I've decided to go character-only on Skritter. Right now I've learned over 3300 characters (including traditional variants), and I'm learning about 400-500 new characters per year, so I need to make my Skritter study as quick and efficient as possible.

For a beginning or intermediate student, writing out entire words seems to make sense, and I certainly have found it helpful up until recently.

But at a certain point, the fact is that, in spaced repetition terms, writing out a multiple-character word is not atomic, i.e., you're being tested on more than one fact. You're being tested on how to write more than one character, and you're being tested on which characters make up the word. This is solvable to a certain degree by manually marking words or characters right or wrong, and skipping the writing of easy characters, but as a whole, as the number of characters and words I learn increases, the number of items in my queue has gotten way too high, and I'd rather be tested on words using Anki flashcards only, imagining the component characters in my mind, without writing them out. Studying both traditional and simplified, as I do, makes this problem much worse.

So I've been banning all multiple-character words that I encounter in my queue. Usually I custom edit at least one of the character definitions at that time to include the word. For instance, I added _金 to the 押 definition when I deleted the word 押金. Also, I might add an Anki card as well, with the pinyin being the front of the card, perhaps with an example sentence.

I'm curious what other Skritter users think about this strategy. If other advanced users want to phase out studying multiple-character words, the Skritter team could perhaps support that learning strategy. For instance, we could have a way to ban the study of a word, while at the same time adding it as an example word to one or more of the characters that make up the deleted word.

Chris, I see the same problem and I fully support your basic idea. However, I also would like to continue with words. If I look back on my character / word study, I went through various phases:
- first I was struggling with a single character and I found it hard, to write a 2 character word. This was the absolute newbie level. At this level, it might be helpful to concentrate only on single characters and add in the radicals, which are used in those learned characters.
- then I went on with simple words, but the focus was still on characters, as the number of characters was still too small to add in a lot of useful words.
- then I came to a phase, where I started mainly to concentrate on words and added in characters found in new words.
Today, I am still learning words, however, from a very specific context (Chinese names and vocabulary, coming from a Chinese history book). I have seen other more advanced users adding vocabulary from legal, chemistry or other study fields.
Therefore, my "wish-list" would be:
- having a character study only mode, where I just can improve my characters' score (which are always behind the word score) and might even over-learn characters,
- adapt the word learning approach: there is no more need for me, to learn and being tested on the pinyin of a word or on the tones. It's a bit a waste of time, if I add e.g. a new name to my list and then I am tested on the pinyin and tone of 王，友，张 or characters like this. What I need to learn here are 2 things: which character is used in this word (therefore, my word writing score is below my character writing score) and the definition of the word.
I could imagine the following steps:
- as a simple first step, introduce a character-only-study mode, just skipping words, when Skritter is looking for the next item to study, but keep everything else, as it is. So as I can select today writing, reading, etc. that I can then additionally select "Character", "Word", "Both".
- Introduce a "Student Role Model", where the user is automatically classified by Skritter into a certain level (Newbie, Elementary, ...) and/or the user can also select his level. Depending on this classification, parameters are automatically set, e.g. for advanced learners, don't test on tones/pinyin on word level, only on character level, set the "add new item frequency" and other parameters. This is just a rough idea about such a role model and definitely would need a lot of more thought, but my feeling is, it would be worthwhile to consider different study needs and strategies at different student levels.

Interesting, I am using the exact opposite idea and I am only skrittering words instead of characters or tones, I'll never remember 圓環遵行方向 unless I skritter it even though I know all the characters. Good luck with your studies.

dfoxworthy, I did the same. But now, I'm at over 3000 characters and over 6000 words, together almost 10.000 items times 4 = 40.000 items which will come for review at some point in time. I'm getting too many reviews and spending too much time on Skritter instead of doing other things, e.g. reading books.
But I still want to use Skritter and I still want to add in new characters and words. So how to do it, I'm having several hundreds of reviews each and every day? If I don't do it every day, the review queue is going quickly up and up into the thousands and becomes no more manageable. Chris is taking here a quite drastic approach by just eliminating all words - I don't want to do it either.
Therefore, I think at such point, I should give a priority to characters, so that I'm better in characters and then learn new words more quickly.
With new words or especially phrases, I am reviewing a lot of simple characters, writing them down, being asked for the pinyin and the tones.... Yes, I could then skip these items, but this is also a strange learning method, as it might lead to wrong learning (I believe I know what character is used, don't write it down and just skip, but I might have been wrong on my assumption; it also doesn't have a smooth learning flow).
Therefore, I'm advocating for different learning strategies at different student levels, so that Skritter is skipping automatically those items, which are no longer necessary for me to review, e.g. the pinyin and the tones of all words, as I should have learned them with the characters already.
Does this makes sense for you?
I would also be interested to hear from others at this level, how they are managing this - thanks.

