Imperial Aramaic and the Languages of Empire

Aramaic appears in writing in tenth- and ninth-century Syria and spreads continuously through the Fertile Crescent thereafter, gaining ground in the administration of the Neo-Assyrian and Neo-Babylonian empires. By the sixth century it was widely understood among scribes and officials, carried an entrenched framework of professional training, and possessed a technical vocabulary and a stock of established forms for letters, contracts and economic documents current in both east and west.^[1] Two texts mark the outer edges of that reach before Cyrus: an eighth-century inscription from Bukān (KAI 320) points to an early presence of Aramaic in Iran, and the Adon letter (TADAE A1.1) shows the language known somewhere in Egypt by the end of the seventh century at the latest.^[1]

Why Aramaic displaced Akkadian across the early first millennium is unsettled. Gzella holds that the driving forces "still defy explanation", and that the sheer diversity of the seventh- and sixth-century Aramaic witnesses "suggests a fairly low degree of conscious language policy", pointing instead to highly mobile speakers such as merchants and craftsmen.^[1] That judgement concerns the pre-Achaemenid spread and does not carry forward to the Achaemenid standard, which he treats as a separate and much more controlled phenomenon.

The usual short explanation, that an alphabet of some two dozen signs beat a syllabary of several hundred, is one Gzella lists among the reasons "suggested by scholars, but are hard to verify", alongside the adaptability of the language itself and the relatively neutral ideological connotations of an idiom that was not the native tongue of the Assyrian conquerors.^[1] Cuneiform administration continued for centuries beside Aramaic, and a scribe still had to learn vocabulary, document forms and professional convention whatever the size of the sign inventory. What the Persians acquired was an existing profession with its training already in place.

The terminology carries several different definitions. Reichsaramäisch, rendered in English as “Imperial Aramaic”, originally denoted the Achaemenid variety but has also been applied to Aramaic from the eighth century onwards. “Official Aramaic” has acquired a similarly broad usage. Gzella therefore chooses “Achaemenid Official Aramaic” to distinguish the chancery idiom from the Aramaic varieties current under Neo-Assyrian and Neo-Babylonian rule. A claim about standardisation depends partly on which texts are included in the comparison.^[1]

Folmer instead uses the chronological designation “Aramaic from the Achaemenid period”. Her stated purpose is “to avoid having the terminology anticipate solutions to the problems addressed in this study”. She also objects to treating “Egyptian Aramaic” as the name of a uniform variety: some texts found in Egypt, including Aršāma's leather letters, were written elsewhere, while the Egyptian material itself contains significant variation.^[3]

Folmer distinguishes three possible meanings of a standard: the variety most widely understood, the one used officially, and the one carrying greatest prestige. She notes that sociolinguistic accounts of standardisation involve selection, codification, elaboration of function and acceptance, and asks whether all these processes can be demonstrated for Aramaic. The question does not deny similarities between documents; it concerns the social history inferred from them.^[10] Rossi likewise qualifies the label “official”, describing Aramaic as a vehicular language while emphasising the continued use of local languages in private texts, provincial administration and some royal proclamations.^[6]

The case for a standard rests on features that recur across the whole documentary range. The most conspicuous is the writing of long consonants with an inserted n, as in ʾnt for /ʾattā/ "you" where older Aramaic writes a single t, as in ʾt. Gzella derives the habit from a nasalised pronunciation in the vernacular underlying the chancery idiom, but concludes that within Achaemenid Aramaic itself it "seems to have been a purely orthographical device", since it also targets sounds unlikely to be nasalised at all.^[1] Such a convention could be transmitted through scribal training. Its presence in documents from distant places does not establish how their writers or intended readers pronounced the words.

Other diagnostics point the same way. A construction using the passive participle with the agent marked by a preposition, šmyʿ ly, "it was heard by me" for "I have heard", is "presumably calqued on Old Persian though still rarely attested in first-millennium BCE Aramaic"; third-person plural object suffixes are replaced by independent pronouns, lʾ ʾytyt hmw, "I did not bring them".^[1] Each is a chancery habit with a contact history behind it, and Gzella hedges the Iranian derivation of the first rather than asserting translation from Persian in each instance.

The limits of linguistic localisation are substantial. Local dialects and earlier written traditions did leave marks, and the corpus carries a measurable degree of variation, mostly in spelling. Even so, Gzella reports that "attempts at identifying regional varieties within Achaemenid Aramaic proved unsuccessful, and the provenance of a text cannot be ascertained on linguistic grounds".^[1] Place of origin, genre and register, and the scribe's degree of formal education all pull at once, and no one of them can be recovered from the language alone.

The control case is the Hermopolis correspondence (TADAE A2.1–7) set beside the letters of Aršāma (TADAE A6.3–16). The Aršāma letters show the chancery idiom in its purest form; the Hermopolis papyri are composed in a distinct regional variety with non-standard phonetic spellings and morphological peculiarities. They nonetheless carry a few traces of chancery orthography, including one instance of mndʿm for mdʿm, "whatever". Gzella reads this as an older tradition being absorbed: the Hermopolis letters "illustrate a tradition of Aramaic letter writing in Egypt that precedes the Achaemenid administration but gradually came under its influence".^[7]

Gzella begins his account with institutional promotion: "When the Achaemenid authorities promoted an Aramaic dialect to the official administrative idiom of their vast and highly heterogeneous empire, this was a sensible if not an obvious choice."^[1] The promotion is an inference from the uniformity of the corpus, and the mechanism behind it is hedged accordingly: the unification follows "presumably as a result of a large-scale administrative reorganization and unification under Darius I and Xerxes".^[1]

Tavernier presses further, to intention. Reconstructing how an order moved between languages, he concludes that Aramaic was placed deliberately between the elite language and the vernaculars:

This implies that the Achaemenids consciously used Aramaic as a kind of empire-wide bridge between the elite language (Old Persian) and the vernacular languages (Elamite, Egyptian, etc.), which points to an active Achaemenid language policy.

