Sumerian Language

An insatiable appetite for ancient and modern tongues

Address comments and questions to:


Name Origin. Sumerian comes from the Akkadian šumeru, of unknown meaning. The  Sumerians called their language eme-gir (eme ‘language’, gir of uncertain meaning, perhaps ‘native’).

Classification: It is a language isolate, no relatives are known.

Overview. Sumerian was spoken several millennia ago in south Mesopotamia (now south Iraq), the land between the Tigris and Euphrates rivers, by the creators of one of the first urban civilizations. After it ceased to be an everyday communication tool it remained a prestige language in Babylonia and across the ancient Near East for a long time.

     It was the first written language, preceding Egyptian by one or two centuries, and it has one of the longest literary records extending for more than three millennia. To the north of the Sumerians were the Akkadians, a Semitic people, with which they had extensive contacts, their languages influencing each other.

   Sumerian has an agglutinative morphology, based on suffixes for nominal marking, and on prefixes and suffixes for making verb forms. It is an ergative language, marking with the ergative case the agent of transitive clauses while their object as well as the subject of intransitive clauses are marked with the absolutive. More precisely, it is split-ergative (i.e. partially ergative) because some pronouns and verb forms follow, instead, a nominative-accusative pattern. Nominal marking (possession, plurality, case) is not directly on the noun but at the end of the nominal clause. The Sumerian verb is complex recapitulating much of the information provided by the nominal construction. Word order is Subject-Object-Verb, noun modifiers following the head-noun.

Distribution. Sumerian was spoken in the Ancient Near East in southern Mesopotamia, the plain bounded by the Euphrates and Tigris  rivers, corresponding to modern Iraq south of Baghdad.

Status. Extinct. The exact lifespan of Sumerian is debated. It is particularly difficult to know when it ceased to be an oral language. It was spoken probably from the late fourth millennium until the early second millennium BCE, at the beginning of the Old Babylonian period. Afterwards, scribes continued to produce Sumerian texts until the late first millennium BCE but Sumerian had become by then a dead language of scholarship and cult.

Varieties. Two main dialects have been identified in the second half of the third millennium BCE. The differences between them changed with time.

-Northern Sumerian (Area: Nippur, Adab, Isin) lacking vowel harmony and using the prefix a to mark the passive voice.

-Southern Sumerian (Area: Lagash, Umma, Ur, Uruk) having vowel harmony and using the prefix ba to mark the passive voice.

Another dialect was:

-Emesal, documented in certain literary and cultic texts dating from the Old Babylonian period or later. Emesal differs from standard Sumerian in vocabulary, pronunciation and morphology.

Periods and Main Documents

  1. Five-thousand discarded clay tablets found in the ceremonial centre of Uruk (c. 3200 BCE), written in Sumerian with a pictographic script called proto-cuneiform, are the oldest written texts known. This early script was logographic (it represented words) and remains largely undeciphered.  The documents seem to be mainly administrative but include also word lists.

  1. Archaic texts from Ur (ca. 2800 BCE). Almost 400 tablets of an administrative nature.

  1. The first literary compositions known plus incantations and administrative and legal documents from Shuruppak (Fara), Abu Salabikh, Nippur and Adab (ca. 2600 BCE).

  1. Old Sumerian period (ca. 2470-2340 BCE). About 2,200 texts from Girsu and Lagash, Nippur, Zabalam, and Adab. They include administrative and legal documents, inscriptions, letters, and fragmentary literary texts.

  1. Old Akkadian period (ca. 2340-2200 BCE). About 3,000, primarily administrative, texts from Lagash, Nippur, Umma and Adab.


  1. Early Neo-Sumerian period (ca. 2200-2113 BCE). More than 200 royal and dedicatory inscriptions from Lagash including 26 on statues and two very lengthy ones on clay cylinders.


  1. Late Neo-Sumerian or Ur III period (ca. 2112-2004 BCE). This period has yielded over 60,000 published Sumerian texts, mostly administrative but also including about two hundred royal inscriptions, three hundred court decisions, hundreds of letters, a few dozen incantations, and some literary texts. They were found in Umma, Lagash, Drehem, Ur, Nippur and Garshana.    

