Turkic vocalism revisited

Previously I wrote about the system of Proto-Turkic vowels as consisting of 8 qualities spanned by features of height, frontness and rounding with three additional phonemes sometimes suggested: /ia/, /e/ and /*ë/.


The first phoneme is reflected identically to *a, except it caused palatalization in Chuvash, so that we have[1]:

  • Chuvash юр : Turkmen gār (: Mongolic *kïrmag) ‘snow’.
  • Chuvash шурӑ ‘white’ : Turkmen sāry (: Mongolic *sïra) ‘yellow’.
  • Chuvash чул : Turkmen dāş (: Mongolic *čïlaxun) ‘stone’.
  • Chuvash шур : Tatar саз ‘swamp’ (?: Mongolic *sïruxaï ‘earth’).
  • Chuvash юм ‘sorcery’ : Tuvan хам ‘shaman’.
  • Chuvash юн : Turkmen gān ‘blood’.

Marcel Erdal proposes instead of reconstructing *ia, that Bolghar Turkic went through the breaking of long *ā and palatalization in Chuvash (and Mongolian) is thus secondary. Generally I reject the idea that Chuvash and Mongolian have any shared innovation among each other and feel that his proposition is falsified by Turkmen-Chuvash pairs such as: тула- ‘to bite’:  la- ‘to rob’, ту : dāg ‘mountain’, gāz : хур ‘goose’. Still, while I reject the change in this direction, it makes sense that *ia would monophthongize into a long *ā in Common Turkic and it seems supported by the data, so the correlation between *ia and *ā should be accepted[2].


Now we can turn to the second phoneme, *e, which is explained by Erdal through a sound change parallel to ā into ia, as raising of long *ǟ to e. Again we can note that there are cases of long ä that aren’t raised, but that *e generally tends to be long. Hence I would propose that *e itself derives from a diphthong *iä that is symmetrical to *ia. Alternatively, perhaps *iä developed into the long low-mid front and the original long *ǟ into the high-mid vowel.

/*ë/ and short /*e/

This *ë (written *ạ in EDAL) stands for the correspondence between ы /ɯ/ in Chuvash and Yakut and /a/ everywhere else. It is the weakest of the proposed extra phonemes since it has no special reflex in Mongolic and can be explained through different processes. Still I think a case can be made for it.

The Chuvash phenomenon is sometimes explained through the regular change of /a/ into /u/ with later delabialization into /ɯ/, however this account cannot explain the case found in triplets such as Hungarian ír : Chuvash ҫыр : Common Turkic *yaz- ‘to write’ or Hungarian tinó : Chuvash тына : Common Turkic *dana ‘steer’ for the simple reason that the raising of *a into Chuvash /u/ took place many centuries after the Hungarians moved to the Pannonian basin.

But we also cannot assume early raising of *ë into /ɯ/, since medieval Volga Bulgarian inscriptions show these words with an /a/. Instead, this is probably a case of a parallel development in Hungarian and Chuvash. Like inherited Finno-Ugric *ë, Hungarian raised this sound to *ï and later merged it with *i.

If the accord of Common Turkic *a with Yakut /ɯ/ really is correlated with Chuvash /ɯ/ (which is not completely certain), I would suggest that it is a development of an early allophone of *a (perhaps /a/ as opposed to /ɑ/) which arose in contexts not alike those in which some of the original short *ä developed into closed *e, perhaps having something to do with palatality, proximity to liquids or length of medial vowels.

Another interesting possibility suggested by the similarity of the ‘steer’ word to Proto-Indo-Iranian *dhainuš ‘heifer’ is that these represent *ai and *äi, but I would find it odd that this state of affairs wasn’t somehow represented in loans into Mongolic.

