Proto-Japonic words are generally polysyllabic, with syllables having the form (C)V.
Consonants The following Proto-Japonic consonant inventory is generally agreed upon, except for the values of and (see below): Scholars agree that the Old Japanese voiced consonants
b,
d,
z and
g, which never occurred word-initially, are derived from clusters of nasals and voiceless consonants. In most cases, the two consonants were brought together by loss of an intervening vowel. A few words display no evidence for a former vowel, and scholars reconstruct a syllable-final nasal of indeterminate place preceding the voiceless obstruent, as in > Old Japanese > Modern Japanese 'grain', > OJ > MJ 'knee'. These nasals are unrelated to the
moraic nasal of later forms of Japonic, which derive from contractions or borrowings from other languages such as
Middle Chinese. The other Old Japanese consonants are projected back to Proto-Japonic except that authors disagree on whether the sources of Old Japanese
w and
y should be reconstructed as glides and or as voiced stops and respectively, based on Ryukyuan reflexes: • Southern Ryukyuan varieties have corresponding to Old Japanese
w, e.g. Miyako 'I' and Yaeyama 'stomach' corresponding to Old Japanese and . Two dialects spoken around
Toyama Bay on the west coast of Honshu also have corresponding to initial in other Japanese dialects. • Yonaguni, at the far end of the Ryukyu island chain, has in words where Old Japanese has
y, e.g. 'house', 'hot water' and 'mountain' corresponding to Old Japanese , and . Some authors, including advocates of a genetic relationship with Korean and other northeast-Asian languages, argue that Southern Ryukyuan initial and Yonaguni are retentions of Proto-Japonic voiced stops and that became and elsewhere through a process of
lenition. However, many linguists, especially in Japan, prefer the opposite hypothesis, namely that Southern Ryukyuan initial and Yonaguni are derived from local innovations in which Proto-Japonic and underwent
fortition. The case for lenition of - > - is substantially weaker, with the fortition hypothesis supported by Sino-Japonic words with
Middle Chinese initials in also having reflexes of initial in Yonaguni, such as 'vegetables' from Middle Chinese (). An entry in the late-15th-century Korean annals
Seongjong Taewang Sillok records the local name of the island of Yonaguni in
Idu script as 閏伊是麼, which has the Middle Korean reading , with glossed in the text as the Japonic word for 'island'. That is direct evidence of an intermediate stage of the fortition - > - > -, leading to the modern name 'Yonaguni'.
Vowels Most authors accept six Proto-Japonic vowels, which are as follows: The vowels , , and have been obtained by internal reconstruction from Old Japanese, with the other Old Japanese vowels derived from vowel clusters. The mid vowels and are required to account for Ryukyuan correspondences. In Old Japanese, they were raised to
i and
u respectively except word-finally. They have also left some traces in
Eastern Old Japanese dialects and are also found in some early
mokkan and in some modern Japanese dialects. The other vowels of Old Japanese are believed to derive from sequences of Proto-Japonic vowels, with different reflexes in Ryukyuan and Eastern Old Japanese: In most cases, Proto-Japonic corresponds to Old Japanese
i2. Proto-Japonic is reconstructed for Old Japanese
e2 in the few cases that it alternates with
o2 (2 seems to be limited to specific monosyllabic nominal stems such as
se~
so2 'back',
me2~
mo 'seaweed' and
ye~
yo2 'branch'.
Prosody The
Japanese pitch accent is usually not recorded in the Old Japanese script. The oldest description of the accent, in the 12th-century dictionary , marked pitch using a variation on the notation for
Middle Chinese tones. The usual interpretation, following
Haruhiko Kindaichi, treats the most common markers, corresponding to the
Middle Chinese level and rising tones, as low and high pitches respectively. The reconstruction of
S. Robert Ramsey inverts these values. Regardless of their values, the different tone patterns in the group words into accent classes that generally correspond to those obtained by comparative reconstruction from modern mainland Japanese dialects. In contrast, Ryukyuan languages share a set of accent classes that cut across the mainland classes. For example, for two-syllable nouns, the has five accent classes, which are reflected in different ways in the major accent systems of mainland Japanese, here represented by Kyoto, Tokyo, Oita and Kagoshima. Ryukyuan languages, here represented by Kametsu (the prestige variety of the
Tokunoshima language), show a three-way division, which partially cuts across the five mainland classes. In each of the modern varieties, the pattern of high and low pitches (in some cases falling or rising) is shown across both syllables and a following neutral particle. Thus to account for the various reflexes, each of the classes 2.3, 2.4 and 2.5 must be split in two, yielding eight classes of disyllabic nouns. In some Ryukyuan dialects, including Shuri, subclass (a) is marked by a long vowel in the first syllable instead of a distinct pitch pattern, which led Hattori to suggest that the original distinction was one of vowel length. == Lexicon ==