Kushan inscriptions from Western and Southern Central Asia (WCA, SCA)

« previous post | next post »

The article I am calling to your attention in this post is of extraordinary importance for its potential to link together many of the themes we have repeatedly investigated during nearly the last two decades on Language Log (see the bibliography below for a sampling of relevant posts).

To make it easier for non-specialist readers, here are a few brief identifications of essential languages and peoples (all late Classical and early Medieval):

Bactrian (Αριαο, Aryao, [arjaː]) is an extinct Eastern Iranian language formerly spoken in the Central Asian region of Bactria (in present-day Afghanistan) and used as the official language of the Kushan and the Hephthalite empires.

The Kushan Empire (Ancient Greek: Βασιλεία Κοσσανῶν; Bactrian: Κοϸανο, Košano; Sanskrit: कुषाण वंश; Brahmi: , Ku-ṣā-ṇa; BHS: Guṣāṇa-vaṃśa; Parthian: , Kušan-xšaθr; Chinese: 貴霜; pinyin: Guìshuāng) was a syncretic empire, formed by the Yuezhi, in the Bactrian territories in the early 1st century. It spread to encompass much of what is now Uzbekistan, Afghanistan, Pakistan and Northern India, at least as far as Saketa and Sarnath near Varanasi (Benares), where inscriptions have been found dating to the era of the Kushan Emperor Kanishka the Great.

The Tocharian (sometimes Tokharian) languages (/təˈkɛəriən/ or /təˈkɑːriən/), also known as Arśi-Kuči, Agnean-Kuchean or Kuchean-Agnean, are an extinct branch of the Indo-European language family spoken by inhabitants of the Tarim Basin, the Tocharians. The languages are known from manuscripts dating from the 5th to the 8th century AD, which were found in oasis cities on the northern edge of the Tarim Basin (now part of Xinjiang in Northwest China) and the Lop Desert. The discovery of these languages in the early 20th century contradicted the formerly prevalent idea of an east–west division of the Indo-European language family as centum and satem languages, and prompted reinvigorated study of the Indo-European family. Scholars studying these manuscripts in the early 20th century identified their authors with the Tokharoi, a name used in ancient sources for people of Bactria (Tokharistan). Although this identification is now believed to be mistaken, "Tocharian" remains the usual term for these languages.

The Yuezhi (Chinese: 月氏; pinyin: Yuèzhī, Ròuzhī or Rùzhī; Wade–Giles: Yüeh4-chih1, Jou4-chih1 or Ju4-chih1;) were an ancient people first described in Chinese histories as nomadic pastoralists living in an arid grassland area in the western part of the modern Chinese province of Gansu, during the 1st millennium BC. After a major defeat at the hands of the Xiongnu in 176 BC, the Yuezhi split into two groups migrating in different directions: the Greater Yuezhi (Dà Yuèzhī 大月氏) and Lesser Yuezhi (Xiǎo Yuèzhī 小月氏). This started a complex domino effect that radiated in all directions and, in the process, set the course of history for much of Asia for centuries to come.

Other essential terms (scripts and languages) are Aramaic (language; script), Brahmi, Gāndhārī, Kharoṣṭhī, Khotanese / Saka, Parthian, Prakrit, Scythian, and Tajik.

Here's the article that prompted this post:

A Partial Decipherment of the Unknown Kushan Script

Svenja Bonmann, Jakob Halfmann, Natalie Korobzow, Bobomullo Bobomulloev

First published: 12 July 2023


with a map, tables, and figures


Several dozen inscriptions in an unknown writing system have been discovered in an area stretching geographically from Kazakhstan, Uzbekistan and Tajikistan to southern Afghanistan. Most inscriptions can be dated to the period from the 2nd century BCE to the 3rd century CE, yet all attempts at decipherment have so far been unsuccessful. The recent discovery of previously unknown inscriptions near the Almosi gorge, Tajikistan, however, allows for a renewed attempt at decipherment. Drawing upon a catalogue of characters and a distributional analysis, we report two identical sequences in the newly found Almosi inscriptions and in the Dašt-i Nāwur trilingual. Based on parallel texts in Bactrian, we suggest to read the name of the Kushan emperor Vema Takhtu in these sequences, accompanied by the title ‘king of kings’ and several epithets. This allows for the deduction of probable phonetic values of 15 different consonantal signs and four vocalic diacritics and the inference that the inscriptions record a previously unknown Middle Iranian language.