Tavernier, in Aršāma and his World, III (2020), p. 96^[4]

Van Bladel attacks a different target. He denies any known founding enactment declaring Aramaic the empire’s official language and stresses the simultaneous governmental use of several other languages.^[5] In the same breath he adopts Gzella's formulation, specifying that Aramaic should be called "a 'highly standardized supra-regional chancellery language'", and he endorses Tavernier's distributional finding: Aramaic "seems to have been used particularly for writing at the satrapal level of administration, meaning that of provincial governors", while "Documents serving lower levels of administration were written in regional languages."^[5]

These arguments address different aspects of the evidence. Inferring institutional promotion from a standardised written corpus is compatible with the absence of a surviving decree that declared Aramaic official. That absence alone neither proves nor disproves deliberate encouragement by the authorities. Gzella argues from the shape of the corpus, Tavernier from the shape of the transaction, van Bladel from the historical linguistics of the successor languages; the pressure van Bladel applies falls on reach, and he narrows even the linking function to "At most, it was a language used to link levels of administration under the Achaemenids, but even this is not certain."^[11] Folmer asks a further question: which processes of standardisation can be demonstrated by sociolinguistic criteria.^[10] A fifth position belongs to Jennifer Hilder, who works from the Egyptian end and reads the choice as coercive: "Thus Aramaic had inherent connotations of Persian power and authority; it was a language imposed on Egyptian administrators, and writing itself was of course a highly specialized skill."^[8]

The Bodleian leather letters of Aršāma, satrap of Egypt at the end of the fifth century, form a substantial surviving set of Achaemenid official correspondence. Hilder's study shows how much of their authority is carried by the shape of the document before any instruction is reached. The address places the senior party first, so that an opening may run either mn ʾršm, "from Aršāma", or ʿl ʾršm, "to Aršāma", according to which end of the exchange he occupies. "By giving himself precedence, Aršāma declares his superiority and asserts his right to privilege and deference. When others begin a letter ʿl ʾršm, they affirm these claims and signal their willingness to cooperate."^[8] The convention is not universal: Hilder notes that in the Persepolis Fortification archive the addressee is always named first, "which may reflect the pragmatic nature of the correspondence".^[8]

The greeting is the sharper instrument, because it is almost always absent. "As the highest-ranking official in Egypt, Aršāma did not give a polite greeting to anyone except Artavanta."^[8] Christopher Tuplin, commenting on the same pattern, judges that "Given Aršāma's high status, Artavanta's receipt of a greeting is in the circumstances quite striking", and the courtesy is taken to mark a man holding a very important position, perhaps acting as satrap during Aršāma's absence.^[12][8]

TADAE A6.7 is one of those letters, and its courtesy sits on top of an unusually concrete demand. Aršāma opens by sending Artavanta "(wishes for) much peace and strength", reports that "(there is) peace in this (place) before me", and wishes the same in return.^[13] He then lists thirteen men by name, one by one, each entry closing with the numeral 1 and the tally recorded at the end: Pariyama, Ammuwana, Saraka, Sadasbinazi, Sarmanazi, Kaʾ, Bagafārna, Piyatarunazi, Asmaraupa, Muwasarma and others whose names are damaged. "(In) all, 13 men. They were appointed (as) pressers (?) among my domains which are in Upper and Lower (Egypt)."^[13] The occupation carries the edition's query, and the Iranian loan behind it, *abišavaka-, is itself a reconstruction.

After, when Egypt rebelled, and the (armed) force was garrisoned, then that Pariyama and his companions were not able to enter into the fortress. After, the wicked [I]n[ḥ]arou seized them, (and) they were with him. Now, if (it seems) like a good thing to you, let an order be issued by you, so that a person should not do anything bad to that Pariyama and his companions. Let them be released. Let them do my work as previously.

Aršāma to Artavanta, TADAE A6.7, trans. Taylor^[13]

The letter shows the whole apparatus at work on a single object of interest. Aršāma calls the thirteen "my slaves"; the request is routed through another man's authority rather than issued directly; and the closing clause states the purpose, which is the recovery of their labour.^[13] The outside of the document repeats the subject in shorter form for whoever handled it, "Concerning the Cilicians [slaves] of mine [who were not] able [to enter] into Miṣpeh (?)".^[13]

TADAE A6.9 orders the provisioning of Nakhtḥor and his party on their journey to Egypt. Seven officials are named, but the geographical information is uneven. Marduk's place is lost; Nabudalani is assigned to Laʿir and Zātavahyā to Arzuḥin. Upastābara is associated with Arbel and two further places, one partly preserved. Bagafārna is at Saʿlam; Frādafarnā and a second official whose name is partly restored are at Damascus. This address list does not establish a complete itinerary between securely identified modern locations.^[14]

The ration schedule distinguishes Nakhtḥor from the others. For each day it specifies two measures of white flour, three of inferior(?) flour, two of wine or beer and a further item whose reading is uncertain, translated “[cheese (?) or lamb (?)]”. His ten servants receive one measure of flour each, and fodder is allocated in proportion to his horses. Three additional servants of Aršāma, two Cilicians and an artisan, receive a measure of flour each. The grain categories, quantities and uncertainties belong to Taylor's edition; the measures are not converted into modern volumes.^[14]

Give them this ration, from (one) official to (the next) official, in accordance with the route which is from province to province, until he shall reach Egypt. And if he should be in (any) one place more than one day, then for those days you shall not give them further rations. Bagasravā knows this order. Rāšta is the scribe.

Aršāma to his officials, TADAE A6.9, trans. Taylor^[14]

Aršāma directs the recipients to supply the provisions from his estate in their provinces. He limits entitlement at any one place by withholding further rations for a stay beyond one day. The closing subscript separates two functions that recur across the corpus: a named man who knows the order, and a named scribe associated with its preparation. The document authorises a specified party over a specified road; it establishes nothing about what an unnamed traveller carrying an Aramaic letter could have claimed.

The same subscript appears on letters of quite different temper. In TADAE A6.8, Aršāma relays a complaint from his official Psamšek that Armapiya and the armed force under his control "do not obey me in the affair of my lord which I am telling them", orders compliance, and closes with a threat: "If Psamše[k] afterwards should send me a complaint about you, you will be questioned forcefully, and a severe sentence will be produced for you." The subscript runs "Bagasravā knows this order. Aḥpepi (or ʾḤWPY) is the scribe."^[15] The same man stands behind a ration schedule and a disciplinary order, which is one reason the categories of public and private resist separation in this dossier.

