I am interested in linguistics and how words spread from place to place. I have seen that there are two language families, and that there are signs that they might be related. Proto-Indo-Uralic is the hypothetical reconstructed language from Proto-Indo-European and Proto-Uralic, the respective ancestors of the Indo-European and Uralic languages. I'm just curious about if the two are related enough to be reconstructed. I'd personally say yes because Proto-Afroasiatic has been reconstructed with some difficulty: https://www.ucpress.edu/book/9780520097995/reconstructing-proto-afroasiatic-proto-afrasian

Link (see "Some possible cognates"): https://en.wikipedia.org/wiki/Indo-Uralic_languages

  • 2
    I don't understand what you mean by "could". I "could" show that Japanese is an Indo-European language if it were one. Are you asking if the evidence for Indo-Uralic is good enough that credible reconstructions can be done?
    – user6726
    Commented Nov 27, 2019 at 2:33
  • Any attempt to show that two language families are related can only be done through a (partial) reconstruction of their shared proto-language.
    – curiousdannii
    Commented Nov 27, 2019 at 3:05
  • 3
    The article you quoted says "most of the supporters of a relationship between Indo-European and Uralic have also supported their relationship to additional language families". i.e. they do not support an "indo-euralic" node within their larger hypothesis.
    – Colin Fine
    Commented Nov 27, 2019 at 8:53
  • indo-european.info/a-song-of-sheep-and-horses.pdf In this paper you can find the Schleicher's fable in Proto-Indo-Uralic.
    – Rock
    Commented Nov 27, 2019 at 16:55
  • Such a reconstruction nears pseudoscience, I would say, at least until new evidence for it is provided Commented Feb 18, 2020 at 17:25

5 Answers 5


Indo-European and Uralic are low-level families, the existence of which is not questioned by linguists and has been recognized for at least three centuries now, even by people with no particular training (merchants, soldiers, etc.), on the sheer comparison of word lists.

The issue of comparing PIE and PU is more challenging, as it involves non-obvious answers to difficult questions. The same is true for about all macro-comparative endeavors.

In my opinion, several answers are possible:

  1. PIE and PU are not related at all. This would mean that both proto-languages emerged out of nowhere independently. This hyper-nihilistic stance, though widespread among linguists, is very little probable.

  2. PIE and PU are close enough to establish a higher-level node: Indo-Uralic. This approach is quite fashionable among current Dutch scholars (Kordlandt, Kloeckhorst, etc). My personal opinion is that this may be based on a number of misconceptions and fake cognates that are borrowings, such as the fake equation PIE *wed- "water" => PU *wete "water". I would recommend caution as regards this theory.

  3. A third option is that PIE and PU are more or less ultimately related but PU is not the closest relative of PIE. That's my personal stance. My working hypothesis is that Hurro-Urartian and possibly also a number of Caucasic languages are closest to PIE. I must acknowledge that this is not the communis opinio among comparatists, but never mind.

In all cases, it must be emphasized that Bomhard's works on Nostratic have already amassed a bulk of data that cannot be easily dismissed. It can be ignored, but not dismissed. In all cases, the last shared ancestor of PU and PIE is of Paleolithic dating, most probably much, much older than 10 000 years Before Present. Which also explains why the exact nature of the relationship is not easily discernible.

  • 4
    What makes you confident that the "water" word is a borrowing?
    – TKR
    Commented Nov 28, 2019 at 19:21
  • That's an interesting question, with at least two facets. One facet is the way PU is reconstructed. The other facet is the sound correspondences between PU and PIE. So, first, I consider that PU *e was in fact a long vowel *e: Classical Uralistics rejects the idea that PU had long vowels, but I disagree for a number of reasons. Next, I believe that the correct comparandum for PIE intervocalic *d is PU affricates *c^ or *c', as in Greek *ked-ros = PU *kac^a "juniper", or PIE *ped- "foot" = PU *pac^a "to cross, move over". So *wed- = *wete is impossible in my system.
    – user23769
    Commented Nov 30, 2019 at 8:18
  • 2
    Interesting, thanks. I can't speak to the first point, but on the second, is there more data for the putative equation PIE d : PU č? κέδρος appears to have no clear IE cognates so could be a Wanderwort. (Also why do you specify "intervocalic"? The d isn't intervocalic in those PIE forms.)
    – TKR
    Commented Nov 30, 2019 at 18:57


