Hello everyone and makers of this amazing website.

I wrote this already on facebook page but will put it here so people who dont have facebook can see this too.


My first time on skritter:
‎2 months, 500 characters learned, undoubtedly the best character learning tool on the planet !

I am mostly learning using my android HTC while commuting to university, so I am basically using time that would otherwise be just wasted for learning huge amount of characters.!

And the most satisfying thing is that now I actually now some characters that even my Chinese girlfriend(she probably knows over 5000), who studied in China didn't know! Thats just unbelievable!


Sometimes I get annoyed if the bus arrives to uni too early and I havent finished all my reviews yet ! :D


Thank you guys soooo much ! Before I discovered Skritter I was learning using pen and paper + flashcards, took me 6 months to learn to only READ under 100 characters..

I hope this will exist forever so I can recommend it to all my friends and even let my future children to use it too !


FEATURES
I ABSOLUTELY love the Mnemonics, Graphs of progress, Words containing a character and the Characters containing the character, its just so amazingly useful and saves me hours and maybe days of time.






*FEATURE SUGGESTION*

One feature I would like to suggest is to add to the character detail section (which contains words containing the character etc.)

Add "SIMILAR CHARACTERS".

For example if you look up 人 ren2 it would show you 入 ru4 under similar characters.. If you look up 已 yi3 it will also show you 己 ji3 under similar characters. And so on there are hundreds like this.

I believe this would be very useful for everyone as so many character are different by only on or two strokes but their meanings can be completely different.

Let me know what do you think about this.



Thank you for reading this humongously long post and happy skrittering !



PS: Some mnemonics are so funny and creative, makes me laugh out loud love this website even more, thank you

Like for example, just from the top of my head, the character xin1 meaning new.
Is made of an axe and dear parts. and one mnemonics is
"Dear and and an axe,thats NEW" ! :D lol I will never forget this character thanks to this mnemonics

Good, isn't it!

Hey jayobird, glad to hear that you're rocking the 汉字 with Skritter. You've already learned more characters than I did in my first year of intensive Chinese study in college. And on your Android, too! Most of the things I hear about the Android version are that it's broken, so it makes me happy when someone mentions it without also bringing up some bugs. (I guess I appreciate the bug reports, too.)

The similar characters idea is interesting. We'd probably have to crowdsource it. I wonder if there's interest from users in manually establishing links between similar characters that are often confused?

Similar characters idea = I LIKE!

poll! :)

Nick, I love studying on the android version more than on my laptop now. It seems like all of the major bugs in Android have been squashed and it just needs lots of little aesthetic tweaks. So I'm glad you're going full bore with the iPod/iPhone/iPad version.

I also think the Android version is pretty good. It kept me sane while I sat through eight or so hours of an excruciatingly boring BEST Robotics tournament yesterday, in which only two teams scored the whole damn day.

The only things I'd like to see are whether or not a word is starred or not, and the "time since last studied," or "New Word" monikers. These are important for my studying regime.

I was also thinking of starting a Skritter-holics Anonymous chapter in Little Rock. Anyone want to join?

Nick, I promote very much the idea of the "similar looking" character study. Actually, I was thinking about this since quite some time. To enter it, is not the big deal, there are numerous books around, who have categorized characters in this way (I'm not too familiar with Heisig, but I thing, he also used such an approach to some degree). I've seen some time ago a research paper, they proposed a mathematical model for calculating a similarity taking into account also pronunciation and meaning.
The tricky part from my point of view would be, how to build a useful learning schedule into today's existing structure to learn (or to drill) similar characters and how to measure whether or not I can distinguish these characters in the different learning modes. So far, I didn't have a good idea about that, so therefore, I never wrote about it.
The more characters I learn, the more often I have this problem of distinguishing between characters; but it only applies to characters.
I definitely would very much welcome a kind of character drill mode in order to get better on characters.

I'm a +1 on the similar characters idea if it could be leveraged into a mini study mode (Skritter Minigame?) which allowed me to study similar characters sone after the other, possibly with an unnecessary amount of repetitions to force my thick skull to accept the differences between 及 and 级 and 极, or 尤 and 优 and 忧 and 犹...

I like the similar characters idea too, though at the moment I just write them on paper and contrast them.

@fatdragon, You're second set of characters are the first that came into my mind when thinking about similar characters that I commonly get incorrect at first glance.

