Elamite was spoken in the southern Zagros regions, which correspond to the ancient cultural-political entities of Elam and Anshan, and expanded into Akkadian-speaking Susiana.


IRAN vii, continued

(3) Elamite, with a Linguistic Sketch


As indicated above, Elamite was spoken in the southern Zagros regions, which correspond to the ancient cultural-political entities of Elam and Anshan, and expanded into Akkadian-speaking Susiana (see map, ELAM i, Fig. 1). Neither its northern nor eastern extensions (middle Zagros, Central Iran) nor its eastern extensions (Karmania, Baluchestan) are certain. Beginning with the turn to the first millennium B.C.E. its area was increasingly reduced by immigrating Iranian-speaking groups, its original center becoming the heartland of the Parsa. The Achaemenids continued the official use of Elamite (see Potts, 1999, for a comprehensive study of Elamite history and culture).

The language is a linguistic isolate, with possible remote relation to Proto-Dravidian (McAlpin et al., 1975; revised arguments in McAlpin, 1981). Four periods can be distinguished: Old Elamite (ca. 2600-1500 B.C.E.), Middle Elamite (ca. 1500-1000 B.C.E.); Neo-Elamite (ca. 1000-550 B.C.E.); Achaemenid Elamite (550-330 B.C.E.). A late survival may have been Khuzi, characterized as the private language of the nobles of Khuzistan by Ebn al-Moqaffaʿ (d. 757 C.E.), cited by Ebn al-Nadim (d. 995 C.E.) in his al-Fehrest (q.v.; comp. ca. 987, ed. Tajaddod, p. 15; cf. Lazard, 1971, p. 363).

The earliest scripts were the yet undeciphered Proto-Elamite (about 1,600 texts, ca. 3100-2900 B.C.E.) and Linear Elamite, which was used only briefly at the end of the third millennium B.C.E.). An adapted Akkadian syllabary system was used from about 2300 B.C.E. to the fourth century B.C.E.

The following sketch follows mainly Stolper, 2004. While there is general agreement on the identification of morphemes and paradigmatic sets, many aspects in phonology, morphology, and syntax remain uncertain, including semantic functions and even lexical meanings; accordingly translations may vary widely. Moreover, from the earliest documentation of Elamite to its later stages, linguistic changes and categorical shifts can be observed, which had become particularly prominent by the time of the Achaemenid texts and were conditioned by the increasing Elamite-Old Persian bilingualism of the authors of these texts.

Phonology. (1) Vowels. There was a system of five (or six) vowels, i e a (o?) u, and a neutralized vowel ə, without phonemic length distinction. Stress was on the initial syllable, which is indicated by frequent assimilation and contraction in final and medial position. (2) Consonants. The basic inventory of consonants used for the conventional broad transcription of Elamite is as follows: Chart 1.

The stops probably did not distinguish voice (voiceless vs. voiced), but tenseness (tense vs. lax), as in Dravidian. This is suggested by spelling variations in Elamite and also in Elamite transcriptions of foreign words and names (Reiner, 1969, p. 115; Khačikjan, 1995).

Spelling variations also suggest the existence of additional phonemes that are not reflected in the broad transcription. These include the following: (1) two affricates, (a) the palatal affricate č [tš], and (b) probably the dental c [ts]; (2) two additional laterals, (a) trilled vs. non-trilled r, and (b) retroflex vs. plain l. The glides w y appear to have been in complementary distribution with the vowels u i ; thus: Chart 2.

Morphology. Typologically, Elamite is an agglutinative language where derivational and categorical morphemes are attached to the right of the root (similar to Turkic). These suffixes are invariable, and express single grammatical functions.

Word classes are defined by distinct sets of suffixes. Transfer between classes is by derivation.

Roots may be either nominal only, or both nominal and verbal. They consist of simple sequences of consonant and vowels, mostly CVC, VCV, CV (kik “sky,” zana “lady,” igi “brother,” ki “one,” da- “to place”). Stems may be derived by the thematic vowels -a, -u, -i. Of these, -i marks both nouns and homophonous noun-verbs (pet-i “enemy, to revolt”).

Gender. The fundamental determinant in both nominal and verbal morphology is the opposition between two genders: animate (person, human and divine), and inanimate (things, concepts). It is marked by personal pronouns, so-called classifiers, and a set of finite endings. Animate gender has three classes: first person (“locutive”); second person (“allocutive”); third person (“delocutive”), and distinguishes two numbers: singular and plural. The delocutive inanimate gender does not mark plural in either nouns or verbs.

