In the previous article, we saw the theory that Nara-period Japanese had eight vowels, more than modern Japanese. Another strong piece of evidence supporting that theory is a regularity called “Arisaka’s Law,” which Arisaka Hideyo found in the 1930s.
Note that the contents here are explained as well-known facts of Japanese linguistics. They contain no quotation from Arisaka’s own paper.
The unit called a “root”
To explain this law, we need the word “root.” A root is, for now, the core of a single word. A word like kokoro (“heart”), which can’t be divided into further meaningful parts, can be thought of as one root. A compound made by joining two words, like yama (“mountain”) + michi (“road”), contains two roots.
What the law says
What Arisaka found is this.
In Nara-period Japanese, within a single root, the Type B o vowel (ö) does not appear together with any of the following three vowels.
| Within a root, Type B o (ö) and… | Do they coexist? |
|---|---|
| the vowel a | No |
| the vowel u | No |
| the Type A o vowel (o) | No |
In other words, a root containing a Type B “o” either stays among other Type B “o” sounds, or is combined with the vowels i or e. It doesn’t ordinarily sit alongside a, u, or the Type A “o” within one root.
For example, the word kokoro (“heart”) is written in Nara-period man’yōgana with Type B o-column kana for all three of its syllables (ko₂·ko₂·ro₂). It is a root in which only Type B o-column sounds line up — an example that fits the law.
Why is it evidence of a difference in vowels?
This law matters because it is a regularity in how vowels combine with one another. If the A/B difference had been a difference in the head consonant, not the vowel, there would be no reason for the “Type B o” to have such a regular relationship specifically with other vowels (a, u, and the Type A o). But if the Type B vowels are independent vowels with their own “compatibilities” with other vowels, the regularity is easily explained. It means that a phenomenon reminiscent of vowel harmony in other languages (where the vowels in a word line up by some quality) existed in Nara-period Japanese too.
A caution
The law holds within a single root. In compounds (two roots joined) and in loanwords that entered later, these vowels can sit side by side. And the exact scope of the law and its exceptions have been debated in various ways by later research. What this article introduces is only its basic skeleton.
Summary
| Point | Detail |
|---|---|
| Content of the law | Within a single root, the Type B o vowel does not coexist with a, u, or the Type A o |
| Example | Kokoro is written entirely with Type B o-column kana — only Type B sounds line up |
| Significance | As a regularity in how vowels combine, strong evidence that the A/B difference was a difference in vowels |
| Caution | Holds within one root; in compounds and loanwords these vowels can sit together |
In a nutshell — within a single root of Nara-period Japanese, there were vowels that never sat alongside the Type B “o.” That regularity backed up the A/B difference as a difference in vowels.