The phonological communication method that originated as the game of "shooting [or 'guessing'] characters" (shezi [phrase omitted]), which used numbers to communicate elements of language according to their pronunciation, was widespread in late imperial China and known under several names. Its ubiquity inspired writers to generalize it as the paradigmatic instance of communicating without speaking or writing. Neo-Confucian philosopher, administrator, and student of Western astronomy Lu Longqi [phrase omitted] (1630-1693, or He Yufeng [phrase omitted], creator of the extant manuscript ascribed to Lu and dating from after 1716), (1) called it the "method of knocking syllables" (qiaoyin fa [phrase omitted]) and inferred its presence in a story of communication through music related in the Confucian Analects:
This [method] is what Confucius, the sage, employed when he was playing the stone chimes in Wei and a man with a straw basket heard it and understood his frustrated purpose. [phrase omitted]. (2) It was not just Confucius who allegedly communicated in this way. Hu Yin [phrase omitted] (fl. 1906), who called the method by a different, common name (that I will discuss presently), saw in it the principles of the telegraph:
Westerners use wires carrying electrical current to manipulate the movement of a compass needle and then observe what letters (mu) the needle indicates, and merge them to form words (zi). Communication is thereby possible instantly, even across mountains or the vast expanses of the oceans. What they are using is precisely this method [of "shooting characters"]. What apparently originated as a game had, by the turn of the twentieth century, become identified with the communications technology of nineteenth-century industrialism. During its documented history of almost a millennium, the game of "shooting characters" was used in several different contexts and discussed in several more. In this paper, I will discuss the game from its earliest mentions in the thirteenth century to the first years of the twentieth, with a focus on the last centuries of imperial rule, from which most of my sources date. I will treat the game's relationship to the discipline of phonology, situate it within Chinese literati culture, survey its proposed usage as a cipher for secret letters or military communications, and finally consider its role as the indirect inspiration for a phonological writing aid for the less educated. Taken together, these actual and proposed applications of one phonological game show that Chinese phonology did not just exist in books, but encompassed distinct non-written practices that were essential to its proliferation. Widely applied in a variety of ludic, pedagogical, scholarly, and esoteric contexts, the study of Chinese phonology is of interest not only for the development of the language itself, but also for the history of cultural and intellectual life.
THE GAME'S PHONOLOGICAL PREREQUISITES
The game of "shooting characters," which commuted syllables to numbers, was comparable to some forms of language play in the European Renaissance. (4) It coexisted in China with games and riddles based on the structure of the script, which, like "shooting characters," were used to write encoded messages. (5) Its closest relatives, however, were arguably the drinking games so common in the late imperial period, as they, like "shooting characters," contained an element of competition and were based on "guessing" (another meaning of she). (6)
The game relied on a series of inventions in Chinese poetics and phonological lexicography. The basis of "shooting characters" was the segmentation of the Chinese syllable into a number of phonological properties and their subsequent arrangement in serial sequences according to those properties. These two related inventions appeared centuries before the earliest documented appearance of the game.
The segmentation of the syllable happened through a series of discoveries. In the fourth century CE, theoreticians of poetry identified a number of pitch tones (see figure 1) among Chinese syllables. (7) At some point, the Chinese syllable was further divided into the two parts of initial (traditionally [zi]mu [phrase omitted], "[syllabic] productive elements," (8) now called shengmu [phrase omitted]) and final (now called yunmu [phrase omitted]). It has been proposed that the analysis of the syllable into two parts might have originated in popular language games, only later making its way into books. (9) Whatever the case, the division of the syllable dates to before the seventh century CE, as dictionaries, "rhyme books" (yunshu [phrase omitted]), from that period make use of it. The segmentation of the syllable was carried out by spelling it using two other syllables. The first of the two spelling syllables shared an initial consonant (sometimes accompanied by a glide) or zero onset with the spelled syllable, and the second shared the rhyme, which included the main vowel, possibly a stop or nasal consonant, an off-glide, or, functionally speaking, a zero final if there was no consonant, and one of the morphemic pitch tones. Using the method, called fanqie [phrase omitted], for the modern standard language, we can spell xian [phrase omitted] as x(u) [phrase omitted] plus (q)ian [phrase omitted].