Since the late 1950s, archaeological excavations in Central Asia have brought to light several dozen inscriptions in an unknown writing system which was dubbed écriture inconnue in French (Fussman 1974: 7) and neizvestnoe pis’mo in Russian (Vertogradova 1982; translated into English as ‘unknown lettering’). The inscriptions, which range in length between fragments of two or three characters and longer inscriptions with several lines of text, have been discovered in an area stretching geographically from modern-day Kazakhstan, Uzbekistan and Tajikistan to southern Afghanistan. Most of them are clustered in the territory of ancient Bactria, situated between the Hindu Kush in the south and the Hisar Range in the north. The longest inscription, Dašt-i Nāwur III, is part of a trilingual which mentions the Kushan emperor Vema Takhtu in its Bactrian section. About 2000 years ago the Kushan dynasty reigned over a large territory encompassing southern Central Asia as well as northwestern India; Vema Takhtu was the Kushan ruler who expanded the Kushan sphere of control into north-western India (Hou Hanshu 88.2921; Hill 2015: 29).

The Dašt-i Nāwur trilingual, dated to the early 2nd century CE (Ball et al. 2019: 346), demonstrates that the unknown script was used by the Kushan administration side by side with more common scripts of the area (Greek script for the Bactrian language, Kharoṣṭhī script for Gāndhārī). Most inscriptions date from the period between the 2nd century BCE and the 3rd century CE, chronologically corresponding to the rise, heyday and fall of the Kushans. Despite several claims of decipherment (mostly focusing on its earliest instance, the so-called ‘Issyk inscription’, e.g. Harmatta 1994: 407–412; Amanžolov 2003: 217–222; Ünal 2019), the unknown script is still considered unreadable at present (Ball et al. 2019: 437). The previous attempts at decipherment suffer from non-reproducible readings as well as implausible language proposals not matching the known cultural history of southern Central Asia.

The discovery of two new inscriptions in the unknown script in the Almosi Gorge (Tajikistan), including a likely bilingual with Bactrian, allows for the substitution of plausible phonetic values for several signs of the unknown script and paves the way for a subsequent determination of other phoneme-grapheme correlations of the writing system. Our results suggest that the script served to record a previously unknown Middle Iranian language at least in Almosi and Dašt-i Nāwur. A newly identified triscriptual provides confirmation of some of the deciphered sound values, indicating that our results are unlikely to be due to circular reasoning or coincidence. In what follows, we will explain our partial decipherment by means of a systematic approach inspired by previous decipherments (Egyptian hieroglyphs, Old Persian cuneiform, Linear B), explaining step-by-step the methodology employed and the reasoning underlying specific interpretations. Section 2 aims at providing the reader with basic knowledge on the linguistic and historical background to follow the discussion, whereas Section 3 focuses on the methodology for decipherments and Section 4 presents the corpus of inscriptions. Sections 5–10 describe the actual decipherment, whereas Section 11 discusses the likely origins of the script. Sections 12 and 13 provide our final readings of Almosi Gorge I and several lines of Dašt-i Nāwur III, whereas Section 14 aims at a first check of the results by means of new material. Section 15 concludes our discussion of the partial decipherment and outlines the wider context and implications of the findings.


The inscriptions in the unknown script have been found all the way from the Ili valley area to regions south of the Hindu Kush, but they appear to cluster in northern Bactria between the Hisar mountains and the Āmū Daryā (Oxus) (see Figure 1). In antiquity, this was the heartland of the Kushan Empire. The Kushans were in origin nomadic people not autochthonous to Bactria. Driven out of their original home in the grasslands of Gansu (Northwestern China) by the Xiōngnú 匈奴 following a major defeat in 176 BCE (Posch 1995: 84–88, Falk 2015: 51–53), a group called Dà Yuèzhī 大月氏 in Chinese sources (and ‘Tocharians’ by Graeco-Roman authors, see Grenet 2012: 7) first migrated into the Ili valley, and then in a second step via the Ferghana valley to Bactria (ca. 130 BCE, cf. Falk 2015: 59).