TADAE A6.11 reports a request, a conditional grant and instructions for its implementation within one letter. Peṭosiri, described as a plenipotentiary, reported that his father Pamun had died "when there was unrest (?) in Egypt", the query being the edition's, and that the household had perished with him, leaving a domain of thirty ardab seed-place abandoned. He asked to occupy it as heir. Aršāma restates the report as a condition and attaches two exclusions:

Now, Aršā[ma] says thus: "If it is thus, in accordance with these words which Peṭosiri has sent [to me,—that] he whose name is [Pamun] his father, that one, when there was unrest (?) in Egypt, perished with the people of [his household, and] his domain, of that Pamun his father, a seed-place of 30 a(rdab), that one was abandon[ed, and to my estate] it was not made over, (and) it was not given by me to another servant of mine,—then I give the domain of that Pamun to Peṭosiri."

Aršāma to Nakhtḥor and his colleagues, TADAE A6.11, trans. Taylor^[16]

The grant is contingent on the report being true and on the land having neither been made over to Aršāma's estate nor allocated by him to another servant, and the beneficiary assumes his father's tax obligation, "in accordance with (that) which Pamun his father was previously paying". Artāvahyā knows this order; Rāšta is again the scribe.^[16]

On the outside, after the Aramaic address, stands a single line in Demotic: "Concerning the fields of Pamun which I have given to Peṭosiri."^[16] It summarises the transaction for a reader working in Egyptian without reproducing the conditions, the exclusions or the tax clause. One object therefore carries two written languages doing two different jobs, which is firmer evidence of bilingual handling than any reconstruction of a lost parallel dossier. Hilder nonetheless keeps that possibility open, allowing that "it could be suggested that a parallel archive existed where Demotic was used extensively on materials which have not survived", or that bilingualism was simply commonplace among the administrative workforce.^[8]

Whether such estate business belongs to the satrapy or to the household is a question Tuplin declines to settle in either direction. Commenting on the term trbṣ in TADAE A6.10, the letter in which Aršāma orders his managers to seek artisans, bring them to his courtyard and mark them with his brand, he argues that the letter is not straightforwardly satrapal business, while cautioning against treating satrapal and personal affairs either as hermetically separate or as simply undifferentiated.^[17] Gzella makes the corresponding observation about the dossier as a whole: the letters reflect the spelling conventions, epistolary phraseology and procedural mores of the Persian chancery "although they are not predominantly concerned with the public sphere".^[7]

The formulae name people, and Tavernier builds his reconstruction from them. His model places a specialist class, the Neo-Babylonian sēpiru and the Elamite teppir, at the point where languages met: a man at that level "knew two or three languages (Old Persian, Aramaic, and, if necessary, Egyptian or Elamite) and at least two scripts (alphabetic Aramaic, Mesopotamian/Elamite cuneiform or Egyptian Demotic script)".^[4] He supports the claim with independent evidence for individuals who commanded two scripts: the Aramaic epigraphs on Elamite tablets from Persepolis, the Aramaic dockets on Babylonian documents, and, more tentatively, Aramaic Fortification tablet 261, on which the personal name ʾršyn appears to use a Greek alpha in place of an Aramaic aleph. The aleph also occurs elsewhere in the same text.^[4]

Tavernier explicitly excludes the lower-level memoranda from a uniform translation procedure:

It should be stressed that it is not the case that the entire Achaemenid administration was situated at the sēpiru/teppir-level. If that were true, it would mean that all the Elamite tongue-shaped memoranda would be translations from Aramaic originals. This is not probable. These memoranda belong to a lower level of administration, totally controlled by simple scribes, such as the ‘Persian puhu copying texts’ (Tavernier 2008: 64–5).

Tavernier, in Aršāma and his World, III (2020), p. 95^[4]

He adds that regional variation should be expected, since "The way it worked in Egypt was not necessarily exactly the same as the way it worked in Persepolis or Bactria."^[4] Rossi notes that Tavernier himself envisages this sequence only for a minority of the tablets, while the more usual procedure probably involved direct translation from Old Persian into Elamite. Rossi's own reservation concerns the crucial assumptions on which the reconstruction depends; he also points to Vallat's different interpretation.^[6] Shaked's proposal about who did the work moves in the other direction again: knowledge of Aramaic among speakers of Iranian was probably confined to the scribes, and the earliest of them may have been native speakers of Aramaic who learned Persian, later replaced by native speakers of Persian who learned Aramaic as a written language.^[6]

The subscripts distinguish related functions without always assigning them to different people. A6.9 and A6.11 name a man cognisant of the order and a separate scribe. In Bactrian A1, Hashavakhshu is described as both the scribe and the man aware of the command. A2 names two people: Daizaka as scribe and Athfiya as the person in charge of the command. These arrangements share an interest in identifying responsibility, but they do not constitute an identical pair of personnel throughout the Bactrian corpus. The spoken language used at each handover, and an intermediate Aramaic draft behind a surviving Elamite memorandum, remain matters of reconstruction.^{[14][16][18][19]}

A more radical proposal has shadowed the question since 1979. Ilya Gershevitch argued that Achaemenid Elamite was Old Persian written ideographically, so that an Elamite tablet was read aloud in Persian and the Old Persian monumental versions were duplications made for display. Rossi's summary is unsparing about its reception: Gershevitch "never revealed the details of this 'alloglottography' theory" and "therefore was strongly criticized by Elamitologists", while Gonzalo Rubio argued in its favour. Rossi refers readers to Wouter Henkelman's extended assessment.^[6]

The single Old Persian administrative tablet from the Fortification archive, published by Matthew Stolper and Tavernier in 2007, put pressure on the theory from an unexpected direction. Rossi's judgement is that the document "actually contravenes stated expectations: at least one scribe wrote Persian language in Persian script and expected someone else to know how to read it", against Gershevitch's insistence that there was never "the slightest chance for Old Persian written in Old Persian script … to have ever been in general use", and against Rüdiger Schmitt's separate conclusion that Old Persian played no part in administration proper.^[6] Rossi allows only hypotheses about the tablet, and notes that it may be the sole survivor of a larger group, or an isolate.^[6]

The question recurs for Aramaic in the east. Naveh and Shaked raised the possibility that the Bactrian letters were dictated in Iranian and recorded on the fly in Aramaic by the scribes; Gzella weighs that against an alternative, that the mixture of Iranian and Aramaic elements "reflects an institutionalized code similar to the Latin adstrate in legal or medical technical language". Comparing the Aramaic Aśoka inscriptions as material from a similarly peripheral area, he concludes that "it seems unnecessary to assume an incipient stage of Iranian-Aramaic 'alloglottography' here".^[2] These are distinct proposals about distinct corpora, and neither should be collapsed into the later Iranian practice of writing Aramaic word-forms as heterograms, which belongs to the post-Achaemenid history of the script.

