Handriting Recognizer Hacks

[GaoLong]

Not sure if the character will come through, but this radical: 尸 (Shi1, Mathews 5756), was difficult to get in until I left out the small line at the top right. You just draw two horizontal lines, then the curving line on the left, and it works fine.

It would be nice to assemble a list of hacks for the current recognizer on this thread while Mike is writing a custom recognizer for the app.

 

[Henry]

I'm pretty sure that the little line on the right is just an extension of the first horizontal line, so there is still a total of three strokes.

Do you have list of the different types of strokes? If so, this would make sense-

1st. heng2zhe2

2nd. heng2

3rd. pie2

sorry I can't seem to write characters, dunno why . . .

 

[GaoLong]

Recognition in general

I'm finding that although some characters come through fine when they are written properly, some are very difficult to get recognized. So I end up writing them 'wrong' in order to get them recognized. This means that the current handwriting recognizer is amusing, but not useful. Hopefully the custom recognizer Mike has mentioned will be better.

I find that only the Pinyin lookup is effective on Plecodict. Radicals are slow as hell on paper or computer and 4-corner lookup isn't coming till Plecodict 2.0. So for the moment I use programs like JWPce and books like Herring's Foursquare Dictionary to look up characters by four-corner, then enter the Pinyin for the character to find them in Plecodict.

I look forward to 2.0, since using 4-corner I can look up 90% of characters in less than 5 seconds per character using a computer program and under 20 seconds from a paper dictionary. Radicals and handwriting recognition are physically incapable of reaching that speed (except recognition of very simple characters of one or two strokes), so Plecodict only functions like the ABC paper edition for the moment (not that I'm complaining, I love the ABC dictionary!).

 

[Jianchuanbrit]

I beg to disagree. I find the handwriting recogniser is very good and extremely useful. It is SO much faster than trying to guess the pinyin, or using the radical tables (which, incidentally, are a big improvement over the old Oxford version). It is also quicker than writing the word by Graffiti when you do know the pinyin.
So you do have one satisfied customer at least!

 

[GaoLong]

Clarification

Please understand, I'm not saying the recognizer is a bad thing. it's a lot of fun and useful as an aid to learning characters (with the temporary inconvenience of not recognizing 'correct' writing all the time). But it cannot be faster than 4-corner lookup for any character beyond five strokes for the simple reason that I can enter the full 5-digit code for a character in the time it takes to write it, or less.

I, too am a satisfied customer, it's just that I have dreamed of having a really good dictionary directly linked to a four-corner lookup system ever since I discovered 4-corner.

I would love to see a contest to determine which lookup system is fastest, since the 4-corner would completely dominate. The only system that comes close is the Rozenburg system from Chinese-Russian dictionaries.

Regards,
Chris Gait

 

[Jianchuanbrit]

How about a contest to see which method is quicker for beginners learning Chinese? My point is, the four corner method may be fast but you have to invet time into learning it and practising it. Frankly, I don't think many people use it compared with using a radical lookup system in a printed dictionary. Compared with that, writing the character is far quicker. Since everyone needs to learn the stroke order anyway if they are learning Chinese, it is simple to learn to write the character on the screen to look it up. Entering the four corner way is only quick if you have learnt it. I have not met any foreign learners of Chiense who use that method (and I have lived in China for more than ten years).

 

[GaoLong]

Four corner vs. other methods

Well, I'm a beginning student who uses 4-corner almost exclusively, but I'm not in China. I'm also self-taught. Perhaps if I learned things 'the right way' I would be better with radical lookups, but I find the whole thing incredibly slow and tedious and full of exceptions. Four-corner does definitely have a learning curve, but then again so does Chinese, and so does the radical method. And 4-corner has exceptions as well, just less of them (when in doubt, it's probably a danged '3', by the way). Four corner also works better in computer programs than paper dictionaries (since you can look something up even if you can't figure out how some of the corners code).
Ultimately, the method is losing out because the majority of native speakers don't use it. There seems to have been an attempt to popularize it in the 1980s (and I have a miniature Xin Hua simplified dictionary from the period that I treasure), but it never took off.
I just found out today that there are two versions of the Kang Xi available with 4-corner indexes, so I'm working on finding those. The Large Chinese-Russian Dictionary (4 vol.) has an index, and that's where I started with it.
I suppose it's one of those things like choice of computer operating system: more religious than logical.

As to a contest: Sign me up!

Regards,
Chris Gait
Nokesville, VA

 

[Henry]

regardless of which is fastest, I prefer to use some sort of handwriting recognition for entering characters because it forces me to practice actually writing the character by hand.

Right now I do more writing on my palm (mostly within Ox. dict) then with pen and paper, but expect that to change and want to be prepared.

I have been using the penpower program for HWR and have not found any problems like the one Chris first mentioned.

It seems that as a teaching tool it would also be very helpful to use radical lookup for entering characters. I find often I can only remember a character by bits, and having the radical first pop into mind could be quite helpful.

I am not familiar at all with 4-corner, Chris do you feel that looking up the characters this way helps you remember them? (or has some other benefit besides speed?)

Henry

 

[GaoLong]

Handwriting and 4-corner

Henry,

Four-corner is a speed system, and that is benefit enough for me, since it allows me to lookup Chinese characters with close to the same speed that one can look up words in a Western language dictionary. I am having better luck with the character recognition now since I have switched (for a time) to reading simplified characters. I have an Yi Jing available in both simplified and traditional characters with much of it available in English, so it's a good exercise for learning both character sets at once.
I'm using a fallback technique. When the HWR fails I shift to 4-corner. If I can't find it there I try to figure out the radical and stroke count. In two days of reading I've only had to use the radical lookup once.
I do find that the characters are easier to memorize if you write them out. But I'm sure my handwriting will be as terrible in Chinese as it is in English and Russian. My primary instructor looked down at my writing one day fairly early in my Russian language course in Monterey and said: "Mister Gait, my father wrote like you. No one could read what he wrote. Not even him. Print!" Unfortunately there's no such thing as printing in Chinese, so I'll have to try simplified characters <g>.
Regards,
Chris Gait