In northern Bactria, one of the five tribes or branches (cf. Grenet 2006 or Falk 2014) of the Dà Yuèzhī, known as guì-shuāng 貴霜 in Chinese sources, kuṣāṇa in Kharoṣṭhī coin legends and kušano in Bactrian, rose to power and eventually founded a syncretic empire combining Hellenistic, Iranian and Indian elements that extended from the western part of the Tarim Basin through modern-day Tajikistan and Afghanistan to northern India (see Yu 2011 and Grenet 2012; but note Kuwayama 2022a, 2022b). During the 2nd century CE, the Kushan empire was at the height of its power, and the Roman, Parthian, Han Chinese and Kushan Empires were Eurasian civilisations of equal rank. The Kushans interacted both with China (via the Tarim Basin and the Silk Road) and with Rome (via the Kingdom of Aksum in north-eastern Africa and southern Arabia and maritime trade routes). The match between the territory of the Kushans (or Dà Yuèzhī) and the find sites of the inscriptions can hardly be a coincidence; the script must have been used to write down either their native language or a prestigious one of their secondary homelands (cf. Rtveladze & Livšic 1985: 36).

In ancient times, the vast area between the Tian Shan (Tien Shan) and the Ili valley in the east, the Hisar Range in the west and the Hindu Kush in the south was a melting pot of different cultures and languages. The dominant languages of the area belonged to the Indo-Iranian branch of the Indo-European language family. This branch, in turn, is divided into three distinct sub-branches, viz. Indo-Aryan, Nuristani and Iranian. In a general sense, Indo-Aryan languages were and are spoken to the south-east of the Hindu Kush, Nuristani languages within it, and Iranian languages to the west and north of it; in antiquity, Iranian languages also dominated the Eurasian steppe and were spoken from modern-day Hungary in the west to Western China in the east. The unattested idioms of intruding nomads from the steppe therefore likely included Iranian languages, but the possibility is not excluded that non-Indo-European languages may also have been among them. This is, however, only a theoretical possibility, as there is no indication (i.e. textual remains in readable writing systems, data from the Nebenüberlieferung or an old layer of toponymic data) that Yeniseian, Turkic or Mongolic languages or a predecessor of the modern language isolate Burushaski were spoken in northern Bactria in the last centuries BCE or the first centuries CE. The available evidence clearly indicates a dominance of Indo-European languages in the region at that time.

The only exception is a Semitic language. Before the Kushans, the Persian Achaemenid Empire had brought with it its administrative language Imperial Aramaic (belonging to the Semitic language family) and its consonantal writing system. The Aramaic script indicates consonants but not vowels; it is a so-called abjad. Most Middle (i.e. post-Achaemenid, but pre-Islamic) Iranian languages, such as Sogdian (spoken north of the Hisar mountains), Khwarezmian (spoken near the Aral Sea), Parthian and Middle Persian are attested in derivatives of the Imperial Aramaic consonantal script as a result of the cultural legacy of the Achaemenid Empire.

A more indirect descendant of the Aramaic writing system is the Kharoṣṭhī script used to record Gāndhārī, the northwestern variety of Middle Indo-Aryan, which was used as a literary language in large parts of Central Asia and is attested alongside the unknown script in several locations. It shares with Aramaic a number of letter shapes and the right-to-left direction of writing. Structurally, however, it is no longer a purely consonantal script, but an alphasyllabary of the Indic type (abugida), in which each character represents a syllable consisting of a consonant and an inherent vowel a that can be replaced with other vowels via the use of diacritics. Another Indic alphasyllabary, the Brāhmī script, was used to write other Middle Indo-Aryan languages (Prakrits), as well as earlier Sanskrit. Sanskrit inscriptions in Brāhmī are attested alongside the unknown script in Old Termez. A derivative of Brāhmī was later employed to record the Middle Iranian Saka languages Khotanese and Tumshuqese, spoken in the western part of the Tarim Basin, as well as the Tocharian languages Kuchean or Tocharian B and Agnean or Tocharian A, attested on its north-eastern rim. The two Tocharian languages represent a separate branch of Indo-European. The origins of the Brāhmī script are uncertain, but unlike Kharoṣṭhī there seems to be little resemblance to Aramaic letter shapes and its direction of writing is left-to-right.

The conquests of Alexander the Great that led to the collapse of the Achaemenid Empire also established Ancient Greek (an Indo-European relative of the Indo-Iranian languages) as a prestige language in southern Central Asia. The Greek alphabetic script was adopted to record Bactrian, the Middle Iranian language of Bactria,1 beginning in the Kushan period. Bactrian in Greek script is attested alongside the unknown script in several locations.

