The Japonic tree

The Japonic family is young by the standards of Indo-European or Austronesian. Its common ancestor, Proto-Japonic, is usually placed in the first millennium BCE, and most scholars tie its arrival in Japan to the Yayoi culture: wet-rice farmers who crossed from the southern Korean peninsula to northern Kyushu. Radiocarbon dates published in the 2000s pushed the beginning of the Yayoi period back to around the 10th century BCE, although many archaeologists prefer a later date and the transition took centuries.

Three kinds of evidence support the link. Archaeologically, Yayoi rice paddies, tools and burials have close parallels in southern Korea. Genetically, mainland Japanese people descend from both the earlier Jōmon population and a later continental population, the 'dual structure' model proposed by Hanihara Kazurō in 1991 and since confirmed by ancient DNA. Linguistically, the internal diversity of Japonic is shallow: a Bayesian study of basic vocabulary by Sean Lee and Toshikazu Hasegawa (2011) dated Proto-Japonic to roughly 2,200 years ago, consistent with a Yayoi-period spread.

Before Japonic, the islands were home to the Jōmon peoples, whose languages are unknown. Ainu, in Hokkaido and formerly in northern Honshu, may be one survivor; others left no clear trace.

The move into the Ryukyus came much later. Archaeology shows farming communities arriving from Kyushu around the 10th–12th centuries CE, in the Gusuku period, and genetic studies point to substantial migration from the north, although dating it precisely is harder. This leaves a gap: the Ryukyuan languages split from Japanese before the 8th century, yet their speakers seem to have reached the islands only centuries later. The usual answer, argued by Thomas Pellard among others, is that pre-Ryukyuan was first a separate dialect in Kyushu. Where exactly it was spoken, and when it left, remain open.

All Japonic languages are strongly head-final. The verb comes last, modifiers precede nouns, and grammatical relations are shown by particles placed after the noun: Japanese Hanako ga hon o yonda 'Hanako read a book' has ga marking the subject and o the object. There are no articles, no grammatical gender and no obligatory plural. Verbs and adjectives take chains of suffixes for tense, negation, politeness and mood, and adjectives inflect like verbs: Japanese takai 'is high', takakatta 'was high'. Speech levels and honorifics are built into the grammar.

The sound systems favour simple syllables of a consonant and a vowel, and timing is counted in morae rather than syllables: Nippon has four morae (ni-p-po-n). Most varieties have some form of pitch accent. In Tokyo Japanese háshi with a fall after the first mora means 'chopsticks' and hashí 'bridge'. Japonic also has rendaku, the voicing of the first consonant of the second part of a compound: te 'hand' plus kami 'paper' gives tegami 'letter'.

Some shared features are best seen historically. Old Japanese had kakari-musubi, in which a focus particle such as zo required the sentence to end in a special verb form. The construction had largely disappeared from Japanese by the 16th century but survives in Okinawan, where the focus particle du is its cognate, as in the saying nuchi du takara 'life is the treasure'.

Ryukyuan has innovations of its own: the raising of mid vowels, heavy palatalisation, glottalised consonants, and in Miyako syllables with no vowel at all.

For writing, Japonic has used Chinese characters from the start, along with the kana syllabaries developed from them. Okinawan court texts such as the Omoro Sōshi were written mainly in hiragana, and modern Ryukyuan writers use kana with extra conventions for sounds that Japanese lacks.

The deepest open question is whether Japonic has any relatives. The strongest candidate is Koreanic. Samuel Martin (1966) and John Whitman (1985) proposed sets of Japanese–Korean cognates, and the two languages are strikingly similar in grammar. Alexander Vovin, in Koreo-Japonica (2010), argued that most of the shared vocabulary is borrowing from a period when Japonic and Koreanic were in contact on the peninsula, and that the grammatical similarity is typological. A wider 'Transeurasian' family taking in Korean, Japanese, Tungusic, Mongolic and Turkic, defended by Martine Robbeets and colleagues (2021) with linguistic, archaeological and genetic data, is rejected by most historical linguists. Proposals linking Japonic to Austronesian or to Kra-Dai, notably Paul Benedict's 'Austro-Tai' (1990), have found little support. For now Japonic is usually treated as an independent family.

Closely tied to this is the question of Peninsular Japonic. Place names recorded in the Korean Samguk sagi (1145) for central Korea include apparent Japonic numerals and nouns. Christopher Beckwith (2004) took them as evidence that the Goguryeo language was Japonic; Vovin and Whitman read them instead as traces of Japonic spoken in the peninsula before Koreanic replaced it. Others doubt that the glosses can bear either interpretation.

Within the family, the reconstruction of Proto-Japonic is still debated. Old Japanese had eight vowels in its writing, but most scholars now reconstruct six for Proto-Japonic, with mid vowels e and o whose reconstruction rests partly on Ryukyuan evidence. How the Old Japanese 'second' vowels arose, and how many diphthongs to reconstruct, remain disputed.

Subgrouping is also contested at the edges. Some linguists see signs that the Ryukyuan languages share innovations with the dialects of southern Kyushu, which would mean that Ryukyuan is a sister of part of mainland Japonic rather than of the whole. The internal division of Ryukyuan into Northern and Southern branches is widely accepted, but the dating of each split is not.