them.

No one posting their "mature success rate" or whatever it's called? Mine is sadly in the high 80s somewhere....

[joshuawbb]

No one posting their "mature success rate" or whatever it's called? Mine is sadly in the high 80s somewhere....

I'd have thought the high 80s in a deck of 13,000 was nicely ideal, it's natural to forget some of the ones that get pushed into long intervals. Mine is about the same, at 90%.

[Silent]

I already agreed that I've to read more:) But sometimes it can be a bit intimidating. However the easier pieces at slow-chinese.com are quite manegable. I ordered graded readers. A few Chinese Breeze readers, an FLTRP reader and the travel in China series. In all ranging from 300-1000 words. I suspect the 300 and 500 words will be a bit too easy when they arrive. But then, its paper, so I will lack the quick mouseover and I don't know about the grammar they use.

Even if it's an easy read it may still be worthwhile as it will be less strenuous and have more entertainment value.

Maybe you're right and should I stop adding cards from the HSK deck and study only the words I catch in the 'wild'. I do have that plan already a long time, but never really got to it. Actually in part motivated by this discussion I created my first cards yesterday. I added 交通，虎 and 龙. The first one because I thought it's usefull. The other two because I've come across these two already several times.

[dnevets]

I seem to remember reading somewhere that the optimum mature card score for an SRS system is 90%. Anything below that, you're either adding too many cards or missing too many reviews. Anything higher than 90%, and you're working below what you're capable of. I suppose this is based on the idea that SRS is used to cram the most info into your brain in the shortest possible time... but then not everyone uses SRS purely for this purpose.

[creamyhorror]

Wow, you guys should share your decks. I've been adding words slowly to my deck from my big list of new words, and I'm only at 364 cards after intermittent usage. I wish I had a nice source of frequency-based words, because my list includes quite a few less common words that I'm not sure I'll ever see in writing. No leeches, not with such a small deck.

I need cards with words + example phrases/sentences in small text; that's how I grasp the word's usage (via collocation). I've done a bit of the method imron described - entered a news article into a parser/lookup tool to break it up into words and look up definitions, then fed selected words into Anki, using the sentences from the article as example. It's a great approach.

[Silent]

On average I doubt sharing decks is very useful. If you add words as you come across them it will be a 'strange' mix of several subjects that pretty much matches personal interests. That's of little use to other people.

If you're after general vocabulary I think it's much better to take some HSK deck. Or a deck based on other courses or frequency lists. Some people have thought about these and made a choice on what is likely to be the most usefull vocabulary. Maybe not perfect, but better than an imbalanced 'random' deck. Decks that are very focused on a certain subject however may be usefull to share. They may give a kickstart in a certain field without too much useless balast.

E.g. a deck with IT terminology, a music deck, a marketing deck, etc.

[Sarevok]

I wish I had a nice source of frequency-based words

It might be worth to have a look at 《现代汉语常用词表》（草案）published by 商务印书馆. It's a frequency-based list of 56008 words in total (based on written corpus). But it's only a list of words written in hanzi ordered by frequency (there is also the same list ordered alphabetically in the second half of that book, the hanzi are accompanied by pinyin in this one), you would have to add the translations and/or example sentences by yourself...

[imron]

Maybe you're right and should I stop adding cards from the HSK deck and study only the words I catch in the 'wild'.

The thing to remember is, many (most?) of the words you catch in the 'wild' will also be part of the HSK deck (after all the HSK is supposed to be representative of what you would encounter in real life usage). So it's not like you're giving up on the HSK deck, you're just reprioritising the order in which to add them to match the frequency you encounter them in a more natural context.

The main benefit I've found of doing this is that you have context to help you remember what the word means, rather than just being a random word in a list.

[WestTexas]

I think making your own deck is the best way to go. I have cards that I don't think most people would find useful, such as 杠铃 (barbell) and 引体向上 (pullup), because obviously I work out a lot. But I think a normal person could probably get through life in China without ever using these words. Using native material to get words to enter is a great idea IMO. Personally when I am reading the news or using QQ, if I see a word I don't know I underline it and later I put it into Anki. I think it makes it much, much easier to remember these as opposed to learning them from a list. Plus, it would seem that these are words you are more likely to see again if you continue doing the same activities.

[jiacheng]

I have a little over 11,000 words in my Pleco pool that I am actively studying. I study each card in 3 directions (characters -> pinyin/meaning; audio/pinyin -> vague meaning/characters; meaning -> pinyin). My success rate tends to be around 93%, probably a bit high for dnevets 90% rule. I may need to consider using more aggressive scoring settings to reduce my daily workload, which is currently around 250 reps per day. I have a separate deck for studying Simplified/Traditional tables, which is 535 cards in 2 directions. Success rate on those tends to be about 90%. Those decks give me about 5 reps per day.

Another thing that I do is that I also use Anki to schedule reviews for my audio lessons. Anytime I study a dialog from class or ChinesePod lesson, I make a card for it, then go listen to the lesson. If I felt like I understood the dialog well, then I mark it correct. The success rate on these tends to be very high, probably over 95%.

In general, I have a lot of confidence in the SRS algorithm, so I have never reset scores. Starting from scratch seems very counterproductive to me and I can't see any good reason to do it. Even if you have a have a lot of useless words that are becoming leaches, then why not just suspend those words rather than interfere with the algorithm?

I love using the Pleco + button to add in cards to my pool that I encounter in the wild.