Other diphthongs

If we accept these diphthongs, we may also wonder about the irregular diphthongization occuring in Chuvash with rounded vowels  such as Chuvash кӑвак, тӑвар, ҫӑвар  : Turkmen gȫk ‘blue’, dūz ‘salt’, ýǖz ‘face’. Forms like these are very common, although they don’t seem to be regular[3].

Thus perhaps the Chuvash diphthongs could be original and (long) monophthongs in the other languages secondary. An interesting comparandum is Mongolic *nixür ‘face’ : Chuvash ҫӑвар, which is sometimes presented as evidence of Proto-Turkic *ń, but could (if real) maybe represent something like **niür.

Altaic situation

Proto-Tungusic had diphthongs such as *ia, *ua, *ui and *ue which tend to monophthongize, and perhaps further research could discover *ie or *iu, at least some words show unexplained alternation between *e and *i.

Mongolic is not reconstructed with diphthongs, but the phenomenon of “i-breaking” and “pre-breaking” and its presence in Khitan (together with umlaut), as well as some oddities of the vowel system (to be covered in a later post), could perhaps be best explain through an earlier diphthongal system.



[1] Raising of *a to Chuvash у (or dialectal о) is regular.

[2] Edal distinguishes between *ia and *iā, mostly based on some assumed irregular correspondences.

[3] It is hard to say what changes are “regular” in Chuvash. Rounded vowels seem to be reflected most commonly as reduced Chuvash ӑ/ӗ, and less commonly as diphthongs or full high vowels, there appears to be some imperfect correlation with Common Turkic length, although I haven’t yet tried to quantify it.

8 thoughts on “Turkic vocalism revisited

  1. This looks very promising to me!

    I presume, though, that you’re familiar with Stachowski’s explanation of the Yakut /ɑ/~/ɯ/ variation as Yakut-internal?

    I forgot: do Turkmen d- and g- imply *front vowels?


    1. I am. Ultimately we need to collect Turkic words with a(/ë) preserved in both Chuvash and Yakut and see if there’s a correlation.

      *k in back voweled words is always reflected as Turkmen g. Oghuz languages (sans Salar, which I’m starting to think isn’t really an Oghuz language) have a voicing distinction for alveolars in all words and velars in front words, although if this is an innovation it could have happened first in front words and only later in back words (only for alveolars).


  2. You might want to check your formatting: there seem to be several stretches of missing asterisks turned into italics.

    *tāla- could be an exception to breaking due to being a loan, either from early Ugric/East Uralic #Tåla- or Indo-Iranian *tsal- (or does II vowel length generally correspond to that in Turkic?). Ditto ‘goose’, if from IE in some fashion.

    Liked by 1 person

    1. ‘goose’, if from IE in some fashion

      That would be particularly interesting for implying rhotacism as opposed to zetacism. Unless it went Tocharian > East Turkic > West Turkic with etymological nativization at the last step…


    2. I wasn’t aware of the Uralic word. There are plenty of other words in Chuvash that demonstrate the proposed lack of breaking, but it is harder to find words which exclude later borrowing from Tatar or similar. This one was interesting because seemingly Chuvash preserves the archaic meaning ‘to harm’ which is found in Old Uyghur but lacking in other modern languages.

      I’m not sure about vowel length correspondence, IIr words in Turkic are somewhat nebulous themselves, there are some obviously Soghdic or Persian words and possibly some Saka in religious terminology, but beyond that seems poorly understood.

      Chuvash probably did go through a rhotacism, native *d went through a mid stage > *z before becoming r (Turkmen aýak : Chuvash ура ‘foot’, Turkmen gaýyn : Chuvash хурӗн ‘birch’), this rhotacism is also found in some Arabic loans.

      Anthropologically, I would expect agricultural vocabulary of this sort to have come from IE into (Pre-)Turkic during the Afanasevo (Pre-Tocharian)-Okunev contacts (or indirectly when Okunev people brought metalworking to Baikal), which should precede any splits in Turkic by two milennia.


Leave a Reply to David Marjanović Cancel reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.