Proto-Ryukyuan is the reconstructed ancestor of the Ryukyuan languages, probably associated with the Gusuku culture in the early second millennium AD.^{[citation needed]}

The modern Ryukyuan languages are spoken on the Ryukyu Islands, from the Amami Islands to Yonaguni. All Ryukyuan varieties are endangered; many speakers are aged late sixties or older, while younger speakers are only monolingual in Standard Japanese.^[1]

Pellard (2009:249-275) gives a list of innovations in mainland Japanese and Ryukyuan, with some redocumented by Pellard (2015:15). For instance, in Ryukyuan, the general word for "body" is *do C, and has been grammaticalized into a reflexive pronoun. The Ryukyuan languages also exhibit a semantic shift "intestines" > "belly" of PJ *wata B. However, Japanese also has some innovations not in Ryukyuan, such as the word otoko "man" < "young boy", kami "hair" < "top".^[2]

There are some innovations shared with Ryukyuan and Kyushu dialects that have not been found in other mainland Japanese dialects. For instance, Yōsuke Igarashi (2018) claims that an innovation of Kyushu-Ryukyuan is to change kami-nidan verbs (-i(2)-) to shimo-nidan verbs (-e(2)-), a grammatical change of -kara from a ablative marker to a locative marker, and some vocabulary items (usually species) only found in such dialects. However, Pellard (2021) attempts refutes the hypothesis, citing typological and cross-linguistic reasons.

The following consonants can be reconstructed for Proto-Ryukyuan:^[3]

• Proto-Japonic *-p- generally lenites to *-w-, as in PJ *kapa 'river' > PR *kawa 'well'^{[citation needed]} It is irregularly preserved in some words, mostly adjectives, which lead Thorpe to suggest a geminate consonant blocking lenition.^[4][5] However, nouns that don't sound emphatic also receive this irregular conservatism.
• Approximants in proto-Japonic preceding a high vowel are merged to a zero consonant in proto-Ryukyuan, such as PJ *upai 'above' > *uwe > PR *ue. ^[6]
• No Ryukyuan dialects preserve the yotsugana distinction; in this case, it means that older *di ~ *zi and *du ~ *zu merge as *zi and *zu.^[7]
  • One possible exception is that a special word for "to take off (clothes)" exists in very few Miyako dialects^[7] (< *padok-): e.g. Hirara /padukɿ/.^[8][9] The regular conventions of proto-Ryukyuan would require an irregular change in the consonant *zu- > *do-.
• In various foreign transcriptions of Old Okinawan, some words exhibit pre-nasalized obstruents, and some dialects also have nasalization before voiced consonants, suggesting that Proto-Ryukyuan had pre-nasalized obstruents.^[10]

The following vowels can be reconstructed for Proto-Ryukyuan:^[11]

All Ryukyuan languages have raised the mid-vowels *e and *o, but not all have merged these sounds with *i and *u. It is even possible the mid-vowels were already raised in Proto-Ryukyuan, but still distinct from the original high vowels. The dialects go through different developments depending on the preceding consonant. In various Northern Ryukyuan dialects, *i will often palatalize the preceding consonant.^[12] To give an example, Shuri ʔitɕi 'pond'^[13] < PR *ike, but Shuri ʔiku- 'how many?'^[14] < PR *eku.

Some Old Okinawan texts can preserve the distinction of Proto-Ryukyuan mid-vowels. For instance, the Old Okinawan anthology Omoro Sōshi records the word for "snow; hail^[15]" as yoki 15 times, while yuki is only recorded once. This may suggest that the proto-form of such word had a mid-vowel *o.^[16]

Proto-Ryukyuan merged the Proto-Japonic diphthong *əi > *e, as in PJ *kəi "tree" > PR *ke "id.", PJ *əkəi- "to get up" > PR *oke- "id."

Proto-Ryukyuan has at least three reconstructed tone classes, classified as class A, B, and C respectively. Class A regularly corresponds to the initial high register in Middle Japanese.

The correspondences of class B and C are somewhat complex. While both can regularly correspond to the initial low register in Middle Japanese, there exists a split that exists for the following low register accent classes in Middle Japanese:^[24] class 2.3, 2.4, 2.5, 3.4, and 3.5. Accent classes 3.6 and 3.7 almost always correspond to class C in Ryukyuan.

Most dialects often have a penultimate tone on a class C noun, such as the Kametsu dialect in Toku-no-Shima, the Nakijin-Yonamine dialect of Nakijin, and the Tarama-Nakasuji dialect of Miyako (in carrier phrases only). On the other hand, some dialects, such as the Asama dialect in Toku-no-Shima, and the Shuri dialect in Okinawa, exhibit long vowels of the penultimate syllable; for Shuri, it is only exhibited in disyllables.^{[citation needed]}

There has been no unproblematic explanation for why there has been a split in classes B and C in Proto-Ryukyuan for the aforementioned accent classes, so this split has been typically projected back to Proto-Japonic.^[25][26] The Kishima dialect of Saga has been reported to have a tonal split in class 2.5 nouns that correspond to the Ryukyuan tone class split.^[25]

Many Ryukyuan dialects have a conclusive and adnominal form that do not correspond straightforwardly to the Japanese ones. They are often palatalized, and such cases have been viewed to be derived from an infinitive *-i + *wor- "to be; to stay". Various scholars attempt to propose a common origin for this and the Japanese form -u, but it is problematic.^[27]

In some constructions in Northern Ryukyuan dialects, the word tends to use a different adnominal form, which can be compared to the Eastern Old Japanese and Hachijō adnominal -o, implying such could be reconstructed at the Proto-Japonic level.^[30]

The origins of the Southern Ryukyuan forms are more difficult to establish.^[30]

As with verbs, Ryukyuan adjectival forms are not cognate with the Japanese ones; they are derived from either the nominalizer *-sa ro *-ku + the auxiliary *-ar- "to be".^[31]

Thorpe (1983) reconstructs the following pronouns in Proto-Ryukyuan. For the first person, the singular and plural are assumed based on the Yonaguni reflex.

• *a, 'I' (singular)
• *wa 'we' (plural)
• *u, *e 'you' (singular)
• *uja, *ura 'you' (plural)

Pellard (2015) reconstructs the following cultural vocabulary words for Proto-Ryukyuan:

• *kome B 'rice'
• *mai A 'rice'
• *ine B 'rice plant'
• *momi A 'unhulled rice'
• *mogi B 'wheat'
• *awa B 'foxtail millet'
• *kimi B 'broomcorn millet'
• *umo B 'taro, yam'
• *patake C 'field'
• *ta B 'rice paddy'
• *usi A 'cow'
• *uwa C 'pig'
• *uma B 'horse'
• *tubo A 'pot'
• *kame C 'jar'
• *pune C 'boat'
• *po A 'sail'
• *ijako B 'paddle'

• Vovin, Alexander. 2024. Reconstruction of Japonic and para-Japonic based on external sources. In Handbook of Historical Japanese Linguistics, by Bjarke Frellesvig and Satoshi Kinsui, 11-37. Berlin: De Gruyter Mouton.

• Kenan, Celik (2024). "UniCog: A Framework Proposal for the Dynamic Compilation of Comparative Data for the Reconstruction of proto-Ryukyuan". NINJAL Research Papers (in Japanese). 26. doi:10.15084/0002000156.
  • Supplementary data doi:10.15084/0002000162 Contains lexical and bibliographic data.
• JR-COGNATES (version 7): list of words attested in Japanese and Ryukyuan, including reconstructed tone classes and the distribution range in Ryukyuan.