One of the oldest splits within Indo-European was between the Centum and Satem languages; they differ in what they did to the "accented velar" phonemes (like * and *ǵ).

However, if I understand right, the Anatolian languages split off some time before the Centum-Satem division happened.

So, what happened to "accented velars" in Anatolian?

3 Answers 3


Several points:
1. some people have put into question the distinction between the k, g, gh series (usually called "velars") and your "accented velars" *ḱ, *ǵ, *ǵh (usually called "palato-velars") in PIE. IMO this is hardly defensible, and the distinction is doubtless necessary.
2. the conventional view is that both velars and palato-velars become k or kk in Hittite. See for example Kloeckhorst's etymological dictionary of Hittite.
3. some people (like Pisani) have challenged that point of view. IMO there's indeed a problem with the conventional approach. And I rather believe that the so-called velars become spirants h and hh in Hittite.

  • Fascinating! Do you have any examples of PIE *k, *g, or *gh becoming Hittite h or hh? If so, that would round out this answer nicely (and I'd happily accept it).
    – Draconis
    Commented Nov 14, 2019 at 15:52
  • for *k, a good example is "blood" *H1es-rk => Hittite eshar vs Old Indian asrk. I think we don't need to posit that Hittite is from *H1esH2r. If *k > h is accepted, then the equation is perfect. Another item is Hittite *huis- "to live", which can easily be derived from *gweyH- "to be alive". Of course, if we accept the spirantisation of velars in Hittite, it changes the picture of Hittite etymology.
    – user23769
    Commented Nov 15, 2019 at 7:23

Whether you believe that *k and *ḱ merged to a single phoneme in Anatolian, or that *k split into *k and *ḱ after the separation of Anatolian from IE - in both cases you will get the same result.

  • Not quite the same result: the three-way distinction still discernible (just) in Luvian is best explained a merger within Anatolian (see Draconis’ new answer). Commented May 10, 2020 at 6:12

Melchert argues that the "plain velars" and "accented velars" remained distinct in Proto-Anatolian, because they show different reflexes in Luvian: *k becomes Luvian k, while * becomes Luvian z.

He cites as an example the Luvian pronoun zi- "this" from PIE *ḱis, cognate with Hittite ki-, Latin cis, OCS , English he, etc.

  • 1
    The claim originated with Melchert (as far as I know), but I don’t think it’s really a case of just “Melchert argues” anymore – in my experience, most Indo-Europeanists accept the three-way distinction in Luvian nowadays. It seems as close to communis opinio as anything in Anatolian ever seems to get. Commented May 10, 2020 at 6:11
  • @JanusBahsJacquet That's fair! In that case I'll probably move the accepted answer to this one—nothing against Fournet's answer, but if there's a decent scholarly consensus on this, that's what I was looking for.
    – Draconis
    Commented May 10, 2020 at 16:51
  • Before you do accept it please read: Melchert, H. Craig "Luvo-Lycian Dorsal Stops Revisited" (Online: UCLA, 2011); Available from Author through UCLA at linguistics.ucla.edu/people/Melchert/melchertopava.pdf (Accessed: 19/3/2021). Melchert has revised his opinions!
    – Ned
    Commented Jun 24, 2021 at 12:59
  • @Ned Very interesting! I need to examine this more closely, but to my understanding he still argues for a three-way distinction in Proto-Anatolian, since the *Ḱ series underwent a conditioned change that the *K series didn't? They merged afterward, but this change can't have happened before Proto-Anatolian or we'd see it in Hittite.
    – Draconis
    Commented Jun 24, 2021 at 17:05
  • 1
    @Ned To me it still seems like solid evidence—you can't have palatalization affect Ḱ and not K unless they're distinct at some point later than Proto-Anatolian, after all. Even if they merged in some environments, the fact that they didn't merge in all environments still requires a distinction to be reconstructed.
    – Draconis
    Commented Jun 25, 2021 at 20:22

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.