  1. Early Old Babylonian period (ca. 2017-1722 BCE). Produced thousands of literary texts preserved by Akkadian scribes as well as many administrative and legal documents.

Phonology (between 2500-2000 BCE)

Syllable structure. Sumerian syllables may be V, VC, CV or CVC. Sequences of two vowels did not occur but consonant clusters are attested in polysyllabic words; they are limited to two medial consonants e.g. ninda ('bread'), absin ('furrow'). Voiceless aspirated stops and affricates (see below) don't occur or are very limited at the end of a syllable while not such restrictions occur in initial position.

Vowels (4). Sumerian had four vowel qualities: i, e, u, a. The role of vowel length is disputed. It was not generally discriminated in the script and most scholars think it was not phonemic. There was vowel harmony in Old Sumerian for some verbal prefixes whose vowels varied depending on the vowel of the following syllable: before a low vowel the prefix vowel was e, and before a high vowel it was i.

    Other vowel changes were assimilation and loss. A prefix vowel tended to assimilate to the vowel of the following syllable while a suffix vowel tended to assimilate to the vowel of the preceding syllable. Assimilation also occurred in open syllables (those ending in a vowel) of polysyllabic stems, the vowel of the first syllable assimilating to the vowel of the second syllable.

Consonants (16-19). Sumerian had two series of stops and affricates. According to some scholars, the contrast was between plain voiceless and aspirated voiceless (as shown in the following table). Around 2000 BCE, the plain voiceless stops became voiced in some environments (word-initially and between voiced sounds) but remained voiceless elsewhere. Other scholars don’t agree that in Classical Sumerian all stops were voiceless, thinking that the contrast was not between voiceless plain and aspirated but, instead, between voiceless and voiced. Consonant length was phonemic when a consonant occurred between vowels.

    The traditional view posits that Sumerian had 16 consonants, as reflected in transliteration. An additional three, usually not represented, are controversial: a glottal stop [ʔ], a glottal fricative [h], and a dental glide [j]. They would have been gradually lost between 2500-2000 BCE. Another debated phoneme, notated as dr or r̆, may have been [tsh] according to Jagersma (2010).



-The transliteration system for Sumerian does not distinguish between word signs and sound signs (see script).

-Word roots and their affixes are joined by dashes. Morphemes are separated by periods. Homophonous readings are distinguished with an accent (acute or grave) or a subscripted number. Unpronounced determinatives (indicators of meaning) are transliterated with superscripted letters. When the reading of a sign or its pronunciation is in doubt, one may represent it in capital letters.

-The plain voiceless stops are transliterated as voiced stops i.e. [p], [t] and [k] are represented, respectively, as b, d and g. In a Sumerian word, g might have been pronounced as [k] or [g] depending on the phonetic environment and historical period (see consonants above). Similarly, b and d may have been pronounced as voiced or voiceless.

-The aspirated voiceless stops are transliterated as plain voiceless stops i.e. [ph], [th] and [kh] are represented, respectively, as p, t and k.

-The glottal stop, as well as [h] and [j] are usually not transliterated.

-The affricate [ts] is transliterated as z, and its aspirated counterpart [tsh] as  dr or .

- [ʃ] is transliterated š and [x] as h or ḫ. Thus, Sumerian h represents the velar fricative and not the glottal one (whose occurrence is controversial and is not usually represented).

- The velar nasal [ŋ] is transliterated ĝ.

-The sounds [m], [n], [s], [l], [r] are transliterated by these same letters.

Accent: Little is known about Sumerian word accent as the writing system doesn't mark it. According to Jagersma (2010), as all Sumerian loanwords in Akkadian are stressed on the last syllable, it is presumed that Sumerian words had a fixed stress accent on the last syllable. However, some clitics might have been unstressed as shown by the loss of their final vowels in some instances (which can only happen in unstressed syllables).