Plausible hypotheses with regard to the language recorded by the unknown script and the origin of the script itself should proceed from the attested cultural and linguistic history of the region at this time. This most likely excludes the hypothesis of a Turkic language, pursued among others by Amanžolov (2003: 217–222), as well as Ünal’s (2019: 174) ‘ancient pre-Proto-Mongolic’ proposal. Fussman’s (1974: 33–34) initial hypothesis that the unknown script at Dašt-i Nāwur represented the language of the Kamboja, a people presumably residing south of the Hindu Kush at that time, which he connected with the present-day Iranian language Ormuri, was universally rejected—including by Fussman himself (Fussman 1994)—once it turned out that the other found specimens of the script were concentrated in northern Bactria (Rtveladze & Livšic 1985: 35–36; Schmitt 1994: 187). Bernard (1979: 253–254; 1980: 439) instead conjectured that the unknown script was created during Achaemenid rule over Bactria in a first attempt to write down the Bactrian language or one of its dialects. According to this theory, the Greek alphabet would have been adopted later, while the older system continued to be used in private contexts for some time, side by side with the new Graeco-Bactrian script. Other researchers were not convinced by Bernard’s proposal (Rapin 1992: 141; Fussman 2011: 133) and some have interpreted the trilingual of Dašt-i Nāwur as counterevidence to the hypothesis (Vertogradova 1982: 167; Schmitt 1994: 188–189), though in principle the existence of a biscriptual monolingual (or triscriptual and bilingual) inscription is not inconceivable. In contrast to other researchers, Vertogradova (1982) did not believe that all inscriptions associated with the unknown script were examples of the same writing system. Instead, she claimed to have identified two (Vertogradova 1982: 167), later three (Vertogradova 2002: 131) separate scripts within the attested corpus (on this, see Section 4 below) and suspected that these represented several different languages. Similar ideas are entertained by Fussman (1994, 2011: 133), Schmitt (1994: 188) and Ünal (2019).

Livšic (1976: 166, fn. 14), and later Harmatta (1994: 407–412), hypothesised that the script might represent a Saka language, that is, an Iranian language most closely related to Khotanese, Tumshuqese and modern Wakhi. Though the “fanciful” (Ball et al. 2019: 437 [643], en. 247) Saka readings proposed by Harmatta (1994: 407–412) could not be reproduced by subsequent researchers, a Saka language is still a priori a likely candidate, since varieties of this Iranian subgroup were probably widespread in the Central Asian steppe in the relevant period. The Saka hypothesis is not incompatible with the idea that the unknown script might represent the language of the Yuèzhī (advocated by Rtveladze & Livšic in 1985: 36), who must have entered Bactria from the steppe zone. Considering the reports that the Yuèzhī had their original homeland in the western part of Gansu, it is also possible that they brought a language native to that region with them to Bactria, for example, a language from the Tocharian branch of Indo-European. It is conceivable in any case that the Yuèzhī spoke a different language initially, but secondarily adopted Iranian languages on their journey first to the west and then to the south. Based on the association of the unknown script with the settlement areas of the Yuèzhī, we thus conclude that the most likely candidate for the language represented in the script is either a representative of Tocharian or Iranian.

As for the script itself, a superficial similarity to the Kharoṣṭhī script was already observed by Fussman (1974: 23), who later also considered Aramaic or mixed Aramaic-Kharoṣṭhī origin (Fussman 1983: 138; Fussman 1994), before settling on the claim that it is “clearly derived from Kharoṣṭhī” (clairement dérivé de la kharoṣṭhī) (Fussman 2011: 133). However, as Livšic (1976: 165, fn. 14) notes, decipherment on the basis of Kharoṣṭhī has not succeeded. A direct derivation from the Aramaic script was considered most likely by D’jakonov, Livšic & Kljaštornyj apud Akišev (1978: 59–60), Bernard (1979: 253; 1980: 439) and Xuršudjan (2011). Ünal (2019) also sought a connection between Aramaic and the unknown script of Issyk and Ai Khanum, but only after implausibly turning the inscriptions upside down. A far-fetched attempt to relate the unknown script to the later-attested Turkic runes via Phoenician and Messapic letters was made by Amanžolov (2003: 217–222), who also read the Issyk inscription upside down. Livšic (1980: 4) found no systematic resemblances between the Turkic runes and the unknown script. Since most successful script decipherments in the past have relied on connections between languages rather than hypotheses about historical relationships between scripts, we left ideas about the script’s origins out of consideration initially. Nevertheless, as the discussion will show, the results which we reach on the basis of linguistic reasoning turn out to be closely compatible with a direct derivation of letter shapes from Imperial Aramaic, with no indication of Kharoṣṭhī as an intermediary, except possibly for the functional structure of the script.

