The vowels el and oi are developed by the widening of the mouth aperture as
e1<
i1 and
o1<
u. Vowel syllables, however, have developed as '
i1<
ye1<
ye and '
u<
wo1<
wo. These sound changes took place not so long before the eighth century.
Since the duration when Old Japanese was available is rather short, older lexical evidences do not always show older forms but may be forerunners of innovation. The widening is proved by pairs of words which have suffered other sound changes. In the case of
e1,
fuki1 (blowing) and
fuye (flute) can be linked with intermediate *
fui1, which is the starting-point of the widening as well as the goal of “
i-reduction.” In the case of
o1 simo1 (below) and the first morpheme of
siduye (low branch) have the common ancestor *
simu, which explains both the widened vowel and the disappearance of
mu with the vocalization of the following consonant. The rules of hiatus avoidance and alliterative epithets also support the inference.
A popular theory of the origins of
e1<
i1+
a and
o1<
u+
a cannot hold good. The only word which is said to be evidenced,
ari (be), is suspicious with the very initial
a, because it violates the hiatus rule. We had better think that an allomorph
ri was used in those compounds and the form
ari is a variant of
ri with a parasitical appendant vowel.
View full abstract