The Persepolis Fortification archive provides a large, archaeologically associated body of evidence for the interaction of Aramaic and Elamite. It includes monolingual Aramaic tablets alongside Elamite records, some of which carry Aramaic ink epigraphs. Other objects, including Aršāma's letter with a Demotic note and Babylonian tablets with Aramaic dockets, also combine languages; Persepolis is especially valuable because annotations can be examined within a substantial administrative archive.^[20][4][16]

Published counts describe different stages in the study of the finds. In 2007 Kuhrt reported at least five hundred Aramaic texts and suggested that the figure might approach a thousand. Henkelman's 2021 survey gives about 840 monolingual Aramaic tablets and 250 or more Aramaic epigraphs on Elamite tablets. He distinguishes these estimates from the initial report of more than thirty thousand tablets and fragments, which is not a count of thirty thousand complete Elamite documents. His estimate for the Elamite material is fifteen thousand or more original tablets, at least seven thousand of them legible and analytically meaningful.^[21][22]

PF 1809 shows the two systems on one tablet. The Elamite text records Parnaka ordering six BAR of flour as rations for two servants at Midush in the village of Himpira, in the seventh month of year 23 of Darius, with the writer and the man who communicated the message both named. Along the left edge, in two lines of Aramaic, stands a compressed reference: "In year 23. To Midush" with the place-name and a following word both carrying the edition's query marks.^[23] The notation identifies the document without reproducing it. PF 1587 carries a fuller one, "Ramakka brought (it to) Persepolis, (for) rations of diggers of stone", and Kuhrt observes that the gloss affects the reading of the Elamite: Richard Hallock took the workers for sculptors, "but the Aramaic gloss at the end suggests that they may be no more than quarrymen".^[24]

PF 1810 puts the writers themselves into the ration system:

21 BAR flour and 12 BAR grain issue to them as rations, to Babylonian scribes on parchment, who are assigned by me. (As) their ration, 7 men monthly are/will be receiving each 3 BAR, and 6 boys are/will be receiving each 2 BAR. Total 13 workers. In month XII, year 23, this sealed document was delivered. Takmaziya wrote. He received the dumme from Yauna.

Parnaka's order, PF 1810, trans. Kuhrt^[25]

Kuhrt glosses "on parchment" as "I.e. writing in Aramaic", and holds it "virtually certain that the Babylonian secretaries in the head-office at Persepolis wrote in Aramaic".^[25][21] The tablet also separates the writer, Takmaziya, from Yauna, the man who supplied the dumme, the message or draft from which the record was written. Kuhrt notes that Yauna's name means "the Ionian", which may indicate that a Greek occupied a high position in Parnaka's service.^[25] The graded rations distinguish seven men from six "boys".^[25] Kuhrt cautions elsewhere that the Elamite category puhu, translated as boys, may designate specially trained servants rather than juveniles.^[23]

Annalisa Azzoni and Stolper's study of the commonest epigraph turns these scattered notations into an argument. Their conclusions are set out as interim and explicitly bounded: until a full edition and functional study of all the Aramaic epigraphs is complete, "conclusions about this single but significantly frequent epigraph are provisional".^[20] Within that limit they hold that "Most or all of the Aramaic epigraphs on clay tablets in the Persepolis Fortification Archive were products of the handling of information, not substantive records of information", and that "A few epigraphs on registers imply that they were written as part of a process of arithmetic verification, by scribes who could read Elamite and write Aramaic."^[20]

One tablet converts an inference into evidence. Gzella writes that some Fortification texts refer to documents on leather, "the language of which, in all likelihood, was Aramaic", reasoning from the unsuitability of cuneiform for ink on a flat surface.^[7] Azzoni and Stolper report that "The epigraph on Fort. 2178-101 indicates that ns(y)ḥ refers to copying documents onto leather, in Aramaic script and language."^[20] That establishes the practice for one identified document rather than for every lost leather record the archive mentions. The study also discusses the Elamite occupational contrast behind Kuhrt's "scribes on parchment": documents refer to "scribes on leather" (tippip kušMEŠ-ukku) and at least one to a "scribe on clay" (tipira halat-ukku), terms that "suggest professional specialization by medium and language".^[20]

Nevertheless, such specialization was clearly not absolute. The epigraphs, again like their Babylonian and Assyrian counterparts, imply functional biliteracy, not a great surprise in the literate support personnel at the polyglot court of a continental empire.

Azzoni & Stolper, ARTA 2015.004, p. 44^[20]

The excavation of Persepolis between 1936 and 1938 produced a second Aramaic corpus from the Treasury building, and its editorial history shows how the interpretation of an established corpus can change. Raymond A. Bowman's editio princeps opens with the count: "the remains of some 269 mortars, pestles, plates, and trays of a distinctive flinty green stone were found in the Treasury building. About 203 of those objects were more or less legibly inscribed in ink in Aramaic writing."^[26] Later summaries give different figures: Gzella refers to 163 objects bearing inscriptions, while Rossi speaks of 163 inscriptions. Neither figure equals Bowman's total of objects or his approximate count of legibly inscribed pieces; the differing summaries do not by themselves explain the discrepancy.^[7][6]

Bowman read the texts as records of a haoma ritual, and his title names them ritual texts. His own preface is markedly more tentative than his title: "Without pretending to have proven a thesis about the texts, I have presented this work simply as what appears to be the most plausible conjecture in harmony with all known facts."^[27] He had already discarded a supporting assumption of his own. Because the objects had been taken for ritual equipment, their shattered condition looked like deliberate breakage, but the recovery of several undamaged pieces and the fact that the royal tableware lay shattered among them "made vandalism a more likely explanation of their broken condition".^[26] A further caution belongs to the objects themselves: they were later dumped in an unused room at the Persepolis Museum and had to be re-excavated and cleaned, after which the faint Aramaic writing almost disappeared.^[27]

The ritual reading did not survive. Gzella states that the ritual interpretation had been abandoned, citing Naveh and Shaked's re-examination of 1973; Rossi states the replacement, that the inscriptions "probably record the manufacture and their payment to treasurers in Arachosia (not their 'ritual' use as believed by its editor Bowman)".^[7][6] A separate caution concerned the older readings of the Fortification glosses: writing in 2007, Kuhrt stated that they had not been collated since Bowman read them. That describes the state she reported at the time; Azzoni and Stolper's 2015 study subsequently re-examined the epigraphic evidence and supplied detailed new readings.^[24][20]