Hi Roland,

In theory you should see a huge decrease in your reviews over time, maybe if you set your target retention rate lower you'd save all the time you need. The study cited through skritter about spaced partition showed that around 85% to 90% to be optimal for learning as much info as possible. That could lower your daily reviews by half or more and over time by 2/3s assuming you are not able to efficiently keep up with the hundreds you get a day. The only problem is you'll have a lot more mistakes which will at least allow you to find what you need to work on more. Obviously your level is much higher so I don't have your problem yet. Best of luck with your studies.

Roland, lowering your retention rate a bit will help. Also, will help just focusing on prompts are important (I don't do definitions/pinyin). With similar stats I get 120-150 words a day if I haven't been adding new stuff recently, and up to 200 if I have. It's manageable. I also study mainly words.

dfoxworthy (or Zhu Geliang???), Byzanti, thanks for your tips, I might try out to lower my retention rate, it's currently at 95%.
Nevertheless, it would be very cool, if Skritter in the long run could implement adaptive learning strategies. Skritter should already have by now a huge database of user data, which could be analysed by statistical or data-mining methods, e.g. identify different learners' clusters and their learning patterns (add frequency rate, number and time interval for reviews, also based on the characters/words difficulty level, etc.). Then compare a single user to these clusters and set parameters in the algorithms to best suit the individual learner. This could for example mean to set the add new word frequency, the review patterns, emphasis on characters vs. words, definitions vs. writing vs. pinyin, etc. all to individual dynamic values.
This could be a huge benefit for the learner and at the same time a huge benefit for Skritter - who else in the world has such a huge database of learning statistics. That could be an unbeatable USP.
By the way, I'm not complaining, as I am part of my own problem. I have added a lot of hard-to-learn items to my queue and I've "received now the bill". Sometimes, I'm also a bit too over-ambitious.

Chris, I am n0t an advanced learner and only have about 300 characters, but in 1975 I knew about 1500. At that time I used my own flash cards in envelopes for SRS. A character went into one envelope for every day revision if I didn't know it, then 2 days, 3 days, 4 days, 5 days, 1 week, 2 weeks and so on up to 1 month.

The sole purpose of this was to learn characters, because one doesn't revise them when speaking. I attended anough classes to know the words - of course you have to know the words. But it was only the characters that needed special treatment as I didn't meet them constantly in reading.

When I started using Skritter I found it really irritating to enter both parts of xiexie, mama, jiejie etc. and also three- and four-character words. So I began to delete parts. Although I could equally well skim over them when they come up. I was unused to learning words this way, because I used to learn and use words in class and revise them later.

Howeer, since I've entered a class once a week as a temporary measure, and they have reached nearly the end of a book, I am skrittering a lot of stuff that is not being revised in class. So I am a bit in two minds about what to do.

Another problem is that the definitions in Skritter often have to cover several meanings, but I may only have met one or two of those this year. So I don't take in all the definitions because I am not going to lean those from Skritter.

In conclusion, for me it doesn't depend on the stage of my learning whether I revise characters or words. It does have something to do with how much I want Skritter to teach me or to use it as a revision tool for stuff I've learnt elsewhere. I still haven't decided how to use it in the long term, but it is a fantastic tool either way.

I've also recently run into this problem since I now have nearly 9,500 words and have had to lower my retention rate since the number of reviews was/is just getting way out of hand. However, I have to say that I think your solution is a little extreme.

Personally, I tend to do phases of learning words and then phases of just learning characters. I think it's a good way to do it (not to mention that when you learn words you are also learning characters a lot of the time). Also I think that by writing the character with other ones it's easier to fix the character in your memory (especially for ones that are easily confused and/or have similar radicals).

Plus of course it's pretty easy to skip words that come up that contain the same characters, like, if you get the first one right then just skip the 2nd, and it's just like learning the character on its own. I guess I don't think there is a big distinction between words and characters TBH, but I can definitely see why lowering the number of reviews would be appealing!

@ Chris Clark

In order for your method to be extremely easy, all the Kings of Skritter would have to do is place the sample words on the main practice area. After that, you could turn off all of your lists and add a 4000 most common characters list to your account. Since the practice data would still be in Skritter, your account would quickly balance out and it would be set.

youjing and I asked the Skritter Gods to do this six months to a year ago. They refused. I think your method is very effective in helping an advanced student reduce their review burden, though. I have turned off tones and pinyin on my account and thought about doing what you are now doing. I decided against it. I instead just stopped adding words for a year. Now my daily queue is from 70-120 items. It used to be from 250-450 items a day.

To advance my studies further, I made anki decks of all of my vocabulary, but had them in traditional characters as opposed to the simplified ones I study on Skritter. Aside from that, given that I could already read books in Chinese and was no longer living in China, I gave a giant shrug and instead headed off to learn Spanish. Since I get the impression you are still in China or Taiwan, I don't recommend the Spanish part.

