CHAPTER 1
Languages of Narrating and Bamboo-Cutter Pretexts
For the ancient Japanese, writing could not have been the familiar process it must have seemed to the ancient Chinese or seems to us today when, despite certain obstacles (e.g., writer's block), putting pen to paper or transferring letters from keyboard to computer screen is as intuitive and self-evident as eating or sleeping. As Raymond Williams has noted, "In modern industrial societies writing has been naturalized. It is then easy to assume that the process itself is straightforward, once the basic skills have been mastered in childhood. There is then only the question of what to write about."
In contrast to the Chinese mainland where a writing system mated to the phonological demands of a native, basically though not categorically monosyllabic language (i.e., one graph = one sound = referent-idea) developed over centuries, the Japanese, content without written language, found themselves confronting a civilization that began to impose itself not through military aggression but through the medium of written texts. Early attempts to adapt Chinese writing to the Japanese verbal ground must have presented seemingly insurmountable problems given the dissimilarity of the two languages. A writing system suited to the largely monosyllabic Chinese language would a priori be eminently unsuited to the agglutinative and inflecting, polysyllabic Japanese language. As contact between Japan and China (often via Korea and Korean immigrants) increased during the early centuries A.D., texts and other inscription-bearing objects (bronzes, mirrors, coins, and seals) began flowing into the islands. Scarcity of sources inhibits accurate reconstruction of the rise of scriptive activity, but judging from extant sources, Chinese writing seems to have entered Japan as early as the first century A.D. It was not, however, until the fourth or fifth century that Japanese began to write using Chinese graphs and, for the most part, the Chinese language—for the most part. The early specimens offer evidence that the Japanese at the very earliest stages were already disengaging phonetic from semantic values as they used Chinese graphs to transcribe native sounds, especially morphemes that constituted personal names and toponyms. Sometimes referred to as "Japanized (wa-ka) Chinese style (kambun)," the early practice probably did not extend much beyond transcription of personal and place names. When the first full-blown text as we know it appeared in 712, soon after the capital was moved to Heijo (Nara) in 710 (the Nara period dates from 710–84), the Japanese had been experimenting with writing for several centuries.
Chinese Writing and Japanese Discourse
Assuming its discursive space in diverse ways, the difficult-to-label Kojiki (Record of Ancient Matters) text immediately raises issues relevant to Japanese attitudes toward writing and the Chinese language. First of all, in contrast to the later Nihon shaki (Chronicles of Japan), Kojiki clearly purports to be a written transcription (selected and edited by O no Yasumaro at the command of Empress Gemmei) of an orally delivered (by Hieda no Are) discourse: "On the eighteenth day of the Ninth Month of the fourth year of Wad? a command was given to Yasumaro: 'You are to select, record, and present to the throne the old materials recited by Heida no Are'" (p. 23). Are was a young man renowned for his prodigious memory and vocal prowess: "One look and he could recite it aloud; one hearing and it was imprinted in his mind" (p. 22). He had earlier been commanded by Emperor Temmu to recite ("read aloud") selected old texts that recorded imperial genealogies and legendary and historical incidents so that a written transcription could be made for posterity. Empress Gemmei revived the project when it was halted with the emperor's death. Here we see an inextricable connection between writing and orality: on the one hand, the written does not, indeed cannot, come into existence without the oral—the oral authorizes the written; on the other hand, the written becomes a "permanent" document that proves the legitimacy of that which authorized it.
A point of controversy is the meaning of "read aloud" (yominaraFu). Some interpret it as signifying that Are somehow clarified the "meaning" of the texts as he recited them aloud. Others argue that, given his performative talents and the common practice of reciting Buddhist sutras, the phrase really meant that Are was commanded to recite the texts in a particular manner, using certain patterns of intonation, and that it was a particular oral rhythm (there must have existed other, competing ones) that Emperor Temmu wanted to valorize. To fix the previous discourses into a specific, orally deliverable mode was tantamount to seizing the essence of the texts. The one who performed the act or had it performed became the legitimate possessor and king of all discourses, that is to say, through a topographic metonymy, king of the country. Cognition of the world, channeled through phonic modulations, might have stopped well short of semantic closure, but political power, as it often does, effected another closure. For the Japanese, as we shall see, written discourse does not easily exist separately as a self-contained entity, but is always positioned vis-à-vis a multitude of "intertexts," whether linguistic stimulus (often, though not necessarily, oral), historical "model," genealogical imperative, narrator, and/or reader-listener. If we agree with the above argument that the text was meant to be intoned, we can conclude that for Kojiki, the written—at the same time that it accomplished the all-important goal of preserving a specific mode deemed proper to earlier discourses—was ultimately dependent on the oral and that the written text existed only in a contingent state that had to be vocally realized each time.