Script. The cuneiform script, the oldest known, was invented in the late fourth millennium BCE. It is a mixture of logographic (word signs) and  phonographic (sound signs) writing. In the beginning (ca. 3200 BCE), the Sumerian script was only logographic but around 2500 BCE phonographic writing was incorporated (phonograms arose from logograms). Lexical words are usually written with word signs and all function morphemes (bases, clitics, and affixes) with sound signs. Word boundaries are not indicated.


Sumerian morphology is preponderantly agglutinative. Morpheme boundaries are easily identifiable and have from one to three syllables each; a few consist of a single vowel or consonant. A Sumerian word consists of a stem which may have one or more  grammatical morphemes attached to it. Grammatical morphemes, which cannot exist independently and are bound to other morphemes, are affixes (prefixes or suffixes) or clitics. Almost all affixes are inflectional; derivational affixes are rare. All clitics are enclitics and most of them are attached at the end of a phrase (like possessive and demonstrative pronouns, plural marker and case markers).

a) Nominal

Nouns are not inflected. They are not marked for definiteness (and there are no articles), gender is not marked on the noun but is reflected in other parts of the sentence, plurality and case are marked by enclitics at the end of the nominal phrase.

gender: Sumerian distinguishes two genders or classes of nouns: human (human beings and gods) and non-human (plants, animals, objects, abstract nouns). This distinction is semantically based and is not dependent on any formal property. Gender class is not marked on nouns, but is shown in third person pronouns (personal, possessive, interrogative) and pronominal verb markers. Besides, gender affects plural formation (only human nouns might take a plural) and case marking (the dative is restricted to humans, the directional to non-humans). Generally, the human form contains an n and the non-human form a b; however in interrogative pronouns the opposite is true.

number: plural marking is restricted to human nouns. The plural marker (e)ne (the initial vowel is lost when the preceding word ends in a vowel) is attached to the last word of a noun phrase, indicating the plural of the head noun. It has a fixed order among phrase-final clitics, following any enclitic pronoun and preceding case markers. Even among human nouns, (e)ne is frequently omitted e.g. when a numeral is used, when a noun is in the absolutive case, when a plurality of humans is regarded as a collective noun.

    Another, less frequent, way to indicate plurality is by reduplication of a noun, adjective or verbal stem. Reduplication is total (the entire word is duplicated) and, in contrast to the plural marker, may occur with both human and non-human nouns.  Sometimes, reduplication implies "all" or distribution ("each of").

case: absolutive, ergative, genitive, dative, directive, locative, terminative, adverbiative, ablative-instrumental, comitative, equative.

    Cases are expressed in Sumerian with enclitic markers.  The case markers are phrase-final clitics.  They follow both the enclitic pronouns and the enclitic plural marker and are, thus, the final word of the noun-phrase signaling its end and helping to clarify the sentence structure. Complex noun phrases may have several successive markers attached at its end.  The core of the noun phrase may be a pronoun, a numeral, or a participle, instead of a noun.

    Two cases are gender specific. The dative is applied only to humans and the directive to non-humans. The absolutive has zero-marking. The ergative and the directive are marked with e, the genitive with ak, the dative with ra, the locative with a, the terminative with še, the adverbiative with , the ablative with ta, the comitative with da, and the equative with gen. These markers experience a variety of phonological changes according to their environment, obscuring sometimes their recognition, especially if the preceding morpheme is written with a word sign instead of a sound sign.

absolutive: it marks the subject of an intransitive clause and the direct object of a transitive clause. In copular clauses (in which the verb is 'to be'), the absolutive expresses either the subject or the predicate. It is also used to address somebody (equivalent to the vocative of other languages).

ergative: it expresses the subject of a transitive clause. Its marker (e) is homophonous with the directive case marker. After vowels is usually omitted.

genitive: it frequently expresses possession but it might also express material, size or content of the referent noun. As the case marker is placed at the end of the noun phrase, the word order determines which is the possessor and which is the possessed. The usual one is: head noun, adjective, dependent genitive (possessor), clitics of the head noun. For example:

  1. gigir sumun ensi-ka-ke                The old chariot of the ruler.

