As the overview article noted, Hashimoto Shinkichi’s other discovery is what’s known as “jōdai tokushu kanazukai” (Old Japanese kana usage). This article checks what it is, using actual words.
First, with our own modern ears
“Yuki” (snow) and “tsuki” (moon). In both, the last sound is “ki.” Any speaker of modern Japanese will feel these two “ki” to be the same sound.
Yet in Nara-period books, these two “ki” were never written with the same character.
There was no hiragana or katakana yet in the Nara period. To write the sounds of Japanese, people borrowed the sounds of Chinese characters and used them like a syllabary — “man’yōgana.” Let’s look at the examples Hashimoto gives in Kodai Kokugo no On’in ni Tsuite (1942).
| Word | Characters used to write “ki” | Group |
|---|---|---|
| yuki (snow), kimi (you), kinō (yesterday), akiraka (clear) | 伎, 企, 枳, etc. | Type A (kō) |
| tsuki (moon), kiri (fog), tsuki (zelkova) | 紀, 奇, etc. | Type B (otsu) |
Hashimoto’s observation is this. For the “ki” of yuki, any of 伎, 企, 枳, and so on will do — but 紀 and 奇 are not used. For the “ki” of tsuki, any of 紀, 奇, and so on will do — but 伎, 企, and 枳 are not used. The two groups never mix.
Not just “ki”
This sorting isn’t limited to “ki.” Hashimoto showed that the same split exists in these 13 kana.
e, ki, ke, ko, so, to, no, hi, he, mi, me, yo, ro
Of these, “e” is a slightly different case — it corresponds to the distinction between the “e” of the a row and the “e” of the ya row. Each of the remaining 12 kana splits into two groups. Hashimoto named these groups “kō” (Type A) and “otsu” (Type B).
Counting this way, Hashimoto says the sound distinctions used in Nara-period texts add up to 87 types in all.
Why can we say the sounds were different?
A question arises. If it’s just a sorting of characters, couldn’t it be a mere writing habit? Hashimoto tested this against grammar.
Take verb conjugation, for example. The “ki” in “kaki” (the continuative form of “kaku,” a four-grade verb) and the “ki” in “oki” (the continuative form of “oku,” “to rise,” an upper two-grade verb) are, by Hashimoto’s findings, written with different groups of man’yōgana. Each kind of word consistently takes its own group of kana; it isn’t the whim of individual authors. If it were only a writing habit, such orderly correspondences wouldn’t arise. It is natural to conclude that because the pronunciations were different, the sorting was kept consistently, word by word and conjugation form by conjugation form.
What the sounds actually were is still debated
So what were those two sounds, really? Here, Hashimoto himself was cautious. In the booklet he notes that several views exist about their identity, that he is researching it himself, and that he can’t yet state anything conclusive (we read this part in the close-reading article). Later researchers interpreted the two groups as a difference in vowels and developed as a leading view the idea that Nara-period Japanese had eight vowels, more than the modern five (that development is covered in this feature’s legacy and influence article). The actual nature of the two sounds, however, still has several views today.
Summary
| Point | Detail |
|---|---|
| Content of the discovery | Kana that sound identical in modern Japanese (ki, ke, ko, and so on) were strictly sorted into two groups of man’yōgana in the Nara period |
| Scope | The 13 kana e, ki, ke, ko, so, to, no, hi, he, mi, me, yo, ro; 87 sound distinctions in all |
| Why we can say the sounds differed | The sorting is kept regularly, by word and by conjugation form |
| Unresolved | The actual nature of the two sounds still has several views today |
In a nutshell — Nara-period speakers told apart, as two different sounds, what we now feel to be the same. The fact was preserved in the way they used their Chinese characters.