For example: In the Romanization of Sumerian, /ŋ/ is written as ⟨g̃⟩ or ⟨ĝ⟩ instead of ⟨ng⟩ or even ⟨ŋ⟩. Also in Sumerian /t͡sʰ/ is written ⟨ř⟩ or ⟨dr⟩. The list goes on with Sumerian.

In Proto-Indo-European (Which I know is technically not a real language) /kʷ/ is written as ⟨kʷ⟩ instead of ⟨kw⟩. This problem mostly applies to Proto-languages actually, but it also applies to some extinct semetic languages, like Ge'ez.

I'll admit I'm definitely a novice in Linguistics, so this could all make perfect sense. But I don't understand and would request clarification.

  • 6
    Why should /kʷ/ not be written ⟨kʷ⟩? It seems like a natural transcription, from an IPA perspective.
    – Draconis
    Commented Mar 26, 2020 at 19:14
  • 1
    I'm saying it's more practical just to write it as a standard letter right next to a letter, instead of the having the little ⟨ʷ⟩. Or that's my perspective at least. Commented Mar 26, 2020 at 23:11
  • 11
    The thing is, PIE (as commonly reconstructed) has a distinction between * (one phoneme) and *kw (two phonemes).
    – Draconis
    Commented Mar 26, 2020 at 23:36
  • 7
    You are asking about an impractical orthography in English !?! ;-) Commented Mar 27, 2020 at 7:09
  • 1
    I am not sure this is tolerate here. So please tell me if I should remove that message. Is there a website where you can learn to at least write and if possible speak sumerian ? Commented Mar 27, 2020 at 7:36

2 Answers 2


First, it's worth noting that these are transcriptions, used by linguists, not actual orthographies used by native speakers. The ancient Sumerians didn't write their word for "god" as diĝir; they wrote it as 𒀭. And the people who spoke something like Proto-Indo-European never wrote it down at all. It's only modern scholars who use the Latin alphabet for this.

And since these are modern transcriptions of extinct or reconstructed languages, we generally aren't sure how the sounds were pronounced! Some scholars think that Sumerian ř was pronounced like IPA /t͡sʰ/, but others disagree, and some argue that it wasn't a distinct phoneme at all (just an artifact of a sound change in progress).

All that we can be certain of is that certain words (like guř "ox") are sometimes transcribed with Akkadian phonetic signs containing /d/, and sometimes with Akkadian phonetic signs containing /r/. So if you want to talk about this new maybe-phoneme without getting into the weeds of how it was actually pronounced, ř or dr are reasonable names to pick.

Or, sometimes it's for reasons of typesetting (like Sumerian ĝ being much easier to type than ŋ until very recently), or to show that something is a single segment (PIE * is generally considered a single phoneme, rather than *k followed by *w). In the end, these transcriptions are mostly only used by specialists in specific fields, who quickly get used to all the idiosyncracies. With no native speakers using these transcriptions in day-to-day life, practicality isn't much of a concern.

  • 2
    Your first paragraph is "obvious" but also really important! You might also mention that the written symbol 𒂷 doesn't necessarily have any internal structure relating it to a vocalization (the way English spelling relates to vocalization). My background is a quick skim of Wikipedia, in justification of my desire to shoehorn in a link to zompist.com/yingzi/yingzi.htm here. :) Commented Mar 27, 2020 at 14:10
  • 1
    Tangentially, I was a bit confused by your use of 𒂷 (ĝa₂) instead of 𒂍 (e₂) for "house", but apparently that's an Emesal spelling. Commented Mar 27, 2020 at 16:46
  • 1
    (Looking at the various meanings of 𒂷, I would guess that the original was probably bisaĝ = "basket, container", for which it may have been a pictograph, which was then adopted as one of the spellings of the common verb ĝa(r) = "put, place", and from there phonetically as a syllabogram for /ĝa/ in various contexts. I'd further guess that the Emesal usage of 𒂷 for "house" is probably one of those phonetic uses, as Emesal in general tended to be written as "eye dialect" using syllabic signs.) Commented Mar 27, 2020 at 16:46
  • @IlmariKaronen Yes, sorry, I should have clarified that better! It was the first sign I could find that had an initial ŋ, but I should have gone with something more standard like diŋir. I'll edit that now.
    – Draconis
    Commented Mar 27, 2020 at 16:51
  • 1
    For a modern, real-life parallel to Sumerian <ř ~ dr>, there is Breton <zh>, which is a phoneme realised in some dialects as /z/ and in others as /h/ (and deriving historically from /θ/, I believe). Commented Mar 28, 2020 at 8:02

There are numerous reasons specific to the language in question. Semitic transcription practices were established a long time ago before the IPA swept the field of linguistics (likewise Finno-Ugric, Bantu, and so on). The letters used are often deliberately vague because there is uncertainty as to the phonetic property of the sound in question (thus Bantu "c" in Guthrie's transcription). Most works before 1985 or so were prepared on a typewriter which had limited capacity for writing phonetic symbols. There is a strong tendency for the first works in a language family to be historical reconstruction (certainly the case in Bantu), so that a description of Sotho is driven by the desire to relate the sounds of Sotho to reconstructed Proto-Bantu – therefore, the same alphabet will be used for talking about the specific language and the proto-language.

In some cases, there is a deliberate choice, for example in IE , the point is to graphically distinguish the labiovelar from a sequence of velar plus glide. That notation is important in understanding the phonology of languages like Tigrinya, since "kw" is not phonologically a consonant sequence.

  • Blooper: Tigrinya is not considered IE, and you likely didn't mean to imply otherwise, though it appears as if you did. Trivial: "kw" is indeed not a con-sonant sequence, it's a grapheme sequence.
    – vectory
    Commented Mar 26, 2020 at 21:48
  • @vectory Does this answer imply that Tigrinya is IE? I certainly didn't read it that way. It's just a language where the difference between kʷ, one segment, and kw, two segments, is phonologically relevant.
    – Draconis
    Commented Mar 26, 2020 at 21:51
  • @Draconis I don't know but don't suppose so. That's why I'm trying to make it clear, to avoid that anyone could get a wrong impression, because the wording, "that notion", isn't very clear unless one already knows what is implied (*kw is a cluster, the other one is a unit? the difference is fleeting nevertheless).
    – vectory
    Commented Mar 26, 2020 at 22:15

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.