Other data heavy sections of the article (all except the last omitted here) are:














Provided that our deductions are correct, it is now possible to read 15 alphasyllabic signs with consonant values (⟨ʾ⟩, ⟨B⟩, ⟨Ɣ⟩, ⟨H⟩, ⟨W⟩, ⟨Y⟩, ⟨M⟩, ⟨N⟩, ⟨K1⟩, ⟨K2⟩, ⟨R⟩, ⟨S⟩, ⟨Š⟩, ⟨T⟩, ⟨D⟩), two ligatures (⟨ŠT⟩ and ⟨HW⟩), as well as four vowel diacritics (⟨ā⟩, ⟨a⟩, ⟨u⟩, ⟨e⟩).7 Even though only around 50–60% of all alphasyllabic signs with consonant values have been determined by now (to judge by the altogether 25–30 signs of the writing system), this makes for a good starting point to re-evaluate the other inscriptions and it will enable other researchers to build on (or falsify) our conclusions as soon as new material is found.

The writing system is of Aramaic descent—the forms of most of the identified letters can be traced directly from Imperial Aramaic prototypes. Previous hypotheses about a development of letter shapes from Kharoṣṭhī could not be confirmed. The creator(s) of the newly deciphered writing system must have known Aramaic writing, but modified it with a new system of diacritic marks. The reconfiguration of the script from an abjad to an abugida-like writing system parallels the development of Kharoṣṭhī, and we consider it likely that both scripts, and perhaps also the Brāhmī script, are the result of a similar historical process taking place in different regions: The adoption of the practice of writing after exposure to the Aramaic script used in the (neighbouring) Achaemenid Empire, without fully adopting the writing practices of Achaemenid chanceries. The diacritic systems of the two scripts are too dissimilar to be formally related, but the idea of vowel diacritics likely spread via interregional contacts.

By all appearances, the language represented by the script—at least at Dašt-i Nāwur and Almosi—is a previously unknown Middle Iranian variety (participation in typical Middle Iranian developments such as vowel shortening or intervocalic lenition of plosives, e.g. *VgV > VγV or *VtV > VdV). The language shares at least one phonological innovation (* > š) with Bactrian and Middle Persian and is lexically closer to Bactrian than, for example, to attested Khotanese (buɣāk(a), baɣ-, šāwu from the same roots as Bactrian ΒΩΓΟ, ΒΑΓΟ, ϷΑΟ vs. Khotanese trāyāka, gyasta, rre etc.). On the other hand, the much later attestation of Khotanese leaves ample space for lexical innovations to have taken place in the meantime, and there are three features that might point to Sakan affinities: The likely preservation of Old Iranian *hu̯ (otherwise found only in Khotanese), an ending for direct objects or rather the accusative singular of a-stems –u (a development typical of Sakan and Sogdian), and the formation of the word for ‘great’—stur(a)—without the *-ka– suffix found in Bactrian ϹΤΟΡΓΟ, Middle Persian sturg, but like Khotanese stura-. Further differences to Bactrian are the absence of the determiner Ι, the formation of ‘saviour’ with an agent noun suffix, and the pronouns udu (demonstrative) and kudu (relative) versus Bactrian ΕΙΔΙΗΛΟ and ΚΙΔΙ.

The ending of the genitive plural cannot continue Old Iranian *-nam because *-am apparently developed to –u; instead, the genitive plural ending may reflect *-nəm, which sets the newly identified language apart from Sogdian, Sakan and perhaps Bactrian as well. At this point in time, it is not clear whether the language participates in the defining sound changes Proto-Iranian *d > l (characteristic for Bactrian as well as modern Pashto and Yidgha-Munji) or rather the exclusively Sakan sound change (found only in Khotanese, Tumshuqese and modern Wakhi) of Proto-Indo-European *ḱw > Proto-Indo-Iranian *ćw > Sakan ʃ(ʃ) (i.e. the change of a cluster consisting of an original ‘patalal’ velar + glide to a voiceless palato-alveolar fricative) and its voiced counterpart, Proto-Indo-European *ǵ(h)w > Proto-Indo-Iranian *(h)w > Sakan ʒ(ʒ).

