Commenting on Friday's post about Japanese braille, Akaki Kuumeri expressed interest in hearing more about kanji braille. Kanji are, predictably, a nightmare to implement in Braille and it seems that most blind people here don't even bother with them much-- but I am bound by the vow I swore on the skull of the very first No-sword blogger (who, for those who came in late, was shipwrecked on this apartment building seven generations ago) never to fail readers seeking information about Japanese orthography, and so I have prepared this post.

Let's do this chronologically. KAWAKAMI Taiichi (川上泰一) was the first to publish a usable system for representing kanji in braille-style dots. He called it kantenji (漢点字), and based it on (4,2) blocks rather than the (3,2) braille standard. This made it incompatible with existing braille equipment, but on the other hand it also meant that it could be used alongside the Japanese braille system rather than simply replacing it -- just like kana and kanji, in fact.

The Japanese Kantenji Association (日本漢字点字協会) has a few examples of how it works on their What are kantenji? page. Here's the basic idea: the bottom three rows of the (2,4) block correspond to the standard (2,3) Japanese braille block, and the two extra dots on top are used to mark "radical 1 [of a two-block kanji]" (*-), "right radical" (-*) or "single-block kanji" (**).

In the example they give, adding the ** above the braille for /ki/ gives the kanji 木 ("tree", native Japanese reading /ki/). Adding the ** above the braille for /me/ gives you 目 ("eye", /me/). Adding the *- above the /ki/ and the -* above the /me/ and combining the two gives you 相 ("mutual", "aspect", etc., never pronounced /ki/ or /me/ but drawn as a combination of 木 and 目).

Naturally, this system could never suffice for all the possible kanji a blind person might want to write or read, and Kawakami's monograph Literature and the blind (盲人と文学) outlines his specifications for expanding the set of possible representations. It isn't for the timid: entirely new radicals are invented to route around certain bottlenecks, and in some ways it seems more like a reinvention of kanji than a mere classification of them.

In the opposite corner, HASEGAWA Sadao (長谷川貞夫) published his "six-dot kanji" (六点漢字) system in 1972. Hasegawa's autobiography is online, and contains a lot of interesting nerd lore about his attempts to get things automated and accessible, plus other interesting anecdotes, like the one about trying to to persuade the National Language Commission (国語審議会) to remove the reading mekura from their official list of tōyō kanji readings. Mekura is an old word meaning, via the blunt etymology "eye-dark[ened]", "blind person"; as Hasegawa explains:

Historically, it was not always used as a discriminatory term (差別語), and I certainly do not believe that it should be removed from dictionaries or anything like that. But it cannot be denied that today it is used [in a discriminatory way]... The tōyō kanji readings (当用漢字改訂音訓表) are the basis of usage in official documents; radio, television, newspapers, magazines and other media; and education... Can this discriminatory term be permitted a place ... in materials so fundamental to the national language's orthography?

Anyway, Hasegawa's system used six-dot blocks, and it took three blocks to write one character. This site has the best explanation I found online of the system. It works like this:

The first block signaled "kanji of type X coming up", where X might be "single-mora kanji" or "kanji whose pronunciation ends in /N/" or perhaps "kanji pronounced /ko:/: variation 2". (There are a lot of kanji pronounced /ko:/.)

The second two blocks usually gave the first parts of the Sino-Japanese and native Japanese readings of the kanji, in that order. So, for instance, to write 子 (pronounced /si/ in SJ and /ko/ in nJ), you write:

     --  *-  -*
     -*  **  *-
     -*  -*  -*
  kanji: si  ko

In other words, Kawakami's system was based on how the kanji look, and Hasegawa's on how they sound. Both are fine approaches to the problem in principle, although it seems to me that Hasegawa's makes more sense for people to whom appearance is irrelevant.

So, friends, those are the two competing Japanese systems for representing kanji in braille. I don't know about you, but as a sighted person who uses only his eyes to read, I now feel like a total underachiever.

(Also, check out Joel's overview of the Braille family tree at Far Outliers, and this position statement about eight-dot braille in which the Braille Authority of North America make not a single bone about their plans to "monitor developments".)