A standard QWERTY keyboard has a few dozen keys. How can Chinese—a language with tens of thousands of characters and no alphabet—be input on such a device? To answer this question, one needs to return to the beginnings of electronic Chinese technology in the wake of World War II, and follow up through to its many iterations in the present day. In particular, we need to examine the development of Chinese “input methods”—software programs that enable Chinese characters to be produced using alphanumeric symbols—and the profound impact they have had on the way Chinese is written. Today’s dominance of predictive text input, along with its unprecedented speed and ubiquity across both ideographic and alphabetic languages, owes its direct lineage to these input methods. The story of Chinese input and of Chinese computing, which I trace in my new book, The Chinese Computer: A Global History of the Information Age, encompasses a cast of brilliant and eccentric personalities drawn from the ranks of IBM, MIT, the CIA, the Pentagon, the Taiwanese military, and the highest rungs of intellectual, corporate, government, and military circles in China.
* * *
In September 2013, I sat down for tea in a neighborhood Starbucks with an octogenarian man and his granddaughter. It was the first time I’d met a Chinese veteran of World War II—the Eight-Year War of Resistance Against Japan, as the conflict is known to many in China.
The conversation came about thanks to a letter-writing campaign. My wife and I had slipped hundreds of letters into mailboxes all along Beijing’s Second Ring Road, addressed to inventors of Chinese input systems I discovered in the People’s Republic of China patent database. Input systems, also known as Input Method Editors (IMEs) or 输入法 (shurufa), are computer programs that enable one to use the keys of a standard QWERTY keyboard to enter Chinese characters on a computer or digital device. Every one of these inventors—including this elderly war veteran, much to my surprise—laid claim to a novel Chinese input method and the certainty that their creation would at last solve the puzzle of computational Chinese text entry.
Jiang Kun spent 15 years working on his system—“Sound Code” (Yinma), as he called it. The system was premised on breaking down Chinese characters into their constituent elements, not unlike many of the systems developed from the 1950s through the 1980s. The system was interesting to read about, certainly—not least because I had never met an inventor this advanced in years, let alone a war veteran—but unlikely to reroute my research in new directions. It might make for a charming footnote, I reasoned; if not, I was still grateful to be there, and especially that he had taken the time to respond to my letter and meet with me in person.
Then his granddaughter turned the wheel sharply on our placid Tuesday afternoon conversation and sent us careening down a cliff.
I think my input system is a bit easier to use.
She whispered the words just softly enough that her grandfather wouldn’t hear.
I’m also working on a Chinese input system.
So is my father.
My language skills were clearly failing me, I thought. Three generations within the same family, working on three different input systems? She confirmed: Not only had her grandfather developed an input system for Chinese, but she and her father had as well. Her father had even quit his job to work on his input system full time, she underscored.
After a pleasant exchange, Jiang Kun stood up briskly, smiled, and reached out to shake my hand. We snapped a photograph, and the conversation was over. I walked out of the café in a haze.
It was that day in fall 2013 when it hit me: The Chinese Input Wars of the 1980s and 1990s—a period in which literally hundreds of inventors competed to develop the fastest, most accurate, easiest-to-use way to input Chinese characters using a QWERTY keyboard—were still raging, and a peace settlement may never be brokered. Input may never settle down.
During the 19th century, and most of the 20th, it was readily apparent why so many engineers, linguists, entrepreneurs, policymakers, and others would have committed themselves to the problem of Chinese in the information age. Chinese-language technologies were objectively slower and less effective than those for English and many other languages. Chinese telegraphy was slower. Chinese typewriting was slower. So was early Chinese computing. These deficits posed monumental problems for state-builders, moreover, along with diplomats, industrialists, researchers, entrepreneurs, military planners, financiers, educators . . . (the list is too long to enumerate). Beyond practical problems, there were also abstract, emotional motivations at play. On a conceptual level, “speed” and “efficiency” had served for so long as both a “mantra” and a “measure of men”—core metrics by which societies, languages, and technologies were compared and judged—that the markedly slower performance of Chinese raised disturbing questions about Chinese cultural and civilizational “fitness.” Were Chinese characters compatible with the modern era? Would China need to abandon them?
But the 21st century is an entirely different story. Chinese-language e-commerce platforms, apps, chat rooms, online literature sites, and social networks are the envy of tech firms and venture capitalists around the world. Market valuations of some Chinese tech companies beggar belief. Barely weeks after my exchange with Jiang Kun and his granddaughter in Starbucks, moreover, Huang Zhenyu stepped into an auditorium in Henan for a national Chinese typing competition and bested his competitors with a speed of 221.9 characters per minute—3.7 Chinese characters every second. Based on any number of metrics, it would seem, Chinese has more than proven itself as one of the most successful writing systems of the digital age. What accounts, then, for this frenetic, unrelenting activity? Why does the number of Chinese input systems—already in excess of a thousand—continue to grow? Did the Jiang family—or, indeed, the dozens of inventors who submit IME patent applications each year—truly think they could outpace a speed of 221.9 characters per minute? At 3.7 characters per second, hasn’t Chinese emerged victorious in its more than century-long conflict with Latin alphabetic supremacy? Why, then, aren’t Chinese engineers, inventors, companies, and users calming down and enjoying a well-deserved victory lap? Hasn’t alphabetic order at last been overthrown?
