14 Aspects of Multilingualism in Turfan as Seen in Manichaean Texts

A particular feature about the German Turfan Collection of texts from Eastern Central Asia is the enormous range of scripts and languages represented in the c. 40,000 fragments brought to Berlin by the four Prussian Turfan expeditions 1902–1914. The impressive number of twenty or even twenty-four scripts and languages has been published and commented upon on various occasions. There have also been various attempts to graphically represent this situation, the most recent ones being the graph and map in the brochure of the Turfan Studies group, or ‘Turfanforschung’ (German 2002 and 2007; English 2007).¹ The main aim of the map is simply to demonstrate the variety of linguistic material available from a particular site and therefore to give an indication of what is present where. Given the scale of the map, the result can only be very general.

The great concentration of material from sites around the modern city Turfan (Tulupan) is immediately obvious but a close look at the languages listed there will show the limits of this kind of presentation. Each language has its own symbol but, since all the symbols have the same size, no indication is given about the quantity of material in a particular language. Toyuq is a case in point. Fragments of texts in the languages Tocharian B, Sanskrit, Chinese, Tocharian A, Old Turkish, Tibetan, Mongolian, (Early) Modern Persian and Bactrian were found there. Some of these languages are represented by quite a number of fragments but Bactrian, for example, occurs in only one fragment in Manichaean script (besides a small number of fragments in Bactrian script) [at Yarkhoto, but not indicated on the map]. Similarly, Greek is represented in Buyalïq at the northern edge of the Turfan Depression by a single line of text on an otherwise Sogdian page from a book used by Christians. This is not to say that the occurrence of Bactrian and Greek in the Turfan area is not highly significant, but it has to be pointed out that the tantalizing information provided by these virtually unique attestations is of quite a different nature to the extensive attestation of many aspects of the other literary languages from the same sites. Nevertheless, it is interesting to look at the significance of the attestations for Greek and Bactrian. The presence of part of a Greek sentence on a Sogdian page is highly significant because, as N. Sims-Williams² writes, it suggests the mobility of the person who wrote it and also shows the particular affiliation of the people using the Sogdian text. The content of the Sogdian text is Christian, a translation from the Psalms that may have been done in Sogdiana, the homeland of Sogdian. It may also have been translated much farther to the east, in the vicinity of Turfan where a Christian community using a Syriac liturgy and various texts in Sogdian and Old Turkish established itself, probably in the ninth century. The Greek quotation shows that this community took note of the Greek text of the Psalms and therefore can be said to be Melkite, that is, affiliated to the Byzantine hierarchy. Equally impressive is the Bactrian fragment.³ It is written in Manichaean script and is therefore the only surviving Bactrian text not written in Greek script. Since Greek script used for Bactrian is defective, for example, in not having a letter for the sound h whereas Manichaean script has a letter h and some other relevant features, the fragment is valuable for its script alone. But its value goes deeper. The fragment consists of a page from a book, folded to a small size, possibly for use in an amulet. It is direct evidence for a Manichaean book and for Manichaean literature in Bactrian and therefore a highly significant link in the chain of missionary endeavors that brought Manichaeism from Mesopotamia, where the religion arose in the third century CE to Turfan where, after the adoption of Manichaeism as the state religion by the Uigur Empire in 762 and the transfer of the centre of this empire to Turfan in 840, Manichaean texts were copied until the beginning of the eleventh century CE.

