The sign 'gi' obviously depicts a plant and its foremost meaning is 'reed' (Labat - Manual d'Epigraphie Akkadienne, 6th). The land is/was largely a reed-swamp. Without a name I would probably call it 'the reed land' or perhaps (referring to the chief or the people being masters of working with reed) something like 'land of the masters/lord(s) of the reed'. But I have not yet come accross such an interpretation. Hayes - A Manual of Sumerian Grammar and texts, 3rd, 2019 p66-67) for example lists some proposed etymologies that, to me, seem very far fetched. So, what is wrong with 'reed'?

  • I might be misunderstanding your question, but gi "reed" doesn't have an R auslaut, does it?
    – Draconis
    May 29, 2021 at 23:56
  • I am from Basra, south Iraq. We still use some Sumerian words in our local Arabic dialect despite claims that the Sumerian language has disappeared. The word 'dingir' refers to a long pointed post... like that found in public squares or in the Ship as a mast or in carriage to hook the animals. This is not very different from 'reeds'. I was of the opinion that this referred to the single creator of the world as 'dingir' refers to the God too.
    – Riad
    Aug 12, 2022 at 16:35

3 Answers 3


There are several signs that are read GI and they have different figures to contrast them. It seems that gir15 is the Sumerian word for saying "Sumerian".
Eme-gir15 = Sumerian language, also written eme-gi-ra, or eme-gi7
Dumu-gir15 = Sumerian man, also written dumu-gi7, or dumu-gi7-ra
gir15 is sometimes replaced by the variants gi7-ra or gi-ra.
The word gir15 has nothing to do with gi "reed".


To add on to Arnaud Fournet's answer:

Sumerian seems to have had a phonological process much like modern French, where most consonants aren't pronounced at the end of words. This means that gi "reed" and gi "Sumerian" may have been pronounced the same, based on Akkadian transcriptions in lexical lists.

But this changes when endings are added, like the genitive -ak. To my understanding, gi "reed" had no "hidden" final consonant (auslaut), so its genitive would be gia, while gi "Sumerian" had one, so its genitive would be gira. This is why the word is sometimes transcribed as gi(r). (And the final -k of the ending is then also not pronounced, unless you add another ending after it, etc.)

Auslauts aside, though, according to the PSD the word for "Sumerian" was actually ŋir; the difference between ŋ and g is concealed by the Akkadian transcriptions, since Akkadian lacked /ŋ/. And as Fournet says, ŋir is written with the sign GIR₁₅ aka GI₇, while gi is written with the sign GI. This doesn't rule out the possibility that ŋir is historically derived from gi, but it implies that the Sumerians didn't think of it that way.

  • Agreed ! Sumerian seems to have plenty of secret consonants, which explains why a number of signs have variable readings like Cv(C).
    – user23769
    Jun 1, 2021 at 3:21

In addition to the difference between 𒂠 respectively 𒄀 "GIR₁₅ aka GI₇, while gi is written with the sign GI" (following @Draconis, above), additional complications arise when considering Sux. 𒊕𒈪 sag̃-gíg and Akk. 𒋗𒈨𒊒 šumeru (following Wikipedia: Sumer): "The first is the pictographic character for "head" [viz. 𒊕 sag̃] the second the character for "night", and for "black" when pronounced gíg" -- caption under an image of an inscribed bust of Gudea.

Wiktionary distinguishes 𒈪 (GE₆, Mi) in Akk. mūšum, mušītum, Sumerian g̃i "night" and Akk. ṣalmum, Sux. giggi "black" -- thanks to the good work of @Sartma. Note that in EPSD and Wiktionary corresponds to ŋ in EPSD2 (The Pennsylvania Sumerian Dictionary Project).

Although saŋ-gíg, literally "head" + "black", sounds like a racial epithet, my impression from the frequent depiction of bald heads in Sumerian iconography begs to differ (note also the recent attempt at night to be derived from naked, Pooth 2015 apud Wiktionary: PIE *nókʷts).

Anyway, the point to be made here is that ŋir, ŋi and gíggi are obviously colexified in some way, but the /m/ of akk. šumeru is difficult to justify as a sound change, and this does a priori not benefit an analysis of gi "reed", not the least because there is nearly a thousand years of difference between Gudea c. 2150 BC, the Early Dynastic Period c. 3350 – c. 2500 BC, and a period of proto-writing c. 4000 – c. 2500 BC. As for possible loanwords in Sumerian, cf. Jagersma 2010: 4, an early layer of borrowed nouns has a suffix -a, yet "An even older layer of endingless forms has been proposed, but remains unproven (Civil 2007; Sommerfeld 2006)."

Incidently, Egypt goes by a couple of names, km.t "the black land", mḥ.t "the Nile Delta" (or north), ḥwt-kꜣ-ptḥ "Egypt; Memphis; the temple of Ptah in Memphis", tꜣ.wj "the two lands" and a couple more. In this case, darkness is often attributed to wet, fertile soil. This might be close enough to to reeds or marshes, conceptually; not that e.g. gꜣš "reed, rush" (c. 19th century BC) would bear any resemblence to the above.

You can find -gir in words that mean farmer, but its not clear if this proves anything.

So, what is wrong with 'reed'?

I don't believe that's the right question to begin with. First you must understand, why gi "noble" (Wikipedia), "Sumerian" (@Draconis) vel sim. should be prefered, by whom, since when, to be able to point out contradictions and gaps in the argument which you are trying to fill -- like so: Noble and Sumerian could easily come from semantic drift after the Ethnonym was already established. A comparable case is frank and free, which seems to be from the name of the Franks applied to a social class in medieval laws. The trivial answer is, of course, not that because this.

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.