Other Middle Iranian languages must have existed beside Middle Persian, Parthian, Bactrian, Khwarezmian, Sogdian and Khotanese as well as the fragmentarily attested Alanic and Tumshuqese, namely—to focus only on Central Asia—ancestors of the modern Pamir languages, Pashto, Parachi, Ormuri, or the ‘Old Steppe Iranian’ variety only known so far through loanwords in Tocharian (on which see Tremblay 2005 or now Bernard 2023). The newly identified language may turn out, pending further decipherment, to be a missing link between Bactrian, Sogdian, the Saka languages, Old Ossetic/Alanic and ‘Old Steppe Iranian’ (and perhaps individual modern Iranian languages), participating in several isoglosses with one or the other of these languages, but with none of them exclusively. If so, it will most likely represent the native language of northern Bactria, areally (and phylogenetically perhaps as well) situated in between Bactrian-, Sogdian-, Saka- and Old Steppe Iranian-speaking territories, in that case suggesting that it may indeed have been a secondarily adopted prestigious language of the new homelands of the Dà Yuèzhī after their intrusion into Bactria. If, on the other hand, the further decipherment reveals that the language has more affinities with Sakan and perhaps ‘Old Steppe Iranian’, this will imply that it belongs to a more easterly, steppe-related environment, befitting the original home of the Dà Yuèzhī/Kushans in the grasslands of Gansu. The analysis of the Issyk inscription will be of particular importance in this regard because it may preserve a small fragment of the original language of the Dà Yuèzhī before their intrusion into Bactria.8 However, a word of caution is in order here, as any answer to linguistic (phylogenetic affiliation of the new language) and historical questions (Iranian ethnicity of the Dà Yuèzhī?) must be weighed against the small corpus of inscriptions in the unknown Kushan script.

To speak of this corpus, we expect an increase because the script can now be defined, delineated and differentiated from random scratch marks. So far unreadable Central Asian (fragmentary) inscriptions mistaken for Kharoṣṭhī characters or Tamgas (ownership marks) may turn out, after close scrutiny, to represent the unknown Kushan script instead. We therefore suggest to review already known inscriptions from the Kushan sphere of influence, not the least because the Socotra inscription represents such an overlooked inscription that could already be shown to contain characters of the unknown Kushan script. Likewise, potential new find sites like Sang-i Navišta, Tajikistan, may contain genuine and unexploited material that needs to be analysed in a close cooperation between linguists and archaeologists.

Since it is not an ‘unknown script’ anymore, we suggest to call the writing system ‘(Issyk-)Kushan script’ from now on, because the writing system is first attested in the Issyk Kurgan, but is also clearly associated with Kushan settlement areas and the Kushan emperor Vema Takhtu. The autonyms of Sogdian (⟨swγδy’w⟩ ‘in Sogdian’), Khotanese (hvatanau ‘in Khotanese’) and Bactrian (αριαο ‘in Aryan [language]’) are all known from texts, and we consider it advisable to postpone the definitive choice of a name for the new language, because its autonym may be preserved in one of the inscriptions as well (perhaps in the as yet unread lines 7–9 of DN III, and possibly with the same suffix *-au > *-u). For the time being, a possible interim name for this newly identified Iranian language might be ‘Eteo-Tocharian’ (inspired by Maricq 1958), not to be confused with the two non-Iranian ‘Tocharian’ languages Agnean and Kuchean. As argued by Ching (in Falk 2015: 49, with references), the term ‘Tocharian’ (e.g. Ptolemy, Geography 6.11.6: Τόχαροι μέγα ἔθνος, following Falk 2015: 118; see also Humbach & Faiss 2012) must have originally applied to the inhabitants of Bactria following the immigration of the Yuèzhī and it may have been the ethnic self-designation of the latter (similarly Grenet 2015: 204–206). Further progress in the reading of DN III may lead to clarification.9

Supporting Information


This recently discovered, partially deciphered Kushan script lies at the conflux of linguistic and cultural vectors linking it to Greece and Rome, Southwest Asia, South Asia, and East Asia.  It is a hub of cultural exchange and demonstrates once again what I and others have claimed about the dynamics of Eurasian cultural interaction during antiquity and the middle period, with Semitic-derived scripts and Indo-European languages, especially Iranian ones, as the leading Kulturvermittlers.


