Not many Westerner seems to appreciate that Chinese Hanzi (and Japanese Kanji) are really words made up of simpler components arranged into squares, instead of written sequentially and separated by spaces. And the shape of the components in the square provides additional encoding information not available in sequentially written languages.
By the same reasoning, using 8-bit ASCII to examine the informational properties of English text is also absurd. English can be comfortably expressed with only 6 bits per character, after all.
Anyway, I did include a GB-encoded version in my calculations. And once you compress the text, the charset-related difference becomes much smaller anyway.
English can be comfortably expressed in 5 bits per character, using 26 letters and the five symbols [. ',"] . Capitalization and numeric digits are nice-to-haves. ;)
I can't provide much in the way of authentic information as to your last question, but anecdotally I find it easier to finger-type Chinese into my phone than English. A couple factors are at work:
- I don't adjust my style for the phone, and it's next to worthless at predicting what I want to say (in English). This leads to a lot of thumb-tiring selection of single letters.
- I don't speak fluent Chinese, so most of what I say is pretty basic stuff, which the predictive input has an easy time with. This leads into the third point,
- Using a pinyin input method, you can call up entire phrases by just putting in the first letter of each syllable, e.g. xx for 谢谢 or bhys for 不好意思 or ng for 那个. This is basically the equivalent of your computer automatically expanding IANAL into "I'm not a lawyer" whenever you type it in, except, for everything. (This is also true for typing on a computer, but I'm comfortable enough on a keyboard that I don't have problems typing English.)