How the Chinese Typewriter Influenced Predictive Text

Caixin: How Chinese Typewriters Led Way to Predictive Text on Smartphones

The first working Chinese typewriter was invented in the 1880s by a missionary named Devello Z. Sheffield, who spent 44 years in Tongzhou, a district in eastern Beijing. Sheffield wrote many books in Chinese and was good at building things. His obituary in The Milwaukee Sentinel noted that he made "his own electrical machine, which sent off sparks of great size and brilliancy."

To build his typewriter, Sheffield first had to determine how many characters were crucial for the project, so he visited typesetting shops and foundries where movable typeface was cut and forged. "He didn't do a statistical analysis," Mullaney explained, "but he fraternized with Chinese printers and asked for their consult and advice." Sheffield decided to use 4,662 characters that he sub-categorized into 726 very common characters, 1,386 common and 2,550 less common, based on how often they were used, and arranged them in a round tray.

Sheffield's typewriter never became commercially available, but his idea of categorizing characters based on their frequency of usage reappeared in 1910 in the first widely sold typewriter, which was co-developed by Zhou Houkun and Shu Zhendong and manufactured by the Shanghai-based Commercial Press. This typewriter had a rectangular tray that held about 2,500 characters etched into moveable metal slugs and divided into two sections according to the frequency of use. Within each of the two regions, characters were arranged by the Kangxi radical-stroke system used in dictionaries.

Compared to a Western QWERTY keyboard, these machines were bulky and a good typist could only hammer out 20 to 30 characters per minute.

However, in 1951, a typist in Luoyang in the central province of Hubei, named Zhang Jiying was lauded by The People's Daily for typing more than 3,000 characters in an hour, or 50 characters per minute. He later reached a speed of nearly 80 characters per minute. Zhang achieved this speed by rearranging his character tray to include 280 common two-character compounds and several common three and four character sequences – words and phrases like "meidi," or American imperialists; "jiefangjun," or liberation army; and "geming," or revolution. This method quickly caught on, and typists eventually began reorganizing much of their trays according to the words and phrases they expected to use most. It is among the first ever uses of the idea of predictive text found in modern times. Many of these character arrangements were prompted by the ceaseless political campaigns underway in the 1950s and the repetitive and formulaic language used to promote them.

The article also discusses how this influenced predictive text today.

LA Review of Books: IT’S TIME TO GET OVER QWERTY — A Q&A WITH TOM MULLANEY ON ALPHABETS, CHINESE CHARACTERS, AND COMPUTING

Is China then in a better position than most other countries when it comes to moving forward technologically in our information age? Or are there things that could lead to the advantages that characters can provide being squandered?

I think China is in a better position, yes. Chinese “Input” is arguably the future of IT — not only in China, but globally. As I put it just now, “input” takes far greater advantage of the QWERTY keyboard than conventional “typing” does in the alphabetic world, with our century-old what-you-type-is-what-you-get mentality. The QWERTY keyboard in China is a “smart keyboard”: a device used to oversee a rapid two-part process of finding and choosing characters from a database, using ever-faster and more sophisticated techniques. This “finding and choosing” aspect of modern Chinese IT is one of the most exciting and surprising discoveries I made in my research, in fact. Namely, modern-day Chinese computing owes a tremendous debt to the work of Chinese library scientists and others back in the 1920s through 1940s — figures like Du Dingyou, Chen Lifu, and others who never knew that the computer would be invented, of course, but who obsessed over the question of how to design faster and faster ways of organizing Chinese library card catalogs, phone books, and filing systems! Little could they have known that, decades later, major companies like IBM, Microsoft, Google, Apple, Sougou, and others would dust off their methods and turn them into the engine of China’s rise as an IT giant!

By contrast, the Latin alphabetic world has spent more than two centuries congratulating itself for “Our Glorious Alphabet,” and yet at the same time has done far less to explore and push the Latin alphabet to its fullest potential. In fact, Silicon Valley is having a really hard time now convincing average computer users in the West that the keyboard is in fact an obstacle — that it’s somehow broken and that average users need to start exploring more of the clever English-language input devices now available on the market — like the absolutely brilliant ShapeWriter system invented by my friend Shumin Zhai.

Aside from predictive text, Chinese is far more suited for compressing information on services such as Weibo/Twitter.