>Is the video representative of texting speed, or is this the fastest human in the world?
From what I understand, he's on the faster end of the normal range.
>Are they making choices that would take effort in a different system, or just clicking the next suggested letter?
With Japanese kanji or Chinese hànzì, there's no practical way to directly input such a large range of characters. Users type a phonetic spelling, then the input method editor presents them with a menu of characters with a corresponding pronunciation. Chinese mostly uses a system of phonetic transliteration based on the Latin alphabet (pinyin), whereas Japanese speakers use both a Latin-based system (romaji) and a native Japanese system of syllabic characters (kana). The flick method shown in the video uses directional gestures to input kana.
For example, if I'm trying to type the Chinese word for bread (??), I'll input the word as it is pronounced, "mianbao". On mobile devices, a list of predicted characters will appear above my keyboard; on a computer, a numbered list will appear beside my cursor. I select the characters I was intending to input by tapping on mobile, or by pressing the corresponding number key or clicking on a computer. The choice of characters invariably requires some amount of human input, because there are many homophones (different words with the same pronunciation).
This method of text input can often be quite slow and cumbersome, so good prediction and correction algorithms are crucial. The input method is constantly guessing which characters you want; if it's not aware of context, it'll make bad guesses and require a lot more manual selection and correction. Good input method software can predict entire phrases and is very resilient to typos.
For Chinese hanzi, some people also use a stroke input method. You are presented with ~10 possible strokes that make up all hanzi, and by selecting them in the correct order you can write a character.
I don't know Chinese but it seems like more difficult method because you have to remember how the character is written instead of just typing the spelling and choosing from a list.
You need to know how the character is written regardless because the stroke order is part of the character. There are some basic rules like working left to right and top to bottom. Also, each component has the same order when written out.
I also seem to recall there's been successful input system like this for Japanese, in particular for handhelds with pen input (think psion and similar early devices).
Optical recognition of Kanji can be though, but with stroke direction it is easier.
correct, that's 'WuBi' stroke input method. for a trained typist, stoke input is much faster than latin-letter based input method such as 'pinyin' since chinese character is structure-based.
From what I understand, he's on the faster end of the normal range.
>Are they making choices that would take effort in a different system, or just clicking the next suggested letter?
With Japanese kanji or Chinese hànzì, there's no practical way to directly input such a large range of characters. Users type a phonetic spelling, then the input method editor presents them with a menu of characters with a corresponding pronunciation. Chinese mostly uses a system of phonetic transliteration based on the Latin alphabet (pinyin), whereas Japanese speakers use both a Latin-based system (romaji) and a native Japanese system of syllabic characters (kana). The flick method shown in the video uses directional gestures to input kana.
For example, if I'm trying to type the Chinese word for bread (??), I'll input the word as it is pronounced, "mianbao". On mobile devices, a list of predicted characters will appear above my keyboard; on a computer, a numbered list will appear beside my cursor. I select the characters I was intending to input by tapping on mobile, or by pressing the corresponding number key or clicking on a computer. The choice of characters invariably requires some amount of human input, because there are many homophones (different words with the same pronunciation).
This method of text input can often be quite slow and cumbersome, so good prediction and correction algorithms are crucial. The input method is constantly guessing which characters you want; if it's not aware of context, it'll make bad guesses and require a lot more manual selection and correction. Good input method software can predict entire phrases and is very resilient to typos.
reply