Two factors are essential to understand the continued restlessness of Chinese input; the first is technolinguistic. Within English-language typing, pushing the key marked “Q,” then seeing the symbol “Q” appear on screen, is a taken-for-granted feature of human-computer interaction—so taken for granted that it has fostered a mythology in which alphabetic textual input is understood by many as an “immediate” act wherein “depression equals impression.” I say “mythology” because, of course, in Anglophone as in Chinese computing there exists no pre-given relationship between any particular button and any particular symbol—to depress the key marked “Q” is simply to close a circuit that could actuate any one of an infinite number of potential operations. Printing “Q” to screen is just one such outcome.
Having operated under this framework for over a century, however—a convention that dates back to the era of mechanical typewriting—Anglophone computing makes it possible for users to be lulled into a deep and (for some) unshakeable complacency: the inability to summon to mind the fundamentally arbitrary nature of all key-symbol relationships. The arbitrariness, and thus plasticity, of all human-machine interaction sinks to the bottom of consciousness, like a pebble sinking to the bottom of a lake. All that remains is the assumption that Q must equal Q.
The situation in Chinese human-computer interaction is structurally different. The key-symbol relationship is just as arbitrary as in English, but in this case everyday users are, so to speak, constantly reminded of this arbitrariness. When there is no possible way to fit each Chinese character on its own dedicated key—when, by definition, what you type is never what you get—the nonidentity between keys and symbols becomes not just “possible,” but the starting point. A kind of persistent, structural mismatch ensues that prevents hypography from settling into the same kind of complacency found in conventional models of human-machine interaction. When “Q” equals “Q,” after all, a “keyboard” and an “interface” are one and the same thing. When “Q” doesn’t equal “Q,” however—when “Q” can be used to describe any number of structural features of a Chinese character, or perhaps the sound of Chinese words—then the keyboard itself is not actually the interface. The keys Q-W-E-R-T-Y all stay in the same places, and yet all one needs to do is change the Input Method Editor (IME) to change the interface. In this way, a kind of buoyancy constantly pushes the intrinsic arbitrariness of language back to the surface of consciousness—no longer a pebble, but prosecco bubbles rising and bursting into the realm of critical awareness, over and over again.
This is not to say that everyday Chinese computer users are all armchair philosophers of language or would-be semioticians. Nor is it to say that input doesn’t engender its own habits of mind and practice. But no matter which Chinese IME users employ, there is no way they can look down at a QWERTY keyboard and pretend to see “Chinese characters” in any straightforward or self-evident way, the way an Anglophone computer user can look down at the keyboard and see “my alphabet.” To input Chinese using a QWERTY keyboard or a trackpad is always a matter of invoking or retrieving Chinese characters from memory. The A-B-Cs of the Latin alphabet are just a means of doing so, one among the infinite number of ways this act of retrieval can be achieved.
The second reason for the continued reworking of Chinese IMEs is sociopolitical, forming part of a nearly two-centuries-long, crucible-like period of civilizational anxiety. This was a period in which China’s position in the world was thrown into radical doubt within a new hierarchy of modernity and civilization dominated by the West. Through so much of this history, China found itself in a position of deficit and compensation—of compensating for shortcomings, whether real or perceived, through means self-imposed or imposed from without. Chinese input is yet another part of this anxiety-ridden history of “compensation”; IMEs were invented, after all, solely for the purpose of circumventing the incalcitrant problem of “fitting” Chinese on a Western-style keyboard. Indeed, each and every one of the components in China’s hypographic revolution originated as some kind of workaround, mod, patch, hack, make-do compromise, or “Plan B” designed to make the best of a bad situation. Even the engineers who pioneered Chinese input never held input up as a formal “alternative” to Western-style technological modernity; never did they exult hypography as a uniquely “Chinese” solution to challenges of the digital era. Even inventors of some of the most successful Chinese IMEs in history never went so far as to suggest that English should perhaps steal a page from Chinese computational input and follow suit.
But then something peculiar happened. Since the turn of the 21st century, all of the many compensatory techniques and technologies of Chinese-language transmission, storage, and inscription—ingenious work-arounds and hypermediations dating back to 19th-century Chinese telegraphy, throughout the period of mechanical Chinese typewriting, and into the early era of Chinese computing—got faster than the mode of textual production they were built to compensate for: English and the longstanding model of one-key-one-symbol, what-you-type-is-what-you-get. Hypermediation, it turned out, boasted a much higher ceiling than “immediacy.”
To achieve 3.7 characters per second, however, is not the same thing as to shift a centuries-old paradigm. It does not, by itself, fully dislodge the legacies of Latin alphabetic dominance that permeate information technology at multiple registers simultaneously. Long after Chinese input has begun to surpass English typing in speed and accuracy, this longstanding sense of deficit—of somehow falling short of one-symbol-one-key, what-you-type-is-what-you-get, and immediacy—forms a belief system that will likely continue to exert influence on the minds of engineers, entrepreneurs, and amateurs alike. For how long, no one can say.