How China Conquered The Keyboard

- ways to support -

- where to find me -

- how i make my videos -

- my courses -

- about -
Johnny Harris is a filmmaker and journalist. He currently is based in Washington, DC, reporting on interesting trends and stories domestically and around the globe. Johnny's visual style blends motion graphics with cinematic videography to create content that explains complex issues in relatable ways. He holds a BA in international relations from Brigham Young University and an MA in international peace and conflict resolution from American University.

- press -

[nek]

Just a little mistake: the Zhuyin (Taiwanese typing) shows the alphabet not of Zhuyin, but actually of Hangul, the Korean alphabet.

[raff]

the fact you put a yellow line showing how much of an ad is left is so appreciated

[bks]

I am Japanese and I am learning Chinese, and pinyin input is very useful for foreign learners.
Such a predictive conversion function is also developing. In Japanese, there are two types of input systems, kana and romaji, and most people use romaji input.

In 1978, the first practical kana-kanji conversion system was completed in Japan. Toshiba launched it as the first domestically produced word processor. Novelist Kobo Abe was one of the first to write literary works on a word processor in the 1980s and participated in the development of NEC's word processor.

[asdfxyz]

I'm an American born Chinese and the Pinyin system is 100% responsible for me being able to browse the Chinese side the the internet (music, news, trends, and even online communication) that otherwise wouldn't be possible. It's quite serendipitous for the Chinese if you think about it.

It's actually fairly easy to use as long as you can speak the language and know how to sound it out into the English alphabet. Google auto suggestion and Google translate is also super helpful.

Pretty cool to see how the keyboard played a role in bringing a country into the modern world.

[astrnox]

While I appreciate Johnny for covering this topic and I did learn a few things from the video, some of the things he shows or talks about are wrong or inaccurate. I attribute it to his not understanding the language and its use.

While Zhuyin is indeed the alphabet used in Taiwan for typing, 11:50 and 12:58 is Hangul, which is the Korean alphabet.

14:50 he describes how Cangjie works, not Pinyin. Cangjie works using radicals, which he shows there. Predictive text didn't change typing Chinese in Cangjie as much as it did in Pinyin. And Pinyin is the primary method that Chinese speakers use to type Chinese. You type the Latin characters of how a Chinese character sounds and it suggests the next character(s) to use. This is important because many Chinese characters sound exactly the same, and the difference when spoken is only known through context.

It should be noted that Johnny probably didn't understand the differences among "radicals", "characters", and "words" in Chinese. For example:
"氵" is a radical which represents water, but we don't use it on its own.
"水" is a character AND a word, which means "water".
"水果" is a word that is made up of two characters, which means "fruit", specifically those we eat as food.
"果" is a character AND a word that means "fruit" in general, we rarely use it alone, and is usually qualified with another character to form words/phrases such as "水果" above, or others like "如果" (if), "结果" (outcome), "果然" (sure enough), "果汁" (fruit juice), "苹果" (apple), etc.

On typing speed, even though they could type 242wpm, this was based on copying an existing text. People do not think that quickly and therefore will not be able to utilise that kind of speeds except say in transcription. English has similar: shorthand/stenography for writing by hand, and stenotype for typing into a machine. Some stenographers can reach 300wpm on such a machine and the official record for American English is 375wpm.

Don't get me wrong. Johnny did a great job with the writing and production. But his lack of understanding the basics of the language was a major handicap here.

[justinlam]

Plus, you don’t always have to type the whole romanized word. Just the first letter of every word sometimes.

For example:
I like you 我喜欢你

This would be typed “Wo xi huan ni”
But if you just type every first letter “w x h n”, the algorithm would predict it and automatically generate the whole sentence—just by typing 4 letters.

[ranulfdoswell]

He keeps showing the Hangul (Korean) alphabet whenever he talks about Zhuyin. Also it's a shame he didn't mention the early handwriting recognition, which was a big thing, at least in Hong Kong in the mid 90s, which ultimately led to the trackpad that's on almost everyone's laptop today. Back then it was a separate device that plugged into a PC and you drew a character with your finger, as if you were drawing with a pen.

[Zhi-zvin]

As a Chinese speaker, I can confirm that we do type slightly faster in Chinese. Let me give u an example, say u wanna type 'I don't know', the English solution would be 'idk' or just type it all out sequentially, but in Chinese for the translation 我不知道 (wo bu zhi dao), we just type in 'wbzd', namely the first letter of the pinyin of each character, then the keyboard would suggest a possible outcome 我不知道, I personally find this extremely useful because it's kinda like typing multiple English words at the same time.

[LoganBeck]

Live here in Taiwan and use Chinese 85% of my day. I type 3-4X quicker in Chinese than I do in English, my mother language, and have always struggled explaining to others how I do so, great video! Also, I look forward to your Taiwan - China video! 加油 - 米漿萬歲！

[nomastersnogods]

The fact that we get free documentaries on YouTube by Johnny Harris is truly a gift. 👍

[maximilianisaaclee]

I think some people have noticed the mistake of Zhuyin being represented with Korean Hangeul instead, which I find kind of ironic. As a Zhuyin user who have learned some Korean, I find Zhuyin typing quite similar to Hangeul typing (with some differences of course).

I sometimes ponder over this technology as well, how it came about, how predictive text technology was developed. We all have come a long way in terms of technology.

[astrospeedcuber]

As an ABC (Australian born Chinese), I can speak and type Chinese, although I'm unable to write the characters. This is all due to the pinyin system, which as the video says quite accurately, allows people to just "sound out" the words; and for me, as someone who has learnt pinyin and can read most Chinese characters, I can simply just type in the pinyin that I associate with the word and find it on the pop-up list of characters.