We have only two proto-languages to compare, and the list of attested roots is already sparse. The set of potential cognates is even sparser and leaves not enough material for reconstruction. It is also hard to decide whether the cognate candidates are real cognates or chance coincidences.

Is there any chance to do better?

Maybe, by throwing in more potentially related language families into the comparison. There is a conspicious cluster of languages in Northern Eurasia characterised by the personal pronouns in M (1sg) and T (2sg). Maybe proto-M-T is reconstructable.

  • "personal pronouns in M (1sg) and T (2sg)" - you mean as in English me and thou?
    – Anixx
    Commented Feb 17, 2020 at 20:34
  • Yes, me and thou are examples of the M-T-pronouns. Because of the underlying method (you counts here as non-T), present day English is no longer M-T in the linked map. Commented Feb 18, 2020 at 12:59


The general consensus among linguists is that there's no compelling evidence for any sort of "Indo-Uralic" language family. Basically, the null hypothesis is that there's no genetic relationship, and nobody's ever presented enough evidence to reject that.

If one wanted to fully reconstruct "Proto-Indo-Uralic", they would need to show a relationship between the two families in the first place. Various people have tried over the years, but none of these theories have been particularly convincing, and it seems likely that the two families are truly unrelated (or rather, not related within the timespan that we can reconstruct with any confidence).

  • This is very much not the general consensus among Indo-Europeanists or Indo-Uralicists in my experience. Most will agree that there isn’t enough evidence to confidently reconstruct much due to the time depth, but the claim that they are completely unrelated is one that I don’t recall hearing anyone in either field make. There’s too much structural coincidence for that. Commented Aug 5, 2020 at 0:51
  • @JanusBahsJacquet agreed. This post seems to confuse lack of conclusive evidence for the affirmative with evidence for the negative. Most would claim that it is probable that they are genetically related, and plausible that they are more closely so than any of the other languages families to either of them, but nothing is indeed conclusive — attempting reconstruction is another matter.
    – Zorf
    Commented Aug 5, 2020 at 2:06

Yes, but only to certain limits.

While its completely obvious that there is no evidence or only very small evidence that Proto-Indo-Uralic actually existed, there are some attempts to reconstruct it. This paper attempts to reconstruct Schleicher's Fable in Proto-Indo-Uralic. Go to the page 30 and you will find this. Keep in mind that either Proto-Indo-Uralic vocabulary has not been accurately reconstructed yet, or atleast I don't know any Proto-Indo-Uralic vocabulary pdf paper.