The Khalili collection, edited by Joseph Naveh and Shaul Shaked and first published in 2012, extended the documentary map to the far north-east: thirty letters and lists on leather, together with eighteen inscribed wooden tallies. At its core are eight letters sent to Bagavant, who resided at Khulmi some eighty kilometres east of Bactra, by a certain Akhvamazda.^[2] The titles need care. Bagavant is titled in the documents, called the governor, pḥtʾ, at Khulmi; Akhvamazda's rank is inferred. The editors write that "It may be deduced from the contents of these letters that Akhvamazda is the satrap of Bactria", and Gzella states the negative plainly: like Aršāma in his own correspondence and Parnaka in the heartland sources, Akhvamazda "is never referred to with that title", but only with the honorific "my lord".^[18][2]

Document A1 is a complaint and its redress rather than a bare instruction, and it shows a provincial chancery quoting evidence into the record. Vahuvakhshu son of Chithrabarzana had complained that Bagavant and the magistrates "removed (things) from the camel-keepers, my apprentice-servants; they despoiled and detained them, and extracted (from them) a tax which they (viz. the camel-keepers) are not obliged (to pay), not letting them guard the camels of the king". The complainant's words stand inside Akhvamazda's letter, followed by the history of the case: Bagavant was interrogated, then left for Khulmi before the decision issued, and the complaint was renewed.^[18]

because you are removing (things from) those men and are imprisoning (them) against my decree, when you come (to me), you will be interrogated. But now, what you have removed, give back to them. […] [D]o not extract (from them) a surcharge (?) more than what they owe. Also, release those men, the camel-keepers, to do their own work. Do not impose on the land [a tax] which they do not have to pay. Hashavakhshu the scribe is aware of this command.

Akhvamazda to Bagavant, A1, trans. Naveh & Shaked^[18]

The external address and dateline complete the parallel with Egypt: "To [blank space] Bagavant and the magistrates, who are in Khulmi." and "… of Marḥešvan, year 6 of Artaxerxes the king. Concerning a tax. Bring this letter."^[18] The subscript invites comparison with Rāšta's in A6.9: both letters identify the personnel through whom an instruction was recorded, although their administrative settings and the wording of their subscripts require separate attention.

A2 shows the cost of that uniformity to a modern reader. The letter concerns a commodity to be taken out of Akhvamazda's house, and the editors report that "The crucial word here is ḥlʾ, which may mean 'sand', but which may alternatively mean 'vinegar'", the two words sharing a consonantal spelling; the contents list accordingly describes the document as concerning "the issue of vinegar (or the clearing of sand)".^[19] The house itself is glossed as "apparently a caravanserai, probably a station on the road in the Persian postal system", which is an inference from context.^[19] A2 closes its instructions with the subscript "Daizaka is the scribe and Athfiya is in charge of the command." Its date is 3 Marḥešvan of year 8 of Artaxerxes.^[19]

The Bactrian material carries a provenance problem that no reading can remove. The documents were bought on the market, and their attribution to a satrapal archive at Bactra is an argument from content. The editors state the reservation themselves: the main part of the group appears to derive from such an archive, but "The other documents in the Khalili collection may also come from the same source, but at the moment it is not possible to establish with certainty that they derive from the same archive."^[28] The place-names are in the same condition. Maithanaka, where C1's supplies were issued, is a locality "on the way from Bactra (modern Balkh) to Varnu (location uncertain)", and the editors offer three competing accounts of how the document reached the archive, including the possibility "that C1 is a copy, sent back to the archive, while the original document was attached to the supplies".^[29][28]

The dates are editorial reconstructions in a stronger sense than the conventional span suggests. The dated leather letters name only "Artaxerxes the king" without a numeral; the editors supply the third of that name, writing that the dates "all refer to a date in the reign of Artaxerxes, which must be the third of this name", and most of the wooden sticks are dated to the "third year of Darius, evidently Darius III".^[28] In the index of royal names the supplements are visible as such, with the numerals bracketed and, for C1, part of the king's name itself restored.

C1 is the most historically charged document in the collection and the most heavily reconstructed. Its header reads "Kislev, year 1 of Artaxerxes, presumably Artaxerxes V, as the royal name assumed by Bayasa / Bessus", and the verso records "Provisions to Bayasa in Maithanaka, when he passed from Bactra to Varnu."^[29] The commentary makes the chain of inference explicit: "In view of the fact that Bessus is mentioned in this document, it seems that the reigning king referred to here under the name Artaxerxes is Bessus himself", and, since Arses had been Artaxerxes IV, "Bessus, who adopted the name Artaxerxes, should be regarded as Artaxerxes V."^[29] The editors hedge twice in the introduction, writing "If this refers, as we think, to the first and only year in the reign of the pretender Bessus" and "It is remarkable that, if our understanding of the document is correct, Bessus is mentioned by name".^[28] On that reading the supply list was drawn up for a man in flight from Alexander, in a language and a format inherited from the administration he had usurped.

C4, the latest dated piece, depends on a chronological convention rather than on an absolute date. The editors assign it to the seventh year of Alexander, "which, by the Babylonian reckoning, is equivalent to 324 bce", and state the alternative in the commentary: "The alternative reckoning for the reign of Alexander, which starts with his Macedonian accession in 336 bce, would give as date 15 June 330, when Alexander had not yet conquered Bactria; this can hardly be relevant here."^[30] The conventional span of 353 to 324 BCE for the collection is therefore the editors' own, resting on restored regnal names, supplied numerals and a chosen era. On the editors' chronological reconstruction, these documents carry related administrative practices across the conquest: a recognisable repertoire of chancery formulae, scribal subscripts and commodity accounting. The inference of continuity depends on the attribution of the dated pieces; similarities of format alone cannot place undated documents on opposite sides of a political transition.

The correspondence of Yedaniah son of Gemariah, a leader of the Judean community at Elephantine, shows how Aramaic conveyed a request through several centres of authority. Two versions of the petition for rebuilding the community's temple, TADAE A4.7–8, survived among its papers. Gzella treats them as drafts with modifications. Their preservation gives access to the petitioners' wording, without proving that every surviving copy was dispatched in precisely that form.^[7]

The petition dated to 407 BCE addresses Bagohi, governor of Judah. It recounts the temple's destruction when Aršāma was away with the king and identifies the local officials and troops alleged to have carried it out. Its opening invokes divine favour for Bagohi before Darius II and the princes. The petitioners then ask him to use his obligations and friendships in Egypt to secure rebuilding, promising offerings and prayers in his name. These are the claims and inducements of a petition, rather than an impartial record of the dispute.^[31]

The closing lines disclose a second approach and address the satrap's knowledge:

Further, about all of this we have sent a letter in our own name to Dalaiah and Shelemiah, the sons of Sinuballit, governor of Samaria. Further, all that has been done, was not known to Arsham.