Nominal system. In nouns, animacy gender is either implicit or marked by derivational suffixes. The basic animacy markers are singular -r, plural -p. These imply an individual or a member of group, and a group or plurality, respectively. Such substantives may be derived from the following word classes: (a) from verb (agent noun), liba-r/p “servant/s”; (b) from toponym, hinduya-r/ hinduš-p “Indian/ Indians”; (c) from inanimate, meni-r “sovereign” (< men “crown”); (d) from adjective, risa-r “the large one.”

Inanimates distinguish several sub-classses, with varying degrees of productivity: (1) -me (abstract, productive): (a) from verb, liba-me “service”; (b) from person, sunki-me “king-ship”; (c) from inanimate, tuppi-me “text” (tuppi “tablet”). (2) -n (weakly abstract; may imply complexes and human collectives and locales, less frequent), huhu-n “wall,” meni-n “sovereignty” (besides meni-me), šuša-n “Susa.” (3) -te (mostly abstract; obsolete since Middle Elamite), hal-te “door,” haltam-ti “Elam.” (4) -(a)š (mostly agricultural terms and Old Persian loans), ara-š “granary,” kurta-š “worker.”

These classifiers correspond to the personal and demonstrative pronouns. Only pronouns distinguish the direct object form (“accusative”) from the unmarked subject/indirect object form (“nominative"/“dative”). The nominal classifiers also mark person/gender in the so-called conjugations II and III which are based on participles, as opposed to the personal endings of the so-called absolute conjugation I. The following table provides an overview of pronouns and nominal gender, person, number, and case marking (Middle Elamite; a. = animate): Chart 3.

By the stage of Achaemenid Elamite the language had developed case distinctions in the first and second persons of the pronouns, and far vs. near deixis. Adjectives are not marked (riša “great”). They can be derived by possessive -na: sunki-na “king-ly”; note the quasi-genitival construction: upat lansiti-p(a) “brickwork of gold” = “golden brickwork” (with plural -p).

Verbal system. The verbal base may be simple (da- “to put, place”); derived by the thematic vowels -u, -a, -i; reduplicated with initial CV- (tallu- > ta-tallu- “to write,” beti- > be-pti- “to rebel”; cf. Indo-Iranian intensives, etc.); or compound (mur-ta- “to put in place” < mur-u- “place”).

Aspect and agency. The fundamental category of the verb system is aspect: perfective vs. imperfective (completive vs. incompletive). Of these two, the marking of the perfective aspect is further determined by agency, which essentially implies transitive (“active”) action of an agent on a patient/object. Transitivity can be expressed in two ways: (1) transitive-active, with focus on the agent, and expressed by the six finite endings of conjugation I; or (b) intransitive-passive, with focus on the patient, and expressed by conjugation III, which is based on the perfective participle in -k. With its intransitive function, conjugation II also marks the perfective aspect of intransitive verbs. Imperfective action is expressed by conjugation II, which is based on the imperfective participle in -n. In contrast to the other two, this conjugation is neutral as to transitivity and focus.

Elamite does not differentiate tense (present vs. past), which is entirely contextual, as in other aspect languages. Thus, perfective transitive action more often than not implies past action, and imperfective action usually present action, but both aspects may refer to either tense. This fact is reflected in translations of forms of any of the three conjugations, and, particularly, modal forms of all three prove the point, as they invariably are rendered as presents (e.g., imperatives).

In addition to the three primary conjugations, there are secondary constructions. These include: (a) the durative or intensive conjugations in -ma, originally derived from all three conjugational bases, but generalized in Achaemenid Elamite; and (b) the so-called nominal construction, marked by the personal classifiers added to the stems of the three conjugations (, -k, -n). Table 4 provides an overview of the marking of aspect, gender, and person in the three main conjugations and the ma-conjugations (ME = Middle Elamite; NE = Neo-Elamite, AE = Achaemenid Elamite).

Examples: (1) primary conjugations (here all 3rd pers. plur.): (a) conj. I, hutta-h-õ “they did”; (b) conj. II, [hutta-k]-p “they have done, did”; (c) conj. III, [hutta-n]-p “they do, are doing”; (2) ma-conjugation (here derived from conj. I; older form with verbal noun and animate -r), pepši-r-ma-h “I renovated.”

It should be noted that these paradigmatic tabulations do not reflect the complexity of a verb system that in reality was very much in flux both synchronically and diachronically (cf. Khačikjan, 1993, and 1998, pp. 63-66, on typology; also Tucker, 1998).

TABLE 4. Elamite verb conjugation.

