range of morphemes to be written and the phonological form of words to be specified
more precisely, which in turn enables us to recognize that the language being written
was Sumerian. Why Sumerian should be the first Middle Eastern language that we can
confidently recognize in writing remains uncertain. The regional term “Sumer”, again
an adoption from Akkadian, identifies the language with the area where cuneiform
writing was invented, although the correlation is less straightforward in Sumerian itself,
in which the regional term is Ki-en-gi, tentatively “Native-language-land,” with gias a
variant of gir 15 (“native”) and ena reduction of eme(“language”). However, this area was
at least to some degree bilingual on the basis of the extensive interplay between
Sumerian and Akkadian, if not multilingual on the basis of ethnographic parallels. The
decision early in the third millennium to start specifying a Sumerian pronunciation
may reflect such factors as status, scale of use or increasing state control, but it may also
indicate that the language was already in retreat.
The introduction of phonographic writing was again a process of abstraction, the
one-to-one correspondence between sign and meaning in logographic writing being
supplemented by the writing of a sound sequence that could include more than one
morpheme or only part of a morpheme. It was also a slow process: isolated phono-
graphic writings of bound morphemes are first attested about 2800 BC(all dates are
highly approximate) but remained limited in use until about 2600 BC. Words were
written more fully from that date onward, but writing conventions were never
standardized and reconstructing a word’s exact phonological form remains difficult, the
relationship between script and pronunciation being complex and ambiguous. In
particular, syllable-final consonants were often underspecified: until about 2400 BC
many consonant–vowel–consonant (CVC) sequences were written CV and it was only
in the eighteenth century that full writings, typically in the form CV–VC, became the
norm. In consequence transliteration (alphabetic representation) of Sumerian
cuneiform rarely reconstructs words, but instead simply links with hyphens those signs
thought to constitute a word.
A further characteristic of Sumerian cuneiform is that many of its signs have
multiple values, because the same sign was used to represent words similar in meaning
or in sound, thus reducing the number of signs that were needed. So, for example, the
sign we refer to as KA, in origin a profile of the head with hatching at the point of
the mouth, was used to represent the semantically related dug 4 “to speak,” inim
“word,” ka“mouth” and zu 2 “tooth,” as well as the phonologically similar zuh
̆
“to steal”
(h
̆corresponds roughly to chin loch). (The subscript numerals again enable modern
scholars to specify which particular sign is used in the original script, so, for example,
dug 3 “to be good” is written with the sign we refer to as H
̆
I while dug 4 is written with
KA. Some scholars use acute and grave diacritics rather than the numerals 2 and 3 in
subscript.) Semantic and phonological principles likewise enabled expansion of the
sign repertoire based on already existing signs. So, for example, the sign KA combined
with the semantically appropriate sign a(“water”) represented nagˆ“to drink” (gˆ
corresponds roughly to ngin sing), and combined with the phonologically appropriate
sign me(“to be”), eme“tongue.”
In addition, a few signs were also used as unspoken determinatives, written either
before or after a noun, to specify the semantic set to which the noun belonged. For
example, the names of deities were preceded by a sign indicating their divine status.
The sign in question depicts a star and represents the Sumerian for deity, digˆir; when–– Graham Cunningham ––