Selected readings

[Thanks to Hiroshi Kumamoto and ZHANG He]


  1. Victor Mair said,

    July 21, 2023 @ 7:05 pm

    From Elizabeth J. W. Barber:

    I was VERY puzzled by a sentence which said the researchers were proposing to call the language "Eteo-Tocharian", right after they claimed it was a Middle Iranian language. But Tocharian (in 2 dialects) is a KNOWN language, and it's not Iranian!! It's a separate branch of Indo-European. So, what gives??
    Another friend sent me a different article (from Live Science) which was composed, clearly, from the same sources. Their paragraph about "Eteo-Tocharian" made somewhat more sense to me. It implies that the Tocharians living within the Kushan Empire may have spoken this dialect of Middle Iranian IN ADDITION to what we already knew as Tocharian:

    "This Middle Iranian language was likely one of the official languages of the Kushan Empire, which sprawled across Central Asia and northwestern India between 200 B.C. and A.D. 700. At the height of its power, in the second century A.D., the Kushans co-existed with the Roman Empire. Ancient Eurasian nomads that originally settled in the Kushan Empire — called the "Tocharians" by Greco Roman authors — may also have spoken the language, which the researchers have proposed to call "Eteo-Tocharian." "

    Calling the Middle Iranian language Eteo-Tocharian, however, which is what this then implies at the very end, still makes no sense whatever to me! Something is still missing here.

  2. martin schwartz said,

    July 21, 2023 @ 8:30 pm

    My Bactrian is rusty, but I think the Bactrians wd. have called their lang. something like *Bāxlīg. It should be mentioned that Bactrian influenced the vocab.of Tocharian A/B (as I detailed with a bunch of theoreticla exx. before much Bactrian was known, "Irano-Tocharica" 1974); later LWs were from Sogdian and Khotanese. Soe Toch. words have been posited to be from Old Iranian. It'll be interesting to see if the MIr. lang. of the newly deciphered script (a lang. esp. close to Bactrian, I think) will help account for some LWs in Toch. Finally, NB Tocharian as an ethnonym may have referred to Bactroid Iranians rather than the speakers of the non-Ir. Agnean and Kuchean. Better experts than I may weigh in.

  3. Elizabeth J W Barber said,

    July 21, 2023 @ 11:30 pm

    Thank you for providing the original text, rather than the redactions by reporters which are all I had been able to find. (My previous comment was written before being able to see this blog.) It is now clear that the researchers are aware that what Indo-Europeanists have been calling Tocharian is not Iranian. But calling the language Eteo-Tocharian will continue to cause a lot of confusion. It will be fascinating, however, to see how this very interesting ongoing decipherment can sharpen our vision of how the "Tocharians" and Iranians interacted– linguistically, culturally, and geographically.

  4. Martin Schwartz said,

    July 21, 2023 @ 11:45 pm

    I hope I'm not being censored. What I had written began with
    saying Wikipedia is often not reliable for Iranistics; Encyclopaedia
    Iranica is much better as a source. I noted that "Arya" [here =
    'Iranian'] does name the B'n language in the Rabatak inscription, where the generic designation /Arya-/ 'Iranian' matches
    /Arya-/ instead of *Pārsa- vel sim. *'Persian' as Darius' designation of the language of his OPers. inscription.

  5. NG said,

    July 22, 2023 @ 2:28 am

    Elizabeth Barber's confusion is exactly why this was a terrible name choice on their part. It has *nothing* at all to do with the Tocharian languages of Xinjiang. They do explain this, but only briefly:

    'For the time being, a possible interim name for this newly identified Iranian language might be ‘Eteo-Tocharian’ (inspired by Maricq 1958), not to be confused with the two non-Iranian ‘Tocharian’ languages Agnean and Kuchean. As argued by Ching… the term ‘Tocharian’ … must have originally applied to the inhabitants of Bactria…'

    They are not claiming that the Kushans ever spoke Tocharian A or Tocharian B, but that they spoke 'True Tocharian' (this is what the eteo- part conveys): that it is their language that really should be called Tocharian, and not the languages we actually call Tocharian.