How did Yasumaro transcribe Are's recitation? Although written with Chinese characters, the Kojiki style, whether in phonological or syntactical terms, is not Chinese, which the Nihon shoki more closely approximates. By the time of Kojiki, proper names and toponyms are not the only items resistant to direct rendering by the Chinese written language; longer discursive stretches, commentary and reading notes, and song-poems are also transcribed in the "Japanized" style mentioned above. Generally called man'yogana, the selected graphs were part of a systematic process of using Chinese characters for phonological value through which the Japanese accommodated Chinese pictoideographs to the specifications of their own language. Take, for example, the opening line of Kojiki:
[TEXT NOT REPRODUCIBLE IN ASCII.]
[When heaven and earth first appeared, the name of the god who went out on the High Plain of Heaven is Ame no Minakanushi.]
After the line a gloss (called kunch? and found throughout the text) is inserted, instructing the reader to "pronounce the graph ten [heaven], which comes after the graph ko [high], as ama; learn from this as you read on" (p. 26). The sounds /a/ and /ma/ are represented by the two graphs [TEXT NOT REPRODUCIBLE IN ASCII.] and [TEXT NOT REPRODUCIBLE IN ASCII.]
Especially resistant to rendering with Chinese graphs, which tend to specify "meaning," were song-poems, since the incantatory qualities in the actual sounds needed to be preserved. Here is one poem (of 112 in Kojiki) considered the ancestor of the thirty-one-syllable tanka form:
[TEXT NOT REPRODUCIBLE IN ASCII.]
yakumo tatu, idumo yaFegaki, tumagomi ni, yaFegaki tukuru, sono yaFegaki wo (NKBT, 89) [the eight-layered clouds rise, in Izumo, land of the eight-layered clouds, to match the layers of my fence, built to confine my newly beloved; what a fine eight-layered fence]
Each graph, whose meaning is largely irrelevant, represents a single syllable.
As the Japanese continued to experiment with writing, they generated a variety of other phonemic-semantic-graphic combinations. Morphemes such as tu, mi, ni, ru, and wo in the above song-poem, what today are called "particles" (joshi) and verbal suffixes (jodoshi), were not represented when the Japanese wrote in a Chinese style. When the Japanese occasionally paid attention to the semantic values of the graphs, the result reads like a "shorthand" notation for transliterating into a Japanese syntactic form a Chinese-like word order. Man'yoshu 2845 is an example:
[TEXT NOT REPRODUCIBLE IN ASCII.]
Only vaguely intelligible to a Chinese reader, the poem requires the following Japanese phonetic realization:
wasuru ya to, monogatari site, kokoro yari, sugusedo sugizu, naFo koFinikeri [to forget about you, I talk about various things and try to clear my thoughts of you; but, try as I might, I find I end up longing for you even more]
The single graph [TEXT NOT REPRODUCIBLE IN ASCII.] is given the expanded Japanese rendering monogatari site, while the segment
[TEXT NOT REPRODUCIBLE IN ASCII.] displays the up-and-down word order necessary when reading kambun in Japanese.
The above examples represent two ways the Japanese adapted Chinese: the first is a type of phonetic use of Chinese graphs (man'yogana) called ongana, where the graph is divorced from its meaning (i.e., kana used for their Sino-Japanese sound values); the second is a way of reading the graphs using "orthodox Japanese sound values" seikun, wheroe both the Japanese pronunciation and the meaning of the graph correspond to its original semantic value. The Japanese also employed another type of man'yogana known as "native Japanese sound kana" (i.e., graphs used for Japanese sound values irrespective of meaning—kungana), with which they experimented at times in radical fashion. Possible combinations include two or three characters pronounced as one syllable or one character pronounced as two or three syllables.
Already evident in the Kojiki text and other documents, the use of kungana reaches radical proportions in the Man'yo collection. When, for example, the syllables ka and mo (an often-used two-syllable particle combination signifying exclamation) needed to be represented, rather than use two ongana, a graph pronounced kamo in Japanese was used irrespective of its meaning. A graph frequently so used is one that means "goose." A spectacular example of kungana and other kana varieties is Man'yoshu 2991:
[TEXT NOT REPRODUCIBLE IN ASCII.]
