kanji experiment

[jgodon]

Hello everybody,
I'm working at Tokyo University doing some research on kanji, and in particular on methods to look-up kanji in dictionnaries & computers.
I've set up an experiment where people have to find some kanji using different methods. It takes a little time (about 30min) but can be a fun introduction to kanji and look-up methods. At the end you're provided with information about your performance and errors.
If you're interested, just connect to http://www.ish.ci.i.u-tokyo.ac.jp/~julian/TEST/ and follow instructions.
Thanks in advance for those participating!
Best,
Julian Godon

[jgodon]

I forgot to say that you don't need any knowledge of kanji or japanese language to complete the experiment(everything is explained). In fact complete beginners are most welcome!

[cacawate]

I enjoyed it. Thanks for posting that.

[nice gaijin]

I don't normally encourage these hit-and-run help-me-with-my-homework/research threads, but this was a bit more interesting than the average one. I'll paste here what I said about the 4 methods you tested

kansuke sucks, period. It is slow and encourages the wrong way to think about how kanji is written.

the last method, skip(?), also sucks. The sample kanji were especially horrible for that method.

components is slightly better, but the organization of the menu is awful. Still a slow way of finding a character, I would only use this if I couldn't count the total number of strokes. Also, mistakes couldn't be erased when the wrong component was clicked. It has potential but its execution in this test is lacking.

strokecount remains the best way to find a character; separating the result list more clearly into bushu would make things easier to find.

If you want to develop a way to help people look up kanji, do it in such a way that encourages them to think about kanji properly. Good habits lead to proper information retention.There was one strokecount character that took me longer to find because I got a phone call, but it was ultimately the most effective method.

[breez]

I'm very much a beginner in kanji, but I found kansuke to be by far the fastest method. Yet ultimately I liked the component method as it wasn't slow either and it relates to how I remember my kanji.

I did the test last night before going to sleep (2AM) and that might be a reason why strokecount and skip methods sucked for me; I got tired looking for the kanji in the resulting huge list :p

edit: How about a combination of skip method and components method? It is quick to pick the form (left-right, top-bottom etc) of the kanji and could help reduce the list so that even less components are needed to be picked.