The roots of the language are gen erally monosyllables though a few longer words are found. These roots are not subject to internal change, since grammatical dis tinctions are made by grouping a number of elements together around the root, not by causing the root to assume different forms. In the case of roots having the form consonant-vowel consonant the final consonant is often dropped when a consonant immediately follows. In itself the root expresses a bare idea only; thus dug means "good," and can appear not only as an adjective but (with the appropriate modifying elements) as a noun "goodness," or a verb "to be good." Nouns are mostly formed by adding prefixes, sometimes the vowels a or u, but more often narn or nig which form abstracts ; thus, from dug, the noun "goodness" would be nam-dug. Sumerian has no gender, which is disregarded in places where it is indifferent and expressed where necessary either by prefixing the determinatives for "male" and "female" or by the use of entirely different words, such as adda "father" and ama "mother." Plural of nouns is not always indi cated, but can be formed when required either by repeating the root (kur "land," kur-kur-ra "lands"), or by the addition of suffixes -ene (used of persons), -mesh or hi-a. There is no special form of the dual, though the number two can be written if essen tial. The relation in which the noun stands to the rest of the sentence is shown by a rather elaborate set of suffixes, which may be treated either as postpositions or case endings. Of these -a generally expresses the accusative, -e the nominative, -da is "with," -ta "from," -ra and -shu "to," -gim "like," while the geni tive is denoted by a suffix -ak, which has the peculiarity of doub ling itself to correspond with the number of genitives to be ex pressed, though this doubling is often concealed by the omission of a final consonant. Finally, the noun always stands at the head of a complex of modifying elements, with which it combines to form a kind of composite unity.
A peculiarity of the verbs is that, besides the simple roots, there are also compounds formed by the prefixing of a noun, the whole making one idea (e.g., de "to pour" and gu-de "to pour out a voice, to speak"). Between the nominal and the verbal part of the compound root are placed such of the accompanying ele ments as normally precede the root. These accompanying ele ments may be classified into three kinds (a) subject prefixes, (b) conjugating elements and (c) infixes. The first merely in dicate the subject, not expressing in themselves either number or person, and the force of the individual prefixes is still insufficient ly clear. Distinction of tense, number, person, mood and voice is effected by the conjugating elements, but the resources of the language for these purposes are not highly developed, and it is clear that some of the grammatical tablets compiled by Semitic scribes for their own use lent to the Sumerian verb a number of artificial refinements which are not observed in the native texts. Extensive use is made of participial constructions, and there is a special form for the optative, and a corresponding prohibitive particle. The infixes, which convey the direct or indirect object, as well as certain adverbial relations, are included in the verbal forms as the result of a strong tendency for the predicate to gather into itself parts representing all the other members of the sentence. The order of elements in the verb-compleic follows a strict rule. The numeral system was sexagesimal and decimal.
The native grammarians knew of five different "tongues" in Sumerian. These are assumed to have been dialects, but in fact nothing is known of them except the eme-sal "broad ( ?) tongue," in which a number of religious texts are composed. This is differentiated from the ordinary form chiefly by a weak ening of certain consonants, particularly of g into m, and by a preference for phonetic writing. Sumerian ceased to be spoken before 1500 B.C., but was long used as a learned language.