Hans of Etymolist has a great post, “Thoughts on PIE *bhag,” resulting from his getting a copy of Nomina im Indogermanischen Lexikon (a title you pretty much have to be an Indo-Europeanist to find exciting, but for those of us with that particular kink, it’s exciting). He says “I embarked on reading it root-by-root. The first one is *bhag (NIL 1-2), and looking at the evidence for nominal derivations listed, I got a few ideas, which I’ll share below.” His summary is “it is possible to eliminate the root *bhag ‘share’ from the reconstruction of Indo-European, if one assumes that the Slavic and Tocharian cognates are actually loans from Iranian and that the Indo-Iranian and Greek cognates actually continue *bheg ‘break’,” but the fun is in the details, and if you like this kind of thing, well, this is the kind of thing you like, so go enjoy.


  1. marie-lucie says

    I enjoyed too

  2. John Cowan says

    Well, a is a doubtful proto-phoneme in PIE anyway, so the fewer of those, the better.

  3. Thanks for the thumbs-up! As I said in my post, I’d be grateful if anyone could point me to any prior discussion of the points I’m making; I can’t imagine that noone before me has come to those conclusions.

  4. marie-lucie says

    JC; a is a doubtful proto-phoneme in PIE

    Why is it doubtful? because it is rare to get definite examples?

    I have been working on a proto-language reconstruction (not in IE) in which only two vowels, and a, need to be reconstructed for the deepest, most ancient level, and a is much rarer than . I suspect that (later evolving into more differentiated vowels according to the surrounding consonants) was the unmarked vowel and a the marked one, as I have found some instances of minimal pairs with similar but not identical meanings. I suspect that this may have been the case with PIE (or pre-PIE) vowels.

    I remember reading about a modern case of a two-vowel system: in at least one language in the Caucasus (I don’t remember which one, or even which family) there are allegedly only two vowel phonemes: /a/ a vowel without allophonic variation, /∂/ a changeable vowel with distinct allophones depending on the consonantal environment.

  5. John Cowan says

    Why is it doubtful? because it is rare to get definite examples?

    Just so.

    in at least one language in the Caucasus (I don’t remember which one, or even which family) there are allegedly only two vowel phonemes

    The whole Northwest Caucasian family is like this. In Abkhaz, there are two vowel phonemes, /a/ and /ə/, and 58 consonant phonemes, including many palatalized and labialized ones. These are written using a following ь (the Russian soft sign) and ә (the Cyrillic schwa) as diacritics. Adjacent to these, /a/ is realized as [e] and [o] respectively, and /ə/ is realized as [i] and [u] respectively, and the orthography reflects this. This state of affairs is not thought to be primitive; earlier /ki/ underwent a shift to /kʲə/, pronounced [kʲi], essentially moving the front quality from the vowel to the consonant. Ubykh, which went extinct in 1992, also had two vowels and even more consonants, 81 to 84 according to different sources.

    There are a few other languages with vertical vowel systems in which front/back is not phonemic, notably Wichita (Caddoan) with three vowels, Margi (Chadic) with two, Arrernte (Pama-Nyungan) with two, and Marshallese (Austronesian) with three (four in loanwords). The short vowel system in Irish is also vertical.

  6. As far as I have heard, it’s just this “Caucasian” consideration that supports the no-a crowd in their definite exclusion of a’s. I think it’s the IE o that corresponds to the vertical a, the IE e to the vertical ə. If you look at it this way, there’s no place for another a at all.

    (The obvious caveat goes – don’t trust me much: I’m no indoeuropeanist in any way.)

  7. Trond Engen says

    Excellent discussion, Hans. I’ll be looking forward to your posts on every root in alphabetical order. Just too bad it’s on Blogger since I only rarely can be bothered to log in to comment. Maybe it’s time to get over it.

    I’m all for getting rid of *a, Or rather, I’m all for getting rid of a phoneme *a independent of *h2. My pet idea about the IE vowel system is that it had gone through diphtongisatiion, maybe similar to Scanian, so that stressed vowels turned into schwa + semivowel. If so, zero-grade simply continued the monophtong. The laryngeals would be the low and central counterparts to *j and *w.

  8. marie-lucie says

    I don’t know the details of the currently acknowledged history of the PIE vowel system, but in the proto-language I am talking about, the oldest reconstructible vowel system is a “vertical” one with ••∂, ••a, but the next oldest system has *e, *a, *o where *e and *o both derive from earlier **∂, the lowest proto-vowel **a having remained largely intact throughout the history of the language family. (Additional complications leading to the more complex vowel-systems of the modern languages can be attributed to the presence of another element, probably a ∂-glide, added in the original vowel nucleus, as well as to the continuing influence of surrounding consonants).

  9. @ Trond Engen: Sorry if I’ve raised hopes, but I’m not planning to comment on all roots in NIL, only on those where I think I have meaningful insights. We’ll see whetherI’ll have any more insights at all. 🙂
    On the PIE vowel system – I don’t have a strong position as to whether /a/ was a phoneme – I have no objections against it being a phoneme and I’d assume it at least for late PIE; as for older stages of PIE, well, a lot of things are possible.

  10. Siganus Sutor says

    In Persian (or Urdu), bagh means “garden”.

    Nothing to do with the matter? Oh, sorry then.

  11. Persian and Urdu!

    Nothing to do with the matter?

    I don’t know; I can’t seem to lay my hands on the etymology of the Iranian word, but it might be related.

  12. John Cowan says

    As far as I can make out, b­āgh is Common Iranian (Persian, Pashto, Kurdish) with no PIE etymology. It has spread into Ottoman Turkish, Azeri, Uyghur, Georgian, Armenian, and (per Nicoll’s aphorism) English, where it implies that the garden has shrubs and trees as well as flowers; the garden of the Taj Mahal is the Charbagh ‘four baghs’. Russian бахча is from the Tajiki Persian diminutive боғча (that’s ge-with-stroke, ayn).

    Of course, confusing *bhag- and bāgh is like writing Ghandi for Gandhi or Bhudda for Buddha: see “Bartholomae, Grassmann, and Grimm” for how to keep them straight.

  13. The word “or” was meant to stand for “as well as”, but now I’m left wondering whether that could be the case in proper English.

    And you could probably add a few other languages from India or from Afghanistan. (Incidentally, Paradise was a Persian garden.) There is a very popular Muslim wedding place in Port-Louis and its name is “Taher Bagh”.

  14. For those who are interested – I’ve posted a follow-up entry, taking into acount a proposal by Hrach Martirosyan to derive Armenian bak from PIE *bhag. Delving into that discussion also provides me with an answer to Siganus’s question – yes, it seems that Iranian bagh etc. “garden” is related to the root discussed.

  15. Excellent, thanks for the update!

  16. David Marjanović says

    The latest I’ve read on the PIE vowel system is this conference presentation.

    Ubykh, which went extinct in 1992, also had two vowels and even more consonants, 81 to 84 according to different sources.

    81 in native words and another 3 in loans, because once you have so many, adding a few more doesn’t seem to make a difference.

    The Moscow School Nostraticists find that the PIE *ḱ/k/kʷ distinction lines up with the vowels they reconstruct for Proto-Nostratic in a very Northwest Caucasian way: **/kʼ/ turns into */kʲ/ next to front vowels, */k/ next to **/a/, and */kʷ/ next to back vowels, while the vowel distinctions all disappear in the ablaut system.

Speak Your Mind