Akkadian (/əˈkeɪdiən/ ə-KAY-dee-ən; Akkadian: 𒀝𒅗𒁺𒌑(𒌝), romanized: Akkadû(m))[7][8][9][10] was an East Semitic language that is attested in ancient Mesopotamia (Akkad, Assyria, Isin, Larsa, Babylonia and perhaps Dilmun) from mid third millennium BC until its gradual replacement in common use by Old Aramaic among Assyrians and Babylonians from the 8th century BC.

It was written using the cuneiform script, originally used for Sumerian, but also used to write multiple languages in the region including Eblaite, Hurrian, Elamite, Old Persian and Hittite.

The Kültepe texts, which were written in Old Assyrian, include Hittite loanwords and names, which constitute the oldest record of any Indo-European language.

Cuneiform was in many ways unsuited to Akkadian: among its flaws was its inability to represent important phonemes in Semitic, including a glottal stop, pharyngeals, and emphatic consonants.

It was, however, notably used in the correspondence of Assyrian traders in Anatolia in the 20th–18th centuries BC and that even led to its temporary adoption as a diplomatic language by various local Anatolian polities during that time.

At its apogee, Middle Babylonian was the written language of diplomacy of the entire Ancient Near East, including Egypt (Amarna Period).

The language's final demise came about during the Hellenistic period when it was further marginalized by Koine Greek, even though Neo-Assyrian cuneiform remained in use in literary tradition well into Parthian times.

The Akkadian language began to be rediscovered when Carsten Niebuhr in 1767 was able to make extensive copies of cuneiform texts and published them in Denmark.

In the early 21st century it was shown that automatic high-quality translation of Akkadian can be achieved using natural language processing methods such as convolutional neural networks.

The reconstructed phonetic value[8] of a phoneme is given in IPA transcription, alongside its standard (DMG-Umschrift) transliteration in angle brackets ⟨ ⟩.

The rhotic ⟨r⟩ has traditionally been interpreted as a voiced alveolar trill /r/ but its pattern of alternation with ⟨ḫ⟩ suggests it was a fricative (either uvular /ʁ/ or velar /ɣ/).

[8][40] The following table shows Proto-Semitic phonemes and their correspondences among Akkadian, Modern Standard Arabic and Tiberian Hebrew: The existence of a back mid-vowel /o/ has been proposed, but the cuneiform writing gives no good proof for this.

[43] According to this widely accepted system, the place of stress in Akkadian is completely predictable and sensitive to syllable weight.

However, even in the earlier stages of the language, the dual number is vestigial, and its use is largely confined to natural pairs (eyes, ears, etc.).

The nouns šarrum (king) and šarratum (queen) and the adjective dannum (strong) will serve to illustrate the case system of Akkadian.

Later, the nominative and accusative singular of masculine nouns collapsed to -u and in Neo-Babylonian most word-final short vowels were dropped.

As is also the case in other Semitic languages, Akkadian nouns may appear in a variety of "states" depending on their grammatical function in a sentence.

Awīl-umman.NOMšū3SG.MASCšarrāqthief.ABSOLUTUSAwīl-um šū šarrāqman.NOM 3SG.MASC thief.ABSOLUTUSThis man is a thiefšarrumking.NOM.RECTUSlāNEGšanānoppose.INF.ABSOLUTUSšarrum lā šanānking.NOM.RECTUS NEG oppose.INF.ABSOLUTUSThe king who cannot be rivaledThe status constructus is more common by far, and has a much wider range of applications.

In general, this amounts to the loss of case endings with short vowels, with the exception of the genitive -i in nouns preceding a pronominal suffix, hence: māri-šuson.CONSTRUCTUS-3SG.POSSmāri-šuson.CONSTRUCTUS-3SG.POSSHis son, its (masculine) sonbut mārson.CONSTRUCTUSšarr-imking.GEN.SGmār šarr-imson.CONSTRUCTUS king.GEN.SGThe king's sonThere are numerous exceptions to this general rule, usually involving potential violations of the language's phonological limitations.

Most obviously, Akkadian does not tolerate word-final consonant clusters, so nouns like kalbum (dog) and maḫrum (front) would have illegal construct state forms *kalb and *maḫr unless modified.

"Alliances which man of Ešnunna (has)")The same preposition is also used to introduce true relative clauses, in which case the verb is placed in the subjunctive mood.

The final stem is the ŠD-stem, a form mostly attested only in poetic texts, and whose meaning is usually identical to either the Š-stem or the D-stem of the same verb.

The same rules as for the maPRaS form apply, for example maškattum (deposit) from ŠKN (set, place, put), narkabtum (carriage) from RKB (ride, drive, mount).

The derived verb then has the meaning of "make X do something" or "becoming X", for example: duššûm (let sprout) from dīšum (grass), šullušum (to do something for the third time ) from šalāš (three).

Body parts that occur in pairs appear in the dual form in Akkadian; e.g., šēpum (foot) becomes šēpān (two feet).

[citation needed] Examples: erbē aššātum (four wives) (masculine numeral), meat ālānū (100 towns).

In the following table the nominal phrase erbēt šarrū dannūtum ša ālam īpušū abūya 'the four strong kings who built the city are my fathers' is analyzed: Akkadian sentence order was Subject+Object+Verb (SOV), which sets it apart from most other ancient Semitic languages such as Arabic and Biblical Hebrew, which typically have a verb–subject–object (VSO) word order.

[51] Further evidence of an original VSO or SVO ordering can be found in the fact that direct and indirect object pronouns are suffixed to the verb.

Aramaic loan words were limited to the 1st centuries of the 1st millennium BC and primarily in the north and middle parts of Mesopotamia.

The completion of this work was hailed as a significant milestone for the study of the language by prominent academic Irving Finkel of the British Museum.