The story of Koreanic
01
A Family of Two, Recognised Late
Koreanic was never discovered in the way Indo-European or Austronesian were. For most of its history Korean was treated as a language without relatives, and the question was not which languages belonged with it but whether it belonged with anything else. The family's second member was recognised only gradually, and its status is still argued over.
Visitors had long noticed that Jeju Island spoke differently. Kim Jeong, exiled to the island in 1520, wrote that he had to learn its speech 'like a child learning a barbarian language', and later officials from Seoul described needing interpreters. Korean scholarship nonetheless classified Jeju speech as a dialect of Korean, in line with the view, held in both Koreas, that the nation is defined by a single language.
The case for two languages rests on mutual intelligibility. In 2007 the provincial government adopted the name Jejueo ('Jeju language') in its language ordinance, and in 2010 UNESCO listed Jeju as a separate, critically endangered language. William O'Grady, Changyong Yang and Sejung Yang then tested mainland Koreans on recorded Jeju speech: listeners who had not grown up on the island understood very little, a result comparable to Dutch speakers listening to Norwegian. Ethnologue and Glottolog now list Jeju separately, while Korea's National Institute of the Korean Language still treats it as a dialect.
The history of Korean itself was put on a firm footing in the 20th century. Ogura Shinpei's work on the hyangga poems (1929) and Lee Ki-Moon's Gugeosa gaeseol (1961) set out the stages still used today: Old Korean, known through Chinese characters; Middle Korean, recorded in Hangul from 1446; and Modern Korean. Because Jeju split from the mainland only in the last thousand years, Middle Korean texts serve as a close stand-in for the ancestor of the whole family.
02
From Manchuria to the Peninsula
Where Koreanic began, and when it reached the peninsula, are among the least settled questions in East Asian prehistory. The earliest texts are late: Silla inscriptions and poems of the later first millennium CE, written in Chinese characters that hide much of the pronunciation. Everything earlier has to be inferred.
Most specialists place the family's origin to the north, in southern Manchuria or the Liao River basin, and think it spread south through the peninsula during the first millennium BCE. One strand of evidence is the place names of the former Goguryeo territory, recorded in the 12th-century Samguk sagi. Some of them appear to contain Japonic words, which Vovin, Whitman and others take as a sign that Japonic languages were spoken on the peninsula before Koreanic displaced them. Others read the same names as Koreanic, or doubt they can be read reliably at all.
Archaeology offers candidate movements but no certain match. Whitman links the arrival of Japonic in Japan with the rice-farming Mumun culture of southern Korea, which would leave Koreanic as the later arrival, perhaps with the bronze- and iron-using cultures that entered from the Liaoning region late in the first millennium BCE. Martine Robbeets and colleagues (2021) instead tie Koreanic to the spread of millet farming from the Liao River region several thousand years earlier, as part of their Transeurasian hypothesis. That proposal combines linguistic, archaeological and genetic data, but its linguistic side has been sharply criticised, and many linguists do not accept that the dates can be read from the vocabulary.
What is clear is that by the time of the Three Kingdoms the language of Silla was Koreanic, and that Silla's unification of 668 made its speech the ancestor of everything spoken today.
03
Verbs Last, Particles After
Korean and Jeju share a grammar that a reader who knows Japanese or Turkish will find familiar. The verb comes at the end of the clause, and everything else precedes what it modifies: adjectives and relative clauses before nouns, and objects before verbs. Grammatical relations are shown by particles after the noun, such as the subject marker -i and the object marker -(l)ul, which the two languages share. Words are built by adding suffixes one after another, and a single verb can carry markers for honorific subject, tense, mood and the relationship between speaker and listener.
That last layer is a hallmark of the family. A Korean sentence ends in one of several 'speech levels' chosen for the listener, and Jeju has an equally elaborate set of sentence enders, including endings that mark how the speaker knows what they are saying, whether they saw it or heard it from someone else. O'Grady and his colleagues give the Jeju question Emeng patang-i sy-e? 'Is mother at the sea?', in which almost every word differs from Korean, although the structure is the same.
In sound, both languages distinguish three kinds of stop at each position, none of them simply voiced: a lax, a tense and an aspirated series. Korean tal 'moon', t͈al 'daughter' and tʰal 'mask' differ only in the first consonant. Middle Korean had vowel harmony, dividing vowels into two sets that could not mix within a word, and traces survive in the choice between verb endings such as -a and -e.
Both languages are written in Hangul, the alphabet devised at King Sejong's court in the 1440s. Jeju spelling keeps the letter ㆍ (arae-a), which Korean dropped in 1933, because Jeju still has the vowel it stands for.
04
Relatives, Dialects and Dates
The largest open question is whether Koreanic has any relatives. The old Altaic hypothesis grouped Korean with Turkic, Mongolic and Tungusic, sometimes adding Japanese. Most historical linguists abandoned it in the late 20th century, arguing that the shared features came from contact and typological coincidence rather than common descent. Robbeets's Transeurasian proposal revives a version of it with new data, and remains a minority view. A narrower link with Japonic, argued by Samuel Martin and John Whitman on the basis of sound correspondences, is taken more seriously, but Alexander Vovin, after long work on the question, concluded in 2010 that the shared vocabulary is better explained by borrowing on the peninsula.
Within the family, the questions are about the boundary between language and dialect. Jeju's separate status is accepted by most linguists outside Korea and by the island's own government, but not by the national language institute. The same question could be asked of some mainland varieties. The Yukjin dialect of far northeastern Korea, spoken in parts of North Korea and among ethnic Koreans in Central Asia, is quite distinct from the standard, but how far it is intelligible to other Koreans has hardly been tested.
The dating of the Jeju split is also uncertain. Jeju came under direct mainland rule in the early 12th century, and its speech was clearly distinct by the 1500s, but whether the divergence began then or reflects older regional differences is unknown.
Finally, Old Korean is poorly understood. Lee Ki-Moon proposed a chain shift in the vowels between Old and Middle Korean, and others have offered competing reconstructions of the earlier vowel system. Because the Old Korean evidence is written in Chinese characters, each of these reconstructions depends on how the Chinese readings of the time are themselves reconstructed.