In rhyme books, syllables were arranged in sequences on the basis of these segmentations. The influential Guangyun [phrase omitted] (The expanded rhymes; eleventh century), (10) following earlier precedents not generally available to later scholars, separated tone from rhyme in a two-layered arrangement consisting first of a sequence of tones that, in turn, organized sequences of rhymes. The sequence of rhymes naturally had a relationship to the phonology of the spoken language. As Chinese speech varied over time and between places, the sequence of rhymes that scholars included in rhyme books varied as well. However, since the literary genre of regularized verse (lushi [phrase omitted]) became, as it name indicates, a fixed form, many rhyme books reproduced roughly the same sequence of organizing rhymes, allowing readers to use it as a reference work for writing poetry. Common rhyme books in this tradition used a sequence of around one to more than two hundred rhymes.
The initials, the existence of which was implied in the fanqie operation, were also used to arrange syllables in sequences. At least by the eighth century CE, scholars had identified a detailed repertoire of initials occurring in Chinese syllables. (11) In the most widespread version, they numbered thirty-six, but writers of later periods often reduced this number to better reflect some variant of the language as spoken in their own day.
When the game of "shooting characters" was recorded in the thirteenth century, a new tool of analysis, the graded rhyme table (dengyun tu [phrase omitted]), had been integrated to the mainstream of Chinese phonology for two hundred years. "Shooting characters" was related to the memorization of sequences from phonological dictionaries in order to facilitate the writing of regularized verse. The graded rhyme tables were probably largely used as a tool to navigate those same dictionaries. By the time of the tables' rise to popularity in the twelfth century, the dictionaries were several hundred years old and represented Chinese pronunciations that were no longer current in the spoken language, making them difficult to use. (12)1 will use Edwin G. Pulley blank's (1922-2013) Late Middle Chinese (LMC) reconstructions to represent the language of the twelfth-century rhyme tables. (13)
The most basic feature of a graded rhyme table was its use of two dimensions. Rather than listing syllables with similar phonological properties in sequences that followed one after the other in a series, the tables used two sequences as the axes of a grid, in which each cell represented a syllable that obtained one property from each axis. The axes commonly contained sequences of initials, rhymes, tones, and the like. It is not obvious that the game of "shooting characters" in its earliest attested incarnation was inspired by the graded rhyme tables, but the use of the sequence of initials suggests it was.
The whole sound system represented in a rhyme book could not easily be made to fit into a single table. Rather, books of tables often contained a number of them, each one listing a smaller set of similar-sounding syllables such as, for example, those included in one of the rhymes of the Guangyun. If the tables had the same layout, syllables belonging to different rhymes but sharing other phonological properties would appear in the same position in their respective table. Zhang Linzhi [phrase omitted] (fl. c. 1130-1203), in the introduction to the table Yunjing [phrase omitted] (Mirror of rhymes; extant version dating from 1197-98 with a preface dated 1161), described the advantages that this table structure had for resolving fanqie spellings: the actual comparing of pronounced syllables could be partially circumvented by moving between corresponding positions in the tables. (14) Later scholars saw in this aspect a similarity to "shooting characters."
EARLIEST MENTIONS OF THE GAME AS PUBLIC ENTERTAINMENT
The earliest evidence of "shooting characters" dates from long after the isolation of initial, rhyme, and tone in lexicography and poetics. The game's earliest mention is found in a collection of jottings dating from 1219-24. Its author, Zhao Yushi [phrase omitted] (conventionally 1175-1232), (15) described it as "a stunt, found among the popular practices, called 'striking the drum to shoot [guess] characters'" [phrase omitted], adding that it was of unknown origin. (16)
The game used two stanzas of regularized heptasyllabic verse (qiyan [phrase omitted]), (17) transcribed in LMC in table 1. (18) The two stanzas of the poem contained syllables that represented the initials and finals...