As a more extensive example of what is represented in the Berlin Turfan Collection let us turn now to the multilingual situation revealed by Manichaean texts in the Turfan Collection. Manichaeism is a religion founded by Mani (216–276 CE) who included missionary activity as a central component of his church and stated that his religious texts were to be translated, a position quite the opposite of that in many other religions. Though we have little direct historical evidence for the activities of Manichaean missionaries we can gather much from the Manichaean texts they and their converts produced in various parts of the world. We know that Manichaean missionaries were active in China in the early eighth century CE because they submitted a Chinese document on Manichaeism to the Chinese court in 731⁴ and achieved tolerance for their religion as one practiced by “barbarians” (Sogdians) in 732. Though this was soon revoked, Manichaean missionaries were still around in the early 760s when, according to the early ninth-century account in the inscription in Karabalgasun, the Uigur Khagan Bügü Khan converted to Manichaeism. Though followed by a backlash, the conversion eventually led to the establishment of Manichaeism as the religion of the rulers of the Uigur Empire who, in 840, moved their political centre to the two capital cities Turfan and Bišbaliq. It is hard to say how deep the conversion went because Manichaeism, with its commandment not to kill, was, like Buddhism, not a religion that the leader of an empire could follow easily. Since Manichaeism had a two-tier system of fully conforming “elects” and supporting “hearers” it was possible for the Uigur rulers to become patrons of the religious community as “hearers.”

From the context of the conversion it seems that the Manichaean missionaries were Sogdians working in China. It seems that they, or the Sogdian community they served, were responsible for the “Chinese compendium” of 731 mentioned above. However, it is difficult to say how close the connections between Chinese Manichaeism and the Manichaeans in Turfan were.

Chinese Manichaean texts were found in Dunhuang and, indeed, some fragments of Chinese Manichaean texts have been found in the Turfan Collection⁵ and they show a high degree of consistency. On the other hand, there is a certain mismatch between the Chinese Compendium and the non-Chinese Manichaean material found in Turfan. The Chinese Compendium states the titles of Mani’s seven books as:

the Evangelium (with a Greek title),

the Treasure of Life (with an Aramaic title),

the *dēwān, a divan, identified both as a “book of discipline” and also as a “book of healing” (with a Middle Persian title),

the book of mysteries (with an Aramaic title, using a Middle Persian loan-word),

the Pragmateia (a Greek title),

the book of Giants (Middle Persian title),

the blessings (Middle Persian), which use Greek, Aramaic and Middle Persian terms.

This mixture of languages in the titles will, at least for the Greek and Aramaic terms, go back to Mani himself. However, a Middle Persian text attributed to Mani by An-Nadīm, the Šābuhragān, is not mentioned in the Chinese Compendium. This is not very surprising in itself because it is a Middle Persian compilation put together by Mani or at his command but without replacing any of his other works or achieving the status of a canonical work. But the omission is important because pages from this book (in Middle Persian in Manichaean script) were found in Turfan. This discovery was one of the reasons why F. W. K. Müller in 1904 was able to so confidently proclaim that the fragments just arrived in Berlin from Turfan contained original Manichaean literature.⁶ Does this indicate that the Manichaean community in Turfan were very different from the community that had the Chinese Compendium made? Or did the Compendium omit to mention the Šābuhragān for the purely technical reason that it was not a canonical book? Similarly, in the Manichaean community in Egypt and further to the west in North Africa, there is no mention of this work. Its presence in Turfan is due to its language, Middle Persian. The Manichaean community in Turfan was composed of speakers of Sogdian and Turkish, but they did not restrict themselves to these languages.

The bi-folio M 172 illustrates this very well.⁷ It contains a Middle Persian text with an interspersed Sogdian version and, on the other page, an Old Turkish text. This bi-folio clearly belonged to a book used by speakers of Old Uigur (Old Turkish) and Sogdian but the Sogdian version of the Middle Persian did not replace the Middle Persian text, it only served to make this text accessible. The Middle Persian text is part of Mani’s canonical work, the Gospel/Evangelium, which he wrote in Aramaic but for which he used the Greek designation Evangelion. What we find in M 172 is not the Aramaic text but a Middle Persian translation and this was obviously held in such high regard that a Sogdian version was added to it. The regard held for the Middle Persian text is significant and shows three things: that Mani’s original Aramaic texts were transferred to Middle Persian; that this transferral was done at an early period; and that the subsequent Manichaean communities in the east ultimately derived from such a Middle Persian community. Given the high regard for the Šābuhragān, a work certainly composed in Middle Persian (as An-Nadīm says) and seen as a work by Mani (which does not necessarily mean that he composed, compiled or wrote it in Middle Persian), it seems likely that other Middle Persian texts such as the passage from the beginning of Mani’s Gospel were also attributed the status of deriving from Mani or from Mani’s direct circle and were therefore retained as originals. These originals could not have been comprehensible to Sogdian-speaking Manichaeans, much less so to speakers of Old Uigur, so at some stage of the journey of Manichaeism eastwards a Sogdian version was added.