The Elephantine petition, as presented in Kuhrt's sourcebook, 17.32(iii)^[31]

A related memorandum, TADAE A4.9, records what Bagohi and Dalaiah had said to an unnamed bearer. It directs that person to speak before Aršāma in Egypt about rebuilding the altar house. The change of documentary form is significant: a petition describing the community's losses is followed by a short record of words to be delivered elsewhere. Kuhrt understands the memorandum as a jotting, perhaps derived from an oral instruction, which could prompt a courier's presentation at the satrapal court. It should not be silently converted into a surviving royal decree.^[31]

The ritual provisions also differ. The petition asks for offerings, incense and burnt sacrifices; the memorandum mentions offerings and incense but omits burnt sacrifices. A further text from five named allotment-holders makes the exclusion of sheep, ox and goat sacrifices explicit and offers payment conditional on rebuilding. Its fragmentary wording records silver and a thousand artabas of barley. Kuhrt interprets the sequence as a resolution involving a restriction on animal sacrifice, while treating the reason for that restriction as uncertain. The texts preserve requests, recommendations and conditions more securely than they document the execution of a final settlement.^[31]

Taken together, they show written Aramaic being used by petitioners, provincial authorities and an intermediary in a transaction crossing between Egypt, Judah and Samaria. The terms did not remain identical as the request moved through that correspondence. The differences can be examined because several documents survive; no comparable completeness should be assumed for an isolated letter from another province.^[31]

Beside the correspondence stands a large body of private law. The oldest datable Aramaic legal text from Egypt is a lease of 515 BCE from Korobis in the Delta, which already shows traces of a preceding Aramaic scribal tradition; the family archives of Mibtahiah (471–410 BCE) and of Ananiah (456–402 BCE) contain bequests, marriage contracts and property transactions that can be read against one another across decades. Twenty-one further loose documents preserve deeds of obligation, conveyances, marriage documents and judicial oaths.^[7] Gzella regards these as the beginning of a durable Aramaic legal tradition, with similar or identical formulae recurring in the Hellenistic and Roman Near East and in the Babylonian Talmud, while noting that the tradition's origin, whether Mesopotamian, West Semitic, Egyptian or Iranian, remains disputed.^[7]

The comparative material is uneven in quality. Twenty-seven fourth-century slave sales from Wadi Daliyeh in Palestine employ "a slightly different phraseology" from the Egyptian contracts, permitting a real comparison of provincial practice.^[7] The 202 papyrus fragments from Saqqāra are so severely damaged that little reliable information can be drawn from them. A group of court records preserves complaints against a former verdict, cross-examination, oaths and decisions, with about half the cases concerning slaves in the surviving legal dossier.^[7]

Base-level bookkeeping survives chiefly on potsherds. Some two thousand ostraca from Idumaea, most recording transfers of goods, form the largest such corpus in Gzella’s survey. He regards them as evidence for the entrenched use of Aramaic in routine accounting within a largely agrarian society.^[7] Porten and Yardeni distinguish about 350 excavated Aramaic ostraca from 33 sites from the large unprovenanced cache known as the Idumean ostraca, which appeared on the market from 1991. Their 2014 introduction records 1,899 objects whose location was known, while other pieces were known only through photographs. It separately counts 913 published photographs and 1,091 unpublished texts. These figures describe different inventories: some objects had been published more than once, and separate photographs of two sides had sometimes been counted as separate pieces. The editors’ account of collection history is therefore essential to interpreting the totals.^[32]

What the language carried at this level was often a private arrangement rather than an order. A letter from Spentadata to Hori and Petemachis opens with a blessing, "May all the gods look after the welfare of my brothers at all times", and turns at once to business: the writer owns a share in a boat held in common with its master, and asks his correspondents to load what Armantidata directs and to pay over his share of the rent.^[33] The instructions for the money are conditional throughout, because the sender does not know how far the transaction has gone: "If you bought grain (with) them and brought (it) to our houses, good. And if not, give the silver to Ar[manti]data"; grain still in their hands is to be handed over for sale.^[33] Hori and Petemachis bear Egyptian names, while Spentadata, his father Fravartipata and Armantidata bear Iranian ones, and the editorial note is careful about the address: "'Brother' is a form of friendly address and does not signify a family relationship here."^[33]

Two fiscal series from Egypt show the language handling recurrent transactions. A customs account of import and export duties dated to 475 BCE and arranged by month (TADAE C3.7) records duties collected from incoming ships and deposited in the royal treasury; Gzella calls it particularly revealing for the economic history of Achaemenid Egypt and for Egypto-Aramaic naval terminology, since the business required Aramaic to absorb Egyptian vocabulary for vessels and cargoes.^[7] The Memphis Shipyard Journal for the years 473 to 471 BCE (TADAE C3.8) supplies similar material in a poorer state of preservation.^[7]

The customs account survives beneath a later use of its papyrus: it was erased and overwritten with the Aḥiqar composition, and the palimpsest has an unexpected editorial use: because the underlying account is arranged chronologically, "the original sequence of the surviving parts of the palimpsest can be determined" from it.^[34] A fiscal record and a wisdom text therefore depend on one another for their reconstruction, without any implication that one institution produced both.

The modern history of this material is still in progress. Nearly all the Elephantine papyri were published in the century after their discovery, but a container of fragments left over from Eduard Sachau's 1911 edition was rediscovered in Berlin in 2014, and James D. Moore's edition sets out to publish the legible documentary papyri from "the over 800 Aramaic papyri fragments discovered in an uncatalogued container known as the Aramaic Box".^[35] The volume edits a selection; the rest remain available chiefly as photographs and preliminary readings.^[35]

At the end of column IV, §70 of the Behistun (Bisotun) inscription states that Darius had his text written down, read aloud before him, and sent out to the lands. The wording of that passage is among the most contested in the Achaemenid corpus, and the dispute over its terms has run for more than a century without resolution.^[6][36]

An Aramaic version of the inscription was found at Elephantine (TADAE C2.1). Gzella describes it as a fragmentary Aramaic version of the res gestae which "has been translated from the Babylonian", supplementing both the rock monument and the Neo-Babylonian copy from Babylon, and leaves its purpose open:

It is unclear whether Darius II had distributed a text celebrating the deeds of his predecessor throughout the empire in order to honor him (in accordance with §70 of the Persian original), or whether the copy discovered at Elephantine served as an exercise in scribal training. Since it closely corresponds to the orthographical and linguistic standards of Achaemenid Official Aramaic, it would, at any rate, have been a suitable model.