  2. chariot old  ruler-GEN-DIR

Two embedded genitive constructions are commonplace:

  1. é dumu lugal-la-ka                        The house of the son of the king.

  2. house son king-GEN-GEN

dative and directive: they both mark the indirect object, the first one used for human nouns, the second applied only to non-human nouns. Both cases are not necessarily equivalent though, what would be expressed with a dative in a human noun phrase may be expressed with the locative in a non-human noun phrase.

locative: marks spatial or temporal location and is applied almost exclusively to non-humans. Besides, the Sumerian locative can indicate the material used to make something, may have a distributive meaning or may occasionally replace the dative.

terminative: signals destination in place or time as well as purpose (the equivalent of 'to', 'towards', 'until', 'for'). It is the opposite of the ablative case. Besides, it may express cause in constructions with nouns like mu ('name') or signify 'in presence of' with igi ('eye'). It is found mostly, but not exclusively, with non-human phrases.

adverbiative: is a controversial case which many scholars consider as a kind of terminative while others, recognizing it as separate, deny that it is a case at all (they regard it as a suffix to make adverbs). It expresses the meaning 'in the manner of'.

ablative-instrumental: indicates the source of a movement or the time from when something happens; also the instrument of an action.

comitative: indicates company as well as other meanings equivalent to 'together with'.

equative: expresses a comparison between two noun phrases ('like' 'similar as', 'equal to'). The standard of comparison takes the equative case but the other member of the comparison is not indicated and has to be guessed from the context.

adjectives: are a closed class with only a few dozen members. They resemble verbs (finite and non-finite) but, in contrast to them, they can't be negated. When there are no adjectives to express a certain meaning, Sumerian may use nouns in the genitive case and, more frequently, stative verbs and participles. Sumerian adjectives may be reduplicated to modify a plural noun but, otherwise, they are invariable. Used attributively they follow their nouns. There are no comparative or superlative adjectives. Sumerian has no adverbs, adverbial meanings are expressed with adjectives, verbal affixes, or noun phrases.

pronouns: personal, interrogative, reflexive, demonstrative, possessive, indefinite

Sumerian pronouns are independent or clitics. The independent pronouns are the personal, interrogative and reflexive ones as well as some demonstratives. Other  demonstrative pronouns as well as the possessive ones are phrase-final clitics. The indefinite pronoun behaves like an adjective.

Independent personal pronouns distinguish three persons and two numbers.

They are infrequent because the subject and object of the verb are indicated on it with affixes. They are used only for humans but demonstratives may be employed as a third person pronoun for non-humans. They function mainly as emphatics. Their basic forms are shown on the table.

The absolutive and ergative forms are the same.

The only frequent interrogative pronouns are a-ba ('who?') and a-na ('what?'). This distinction between human and non-human is, formally, the opposite of that found in other pronouns in which n refers to humans and b to non-humans.

Two other interrogatives appear in late texts: me ('where?') and en ('when?'). A fifth interrogative, a.gen7 ('how?') , is only attested once. They behave like nouns, take case markers, and are positioned immediately before the verb.

The reflexive pronouns and ní-te ('self') are independent and take case markers. They are followed by a possessive pronoun which specifies the person referred to. Ní-te is used before a third person human possessive and before all other possessive pronouns.

Demonstrative pronouns are of two kinds, clitics or independent. The first type, which is sparsely attested, recognizes three degrees of distance: e/be ('this'), še ('that visible'), re ('that invisible'). They don't distinguish gender or number, they are not marked for case, and are attached at the end of the noun phrase, before the plural and case markers.

Only two independent demonstratives are attested: nen ('this') and ur ('that'). Like the enclitic demonstratives, they don't distinguish gender or number but, in contrast to them, they may be head of a noun phrase and take case markers.

Possessive pronouns have the following basic forms (which often experience changes according to the preceding or following element):

They are phrase-final clitics which precede all other clitics of the same noun.

For example:

  1. ka dumu-ne-ne-k-a                    in the mouth of her children

  2. mouth child-her-PL-GEN-LOC

Note: ane becomes ne, ene becomes ne, ak becomes k.

