It is well known that pictographic languages that use Hanzi, like Mandarin, are difficult to work with for computer input and output devices. After all, each character is a tiny picture that represents an entire word, not just a sound. But did you ever wonder how China used telegraphy? We’ll admit, we had not thought about that until we ran into [Julesy]’s video on the subject that you can watch below.
There are about 50,000 symbols, so having a bunch of dots and dashes wasn’t really practical. Even if you designed it, who could learn it? Turns out, like most languages, you only need about 10,000 words to communicate. A telegraph company in Denmark hired an astronomer who knew some Chinese and tasked him with developing the code. In a straightforward way, he decided to encode each word from a dictionary of up to 10,000 with a unique four-digit number.
A French expat took the prototype code list and expanded it to 6,899 words, producing “the new telegraph codebook.” The numbers were just randomly assigned. Imagine if you wanted to say “The dog is hungry” by writing “4949 1022 3348 9429.” Not to mention, as [Julesy] points out, the numbers were long driving up the cost of telegrams.
It took a Chinese delegate of what would eventually become the International Telecommunication Union (ITU) to come up with a method by which four-digit codes would count as a single Chinese character. So, for example, 1367 0604 6643 0932 were four Chinese characters meaning: “Problem at home. Return immediately.”
Languages like Mandarin make typewriters tough, but not impossible. IBM’s had 5,400 characters and also used a four-digit code. Sadly, though, they were not the same codes, so knowing Chinese Morse wouldn’t help you get a job as a typist.
A few years ago I was reading about alternate number systems. Humans , of course, are typically base-10 machines (no doubt because of 10 fingers) while computers are base-2, where the digits zero and one are conveniently represented by voltage-on and voltage-off. But what are the relative merits of x-symbols vs y- digits?
If you have a small symbol set (like binary) then the number of digits you need to express a given quantity goes up. Conversely, you can reduce the number of digits needed by increasing the number of symbols, but now you have a potentially cumbersome symbol set. It strikes me that these same principles probably apply to alphabets, and that pictographic systems like Chinese correlate to the latter case.
If I recall, mathematics say the most efficient balance between symbol set size and number of digits is Euler’s number… implying a symbol set of three. Indeed base-3 computing hardware would seem to promise improvements in power consumption and performance… but I wonder if similar math/logic applies to written language.
Culturally interesting perhaps, but 10’s of thousands of pictographs is not a practical way to navigate the modern world. My question is, what size would the most efficient alphabet be?
Ternary is fun, but seximal is the future!
I wonder if that article might be related to the crypto challenge in the next badge for HAD conference.
I don’t know why but the reply button to the main thread sometimes place the comment beneath another.
Is it not written, “Hexapodia is the key insight”?
I regularly use binary (computer stuff), octal, hex, decimal of course, base 12 (hours, time, eggs haha), and sexegesimal (time again, celestial nav hobby). Alternate base systems are more common than you’d imagine.
Time and eggs don’t use base-12, they use the same character set as base-10. You would need dedicated symbols to represent 11 and 12 for it to be in base-12. If you do in fact have distinct characters for your egg-counting, please share; that sounds fascinating.
Oh, a “shocked face” youtube video. I’m definitely going to not watch it 🙂
are there youtub videos without it?? who would have guessed. ( i use the clickbait remover for firefox, so i never see the title image of a video)
But you do see the clickbait title or is that blocked too (that would really help).
yes, for example
danooct1channel10k dictionary, using 10k worth of numbers is still going to be an insanely difficult memory issue for the operator imho.
This reminds me of the old telegraph code books.
Its mainly designed for telegraphy
Thousands of years of using such ineffective system of writing makes it nearly impossible to change it for something better. It is also interconnected with speech system, which uses pitch to change meaning.
Hard to solve.
Reminds me of nations which for similar historical reasons keep strange measurement units:) But they minimally united nice part of world with using simple and effective language with plus-minus adequate writing system. Thanks and go on!
1131 1344 7024 1639 1034 1129 1458 3010
In shanghai near the Bund is a great museum called the Shanghai Telecom Museum, on yan’an road. Must see for geeks, theres even a semaphore signal tower near it. While your are at it, visit the gutzlaff signal tower 5mins from there, a semaphore style weather info tower from the early 1900’s.
better question is where is simple (short) sample of dictionary in any language.
esperanto need 3K , simple basic english (using in other place uses english 8K)
meybe we make smalest dictionaty?
We should all switch to esperanto.
It was a thing by the turn of the century, I remember.
Some websites had an Esperanto version at the time.
Esperanto basically was seen as a new version of Latin language, as a world language.
From a political point of view it would be rather neutral, too.
English is being associated with western society, by contrast.
So it would make sense if a basic set of Esperanto would be thought worldwide in schools, maybe.
It would maybe build bridges about different societies, because it feels more like the true language of a free world.
There is no reason for this to involve Chinese at all. The random number is directly representing a word there is no need to go through an intermediate character.
My Chinese is a little rusty but I’m pretty sure it isn’t one-character-one-word. There are plenty of words that are compound ie two or more characters linked together to make a word but like English here.
Reminds me of some vintage RTTY decoders of the 1970s and 1980s.
They could do Japanese RTTY with JIS encoding. Such as Tono Theta 777.
It mainly were models by Japanese manufacturers. The Japanese were big in amateur radio in late 20th century.
Yeah, Chinese telegraph code. I have a reference book on that. Realistically someone who knows 2,000 characters is literate but they did a thorough job. To speak about 50,000 is blowing smoke because that includes archaic and no longer used characters. 11,000 would be very impressive. My first dictionary which I still use was called the 5,000 Dictionary. Blue paperback. These days Unicode handles it: each character has its own number. Everybody calm the heck down, won’t you?
Lin Yutang invented a Chinese typewriter, I don’t know if that’s the one mentioned in the article.