How to identify similar characters? I think, there might be 2 methods:
1. Laurence Matthews approach: Chinese Character Fast Finder (Tuttle), this could cover fatdragon's examples.
2. Use of a handwriting recognition engine to find characters, which consist of different elements, but for a human being look similar.
I could imagine, the best result could be obtained from a combination of both methods. e.g. take Matthews as a basis and then let also a hwr engine run over the 3000 most used characters and see, what the hwr engine proposes. Then one could build a database, which defines for each character the most similar 10 or 20 characters (or build a 3000 x 3000 table and put an index in there, how close a given character is to the other character).
How to learn? Here I could imagine the following: 1) Once I see a character, where I know, that I sometimes mix it up with other characters or 2) I'm going to a special character similarity learning mode, then:
Skritter presents me a set of max. 10 (or whatever number) of characters with a high similarity index and let me decide, how good I can distinguish character recognition, pinyin and definition. Based on this result, Skritter could build up user specific "learning-sets", which present all their set-members on one page together for learning, where we have to write them down one-by-one or the pinyin or the definition. For example: we get a set of 10 characters displayed and have to input the pinyin and define, whether we know the definitions. The next round, we get the 10 definitions displayed together and have to write down one-by-one the character. This will be repeated, as long as the set is not properly memorized (with the normal SRS algorithm, however, applied to the whole set).
Characters chosen should only come from the set of characters, the user is learning. The question, what happens, if this set is expanding during the learning period must be also somehow addressed.
I think, it would be very important, to see all characters of a given set at the same time, so that we really can learn to distinguish them. If they come one after another, we have more or less, what we have today (might be somehow better, but when I encounter problems like this, I write all of the characters together with their pinyin and definition on a sheet of paper and soon after a few learning rounds, I really know how to distinguish them). The second problem I have is, that I know that I'm mixing up a character, but I have forgotten, with which one. If Skritter could show me such kind of similarity sets, I think, it would be a tremendous help to find out, where my problem is.
What do other Skritterers think, does it make sense? What do Nick/Scott think, implementable with affordable effort?

We could also have a way for Skritter users to add to the list of similar characters, like we can add mnemonics.

James

Hmm, starting with an established list like the Tuttle book would be a good base, but I agree with Nick and James that crowdsourcing it would probably be the most complete way to go, though.

A little bit of pondering aloud here: it might be helpful in the long run to allow individual users to tweak a list like this to their own needs; there may be a dozen users who have a hard time with 龙 versus 尤, but I don't need that. The other issue is that really, studying the characters against one another isn't really that important to me - what I would benefit from would be studying words including those characters (particularly the two sets I mentioned above) and pinpointing which character goes with which word. That's something I can study manually, of course, but if I were designing a similar character study method for Skritter, that would be a priority for me.

I am very happy to see that there is very positive response for the new feature I suggested. Thank you guys also for the ideas hot to pull it off. I guess, now its up to the creators to consider it.

Good ideas here. There isn't time now, but down the road, I'd love to build something like this.

Roland, do you have a reference for that paper you mentioned, about the model for calculating character distance based on stroke, pronunciation, and meaning similarity?

I just want to +1 on your statement, "I love Scritter!". :) I too started two months ago. As of today, I finally passed the "500" kanji mark. :D With a lot of effort using other study materials, I feel I might be able to pass JLPT 1kyu next year. My progress is frighteningly fast!!! :D :D :D Scritter = <3

Nick, I don't have it, I googled it today quickly, but couldn't find it - I'll do this again over the weekend or next week.
James, Fatdragon, sorry I think, I was not to clear and mixed a little bit user behavior and implementation. My basic idea is:
1. There is an algorithm, which determines a similarity index of a given character against all other characters.
2. Skritter chooses then a set - for a given character - consisting of the 10 characters with the highest similarity index and presents it to the user.
3. The user either chooses or evaluates through testing his own, user specific set, which will then be taken for him/her for learning.
4. The user can select additional characters and add them to his/her set.
5. Skritter will evaluate all user results and also takes into account the characters, which users have added to their set, to re-calculate the similarity index.
As such, the "Skritter Standard Set" for a given character is not static, but dynamic and algorithm resp. the similarity index is adaptive, based on user experience.
Fatdragon, I see your point with the words, but I think, these should be 2 different scenerarios. I've never thought about this.

@nick - whoops, didn't see your crowdsourcing comment earlier. Great minds think alike.

By the way, to add to the "I love Skritter" theme, today on the subway I was reading over my neighbor's shoulder and I was pleasantly surprised that I was able to figure out what "大麻合法化" meant. ;)

James