Also the "predictive" part of the simiplified Chinese pinyin keyboard is also very smart; something that I had not noticed previously. For example, the Chinese characters for "I will go to watch the movies" is "我去看电影". The pinyin, which I will just sound out here, would just be "wo qu kan dian yin". Now that would seem pretty normal when you look at it, but as the video also mentions, the keyboard "predicts", though it's really more like "guessing" what you want to say. One really smart and time-efficient thing is that you can simply type the starting consonant (doesn't always work for words that start with a vowel) and a bunch of words that the computer thinks you want to say (based on the most popular words used in the cloud system and also from your own typing) will pop up. So instead of "wo qu kan dian yin", as it is a pretty common sentece/phrase to say, you could, and most people would, just instead type "wqkdy", which if I try to type in the keyboard right now, "我去看电影" is exactly correct.

Very smart indeed.

[sujalhansda]

This guy makes documentaries on random topics so interesting. Appreciate the hard work !

[wheatonrecurrence]

Me: How do I make this Chinese dish?
Johnny: Before that, we can't talk about Chinese food without geopolitics

[ThisisJackyNg]

As a person from Hong Kong (21 currently, things might have changed to younger generations, I don’t know), we’ve taught to use Cangjie and Sucheng (the simplified version of Cangjie, basically just type the first and last buttons) when we were in primary school. I found both Cangjie and Sucheng are extremely complicated since I need to remember all the characters which weren’t even printed on the keyboards mostly. All students in Hong Kong have compulsory Mandarin-Chinese classes (we are taught Cantonese-Chinese on Chinese lessons and the pronunciations of Cantonese-Chinese and Mandarin-Chinese are different) as well, we have to learn the phonetic alphabets of Mandarin-Chinese so as to know how to pronounce Mandarin-Chinese probably, which are Pinyin. Therefore, I gave up using those complicated Cangjie and Sucheng typing and get into Pinyin typing, which is far much easier. It’s basically just like typing in what the Pinyin that we’ve learned in classes, as simple as that. No extra things need to be memorised.

[yuichilee]

Another thing about cloud based predictive input is that whenever you type in a word that is not in a dictionary ( i.e. names, and some names are so rare you have to choose the right character every single time), it remembers it on the cloud and the next time you type using the same IME, even on a different computer (I use Sogou), it would suggest that name that you've once typed somewhere on some other computer

[mfaizsyahmi]

This video feels half-cooked. It's missing, or plain wrong, on several major points:

1) [OMISSION] The pinyin is not a 1-1 match with the latin alphabet because the Chinese language(s) has tones. It's those accent marks in the vowels of pinyin. There's no quick analogue of it on the QWERTY keyboard, so they adapt for it by assigning a number to each tone, and add that at the end of the sound you typed.

2) [OMISSION] A far larger technological hurdle they faced, that is not touched at all in this video, is how to represent the 1s and 0s of digital data to the many tens of thousands of characters in the language. Whereas our latin alphabet, in both uppercase and lowercase, plus numbers and symbols, can fit within 7 bits, and we agreed on a standard called ASCII to ensure the byte code for, say, the letter A, is the same across systems. They had to create their own encoding, and they had to figure out which characters can fit on 1 byte, and which needs 2 bytes, and how to signal the decoder to know which is which, and then how to draw them on screen, and design the typefaces and so on. And after all that, they need to convince each other to adopt the same encoding, which isn't a thing between countries since each country develops their own encodings. But finally Unicode comes in to the rescue, and we got emoji.

3) [ERROR] The chart you're showing for zhuyin is actually for Korean Hangul, a completely separate system. Hangul is a proper alphabet for general use in the Koreas, not just a learning tool to speak a language like pinyin/zhuyin.

4) [ERROR] The character prediction you're showing is for handwriting, not typing, Chinese Hanzi/Japanese Kanji/Korean Hanja (CJK) (they're all really similar). A vast majority of CJK characters is formed by combining 2 parts, a radical for general meaning, and another part for sound.
When an IME receives the handwriting input for the radical, which conveniently is usually written first, it suggests to you full characters with that radical, usually sorted by the number of strokes in the characters. As you write more strokes, it narrows down further.
When you are typing in pinyin/zhuyin, it doesn't figure out by the radical, which is not encoded in the sounds of the language because it only conveys the general idea of that character. It instead looks up an internal homophone table, basically all the list of characters that has the same pinyin.

5) [Extra] The later part of character prediction in the video is down to how many compound words the Chinese language has, largely composed of two or more characters, to convey more complex ideas and reduce homophones. When you type the character for lightning (電), you're more likely to talk about things like electricity (電気) or a train (電車) for example, both compound words in both sino- and sinoxenic languages. So, the IME suggests them.

Johnny's journalistic vigour is pretty limp on this video. Sorry.

[magma_chicken]

As an ethnic Chinese who has English as my first language, I have to say that pinyin is incredibly useful in helping me learn how to pronounce each character (especially the more complicated ones).

[dcaf]

I’ve been typing Japanese since age 7. We have a very similar system of typing the sounds of the words and characters, and letting the computer render the native Japanese characters and Kanji. I’ve noticed in the last decade that Japanese keyboards on Windows and Mac have gotten better at predictive Kanji rendering of proper nouns, such as movie titles and celebrity names, as if it knows what a lot of people have been typing recently.

[iwavns]

3:24 “看”does not mean "gratitude", it means "look" or "see", actually “作” doesn't mean "take" too, it has many meanings, depending on the character that it is paired up with.