Though there are many theories grouping Korean and Japanese in the same family, none of these are widely accepted by linguists.

Yet the grammars of these two languages are extremely similar in many ways:

  • topic + comment
  • word order
  • use of particles
  • agglutinative verb morphology but simple noun morphology
  • two kinds of adjectives: noun-like and verb-like
  • no definite or indefinite articles, no distinction between singular and plural, use of counters
  • complex honorifics systems

Are there other cases where two languages not considered by most linguists to be in the same family are grammatically similar to such a degree, or are these two languages special?

  • 2
    Define the set of properties (parameters maybe), find that data for all world languages, define a distance measure using those properties, compute the 'distance' between every pair of languages, then do principle components analysis. Then you'll see how the languages clump together and if there is something still special about Korean/Japanese.
    – Mitch
    Sep 20, 2011 at 22:23
  • 1
    I notice you tagged the question with sprachbund. Perhaps the examples at en.wikipedia.org/wiki/Sprachbund would be relevant? Sep 20, 2011 at 23:26
  • @Mitch: Yes of course that's the simple method I used to discover that Japanese and Korean were similar. I'm surprised nobody noticed it before (-; Sep 21, 2011 at 7:27
  • 1
    @Anschel: One reason I'm noticing this is because I'm travelling at the moment, currently in the Balkans which possible has the most famous sprachbund, but here the similarities are just bits around the edges compared to the Japanese/Korean case. Sep 21, 2011 at 7:29

3 Answers 3


The following features are characteristic of Insular Celtic, a subfamily of Indo-European spoken in Great Britain and Ireland. They are also characteristic of the so-called Hamito-Semitic languages of North Africa, a subfamily of Afro-Asiatic. Not every Celtic or H-S language has all of the features, but they are fairly typical of the family as a whole, especially in their most archaic variants (e.g. Old Irish).

  1. Conjugated prepositions: [Prep + Pronominal Obj] is a single word.
  2. Word order: VSO, N - Modifier, Prepositions
  3. Relative clause linker: invariant particle, not relative pronoun
  4. Relative clause technique (oblique): copying, not gapping, e.g. "the bed that [ I slept in it ]".
  5. Special form of the verb peculiar to relative clauses.
  6. Polypersonal verb (subject and object both marked).
  7. Infixing/suffixing alternation: Object marker is infixed to the verb if there is a preverb, suffixed otherwise.
  8. Definite article in genitive embeddings may occur only on the embedded noun: "house [the-man]" = the man's house.
  9. Nonconcord of verb with full-NP subject: verb can fail to agree with the subject, depending on word order.
  10. Verbal Noun (VN: object in genitive), not Infinitive (object in same case as with finite verb).
  11. Predicative particle: in copular or nominal sentences, the predicate is marked with a participle homophonous to a "local" preposition: "He (is) in a farmer" = He is a farmer.
  12. Prepositional periphrastic: BE + Prep + VN, e.g. "He is at singing".
  13. DO periphrastic: DO + VN, e.g. "He does singing".
  14. Notional adverbial clause expressed as "and" + finite clause.
  15. Nonfinite forms usable instead of finite main-clause verb.
  16. Word-initial phonological change, expressing a variety of syntactic functions.
  17. Idiomatic use of kin terms in genitive constructions, e.g. "son of sending" = messenger, "son of land" = wolf.

(Source: Orin Gensler, A Typological Evaluation of Celtic/Hamito-Semitic Parallels, Berkeley, 1993.)

Note that this distinctive feature-complex is not typical of other Indo-European languages, and does not seem to be present in Continental Celtic. This has led to the supposition of a substrate/sprachbund situation in the ancient British isles: the language spoken by the inhabitants before the arrival of Celtic was in heavy contact with Afro-Asiatic speakers and shared these distinctive features, in the usual sprachbundy way; this pre-Celtic language influenced the Celtic language imposed by the Indo-European conquerors in the usual substratal way.

No other language groups of the world seem to share this set of 17 features to any statistically significant degree; this is unlikely to be the working-out of typological universals.

  • In item 3 of your list, you mean "particle" and not "participle," right? Also, could you say more about what you mean by "statistically significant" in the last paragraph?
    – Aaron
    Sep 21, 2011 at 6:52
  • 1
    "Statistically significant": There are two things we have to look at. First, there are certain correlations that seem to occur across the world's languages. These are called "typological universals," but they're really more high-probability tendencies of features. The classic example is that VO languages tend to have prepositions, OV languages have postpositions. If any of these 17 features tended to "group together" in this way, they would be less useful as independent markers of similarity.
    – librik
    Sep 21, 2011 at 7:08
  • The other important thing to look at is, of course, whether there are other languages that share the whole feature-complex. This is the question of coincidence. Gensler studied grammars of 78 languages across the globe and scored them on the presence of these features. The Celtic and H-S languages were grouped at the top of the list sorted by score, but at #8 was Chumash, an extinct Native American language of Southern California. By doing the statistical analysis on the probability of each feature's occurrence, one can conclude that you'd expect one such language just by random chance.
    – librik
    Sep 21, 2011 at 7:17
  • 2
    Do you know if any correction was done for having selected these 17 features to compare with the foreknowledge that they were the same in Celtic and H-S? That would definitely affect the significance level of any tests.
    – Aaron
    Sep 21, 2011 at 7:29
  • 1
    @librik: This is the most fascinating answer I've seen on the site since it began. Thank you. Sep 21, 2011 at 7:32

Agglutinating Old Turkish (Uyghur) and Mongolian are similar to the extent of affixes being the same though quite sure these languages are not in any way related. Being no expert of Mongolian, I can't provide details.

  • What do you mean by "affixes being the same"? Do you mean moreso than when also compared with Hungarian, Finnish, Korean, and Japanese? Apr 24, 2014 at 0:11
  • I know neither of these languages, so I wouldn't know, but what I mean is that though there can be no relation proven, certain affixes are phonetically same or near-same. Since this was just a somewhat passing reference in a course on Uyghur Turkish, I don't recall the details.
    – zwiebel
    Apr 24, 2014 at 7:20

A sort-of example is English and French (or English-Spanish or English-Italian, ...)

This isn't a truly valid example because they're both Indo-European languages. But the Germanic and Romance languages are from completely different branches of the Indo-European family. Genetically, English and French are as unrelated as English and Hindi. Indeed, the ancient Romans and Germanic tribes were not aware that their languages were related at all.

The strong similarities between English and French (and between the Germanic and Romance languages in general) are entirely due to 2000 years of contact.

However, it still seems the case of Korean and Japanese is particularly interesting since there wasn't as great a level of contact between ancient and medieval Japan and Korea beyond the common influence of Classical Chinese. It's also interesting that Korean and Japanese are far more similar to each other than either is to modern Chinese, when all three are unrelated languages.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.