Gzella, "Aramaic Sources" (2021)^[34]

The papyrus is a single document, and its final column is what complicates the picture. Hassan Rezai Baghbidi explains that "The lines 66-69 of the Aramaic version found at Elephantine are in fact the equivalents of the final lines (lines 50-60) of Darius's second tomb inscription at Naqsh-e Rostam (DNb)."^[37] Kuhrt also identifies material from Darius's tomb inscription at Naqsh-e Rostam in the papyrus.^[36] The passage as printed by Baghbidi reads:

[. . . Very much] make known how you act and of what kind your conduct is. [. . . which one] may say in your ear. Hear what one may say besides. [. . . May] do; observe also that which a poor man may do. [. . . nor in] your well-being be without fervor (in counter-attack) . . .

Aramaic Behistun papyrus, lines 66–69, as printed by Rezai Baghbidi^[37]

The same passage qualifies the transmission history. "The existence of two Old Persian loan-words in this part of the Aramaic version shows that it was not translated from Babylonian", the words being prtr, from Old Persian paratar, and ʾymnš, from ayāumaniš.^[37] Baghbidi cites Nicholas Sims-Williams’s 1981 study when discussing these loanwords. On Baghbidi’s analysis, they indicate direct contact with Old Persian in the inserted passage. The argument concerns those lines and should not be extended automatically to the translation history of the entire column.

Baghbidi's explanation of how the passage got there is his own hypothesis and should be kept separate from the identification. On his reading, Darius ordered the text copied onto clay tablets and parchment and circulated to the provinces, and two items were added at that stage, which he identifies as the genealogy and the statement of royal virtues represented by the two tomb inscriptions.^[37] Three dates are involved and none entails another: the carving of the monument, the promulgation of copies, and the date of the surviving Elephantine papyrus. The surviving object has its own later history. Kuhrt dates its chancery script to the late fifth century and notes that a record of memoranda on the back of the reused papyrus begins in 417 BCE. She presents commemorative recopying and recirculation under Darius II as possibilities, not established purposes.^[36]

The trilingual stele found in the Leto sanctuary near Xanthos in Lycia places Aramaic beside Lycian and Greek on one monument. The texts concern a cult foundation, its priest, property and continuing provision. They occupy different faces: the Aramaic is on a narrow side, while the longer Lycian and Greek texts occupy the broad faces. Kuhrt describes the Lycian and Greek versions as closely related and the Aramaic as a shorter endorsement. For Briant, the physical disposition is one reason not to assume that the Aramaic supplied an authoritative original of which the other texts were merely translations.^[38][39]

The financial clauses illustrate both correspondence and uncertainty. The Greek gives three half-minas annually, and the Aramaic one and a half minas of silver: these amounts agree arithmetically. The Lycian numeral is less secure. Kuhrt prints 120(?) adas and records alternative readings of 18 and 22.5, with the proposed conversion into minas depending on the reading chosen. The languages should therefore not be described simply as fixing different payments. Some differences arise from the state of modern understanding, while other provisions are genuinely expressed differently or omitted.^[38]

The Aramaic opening calls the place the fortress of Orna, but its next clause still attributes the decision to Orna's inhabitants. This is not the disappearance of civic agency from the Aramaic text. The key interpretative disputes concern what the satrap Pixodarus is represented as doing, how the closing provisions relate to property, and how the Greek designation kyrios should be understood. Dupont-Sommer supplied a verb meaning “said” after Pixodarus in the opening, creating a satrapal declaration. Kuhrt rejects the addition and reads a formula fixing the resolution's time and place.^[38]

The closing Aramaic property clause is also disputed. Kuhrt adopts Teixidor's interpretation, in which the written decree conveys title to property; her note contrasts it with Lemaire's property-holder writing the law and Dupont-Sommer's satrap inscribing it as master of the decision. This uncertainty cannot be settled by assigning all Aramaic texts to a superior administrative level in advance. In Briant's interpretation, the city made its own cultic regulations and invoked Pixodarus as their guarantor and a judge in disputes. The Aramaic records his acceptance of that role, without giving the civic text its force of law.^[38][39]

That interpretation has been challenged. Lisbeth Fried questions whether Greek mainland uses of kyrios supply adequate parallels for a Persian satrap's position in Lycia. She compares the Xanthos text with cultic regulations from Lagina that name no satrap, arguing that Pixodarus's involvement was exceptional and connected with his control of Lycia. The disagreement concerns institutional context as well as translation. The coexistence of three languages on the stone establishes multilingual publication, but does not by itself determine the legal relation between the versions.^[40]

The royal date is likewise interpreted, not fully specified on the stone. The Aramaic names the month Siwan and year one of Artaxerxes, without a numeral distinguishing the king. Dupont-Sommer's attribution to Artaxerxes III led to 358 BCE; Badian proposed Arses, who took the name Artaxerxes, and 337 BCE. Briant reports the latter as the majority view, and Kuhrt treats it as virtually certain. Fried also favours Arses but gives the date as 338/337. Thus the inscription's regnal wording and the historical conversion should remain separate, even when using the influential Artaxerxes IV attribution.^[39][38][40]

Aramaic also appears in much shorter records from western provincial administration. Twelve clay bullae from Daskyleion bear Aramaic inscriptions beside seal images. Röllig's edition explains that incomplete and superficial impressions make many readings difficult; he worked from drawings, photographs and prints rather than examining the originals himself. The inscriptions include personal names, often Iranian, which he assessed against comparanda elsewhere in Anatolia. The material adds evidence for Aramaic alongside sealing practices, but the letters originally secured by the bullae do not survive. Cilician coin legends provide another, distinct provincial use of written Aramaic.^[41][7]

Sustained administrative contact left Iranian vocabulary throughout the Aramaic documents, and the density of it is itself a register marker: many Persian loanwords appear in the Aršāma correspondence and are absent from the private Hermopolis letters.^[1] The terms cluster where the institutions do, in offices, orders, estates and personnel. A6.9's phrase for the road the traveller follows rests on an Iranian loan, and the occupation of the thirteen Cilicians in A6.7 is a further one, both reconstructed from the Aramaic spelling.^[4]

Borrowing ran in more than one direction and over more than one period. Aramaic had absorbed Akkadian material long before the Persians; the Egyptian documents take up Egyptian vocabulary where the business is local, most visibly in the naval terminology of the customs account.^[7] Contact reached grammar as well as lexicon, though the evidence there is thinner and Gzella's claim correspondingly guarded, since the passive-participle construction he attributes to Old Persian influence is "still rarely attested" in first-millennium Aramaic.^[1] A borrowed office-name shows that an institution travelled; it does not show that the people using the word spoke the language it came from.

