A comment on an answer to anoher question about Lithuanian suggests that 'quite a large number of words was borrowed from non-IE languages'.

While some words in Sanskrit indeed seem to have Fenno-Ugric (or Paleo-Asiatic) cognates, I am curious about the rest of the language families.

What is approximate distribution of the suggested loanwords in Sanskrit by language families?

  • 1
    T. Burrow's The Sanskrit Language has some words on this, IIRC. And my memory is that even with the largest estimate of speculated words of non-IE origin, it would be absurd to say that this is any significant fraction. Commented Jun 27, 2017 at 15:27
  • There's another question specifically about Uralic (Fenno-Ugric etc) loans into Indo-Aryan (or Iranian), without a positive answer (none), so I paid some attention to conclude so far that any mention I saw speaks of loans into--not from--a Uralic branch.
    – vectory
    Commented Feb 11, 2020 at 20:53
  • Are there any directly attested IE languages without significant non-IE substrate? Commented Feb 14, 2020 at 5:19

2 Answers 2


Sanskrit and wider the Vedic language really had a non-IE substratum, or, to be more precise, no less than three substrata: Dravidian, Munda, and one more called "unknown Language X". Apart from T. Burrow's The Sanskrit Language starting from page 374, here are more sources on that:

  • an article on Dravidian substratum. The page navigation is in Tamil, the "next page" button is "அடுத்த பக்கம்" in the bottom right corner of the page;
  • a list of Sanskrit lexemes of Munda origin.
  • 2
    The question was about the number/proportion: what's given at the first link ranges from 4% to 2%, argued by some to even be 0%. Commented Jun 28, 2017 at 23:25

Adding on to Yellow Sky's answer, at the Indo-Iranian stage there is theorized to have been a substrate language spoken where the Bactria–Margiana Archaeological Complex (BMAC) is located in Central Asia, through which migrating Indo-Iranians passed. A good selection of religious terminology, agricultural names, and taxonomic names present in both Old Iranian (Avestan, Old Persian) and Vedic Sanskrit are pinned to the BMAC substrate language. Notably sóma (Avestan haoma) is thought to have come from there.

The Indo-Iranian Substratum by Alexander Lubotksy further answers this question.

  • Despite having "openaccess" in its site name, the Leiden repository does not allow access to the linked paper. And it does not speak clearly about this, the interface just calls itself again and again ... Commented Feb 11, 2020 at 15:34
  • @jk-ReinstateMonica when I click the link a PDF downloads. Try a modern browser?
    – Aryaman
    Commented Feb 11, 2020 at 16:45
  • @Aryaman, your comment is pointlessly snarky and factually wrong (I replicate the problem on my machine, and my browser is very modern). However, I think the problem is that the leiden page just pushes the file to your machine without asking or telling.
    – user6726
    Commented Feb 11, 2020 at 16:57
  • @jk-ReinstateMonica No snark was intended (and I think the content of my answer has enough merit to not be downvoted?) Regardless, maybe openaccess.leidenuniv.nl/handle/1887/2691 is better accessible for you. Leiden is very much an open access institution.
    – Aryaman
    Commented Feb 11, 2020 at 17:24
  • 1
    @user6726: What a user experience! Now I have a dozen copies of the paper in my Downloads directory. I'd expected a normal download dialogue, with the possibility to immediately display the pdf ... Commented Feb 11, 2020 at 17:36

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.