Calques on Old Persian increasingly influenced the system. For example, Old Persian subjunctives with future connotation are all rendered by conjugation III (imperfective), and Old Persian presents are mostly done by the imperfective durative constructions with -ma-n of conjugation III. Past marker. The marker of pluperfect tense, -ta, is found in Achaemenid Elamite, which developed from an earlier completive nominal clitic (akka makuš šari-š-ta “what the magus had destroyed” (cf. OP participial perfect 1st sing. -tá āham).

Mood. Non-indicative moods include two. The imperative had 2nd com. -t of the conjugation I in Middle Elamite (ume hap-ti “hear my prayer!”), but 3rd sing. -š in Achaemenid Elamite (mite-š halpi-š “go, defeat!”). The optative is marked by enclitic -ni (NEl hutta-h-š-ni “may they do,” conj. I; Ach. El. kata-k-t-ni “may thou live,” conj. II).

Negation is marked by in-. It is nominalized by the personal classifiers at earlier stages (in-ki, in-ri). Prohibitive is marked by anu (anu u i-r turna-n-pi “lest they (i-r-p) know me (u),” conj. III).

Syntax. Noun phrase. Elamite is a head-initial language and is characterized by Suffixaufnahme. That is, the head noun normally precedes dependent units, which are marked by the gender/number suffix showing agreement with that of the head noun. The range of dependent units thus marked includes the following: numbers; adjectives; pronouns, dependent nouns and noun phrases; adverbial phrases; as well as appositions (which allow agreement with all persons). In case of dependent noun phrases, there may occur “double case,” as the suffix of the dependent phrase is itself immediately followed by the suffix of the head noun (for adverbial agreement, see Adpositions below):

(1) Number. liba-r in-ri ki-r lit. “servant-3s [3rd sing.] not-3s one-3s,” i.e., “there is not a single servant”; this example also exemplifies the agreement of the negative particle in- with the gender/number of the subject of the clause. (2) Adjective. Peltiya, napi-r risa-rra "Beltiya, the goddess-3s great-3s.” (3) Dependent nouns. (a) possessor. Inšušinak, napi-r u-ri “Inshushinak, god-3s (of) I-3s”; (b) names and kinship terms, unmarked, DILBAT, zana-ø šušu-n-ra "Dilbat, the Lady of Susa-3s”; but -n with some kin sets, huhu-n nika-me “descendents-nt (of) we-nt”; (c) inanimate head (nt), siya-n Manzat-me “temple-nt (of) Manza-nt.” (4) Apposition. u, sunki-k anzan šušun-ka “I, king-1s (of) Anzan (and) Susa-1s”; sunki-r peti-r ak tari-r "(a) king-3s, (whether) enemy-3s and/or ally-3s.” (5) “Double Case.” riša-r [nap-ip]-r-a “the great(est) of the gods” (phrasal marker -a).

Beginning in Middle Elamite, and prominent in Achaemenid Elamite, is the replacement of these by the generalized ending coordination marker -na: par-ø Makištarra-na “family-3s Makištarra-na,” i.e., “M.’s family.”

Adpositions. There is a relatively small set of postpositions which had considerably increased in Achaemenid Elamite: -ma “in, on,” -ikki “to,” -mar “from,” -me “after.” Others are spatial-temporal nouns, some from verbs, and show agreement: peti-p [pat-p u-p] rabba-k-na “may (opt. -na) enemies (peti) be bound (rabba-k) beneath (pat “foot”) me (u)”; i-r [pat-r u-r] ta-t-ni “may thou (2nd common -t) put (ta-) him (i “he”) under me.” There is one preposition, kuõ “towards.”

Word order. Typologically, Elamite is a topic-prominent language. The basic order is subject-object-verb. However, the subject proper is often preceded by the unmarked topic (or it is the topic itself, or secondary topic). Typical are resumptive pronouns in immediate preverbal position: 1st sing. n-, 2nd sing. n-, 3rd sing. r-, 3rd plur. p-, as well as i-n for inanimate. In quasi-ergative function, these functioned as objects in the perfective-transitive conjugation I, but as subjects in both the intransitive-perfective conjugation II, and the neutral imperfective conjugation III: Chart 4.

In Middle Elamite up to three pronominal elements could form the sequence indirect object-agent/subject-direct object: ap u i-n dunu-h, lit. “they I it gave-I,” i.e., “I gave them it (the temple; perfective conj. I).

Complex sentences. A distinct boundary marker of phrases and clauses, and connector, is the clause enclitic -a, as is seen in the following examples for various types of subordination.