taratine no, FaFa ga kaFu ko no, mayogomori, ibuseku mo aru ka, imo ni aFazusite[breasts drooping, mother raises silkworms; they spin themselves into confinement, and I, my heart confined, am wrapped in gloom, unable to meet my beloved]
The written version, which contains the three basic ways of using Chinese graphs noted above, can be disassembled into the following components: seikun (tara) + two kungana (ti and ne) + ongana (no —the initial segment, taratine no, forming a "pillow word" for "mother") + three seikun (FaFa, ka[Fu], and ko) + ongana (no) + seikun (mayo, mod. mayu —here incorporating further graphic play, since the character, which means "eyebrow," was often associated in China with cocoons) + kungana (i —represented by two characters meaning "horse voice") + kungana (bu —also with two characters meaning "bee sound"; the pronunciations seem to be onomatopoeic versions of a horse's neighing and a bee's buzzing, thus adding an aural quality to the poet's psychic state) + another kungana (se, represented with two graphs meaning "rock flower") + seikun (kumo —like kamo above, but here given two graphs meaning "spider") + kungana (a[ru] —"rough or wild") + two ongana (imo —written with two characters that mean "different mother," the written version adding another dimension to the poem, since it suggests that the "beloved" in question may be a half-sister or that an actual confinement [by a mother, though not the poet's real mother] has prevented the poet from seeing the woman) + ongana (ni) + three seikun in kambun word order (aFazusite).
Such extreme instances of linguistic play, which would not appear again, demonstrate that the Japanese, far from despairing over the monumental task of transcribing their language using graphs suited to the mono(or at times a pleonastic bi-)syllabic Chinese or viewing transliteration simply as a mechanical problem, actually relished the potential in the phonemic-semantic-graphic combinations for extravagant or obtuse modes of expression. In other words, they showed a strong predilection for linguistic experimentation and play—an attitude closely tied to a belief in the spiritual efficacy of words (kotodama)—that continued into the Heian period, when simpler forms became available for transcribing the native language. It should be clear from the above that by the late eighth century the Japanese did not employ Chinese writing solely to transcribe their native (oral) language but took full advantage of the linguistic differences and saw in the mainland writing system a distinct and powerful aspect of discourse that took on a life of its own, so to speak, and often resisted easy reclamation by native speakers unfamiliar with the rules of the game. Scriptive anarchy was prevented, of course, since without vocal and syntactic reclamation, the written text was apt to be quite meaningless (the vocal aspect, often in ritualized settings, was emphasized because of the valorization of kotodama).
Belief in the magical power of poetry might help explain why a particular poem appears in different ("prose") contexts. We shall see that in most premodern Japanese texts poetic forms coexist discursively with "prose." Prose-poem "hybrids" are not uncommon in the West but discursive or generic boundaries tend to be preserved, and the phenomenon occurs with much less frequency. I wish to emphasize not only that song-poems appear together with "prose" but also that a particular poem or variation often appears indifferent "prose" contexts. An early example of what I mean occurs in Kojiki (NKBT, 315):
On another occasion (mata arutoki), the emperor [Yuryaku] climbed Mt. Kazuraki; here (koko ni) a great wild boar appeared; when the emperor immediately took an arrow with a whistling tip and shot at the boar, that (sono) animal became angry, and charged straight at its enemy [or, alternatively: let out a roar]—read the three characters U-TA-KI for their sound value—sthereupon (kare) the emperor, fearing that (sono) boar's charge, climbed up a hari tree; here (koko ni) he recited a poem [sang]: Stately ruler of the land, I, an illustrious sovereign, was out hunting; frightened by the charge of a wild boar, a wounded boar, I climbed up into the branches of a hari tree, high upon a hill. And thus he sang. [yasumisisi/waga oFokimi no/asobasisi/sisi no yamisisi no/utaki kasikomi/waga nigenoborisi/ariwono/Fari no ki no eda]
The Nihon shoki text, couched in its usual kambun style, gives a different version of the incident. While the emperor is out hunting, a strange bird signals a warning to the participants. At that moment an angry boar appears, and as the other hunters climb trees in fright, the emperor orders an attendant (toneri) to shoot the animal. The attendant also climbs a tree and loses his senses. Meanwhile, as the boar charges at him, the emperor takes his bow, stabs it, and then raises a leg and stomps it to death. The emperor orders an end to the hunt, and as he is about to kill his cowardly attendant, the latter composes the above song-poem and recites it as he faces his death. The empress is deeply moved, and in lofty language persuades the emperor not to kill the attendant. On his way back to the palace, the emperor happily states, "Everyone else hunts birds and beasts; I return home having obtained marvelous words." The Nihon shoki version transforms a cowardly (therefore eminently human) Kojiki emperor into a wise and benevolent Confucian sage who listens to those around him and favors the acquisition of verbal lessons over wild game. Its kambun rhetoric incorporates an impressive array of citations from Chinese texts. Modern scholars rescue the emperor in the former version from a comical reading by claiming that rather than making fun of the emperor, the passage actually praises the hari tree for its fortuitous presence.
