According to the Wikipedia article on Koreanic languages:

Among extant languages, Korean is considered by most linguists to be a language isolate and by others as part of the widely rejected Altaic family or the Dravido-Korean languages.

And the Wikipedia article on [the Altaic family] states that:

Altaic (/ælˈteɪ.ᵻk/) is a proposed language family of central Eurasia, now widely seen as discredited.

Yet from my personal experience living in Japan, Japanese and Korean seem very close. Korean speakers tend to learn Japanese very quickly, and the usual explanations given are because they are very similar languages, or because they have similar grammars.

Moreover, I speak Japanese, and I usually recognize Korean when I hear it because I always think it is Japanese being spoken until I realize that I don't understand what is being said.

So why aren't Japanese and Korean considered part of the same group? Why is the Altaic hypothesis dismissed? Why is Korean considered an Isolate?

  • 6
    I would credit the areal influences that sumelic mentions, and even include simple coincidence. Areal influence is particularly strong in phonetics. Khakas and Tuvan sound a lot like Mongolian, Korean and Chukchi, not like Turkish, but they are in fact fairly closely related to Turkish, and not related to these other languages. – user6726 Dec 5 '16 at 23:01
  • 5
    They appear to form at least a Sprachbund, but any "genetic" relation must predate the bounds of the comparative method, which is about 6-8000 years. – jlawler Dec 6 '16 at 4:23
  • 2
    Japanese and Korean both have lots and lots of loanwords from Chinese. – fdb Dec 6 '16 at 11:05
  • 2
    Just a note regarding wikipedia. The wikipedia article seems to be very biased and leaning towards the opposers of the Altaic language family. It says it is discredited and gives a feeling that nobody believes in it anymore. In reality there are quite many major publications (e.g an Altaic Etymological dictionary from Brill) out there. It is not uncommon that Wiki-articles do not coincide with the cocensus or that the information is outdated. – Midas Dec 7 '16 at 8:25
  • 2
    It is definitely widely seen as discredited. Widely does not necessarily mean a consensus or even a majority view however... – curiousdannii Dec 7 '16 at 11:23

Remember "isolate" doesn't mean "shown to be unrelated to any other language". It means "not shown to be related to any other language" (in a sufficiently convincing manner to establish a consensus).

Basically, Korean is considered a language isolate because modern linguists expect relatedness to be demonstrated by showing there is a significant amount of vocabulary (possibly including morphemes, not just entire words) in each language that can be derived from a common source via a regular set of sound changes for each language.

Figuring out these sound changes, and determining the form of the common vocabulary, is the process of reconstructing a Proto-language.

For Japanese and Korean, despite the superficial similarities, nobody has constructed a convincing proto-language. Wikipedia mentions some of the criticisms of proposed Altaic sound correspondences between languages.

There are good reasons for requiring evidence of common vocabulary. Areal effects have a large influence on the pronunciation of languages, and they can also affect grammar.

For example, the Basque language doesn't have any common ancestry with Spanish, as far as we can tell, and if it did they certainly must have diverged far earlier than Spanish and Romanian. However, Spanish pronunciation shares some features with Basque that it doesn't share with Romanian:

  • Spanish and some dialects of Basque have 5-vowel systems /a e i o u/. (Some Basque varieties have vowel systems more similar to French, with /y/ and nasal vowels. This is another example of areal effects.) Romanian has 7 main vowels /a e i o u ə ɨ/.
  • Spanish and Basque contrast a trilled rhotic /r/ and a tapped/flapped rhotic /ɾ/. Romanian has a single rhotic phoneme /r/.
  • Basque and some varieties of Spanish have the retracted sibilant [s̺].

An example of an areal grammar feature is the "have"-perfect in Western European languages. The verbs involved (reflexes of Germanic *habjaną and Latin habeo) are not cognates, so it's impossible for the structure to be inherited from a common ancestor.

