I was reading Wikipedia's introduction into the Elamite language, where it says that it had a vowel inventory of /a/, /e/, /i/ and /u/. “What a coincidence,” I thought, “just like Akkadian!” Now, Jagersma reports in his Descriptive Grammar of Sumerian that some scholars postulate the existence of an /o/ phoneme in that language (p. 59) but he himself rejects that hypothesis. However, surely some languages which are written in Cuneiform script must have had an /o/ phoneme and thus some way to encode it in syllabograms.

Because Sumerian and Elamite (and Hurrian) are, as far as I'm aware, language isolates, it's hard to argue what kind of phonetic inventories they had aside from what we can read of the Cuneiform, but perhaps Old Persian or Hittite or one of its Anatolian cousins can be shown to have a vowel inventory which does not conveniently match the typologically-unusual four-vowel inventory of Akkadian. So do we know some language with an /o/ vowel which is written in Cuneiform, and if so, is that vowel given a separate set of syllabograms like the other four?

If that is the case, then I can perhaps be more confident that Elamite did have a vowel system matching Akkadian's, and this is not simply an artefact of the writing system.

  • 2
    It’s certainly commonly believed nowadays that Hittite had /o/, and it seems to have been at least proposed that Akkadian did too. Commented Sep 21, 2023 at 13:26
  • Whilst Old Persian's script was cuneiform (in the sense of wedge-shaped), it wasn't Cuneiform (in the sense of in the Sumero-Akkado-Elamo-Aramaeo-Hittite-etc family of scripts). Likewise Ugaritic. Old Persian had no /o/ (of any length), but Ugaritic did (albeit only long /o:/ from monophthongisation of the diphthong *aw). There are some Ugaritic words or names attested in Akkadian texts though; I don't know how /o:/ is represented though
    – Tristan
    Commented Sep 21, 2023 at 13:43
  • 1
    Delete "Aramaeo". @Tristan
    – fdb
    Commented Sep 22, 2023 at 17:31
  • quite right (although it's too late to correct that). I'd confused myself and misremembered aramaic heterograms being in cuneiform rather than in middle iranian - and so assumed there were some old aramaic texts in cuneiform but that wasn't right
    – Tristan
    Commented Sep 22, 2023 at 18:13

2 Answers 2


It's likely that Hittite had an /o/ (and represented it in cuneiform!), but most transcriptions still don't reflect this.

In Hittite cuneiform, the signs U and Ú were both used frequently for phonetic purposes. This isn't too weird—Akkadian did this too—and for a long time it was assumed their use was fairly arbitrary, like the distinction between TA and DA in Hittite. They both represent the same sound, and the choice to use one versus the other comes down to the individual scribe and certain conventions.

But more recently, corpus analysis suggests that U and Ú were much less interchangeable than TA and DA. Most words are almost exclusively spelled with one or the other. For example, lūli- "pond" is always spelled lu-ú-li-, and lūri- "disgrace" is almost always spelled lu-u-ri-.

Similarly, while the morpheme u- "hither" is spelled both ways, within a particular verb it's entirely consistent which one is used. The verb ūnna- "drive hither" is always spelled ú-un-na-, while ūššiya- "open (curtains)" is u-uš-ši-ya-.

In that case, which is which? Well, if we compare the paradigm of the ablauting verb au/u- "see" against pai/pi- "give", we see, for example, first singular u-uḫ-ḫi (ūḫḫi with U) and pé-e-eḫ-ḫi (pēḫḫi with E), but first plural ú-me-e-ni (ūmeni with Ú) and pí-i-ú-e-ni (pīweni with I). Kloekhorst considers this conclusive proof that the u~ú alternation happens in the same environments as e~i alternation.

(For some reason, in his argument Kloekhorst uses the spelling pí-ú-e-ni, but that could just as well be read pé-ú-e-ni. I don't know why he doesn't cite the clearer form pí-i-ú-e-ni, which is also attested and is unambiguous.)

Based on this, it seems the U sign was used for /o/, and the Ú sign for /u/. This theory has been gradually catching on, but transcriptions are slow to change, and still write lūri- instead of lōri-. It didn't get its own set of syllabograms—that is, there was no DO sign separate from DU, and so on—but plene spellings could be used to clarify it, just like with the ambiguity between /e/ and /i/. After all, there was no DE sign separate from DI either! (Side note: both ŠU and ŠÚ were commonly used in Hittite. I wonder if they were ever repurposed to write this distinction? I'm guessing not, but that might be an interesting study.)

For more details, see Kloekhorst's Etymological Dictionary of the Hittite Inherited Lexicon; a lot of his phonetic conclusions in that work are controversial to say the least, but he provides a lot of data on the U vs Ú distinction and that part at least seems well-accepted.

  • Can you explain what you mean by the signs being used "for phonetic purposes"?
    – TKR
    Commented Sep 21, 2023 at 16:43
  • 2
    @TKR As opposed to logographic purposes—Ù (aka U₃) was used as a logogram for "and", for example, and U₄ as a logogram for "day", but not to write the sound /u/ in arbitrary words.
    – Draconis
    Commented Sep 21, 2023 at 16:51

According to this article (esp. note 8) Middle Elamite had the phonemes /o/ and /u/, which in Achaemenian Elamite shift respectively to /u/ and /i/.


In the transcription of Old Persian names and words, Achaemenian Elamite uses the sign u for Persian /au/, and the signs ú and hu for Persian /u/ and /ū/.

  • 1
    That is an extremely interesting article, thank you for pointing it out to me. I'll read it at length and probably add a note in the Wikipedia page for Elamite that its vowel inventory may not simply be /a/ /e/ /i/ /u/, after all... I'm sorry that only one answer may be accepted.
    – Wtrmute
    Commented Sep 22, 2023 at 23:26

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.