John Cayley reviews The Chinese Computer

Thomas S. Mullaney. The Chinese Computer: A Global History of the Information Age. Cambridge, Mass.: MIT Press, 2024.

Review by John Cayley

From Typewriting to Hypowriting?

Despite its subtitle, like Thomas Mullaney’s earlier The Chinese Typewriter, most of this sequel is taken up with meticulously researched Chinese technological history.^{^[1]} The book does situate its history in a global context, although not in a manner that one might expect. Mullaney recognizes that the Chinese information age was subject to a “foundational provincialism” (p. 183). But this characterization of the postwar anglophone regime of computation is evoked to highlight the ingenuity of Chinese engineers obliged to mod their technologies to work with what may be more readily recognized as a foundational imperialism. What others might see as an agonistic neglect of China’s information age is handled by Mullaney’s telling us that Chinese computing is “inherently compensatory” (p. 11). This, one of his “axioms of Chinese computing,” dictates that it work to compensate for the Anglo-literate limitations of foundational technologies granted to it by the West (p. 3).

Mullaney’s other axioms confine his book to a relatively safe, apolitical narrative of technological progress. Apart from those that relate to “compensatory” hardware—keyboards and peripherals—the others pertain to software: Chinese Input Method Editors (IMEs). These include: “The addition of [IME] mediation can lead to the subtraction of time”; Chinese characters have “an infinite [sic] number of input sequences . . . in theory”; and People’s Republic of China-sponsored phonetic-driven IMEs come to predominate (pp. 9, 12, 19). What brings together his brilliantly detailed associated histories is Mullaney’s first axiom, “In Chinese computing, what you type is never what you get. . . . users . . . operate entirely in code all the time,” exemplified by chapter one’s histories of human, notably female, operators who memorized and embodied vast tables of entirely arbitrary codes for characters (p. 7). From this and the other axioms, Mullaney derives an overarching thesis, interpreting the inscriptional practices that emerged from Chinese computing as a phenomenon he christens hypography, “writing that operates in the service of conventional writing . . . at a register beneath” (p. 20). This leads to a rather grandiose claim, “This book is a history of Chinese computing. More than that, however, it is a preliminary roadmap to a new era in the history of writing” (p. 28).

The question of writing, its ontology, and its relationship to language is implicitly evoked, albeit in a context where “conventional writing” is taken for granted. As text, computation does indeed understand writing conventionally and functionally. Text consists in strings of alphabetic data representing language. Any problems with the what or how of this representation is of little concern to engineers, whose science knows computation rather than language. Mullaney’s definition of hypography brings Aristotle to mind: speech as representation of thought, writing as representation of speech, and hypography as representation of (conventional) writing, with built-in hierarchical ordering, “in [its] service” (p. 20).

Mullaney’s axioms resolve to structuralist linguistics. “Infinite” input sequences are a function of the sign’s arbitrary nature. Placing the hypograph in a “register beneath,” etymo-metaphorically, doesn’t render it different in kind. And any writer does so “entirely in code all the time” (p. 7). Mullaney maintains that because Western keystrokes correspond to letters, they have an “immediacy” that Chinese computing can only long for (p. 11).^{^[2]} The theoretical foregrounding of any such “immediacy” is crucially at stake for our understanding and appreciation of language as such. No traces of language can be immediately identified with what they mean.

In the second decade of the twenty-first century the way we model language changed, coinciding with the arrival of auto-completion assisted by post-statistical machine learning. Its foreshadowing by what Mullaney identifies in hypographic Chinese IMEs is striking, as analogous to the prompt and response transaction with Large Language Models. Culture’s feverish archival shift from libraries of texts to search-mediated databases is widely acknowledged. It is not so often recalled that a language model “is a database in the form of a neural network.”^{^[3]}

Mullaney suggests that hypography may be a more efficient future for writing, but when he discovers its emergence in the inscriptional practices of Chinese computing this is, in part, as a function of diagnosing one of its pathologies, a character amnesia that is remarkably prevalent throughout otherwise highly literate constituencies. Chinese IME computing makes people forget how to write. “The brush is raised, but the character is forgotten,” tibiwangzi, the phrase for this syndrome, reminds us that Chinese writing is integrated with very different practices of language in the sinosphere. The struggle of Chinese computing to accommodate characters was influenced by a long history of written expression entangled with embodied practices that encompass China’s highest visual art, calligraphy, the powerfully resonating written gestures of its cultural and political elite.

Hypography was not free to choose the most efficient sequences for Chinese retrieval-typing (see pp. 216–17). Users chose, and developers vied, to provide IMEs that were best integrated with an embodied, enculturated disposition to inscribe with computers. If hypography is an ascendent principle—and a poststructuralist argument could be made that all inscription is hypography—it is never a free-floating abstraction. It must be integrated with the embodied practices that we apply to language and use to create it. The extension of Mullaney’s hypographic paradigm to language model prompt and response is powerfully suggestive. This has already become some part of how we write now and will write in the future. But, for the present, the large models work with text and the question of how our writing with models will be integrated with our embodied practices of language may prove existential for human language as such.

^{^[1]} Thomas Mullaney, The Chinese Typewriter: A History (Cambridge, Mass., 2017).

^{^[2]} Mullaney’s usage and the context of IME mediation revivifies the sense of without mediation that is latent in an English word usually referring to time.

^{^[3]} Aden Evens, The Digital and its Discontents (Minneapolis, 2023), p. 189.