The Nanbanjin Nikki


Beyond man’yôgana: characters as art in the Man’yôshû

There’s a widespread but simplistic notion of the evolution of writing in Japan that goes like this:

This narrative misconstrues or obscure several facts:

This is not an scholarly essay; I just want to cite some illustrations of the last bullet-point, because I find them as perplexing as they’re fun. Readers who need more are pointed to the references.

Basic sinographic techniques in the Man’yôshû

Though the writing of the poems is intended to convey Japanese utterances, the overall framework of the collection (commentaries, notes, settings, prose passages) is in Classical Chinese, then the standard form of writing in Japan. Æsthetically, this must have caused the impression of an harmonic combination of registers, like a musical composition with alternating tempos. I believe it’s still a matter of debate whether the Chinese parts were read as Chinese, within the tradition of ondoku “sound-reading” recitation, or if they were read as Japanese by using the kundoku techniques for simultaneous translation, or both; in any case, the difference in inscriptional style is not meaningless.

The two essential techniques used for the Yamato register were morphography and phonography, the latter divided in ongana and kungana (see below). The inscription style of the Man’yôshû flows and changes notably in different sections; phonogram writing predominates in books 5, 14, 15, 17, 18 and 20, totalling 1033 poems (out of some 4500+). The remaining poems are in mixed script: morphograms for lexical items and phonograms for inflexions and particles. This follows earlier usage such as senmyôgaki, and is also a predecessor of the modern kanji-kana majiribun practice in Japanese writing.

Ongana (“sound kana”) refers to the Sino-Japanese phonographic readings already established in the period, then based on a relatively late form of Early Middle Chinese (EMC) with some Korean (Paekche) influence (Miyake’s “System D”)—the same system of the earlier Kojiki phonograms. These readings were used to approximate the sounds of two varieties of Old Japanese (OJ): Central, and the Azuma (or Eastern) dialect. As a typical example of mixed writing with morphograms and ongana, poem #1118 has the sequence 「有險人母」。 It reads like this:

Type Morph. Phon. Morph. Phon.
EMC reading (*wuw) *kɨamʰ (?) *məw
Read as ari kěmu pito mo
Meaning exists suffixes: -kǐ [+past] -mu [+probability] person particle: [+emphasis]

The other kind of phonogram, kungana or “translation-kana”, is a somewhat roundabout way of representing sounds, necessitating a two-step mental operation. To illustrate, imagine a reader working with #81. At a certain point she meets the sequence 「見鶴」. The first character is employed morphographically for the verb mi(ru) “to see”, as in modern Japanese, but interpreting the second one takes more work. The reader must first think of the Chinese character as a morpheme, and translate it to a Japanese equivalent; 鶴 is taken semantically as “crane”, which in contemporary OJ was turu. Next, the reader discards the meaning, and reinterpret the phoneme sequence as some other morpheme altogether; in this case -tu-ru, a past suffix inflected in attributive (rentaikei) form. The whole thing read mituru “did see [+attributive]”. Polysyllablic words can be written with a composite of kungana: 「酢堅」 is “vinegar, hard”, which translates to su+kata = sugata “shape” (#778).

The kungana are in minority in the Man’yôshû, and hardly ever appear in other works. In general, and in contrast to later kana, both ongana and kungana can stand for more than one syllable or mora; e.g. ongana 足 (EMC *tsuawk) for the two syllables suku in 足尼 sukune “Lord”, or kungana 村 “village” for mura- “group of” in 村山 murayama “many mountains”.

As for morphographic writing, it’s most often used to represent native Japanese words, like the examples above of 見 for mi(ru) “to look” or 山 for yama “mountain”. Rarely, they’re also used for Sino-Japanese (SJ) words: 餓鬼 gakï “hungry ghost” (#608), 布施 puse “alms” (#906). The reason SJ was avoided is that it still amounted to Han loanwords, and the register here was of Yamato songs, so it was a bit of a stylistic faux pas. (On an unrelated note, these two particular loans are traceable beyond Chinese, to Sanskrit preta and dāna respectively.)

Playful writings

The task of deciding, for each character, whether it should be read as a morphogram, ongana or kungana would be difficult enough as-is, but the Man’yôshû hardly stops there. We should keep in mind that “writing” (sho 書) meant something different for this civilization than it does for us; it was an integrated ensemble of textual composition, calligraphy, inscription, and even papermaking and scroll-mounting. As such, things like the pun-like nature of kungana or the puzzle of deciphering intended uses are just the tip of the iceberg; what to us are inscription-level flourishes were an integral and important part of the classical poetics. LaMarre has noted numerous parallels between techniques at different layers of expression; things like juxtapositions, pivots, and pillow-words were practicised in the same spirit as the incriptional play with characters. It’s a shame that, for reasons of convenience as well as familiarity, modern editions tend to reduce the Man’yô to its strictly linguistic/phonetic register (and even “update” the original OJ phonetics!), which does violence to the whole thing.

The “playful writing” of the Man’yôshû is called in modern Japanese gisho or tawamuregaki; Seeley calls them “derived writings”. Ômori et al. is one attempt to classify gisho play, proposing as its main categories:

Other than these three Ômori categories, there are even more ways of writing in the Man’yôshû, often requiring subtle cultural inferences to work. A few examples from Seeley include:


In light of the above, I find it hard to agree with theorists who say Chinese writing was an obstacle, hindrance, or handicap to the Japanese, which was supposedly overcome only with the creation of kana. Clearly Yamato scribes were quite at ease with Chinese characters.

The alphabetic conventions used in this post are:


I also consulted these helpful online editions of primary sources:


This is nice. I totally support your opening comments. The notions about the development of Japanese writing that you present seem to me to have their origins in ideology and politics. It’s a view that adopts an idealised picture of growth from Chinese beginnings to the final development of a Japanese writing system that can (superficially at least) be ‘divorced’ from its Chinese roots and stand on its own two feet. Like all nationalistic narratives, it needs to simplify history to make the facts fit the ‘story’.

Personally, it actually took me quite a while to realise how intertwined the Chinese and Japanese writing systems are. Like many others, I took at face value the neat way that the modern Japanese language has been systematised. It was only after I came to China that I gradually came to realise how much of what is treated as quaint or irregular in Japanese is actually a direct reflection of Chinese. There is a lot of character usage that can’t be understood without reference to Chinese. The whole situation is summed up best by the Kanwa Jiten — that very strange beast, a dictionary of Chinese, written in Japanese!

At any rate, since I’ve never studied much about ancient Japanese writing (it always seemed too old or dusty or difficult), so it’s nice to see what you’ve presented here.


[...] from ichibigo, of unclear meaning but definitly unrelated to covers or trays). Giving the courtly taste for complex writing and Shônagon’s notorious affinity with Chinese characters, I’m guessing that it must be [...]

[...] so crudely utilitarian. Au contraire, from the beginning, it was precisely the complexity that delighted the Japanese (or they’d have long switched to kana or bonji). And remote classical poetry [...]