Two enclitic pronouns cannot be used together and, thus, a possessive pronoun can't combine with an enclitic demonstrative pronoun. But it may be used in conjunction with an independent demonstrative.

Sumerian has no articles but indefinite meaning may be conveyed with the indefinite pronoun na-me ('any'). It is always used attributively, doesn't take case markers and doesn't distinguish gender.

compounding and derivation: compounds in Sumerian may be right-headed or left-headed. Most noun-noun compounds, but not all, are of the latter type. For example, the first three compounds shown below are left-headed, the following two are right-headed, and the final two lack a head i.e they are coordinative (as if both words were joined by 'and'):

  1. é-muḫaldim (‘kitchen’) from é (‘house’) + muḫaldim (‘cook’)

  2. ereš-diir (‘highpriestess’) from ereš (‘lady’) + diĝir (‘god’)

  3. dumu-saĝ (‘firstborn’) from dumu (‘child’) + saĝ (‘head’)

  4. an-šár (‘horizon’) from an (‘heaven’) + šár (‘circle’)

  5. šu-si (‘finger’) from šu (‘hand’) + si (‘horn’)

  6. zíd-munu4 (‘beer’) from zíd (‘flour’) + munu4 (‘malt’)

  7. ú-šim (‘plants’) from ú (‘grass’) + šim (‘herbs’)

Most adjective-noun compounds are also left-headed (many include adjectives meaning big or great):

  1. lugal (‘king’) from (‘man’) + gal (‘big’)

  2. ur-maḫ (‘lion’) from ur (‘dog’) + maḫ (‘great’)

  3. kù-sig17 (‘gold’) from kù.g (‘precious metal’) + sig17 (‘yellow’)

Nouns may also be combined with a present participle. In the most common type the noun is the head and the participle behaves like and adjective (left-headed):

  1. ki-tuš (‘dwelling’) from  ki (‘place’) + tuš (‘sitting’)

  2. bar-dul5 (‘fleece’)  from bar (‘exterior’) +  dul5 (‘covering’)

In other compounds of this type the participle is the head, refering to a person that performs an action (right-headed):

  1. dub-sar (‘scribe’) from  dub (‘tablet’) + sar (‘writer’)

  2. kù-dím (‘goldsmith’) from  kù.g (‘precious metal’) + dím (‘shaper’)

Derivation is infrequent but the language may have two derivational prefixes (though  their status is controversial). One (nam) forms abstract nouns from concrete ones e.g. nam-lugal ('kingship') from lugal ('king'). The other (niĝ) is deverbal converting verbs into nouns e.g. niĝ-ba ('gift') from ba ('bestow').

  1. b)Verbal

Sumerian verbs form a closed class with a few hundred words (compounding and derivation don't occur). New verbal stems can only be created through stem reduplication (full or partial). In action verbs, reduplication may express iterativity, in stative verbs intensity.  The combination of a verb and a noun may express a new verbal meaning for which does not exist a stem (phrasal verb). Another way is by using certain prefixes that change the meaning of the verb.

A Sumerian verb may mark subject, direct object, indirect object, case, modality, negation, voice, number, and aspect. There are no real tenses.  A verb clause may stand on its own; nominal clauses are not essential. In other words, a verbal form alone is sufficient to make up a complete clause. When there is a nominal clause, its information is repeated in the verb (coreferential marking).

verb structure: up to nine prefixes and three suffixes can be attached to the stem in the following order:

preformatives: they precede all other verbal prefixes. Up to ten are attested but many are mutually exclusive. They play different roles. One of them is negation:

nu negates statements, na(n) negates wishes, requests, or commands, and bara negates assertions. The prefix u is only found in perfective forms and indicates that the action is prior to that expressed by a following verb; it can also be used to mark the passive voice. For the latter function two other vocalic preformatives, i and a, are commoner, but in contrast with u they are restricted, respectively, to the Northern Sumerian and Southern Sumerian dialects (see voice). The prefix a expresses modality like assertions, wishes, or commands. Ga expresses either a transitive subject or an intransitive subject of the first person; it is always the first morpheme of the verbal form.

