Dating proto-languages is obviously something we can't do precisely, but we can offer reasonable ranges. For example, Proto-Indo-European can't really be much younger than 5 millennia, and let's say that 9 is an upper bound that would satisfy even the Indo-Anatolian and Anatolian Urhemeit proponents.

For Proto-Afrasian, Wikipedia gives an incredibly wide range of 7500 BC - 16000 BC, with dates before 10000 BC seemingly contested by lexical evidence. In any case, this is certainly older than PIE. Among generally accepted and proven language families (eg. not anything strictly containing Indo-European) what are the others with a remarkable time depth? Are there any which are possibly as deep or even deeper than Afrasian?

  • 1
    The Trans New Guinea 'phylum' has been hypothesised to have begun spreading around 10,000 BP, in association with the first development of agriculture on the island of New Guinea. Aug 24, 2015 at 11:33
  • 1
    I find this exercise a bit odd. Any proto-language we can name had ancestors we cannot name, and none of those we can name is dated to anywhere near the theorised time of origin of human language. Feb 13, 2018 at 16:48
  • 2
    @A.M.Bittlingmayer: It wasn't an excercise in naming proto-languages though, but rather a question about the greatest depth currently achieved by the comparative method. This is why I specifically asked about well-established language families and not about their named or unnamed ancestors.
    – user54748
    Feb 13, 2018 at 21:58
  • 1
    @user54748 Makes more sense. In that case then it is largely a function of how many divergent branches of the primary language family happened to survive or at least leave a record. Feb 15, 2018 at 10:25

3 Answers 3


For the true believers there is proto-world (known under different terms, too) dated somewhere between 100.000 and 200.000 BC. Greenberg and Ruhlen even dare reconstruct some proto-world words like *dik "finger".

If you buy Greenberg's idea of Amerind, it must be around 15.000 BC.

Proto-Niger-Kongo (not yet constructed, but assumed to be feasible to reconstruct) is ca. 10.000 BC.

On more secure grounds, there are:

Proto-Sino-Tibetan: 6.000 BC
Proto-Austronesian: >4.000 BC

I could not find dates for proto-Tai-Kadai or proto-Pama-Nyunga. Proto-Dravidian is surprisingly young, estimated at 500 BC.

  • 3
    Various authors have suggested dates for proto-Pama-Nyungan ranging from 4,000–6,000 years BP, so a long way short of the time depths the OP is interested in. Aug 25, 2015 at 10:40
  • 1
    If it existed, Proto-World is oldest by definition, and in the same manner is any hypothesized proto-language older then it's descendants :) Thank you for the data on Proto-Niger-Kongo and the rest though, and @Gaston for the Pama-Nyungan. I thought it would be older than that too.
    – user54748
    Aug 25, 2015 at 16:32

Groups that could possibly contest such dates could be: a) "Old European" (call it whatever you like) in the sense of a group of languages being spoken in central and west Europe before the spread of Celtic and Germanic. b) The group that included the languages (eastern mediterranean substratum?) that were spoken by the first farmers of Anatolia who also crossed into Europe (esp. Crete and the Greek mainland). However, let me raise a caution as we have not enough material to reconstruct those yet, at least not in the same level as we have for Proto-Afrasian. All you got is the pre-Celtic substratum (100 words?) on one end and pre-Greek / Linear A / Eteo-Cypriot on the other end (good luck understanding those). Any dating would be hypothetical. Also, I would be very careful with whatever theories include Basque and Etruscan, their grouping and dating.


Check Appendix B in The Tree of Language by B. Pelletier at Amazon Look Inside. It's all there.

  • 3
    We are looking for self-contained answers here, so please add the most relevant points from the reference to your answer. Sep 10, 2018 at 11:49

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.