The relative pronoun is animate akka, acc. appa “who” and inanimate “what”: sunki-r akka ta-š-t-a “the king (sunki) who has put” (ta-, 3rd sing. conj. I; pluperf. -t). However, the dominant construction in Midle Elamite was clause nominalization, marked by gender agreement: siya-n [in-me kuši-h-ši]-me-a “the temple (siya, inan. -n) they (3rd plur. -h-š, conj. I) did not (in-) built (kuši-).” Nominalized clauses: from conj. I, with pluperfect -t, [kuši-š-t]-pe “having given birth” (3rd sing. -š, 3rd plur. -pe); u-n beša “he (who) created me” (acc. u-n).

The relatives introduce subject and object clauses: object clauses are nominalized, with inanimate agreement -me; here with locational relative pronoun muru “where,” Mid. El. [muru huma-h-š-ta] in-me durna-h “where they took [it], I (1st sing. -h) do not (in) know”; Ach. El. mur halmarriš hi kuši-k-a “where this fortress is built.”

Adverbial clauses. Various types of adverbial conjunctions are found in Achaemenid Elamite: sap “as, when,” kuš “until,” anka “if, when”; phrasal: sap innu “as long as,” mani sap innu “after”; with appa: appa anka “as,” sap appa “when.”

Quotations. Direct speech is indicated by verbs of speaking (turu-/tiri- “to tell, to speak”; na- “to say”) and is marked by quote-final -ma, with speaker-agreement: [i-r unsa-h-a] ma-r-a tiri-n-r-a “PN says ‘I exchanged (unsa) with him (i-r)’”; note enclitic -a.

Calques on Old Persian. Examples of calques from Old Persian abound in Achaemenid Elamite. Those include calques on the OP “eżāfa"-construction of OP N haya N: PN akka makuš “PN the maguš”; taššu-p appa PN-na “the troops (who are those) of PN”; PN akka GN-ma kurdabattiõ "PN (who is) the chief worker at GN.” Probably most radical was the shift in the verbal system from essentially aspect and nominal towards a tense system.


Bibliography of Elamite:

Gene B. Gragg, “Elamite,” in Ronald E. Asher and J. M. Y. Simpson,, eds., The Encyclopedia of Language and Linguistics, 9 vols., Oxford, 1994, III, p. 1104.

Françoise Grillot-Susini, Éléments de grammaire élamite, Paris, 1987.

Eadem, “Une nouvelle approche de la morphologie élamite: racines, bases et familles de mots,” JA 282/1, 1994, pp. 1-18.

Eadem, “Elamite Language,” in EIr. VIII, pp. 332-35.

Margaret L. Khačikjan, “To the Typological Characteristics of the Elamite Language,” in J. Zablocka and S. Zawadski, eds., Everyday Life in the Ancient Near East, Sulmu IV, Adam Mickiewicz University, Seria Historia 182, Poznan, 1993, pp. 143-50.

Eadem, “Notes on Elamite Phonology,” Studi Micenei ed Egeo-Anatolici (Rome) 35, 1995, pp. 105-9.

Eadem, The Elamite Language, Documenta Asiana 4, Rome, 1998.

Gilbert Lazard, “Pahlavi, pārsi, dari: les langues de l’Iran d’après Ibn al-Muqaffa’,” in Clifford E. Bosworth, ed., Iran and Islam: In Memory of the Late Vladimir Minorsky, Edinburgh, 1971, pp. 361-91.

Florence Malbran-Labat, “Le morphème ir en élamite,” Actances 7, 1993, pp. 139-60.

Daniel T. Potts, The Archaeology of Elam: Formation and Transformation of an Ancient Iranian State, Cambridge and New York, 1999.

E. Reiner, “The Elamite Language,” in Keilschriftforschung und alte Geschichte Vorderasiens. Altkleinasiatische Sprachen, HO 12/1-2, Leiden and Köln, 1969, pp. 54-118.

Matthew Wolfgang Stolper, “Elamite,” in Roger D. Woodard, ed., 2004, pp. 60-94.

Elizabeth Tucker, “The ‘Nominal Conjugations’ in Achaemenid Elamite,” in Maria Brosius and Amelia Kuhrt, eds., Achaemenid History XI, Studies in Persian History: Essays in Memory of David M. Lewis, Leiden, 1998, pp. 165-94.



(Gernot Windfuhr)

Originally Published: December 15, 2006

Last Updated: March 29, 2012

This article is available in print.
Vol. XIII, Fasc. 4, pp. 386-390