ventive prefix mu: signals that an action or state is oriented towards the speaker or occurs close to the speaker. It is often associated with a noun phrase in the ablative or terminative case which tells ‘from’ where or ‘towards’ where the action takes place. When this prefix occurs with a verb of motion it changes its direction. Thus, the verb ĝen means ‘go’ without the ventive prefix, but ‘come’ with it.

prefix ba: it has several different functions expressing either a non-human indirect object, a change of state or the passive of intransitive verbs. Besides, it may express the indirect reflexive in which  two participants, the indirect object and the subject, are one and the same.

initial person-prefixes: indicate the referent of a following dimensional prefix (case-like marker).

dimensional prefixes: occur between an initial and a final person-prefix and are cognate with the case markers. They mark the indirect object (a, ra), the comitative (da), ablative (ta), terminative (ši), and location (ni, e). Forms with two dimensional prefixes are quite frequent and some forms have three of them. The primary use of the dimensional prefixes is to refer to some participant which has a role in the action or state expressed by the verb. As a rule, the first dimensional prefix of a verbal form is used together with an initial person-prefix, which specifies the gender, number, and person of what or whom the dimensional prefix refers to.

When there is more than one dimensional prefix, the second and third prefixes lack an initial-person prefix. Those dimensional prefixes that mark the indirect object are fused with the initial person-prefixes as follows:

final person-prefixes: in perfective verbs they express the transitive subject or the indirect object, in imperfective verbs they express the direct or indirect object. They do not have separate singular and plural forms. Plurality may be indicated by attaching a plural person-suffix (see below) to a verbal form in addition to a final person-prefix.

verbal stem: an unmodified, simple verbal stem consists of one or two syllables. Most verbs have the same stem in the perfective and in the imperfective but some verbs have a special imperfective stem (suppletive or partially reduplicated). Verbal plurality is indicated either by full reduplication or by a different stem which is lexically singular or plural. Verbal number quantifies actions and states signaling that they occur at multiple times and/or multiple locations.

imperfective suffix: the imperfective aspect is marked in many verbal forms by adding the suffix ed immediately after the stem.

person-suffixes: they are present in all verbal forms. In the imperfective they express the subject, in the perfective they express the subject in intransitive verbs and the direct object in transitive ones.

Thus, person-suffixes don't overlap in function with final person-prefixes (see aspect).

For comparison purposes we show here the three kinds of person-marking affixes in the Sumerian verb:

IPP = initial person-prefix; FPP = final-person prefix; PS = person-suffix; nh = non-human               

1. e and its plural ene are used only for transitive imperfective verbs; they express the transitive subject.

the nominalizing suffix a is used for verbs in subordinate clauses and to form the past participle.

aspect: both transitive and intransitive verbs may have perfective and imperfective aspects.  Aspect in transitive verbs is distinguished by having different systems of subject and object marking. Aspect in intransitive verbs is distinguished by a different stem (there is no differential inflection):


The imperfective stem is usually marked by the suffix ed or, less frequently, by  reduplication; some verbs have a special imperfective stem unrelated morphologically to the perfective one (e.g. 'go' has perfective stem ĝen and imperfective stem du).

The perfective expresses a completed action and frequently refers to a past action or a timeless action. It can also refer to the present but only in some types of subordinate clauses. In sentences expressing wishes it may have a future sense. Stative verbs are always in the perfective. The imperfective expresses an incomplete action and can be applied, thus, to the present or the future. It can also express a past progressive or simultaneous action or a past action that is relevant to the present.

voice: active, passive. The passive voice is marked in Southern Sumerian with the prefix ba; it always refers to an event, not to a state (dynamic passive). A stative passive can be formed in this dialect with the preformative prefix i. Northern Sumerian employs, instead, the preformative a to mark both a dynamic or a stative passive. Sumerian passives are always impersonal, they never include an agent. The preformative u can also be used to express a dynamic passive in both dialects.

non finite forms: present participle, past participle, imperfective participle.