As Midas points out, different people have different viewpoints, and "dismissed" probably has an overly strong connotation of a final judgement. It's appropriate for Wikipedia and any other source trying to be "neutral" or conservative to classify Korean as an isolate, since the mainstream view (apparently; I'm not an expert) is that it has not yet been shown to be related to any other language group with a high degree of certainty. But different linguists have different leanings toward "splitting" languages apart or "lumping" them together. It's possible and perfectly legitimate for someone to think the languages are likely to be related even though we can't definitively demonstrate it yet.

| improve this answer | |
  • Thanks. Most of the responses including yours, seemed to focus on the fact that I think that they sound similar. But there's also ease of learning: doesn't that hint at some genetic relationship? To use your example: Spaniards find Romanian much easier to learn than Basque. – Alex Kinman Dec 8 '16 at 19:43
  • 3
    @AlexKinman: Well, ease of learning can be based on shared grammar features that are caused by areal convergence. It's not obviously related to genetic relationships. Most English speakers seem to find Spanish easier to learn than German even though German is more closely related to English from a historical perspective. – brass tacks Dec 8 '16 at 19:48

My mother tongue is Chinese and I speak neither Japanese nor Korean yet I don't think they really sound close. I will definitely not mistake Japanese speakers as speaking Korean.

It's very easy for "close" languages to borrow and share a lot of features. An example I remember quite well is the pronunciation of "r" in became quite similar in French and German over time (for a lot words at least), even though those languages fundamentally don't have much to do with each other, however that doesn't make them belong to one language family.

So the issue in a large part is how do you separate features borrowed from neighboring, but essentially unrelated languages from features that stemmed from a common language family. It's true that Japanese and Korean share certain common features (e.g. grammar) between themselves and also with some other languages in the area, for example Manchu Language, but the sources of such similarities are debated.

Also, the Altaic hypothesis is largely based on the idea that a large portion of the ancestors of modern Korean and Japanese came from north/northeastern Asia (and then they mixed with the population coming from further south to form their respective races), which might be true to a large extent. However, even if that's true, since the geographical conditions of those two places were largely isolating (especially Japan), and also since there were another large portion of the population who came from completely different places (e.g. from Southeastern Asian islands in the case of Japan), they might well have developed a lot of their own features during the tens of thousands of years so that it would be hard to still classify them as belonging to the same language family. Apparently all human beings eventually came from the same place... So at what point of a language's evolution do you draw the line on language families is a somehow subjective issue actually.

| improve this answer | |

The Altaic hypothesis is far from dismissed. There are publications out there regarding the Altaic family and there are still voices within major institutions (e.g. Brill) speaking of it. Wikipedia as a source for linguistic concensus has deteriorated over the past years. I am not for or pro the Altaic hypothesis, I am just stating that "dismissed" is largely exhagerated. You could say that the Altaic hypothesis containing Turkic, Mongolian and Tungusic is far more accepted than the Macro-Altaic hypothesis including Japanese and Korean.

| improve this answer | |

Combined with molecular anthropology, these language families have one super cognate"eye". Tai-Kadai:mata Austronesian:mata Austroasiatic:mat Japonic:ma Hmong-Mien:muejh Sino-Tibetan:muk

Mogolian:nud nivkh:nangay Koreanic:nun

| improve this answer | |
  • 2
    Welcome to Linguistics! This post would benefit from adding further details. Being a one-line post, it may attract downvotes and criticism. Please edit it to add further relevant information — preferably with references to credible sources. – bytebuster Dec 7 '19 at 17:08
  • you need to know that "cognate" by definition excludes "loan"; this is a difficult difference even for Indo-European. To answer this question with a oneliner there are however not enough similarities, they say. The word mat- is so widespread without any transformation that it's likely a loan. For n there is: Arabic ayn, Akkadian inu; Apalai onu, Quechua nawi, Nawajo anaa, Newar na kha, Cherokee agadoli (Marathi dola, Georgien tvali); Ewe ngku; There's even Gujarati nen, Hindu nain, Sanskr. nayana, or AGr omma, mata, ophthalm-, also Sumerian IGI. This is fun! – vectory Dec 10 '19 at 3:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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