  • Page 4 of the book is more important for the question; to quote a quote in a quote: "As far as the Indo-Uralic hypothesis is concerned, it is easily far more promising than most other hypotheses recently debated in [the Journal of Indo-European Studies], since even its alleged opponents call it “plausible but inconclusive” (Campbell & Poser 2008: 162), telling us that “you can believe in it if you want” (Koivulehto 1993: 189). (…)"; quoting [Kallio 2015]; But because the chosen citation formating of the book, I'm not sure which of the two [Kallio 2015] items in the books bibliography it is
    – vectory
    Commented Nov 27, 2019 at 20:25
  • . Also, this question would be worthwhile as more than a comment if a summary of reviews of the book were mentioned; unless there was none that's not mere polemic bile and gall. At least, it would be necessary to mention and compare concurrent efforts. Also, it doesn't hurt to go into detail: For example, one could excuse the stance admitted a few paragraphs earlier in the introduction about searching for an actually spoken language opposed to the algebraic approach, so to speak, as a necessary stance to justify ones effort.
    – vectory
    Commented Nov 27, 2019 at 20:47
  • No ad hominem: The collumn layout for abbreviation table, and the words accidentally \ in some sentence are mistakes (per Scott Aaronson about prejudices of quality work, who stresses more than one factor)--I say that as a guilty offender. The syncope p. 4 l. 1 looks like typographic shortcut, but same page implies multiple rewriting attempts: "In the nineties a decline as seen in the theory which proposes at least two strata of Indo-European (with the archaism of Hittite barely mentioned), with the most commonly used manuals barely presenting the effects of gradual dialectalisation."
    – vectory
    Commented Nov 27, 2019 at 21:09
  • 1
    The quoted paper crosses the boundary between serious scholarship and fantasy conlanging. I see it already on the conlanging side, despite the presence of a lot of serious reconstructions and scholarly citations. Commented Nov 28, 2019 at 13:20
  • 1
    @Rock: Among the conlang-ish elements in their Proto-Indo-Uralic I just want to mention the thrown-in word for "horse". This is far beyond the amount of speculation and imagination usually used in linguistic reconstruction. Commented Nov 28, 2019 at 14:06

From a cursory reading of a summary by Kortlandt (either in The expansion of the Indo-European languages, 2018, or something similar), I gained the impression that the two families were initially unrelated but later came into contact. That's my own impression, I don't mean to misrepresent the author. He is noted as Indo-Uralic author in Wikipedia's respective category--which doesn't mean much. He seems very careful not to submit to any premature judgement, but certainly shows interest in the comparison.

@Annix submitted under another question that proto-uralic had no number words, and took these from IE's.

@... (Midas?) has an unanswered question here asking for evidence of Uralic loans into Indo-Iranian (or later); Loans into the opposite direction are known. This probably has something to do with assuming that Indo-Iranians moved from the Steppe westwards around the Caspian--which is an unsettled matter of debate. Kortlandt also references indications that Uralic folk moved westwards into the Steppes "at the same time". Personally I'm not giving this two much weight for two reasons: a) The single Mother-Hypothesis is untennable, one dimensional thinking is but a necessary simplification--look for example at the account of Kurdish language in the Encyclopedia Iranica for a good deal of assymetrie b) The Steppe Hypothesis is convincing, but not the last word, because, how, while presence of artifacts is certain, Urheimat should be more than a nebulous idea, I don't know; Also, recent results (cf Gamkrelidze, ejectives, 2010) again favour early South Carpathian presence drawing on DNA analysis--which Kortlandt did not mention, iirc--of course, one shouldn't confuse Language with Genetics (incidentally, Kortlandt has written on the mathematical side of things, too, which I just discovered writing this question).

As an aside, you should ponder the question, if a child can come from two parents.

@Curiousdannii is correct in the comments, insofar any proof must imply that a proto language can be reconstructed, if you understand relation as single origin. However, since you seem to be aware that no such reconstruction has been achieved--and to be fair let's say that I have yet not read past the intro of "A Storm of Words", that wad linked by @Rock, and I would not understand half of it, but have to rely on the approximate judgement, in which I follow @Draconis' reservations--you are rather asking about methodology, and whether a proof can be constructed by other means.

Ultimately, it is programatic. Given the necessary difficulties that had to be expected, if the hypothesis were true in one form or other weaker ones, wonder, are enough people working on it to find out?

!image, now spread all over the web, originally from the guardian, I believe

You might as well wonder for the time being, in an absurd sense, when you look at your own family tree and how it seems to grow exponentially along the time axis, considering that twohundred generations back saw only ten 10, or perhaps 100 billion people on the planet, whether your parents were siblings and if you can reconstruct Askr and Embla (or Adam and Eve) from that.

  • 5
    An ouch! to the picture. There is no binary split between Indo-Iranian and "European" proposed by any serious historical linguist I know. "European" in the term Indo-european was always meant as a purely geographic designation, not as a proposal for a node in language evolution. Commented Nov 28, 2019 at 13:16

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.