Sumerian participles can be used both as verbal adjectives and as verbal nouns, in an active or passive sense depending on context. They can be negated with the prefix nu and their verb roots may be reduplicated to express number.

The present participle is the same as the perfective verbal stem and thus consists of the verb root without any affix with the possible exception of nu. It expresses a non-specific, timeless, action or state.

The past participle derives also from the perfective stem and is formed by adding the nominalizing suffix a to it. It normally expresses a specific action or state, usually a past one.

The imperfective participle consists of the imperfective verbal stem, formed by adding the suffix ed to the root or to a special imperfective stem if a verb has one (suppletive or reduplicated). It usually expresses a future action and may express necessity or obligation. Sometimes, it expresses a present or ongoing action or a past progressive action.

For example:

  1. stem: su.g (repay)

  2. present participle: su (repaid)

  3. past participle: gu suga (ox repaid)

  4. imperfective participle: suged (to be repaid, repaying)


    The syntactic function of nouns is usually conveyed by an enclitic case marker but word order plays a role as well. Sumerian has a rather free word order in clauses, but the verb is  always the last word of the clause, being preceded by a number of noun phrases. In a transitive clause the order is usually Subject-Object-Verb.

    The order in the noun phrase is more strict  beginning with the head noun and ending with one or more enclitic markers:

head noun-attributive adjective or participle-numeral-noun phrase in the genitive case-enclitic possessive or demonstrative pronoun-enclitic plural marker-enclitic case marker.

    The only essential elements are the head-noun and a case marker (though the latter may be zero).

Sumerian is a split-ergative language. Case marking is ergative as well as the inflection of perfective verbs (the subject of transitive and intransitive verbs are labeled with different markers). The inflection of imperfective verbs follows mainly a nominative-accusative pattern (there is no difference in subject marking of transitive and intransitive verbs).

    Let's analyze a couple of sentences:


  1. Eridug-a        é                        gu-a                    bi-n-du      

  2. Eridu-LOC    house-ABS    riverbank-LOC    CR-S 3hs-built

  1. He built a house by the riverbank at Eridu.

LOC: locative marker, ABS: absolutive marker; CR: coreferential marker, S 3hs: subject 3rd person-human singular.

-In this sentence there are three simple nominal phrases and one verbal phrase. The order is OV. The subject is not expressed by an independent pronoun or by a noun but is marked on the verb. The verb root is du ('build') and it expresses the perfective aspect because there is no imperfective marker or imperfective stem. The perfective refers usually to a completed event and is, thus, equivalent here to the English simple past ('built').

-The subject of a transitive verb in the perfective is marked by a final person-prefix (which should be, as here, immediately before the stem) and its object by a person suffix. Here, the final person-prefix is n indicating that the subject is 3rd. person human; we assume it is singular ('he') because there is no plural marker. There is no person-suffix,  i.e. object marking is Ø corresponding to 3rd person singular (either human or non-human). The prefix bi is a locative dimensional prefix coreferential to a locative marking in a noun-phrase.

-The noun é has Ø marking for case, i.e. it is in the absolutive which in a transitive clause, like this one, signals the object. The other two noun-phrases are marked with the locative clitic a to indicate place. Their relative order is of little importance.

-In summary, the subject of the sentence is 'he', the verb is 'built', its direct object is 'a house', 'by the riverbank' and 'at Eridu' are adjuncts.


  1. an-ta                éĝál                        a-mu-ra-ta-ĝen

  2. heaven-ABL    abundance-ABS     MOD-VT-IO 2s-CR-go

  1. ‘May abundance come from Heaven for you!’

ABL: ablative marker, ABS: absolutive marker; MOD: modal prefix, VT: ventive prefix, IO 2s: indirect object 2nd person singular, CR: coreferential marker.

-The verb ĝen ('go') is in the perfective (the imperfective uses the suppletive stem du) and is intransitive. The ventive prefix mu changes its direction to 'come'. The modal prefix a expresses a wish which in combination with the perfective aspect refers to a future event.