The hardest question is what any of this implies about what people said. Van Bladel's argument is that the documentary record cannot bear the weight usually placed on it, and he assembles his case from the later linguistic history rather than from the Achaemenid corpus. Aramaic "did not become a common language of any non-Aramaean population, Iranic-speaking or otherwise, under the Achaemenids, except the Judaeans and Babylonians, peoples who already had intensive contacts with Aramaic-speakers under the Assyrian and Babylonian Empires".^[9] Xenophon's party used a Persian-speaking interpreter to deal with an Armenian village headman rather than an Aramaic-speaking one. Armenian is full of words adopted from Persian and Parthian rather than from Aramaic, which he reads as a sign of unilateral bilingualism; and Middle Persian shows practically no Aramaic loans outside Manichaean texts, many of which are translations.^[9]

He also corrects the term itself. "The term lingua franca, however, usually refers to a simplified common vernacular of communication between groups, not to an administrative link language used in writing by a specific stratum of administrators."^[9] His conclusion narrows the claim on both axes: outside Syria and Iraq, where Aramaic was already becoming the common vernacular before Cyrus, it was "as far as the government was concerned, no more than a language of professional scribes and secretaries", and even the linking function is held with reserve.^[11] What he does not dispute is the documentary fact: "there is no doubt that Aramaic was used for correspondence, record-keeping, and formal documents. Such texts do survive in sufficient numbers to be sure of this."^[11]

These are arguments from historical linguistics, and they require assessment beside the documentary and Aramaic specialist literature rather than in place of it. The distribution van Bladel accepts is Tavernier's, and the standard he accepts is Gzella's; the disagreement concerns what a supra-regional written medium tells us about speech. Two further cautions belong here. A name settles nothing, since Iranian names were borne outside Iranian-speaking families and individuals could carry more than one. And competence in the written standard varied even among those who used it: Gzella observes that the Elephantine papyri show more spelling variation than the Aršāma letters, which indicates that they were produced by Aramaic-speaking inhabitants rather than by chancery-trained professionals.^[7]

Documentary material dominates the record, but three literary texts on papyrus survive from Elephantine. The story of Aḥiqar, an adviser at the Neo-Assyrian court who survives a plot against him, has a collection of older wisdom sayings attached to it (TADAE C1.1); a fragmentary tale about Bar Puneš seems to be set at a royal court (TADAE C1.2); and the Aramaic Behistun version is the third.^[34] An ink inscription on the walls of a burial cave near Sheikh Fadl (TADAE D23.1) may be a fourth, though it is too fragmentary and palaeographically difficult for its contents to be determined.^[34]

Aḥiqar is linguistically layered. The narrative frame is in Achaemenid Official Aramaic and served as an introduction to the sayings, which are "entrenched in older lore and composed in a pre-Achaemenid variety of Aramaic the provenance of which remains elusive"; scholars divide between eighth- or seventh-century southern Syria and seventh- or sixth-century north-western Mesopotamia.^[34] Gzella suggests the composition may have served both to teach a formal style to administrators in the making and to depict the moral and intellectual ideal of a loyal court official, while stressing that the true extent of the literary tradition and its social reach cannot be assessed: whether such texts were known to significant parts of the population, or confined to scribal education, "must remain unanswered".^[34]

One long-standing model has been revised. Jonas Greenfield proposed a supra-regional "Standard Literary Aramaic" existing alongside the chancery idiom; Gzella holds that it "did not exist alongside Achaemenid Official Aramaic … but formed a subset of it".^[34] The Aramaic portions of Ezra and Daniel preserve recoverable roots in the chancery idiom, but their surviving forms show contact with the local Aramaic of Judaea and successive phases of redaction around what may have been a fourth-century core, so a chancery formula in those books attests the durability of a register rather than the authenticity of a narrative.^[34]

The conquest did not immediately end the use of inherited administrative writing. On Naveh and Shaked’s chronological interpretation, Bactrian C4 belongs to the seventh year of Alexander, in 324 BCE. It records the collection and disbursement of commodities and names officials responsible for supplies at Araivant and Varaina. This is evidence for Aramaic accounting under the new reign, rather than an epistolary formula identical to the earlier letters to Bagavant.^[30] Gzella describes the broader linguistic development differently: once the Achaemenid centre disappeared, "elements of the official scribal tradition could interact more freely with regional forms of Aramaic, and linguistic innovations could spread more easily across the speech area".^[34]

The later history is regional rather than uniform. Aramaic faded quickly in Anatolia and lasted only into the Ptolemaic period in Egypt, while conservative varieties continued at the peripheries, in northern Arabia and in Parthia. In Hellenistic and Roman Syria and eastern Mesopotamia, local dialects were promoted to the chancery idioms of small kingdoms at Palmyra, Edessa and Hatra, though heavily marked by inherited spelling conventions; in Judaea an Aramaic literary heritage was gradually transformed into a new Jewish literary Aramaic between the fourth and second centuries BCE.^[34] Gzella's conclusion is that "the end point of Achaemenid Official Aramaic is hard to determine".^[34]

One strand of that afterlife is regularly confused with a different phenomenon. Middle Persian was written with Aramaic word-forms, heterograms, read aloud as Persian words: a matter of orthography in a script derived from the Achaemenid secretarial hand, and distinct from Gershevitch's proposal that Achaemenid Elamite tablets were themselves Old Persian in disguise. Van Bladel notes that all the Iranic languages written in antiquity, with the possible exception of Bactrian, were rendered in regional varieties of the Aramaic secretarial script of Achaemenid times, which evolved into distinct letter-shapes under the successor powers.^[5] These later writing systems preserve an Achaemenid inheritance, while the survival and transformation of the language itself followed different regional courses.^[5][34]