According to Gelb 1961, the famous Sumerian sign É ("house, building") was originally pronounced /ħa/ (or ḥa in Semiticist transcription). The main evidence for this is loanwords into other languages: É.GAL ("temple", literally "great house") > Hebrew hēkhāl, Aramaic hēkhəlā, Arabic haykal. The pronunciation would then have shifted to /e/ through dissimilation and pharyngeal loss (both of which are well-attested in Akkadian and seem to be widely accepted).

However, this source is half a century old, and I haven't seen any mention of /ħ/ in more recent discussions of Sumerian (like Foxvog's grammar).

Is it generally believed nowadays that Sumerian had an /ħ/-like phoneme, which doesn't show up in Akkadian due to the famous loss of pharyngeals? If so, are there any other signs whose Sumerian values should be reinterpreted as a result (like the Sumerian word for "house" being ḤA instead of É)? And if not, how do modern scholars explain the loans like hēkhāl?

2 Answers 2


This is an interesting question.
On the whole, Sumerologists read Sumerian thru the window of Babylonian phonetics. Quite clearly, this means that Sumerian as per Sumerology is not genuine Sumerian.
I've been working on reconstructing the real phonetics of Sumerian. It seems that Sumerian had two "secret" laryngeals, not just ḥ which is e-coloring, but also h which is not e-coloring.
Extremely archaic cuneiform like that of Eblaite shows that MA2 "boat" was probably *mah and A2 "arm" was probably *ah.
Sometimes the "secret" laryngeals of Sumerian are visible in the other languages, as you mention with the case of É.GAL which was probably pronounced *ḥay-kal, which also gives Hurrian haikalli "palace".

  • Very interesting! Do let me know if you publish or release something on this.
    – Draconis
    Mar 29, 2021 at 20:17
  • One quick follow-up—do you agree with Gelb that A2, E3, I3, U3 also had an onset (distinct from E2's /ħ/) that disappeared in Akkadian use? And if so, what do you reconstruct that onset as? (Your mention of A2 as *ah implies otherwise, but I want to check.)
    – Draconis
    Mar 30, 2021 at 20:18
  • 1
    In the case of I3 vs i, it seems that i3 was 'i (aleph i) while i was ḥi (heth i) u3 was 'u. These signs did not have the same initial.
    – user23769
    Mar 31, 2021 at 7:10

Jagersma agrees with Gelb that Sumerian had "hidden" phonemes /h/ and /ʔ/, but disagrees about /ħ/.

He points out that transcriptions of É (or É.GAL) generally use /h/, even in languages where /ħ/ was available: see the Hebrew, Arabic, and Aramaic loans mentioned in the question, but also Hurrian haikalli (mentioned by Fournet) and Ugaritic hkl. Based on this, he reconstructs a pronunciation /haj/; the later reading as /e/ comes not from Akkadian dissimilation, but from simplification of diphthongs. (See also the early Akkadian loan ayyakku < *hayyanaku < É.AN.NA(K) /haj ʔanak/.)

According to Hasselbach, it's true that the sign É is used for /ħa~ħe/ in Sargonic cuneiform; she attributes this not to a Sumerian /ħ/, but to diphthong coalescence. Since the classic Akkadian dissimilation of /a/ to /e/ near pharyngeals was already starting to take place, it would make sense for Akkadian scribes to use the sign /haj~he/ (É) for the H that triggered the dissimilation and the sign /ha/ (Á) for the H that didn't.

If so, are there any other signs whose Sumerian values should be reinterpreted as a result (like the Sumerian word for "house" being ḤA instead of É)?

Many. For a few examples, Jagersma reads Á as ha, MÁ as maʔ, and AN as ʔan, though he notes that it's difficult to tell the difference between /ʔ/ and /h/ without the help of foreign loans (since they both disappeared fairly early, and are mostly visible indirectly by how they trigger postconsonantal forms of suffixes and preconsonantal forms of prefixes).

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.