-The prefix ra signals the indirect object (the beneficiary) for the 2nd person singular ('you') while ta is coreferential with ablative marking in the noun-phrase. In intransitive verbs, like this, the subject is marked by a suffix which here is Ø i.e. 3rd singular.

-There are two noun-phrases, one with Ø marking corresponding to the absolutive (i.e. marking the subject of intransitive clauses) and the other in the ablative indicating source ('from').

-In summary, the subject of the sentence is 'abundance', the verb is 'may come', the indirect object is 'you (singular)', 'from Heaven' is an adjunct.


  1. bara-ba-řú-d-e

  2. NEG-MM-hold-IMP-S 1s

  1. ‘I will not hold her back’

NEG: negation marker, MM: middle marker, IMP: imperfective suffix, S 1s: subject 1st person singular.

-The sentence consists just of a verbal phrase. The verb řú ('hold') is marked with the imperfective suffix d (originally ed). In imperfective verbs the subject is indicated by a person-suffix and the object by a final-person prefix which are, respectively, e (from 1st. sg. en) and Ø (originally 3rd. sg. n).  The preformative bara expresses a strong negative statement.

    Main clauses are simply juxtaposed; only exceptionally the coordinating conjuction ú ('and also'), which is an Akkadian loanword, may be used to join them. Most subordinate clauses are nominalized and include a finite verb form with the nominalizing suffix a; they behave like nouns and take case markers. Subordinating conjunctions are also rare though three are attested: enna ('until'), uda ('if'), tukumbé ('if').

    Modality may be expressed by attaching the prefix a to the verb form but there is also a special imperative form in which the prefixes are placed after the stem instead of before. Yes/no questions have the same structure as declarative sentences from which differentiate by intonation.


Sumerian started to borrow words from Akkadian since the third millennium. Conversely, many Sumerian words entered Akkadian; in fact about 10% of Akkadian lexicon is  probably of Sumerian origin. Only few of these borrowings are attested in Old Akkadian, most of them are found in Old Babylonian. A small number of Sumerian words passed into Hebrew, some directly from Akkadian but most of them indirectly through Aramaic and other languages.

Basic Vocabulary

The Sumerian number system is sexagesimal. Numbers from 1 to 5, 8, 10, 20, 60, and 3600 are expressed by simple words. All other numbers are compound numerals combining two or more simple numerals. They can take case markers.

one: dis̆

two: min

three: es̆

four: limmu

five: ja, i

six: aš

seven: umin (earlier form i-min=5+2)

eight: ussu

nine: ilimmu (earlier form i-limmu=5+4)

ten: u


thirty: ušu

forty: nimin

fifty: ninnu

sixty: ĝešd

120: ĝeš-min

240: ĝeš-limnu

600: ĝešd-u

1200: ĝeš-u-min

3600: šar


mother: ama

brother: ses

sister: nin

son/daughter (child): dumu

head: saĝ

mouth: ka.g

eye: igi

hand: šu

foot: ĝiri

heart: šà.g

tongue: eme

Key Literary Works (forthcoming)

  1. © 2014 Alejandro Gutman and Beatriz Avanzati

Further Reading

  1. - A Descriptive Grammar of Sumerian. A.  H. Jagersma. Doctoral thesis, Leiden University (2010). Available online at:

  2. -Introduction to Sumerian Grammar. D. A. Foxvog (2013). Available online at:

  3. -The Sumerian Language: An Introduction to its History and Grammatical Structure (3rd edn.). M-L. Thomsen. Akademisk Forlag. (2001).

  4. -'Sumerian'. P. Michalowski. In The Ancient Languages of Mesopotamia, Egypt, and Aksum, pp 19-59. R. D. Woodward (ed). Cambridge University Press (2008).

  5. -'Sumerian'. G. Cunningham. In Concise Encyclopedia of Languages of the World, pp 1022-1025. K. Brown & S. Ogilvie (eds). Elsevier (2009).

  1. Top   Home   Alphabetic Index   Classificatory Index   Largest Languages & Families   Glossary