It seems that the pressing concern here is study time. I am not at the same advanced level as you all are, but I have made a few small changes to my Skritter habits so that I am more efficient. When I am "in the zone," I average out about 600 reviews per Skritter study hour (more like 80 real minutes, with breaks to surf the net and check e-mail). Basically, I write everything by hand and use the keyboard shortcuts.

Here's what I do:

1) Writing out by hand. When I first started learning Chinese, it was very helpful to have Skritter correct my stroke order and direction. But I don't need that feedback anymore.

Now when I study, I don't use the Wacom tablet or mouse to write. These are so slow. Instead, I write the characters out on paper. This is so much faster, and sloppier :) But it serves it's purpose.

Under the characters, I even draw out the tones:
男人
/ /

2) Keyboard shortcuts. While one hand is writing out the character, I've got my other hand on the keyboard. I use the keyboard shortcuts to quickly go through the prompts.
spacebar - next prompt
s - reveal (will also mark the prompt as "wrong"）
v - changes prompt from "wrong" to "correct", and vice versa
1,2,3,4 - grade each prompt

Usually I will see a prompt, write it out by hand, quickly tap 's' and 'v' (S to reveal, and V to mark the prompt as correct). If it matches what I've written, I hit space bar to get to the next prompt.

I use the 1234 buttons if I need to change the prompt from correct to a different grade.

Like I said, 600 reviews per Skritter hour when I'm in the zone. Writing with a mouse or a Wacom tablet really slow you down. I'm testing with the iPhone app, and while it's not as fast as writing with your hand on paper, it is much quicker than a Wacom tablet or the mouse.

I tried the paper method, and it was faster for me to use the Wacom tablet. I think most of the difference was in not having to take my eyes off the screen or grade myself.

The first thing to try for reducing review load is to lower the retention rate (and wait a few weeks). Banning leeches and unimportant words is also a good idea. Upping your review speed in manners similar to atdlouis's descriptions is good. Skipping over things when you recognize you know them (rather than writing them) is good. Turning off some parts of study entirely that you don't need helps.

Being able to change which parts you want for characters vs. words would help, but that's a bit tricky for us to do. Changing the parts based on which list, though, could get at the same thing, and is on the to-do list (for after the iOS app is done).

Putting sample words on the study page itself would be awesome, but we need a study page redesign before we can fit it in, and that's not top priority yet either.

The reason why I find writing out on paper faster is because I have started to reduce the number of strokes in radicals and characters the way the Chinese will when they write by hand.

For example, the radical 夂 should be 4 strokes. I write it in 2 or 3 - sometimes just 1 stroke. Skritter does recognize some reduced strokes, like 不 when it's written in just 3 strokes. And I still use the stroke animation when I'm not sure about order; for example, I recently had to see the animation for 潇.

But the more I write by hand, the less often I lift my pen up from the paper to make a new stroke. It's a lot sloppier, but it's faster. If I write 夂 or 王 in one or two strokes with the Wacom, Skritter doesn't recognize it. Nor should it; it teaches standard writing, and it's very good at that. When I write neatly, many Chinese compliment me on my handwriting - I learned it all from Skritter.

On the other hand, I suppose I've been doing the write-by-hand for some time, and using the keyboard is like second nature now. If I had put as much effort into mastering the Wacom, I could probably use it a lot faster.

Thanks everyone for your suggestions on reducing study times.

At the same time, @roland's math makes sense. Anki and Skritter are useful tools, but spending too much time on them pushes out richer and more enjoyable ways to learn Chinese, like writing stories and essays, making friends, reading books, etc. Writing out full words no doubt improves word-writing accuracy, but I don't see it as worth the time investment.

I also want to re-emphasize the SRS point - writing out an entire word is not atomic.

@Chris - I suspect the issue here is that you've advanced to a certain point where you don't really need too much practice to internalize words, as you've already internalized the different meanings of the characters. You're right, you should probably be speaking, reading and writing Chinese instead of sitting at the computer. ;)

I think this will only be a problem for students at a pretty advanced level. I'm at a much more basic level; since I find I remember things better when I write them out, and since my main Skritter goal is learning lots of words, I want to write whole words out. I find word study much more efficient than studying individual characters, but who knows, maybe in ten years I'll agree with you. ;)

Another point is that, in my experience, self-grading after imagining the answer is highly unreliable. I often manage to convince myself that I really knew something, but when I have to write it I realize I didn't really know it after all.

James

Just to update everyone, going wordless on Skritter has been awesome. I give this strategy my wholehearted recommendation for advanced learners who have learned 2500 characters or more, especially those that study both traditional and simplified.

Right now I'm averaging less than a half hour per day on Skritter, even though I've added 54 characters this week.

Some notes:
1. I use dashes now when including words in character definitions: 聚-. I include the pinyin and/or definition when I believe it would be helpful.
2. I only delete words if I get them wrong - it takes some time to delete words, and the words I get right are not going to be the words bogging down my study time.