    Be that as it may, the practical result is to confuse endonyms and exonyms and create a deeply confusing situation. An unfortunate decision, and hopefully one that will not gain any traction in the literature.

  6. Philip Anderson said,

    July 22, 2023 @ 3:41 am

    @ Elizabeth J. W. Barber:
    The Tocharian languages were called that because scholars initially associated them with the historically-attested Tocharian people, but I think this is generally rejected now.
    The theory is that this unknown language was the one that the historic Tocharians really spoke, so NOT “in addition”, as made explicit here:
    ‘ "Eteo" is a prefix used by modern scholars that means "true" or "original." ‘
    This use of related names for unrelated languages is clearly confusing, but there is the precedent of Hittite and Hattic (and Biblical Hittites).

  7. Philip Anderson said,

    July 22, 2023 @ 8:34 am

    Trying to associate a newly-discovered language with a historical tribal name seems like a risky approach at best, although there’s a natural human dislike of loose ends that makes even academics want to replace a nation-less language and a language-less people with a simple solution, rather than accept that there must have been plenty of unrecorded tribes and languages in the vast spaces of Central Asia.
    Even using endonyms can be confusing, witness the arguments over the name of the Former Yugoslav Republic of Macedonia (FYROM), now North Macedonia to distinguish it from Macedonia in Greece.

  8. Peter B. Golden said,

    July 22, 2023 @ 9:04 am

    This is an excellent summary of a much discussed issue. As a side note, I would add that the Brāhmī script was also used in the Bugut Inscription (the oldest known inscription from the first Türk Qaghanate), dated to the 582 or perhaps 587, which was written in Sogdian. The Brāhmī text at the base of the monument is used to write passages in Mongolian, the language of the 柔然 Rouran/Asian Avar Qaghanate overthrown by the Türks in 551/552. It would seem that Rouran Mongolic was still being used for some official purposes. See the important work of the late Alexander Vovin, "A Sketch of the Earliest Mongolic Language: the Brāhmī Bugut and Khüid Rolgoi Inscriptions" International Journal of Eurasian Linguistics, 1/1 (2019): 162-197.

  9. David Marjanović said,

    July 22, 2023 @ 5:36 pm

    "A Sketch of the Earliest Mongolic Language: the Brāhmī Bugut and Khüid Rolgoi Inscriptions"

    Typos for Khüis Tolgoi. The paper is here on academia.edu; just scroll down.

  10. Victor Mair said,

    July 24, 2023 @ 2:09 pm

    "Ancient ‘Unknown’ Script Is Finally Deciphered"

    Researchers have decoded more than half of the characters in the so-called Kushan script by comparing them with inscriptions in a known ancient language called Bactrian

    By Sascha Pare, LiveScience on July 19, 2023


    More accessible to non-specialist readers. With clear photographs of the inscriptions at Almosi Gorge.

  11. Victor Mair said,

    July 31, 2023 @ 4:14 pm

    From Lucas Christopoulos:

    "The Greek alphabetic script was adopted to record Bactrian, the Middle Iranian language of Bactria, beginning in the Kushan period."
    … "founded a syncretic empire combining Hellenistic, Iranian and Indian elements that extended from the western part of the Tarim Basin through modern-day Tajikistan and Afghanistan to northern India"

    No, it already happened during the three hundred years of Greco-Bactrian and Indo-Greek period (see Greco-Bactrian coins) before the Kushana. Greek alphabet was used by the elite for literacy and administrative purposes to rule the kingdom. Also, the term "Hellenistic" in the region already indicated a mixture or Greek, Indian and Iranian elements from the same Greco-Bactrian and Indo-Greek period. It was not separated, and that is because of the "universalism" of Alexander the great (who had a Sogdian wife) that they were able to mix together (he was famous for having said: "there are no Greeks and Barbarians, but only educated or uneducated people"). The mixed-blood Greek-Bactrian and Indian kings and the first known king of the Kushana, Kujula Kadphises (Greek Κοζουλου Καδφιζου, or Κοζολα Καδαφες, Chinese : 丘就却) the grandfather of Kanishka I (78-144), together with the later Kushana kings, simply continued and developed that same way of ruling with mixed cultural references and elements from these three civilizations. It stopped when the Sassanid Persians conquered the area of Central Asia ans Southern Central Asia, reviving nationalistic Iranian values of ruling and seeing themselves as the heir of the Achaemenids

RSS feed for comments on this post