However, the linguistic composition of Manichaeism is even more complicated. This concerns Aramaic and Parthian. One Aramaic text has survived in the fragments found at Turfan and Aramaic terms and phrases also survive in the Chinese Manichaean texts.⁸ Mani was a speaker of Aramaic and composed all of his canonical works in that language. He may have spoken Middle Persian, possibly even Parthian, although Middle Persian and Parthian Manichaean texts are not really evidence for this, but rather for the activities of his entourage and Manichaean missionaries.

All of this gives us a structure which can be graphically represented as follows:

The table signals the following: Aramaic—the language of Mani’s texts—is retained in some formulae but is not added to after Mani’s life.

Middle Persian, the language of one text, the Šābuhragān, closely associated with Mani and possibly with others—the extent of the Šābuhragān in the Turfan collection is disputed because it can be defined as texts with the specific headline or texts with a content likely to have been in the Šābuhragān—is not only retained but added to in order to translate Aramaic texts by Mani, surely already in Mani’s lifetime—such as the Middle Persian version of the beginning of Mani’s Gospel, the Middle Persian version of the beginning of the qšwdgʾn ʾfrywn⁹ and the Middle Persian version of the canonical books such as Giants and Henoch. All this became the base of the first Middle Persian-speaking Manichaean communities who, during Mani’s life or after it, added to this body of Middle Persian material. How much Middle Persian material was subsequently produced is not clear. Some texts were later modified to accommodate and justify a designation (dēnāwar) adopted by Manichaeans in Central Asia during a schism in the seventh century.¹⁰ It seems likely that, otherwise, Middle Persian material ceased to be produced until the eighth and ninth centuries when some late Middle Persian material was made.¹¹ One of these texts¹² suggests that the use of the word ṯʾjygʾnyy “Tajik” to characterize the melody of a Middle Persian hymn must point to a late composition. These texts are connected in content with the Uigur support for Manicheaism. Here the boundary between Middle Persian and Early Modern Persian was already being crossed, and some specifically Early Modern Persian texts follow, demonstrating the spread of Modern Persian as a lingua franca over a large area, but also the presence of Modern Persian speakers in the community in Turfan. They bring with them not only Manichean texts but also non-Manichaean poetic works in Manichaean script.

The reason for the gap in the production of Middle Persian texts (at least as reflected in the Turfan collection) is the Manichaean mission to Parthian-speaking areas and subsequently to areas further east. This began during Mani’s lifetime, as the text M 2 tells us.¹³ The mission has some remarkable aspects. Parthian is the language of the northern part of Iran and, as the language of the Arsacid dynasty, which the Middle Persian-speaking Sasanian dynasty replaced in 224 CE, it was also a language with high prestige, a fact testified to by the high number of Parthian loanwords in Middle Persian (and Armenian). However, the Parthian Manichaean community seems to have reversed this relationship in the sense that it retained Middle Persian texts. The prestige of these texts seems to essentially derive from their descent from Mani himself, or from his entourage. Since there are apparently no Parthian translations of many of the core Manichaean texts that have survived in Middle Persian in the Turfan Collection, it seems that this was quite deliberate, that a Parthian version of the Šābuhragān and other texts was not deemed necessary or desirable. This may have been in part because the community was bilingual in Middle Persian and Parthian, but it also seems to be the result of reverence for a language closely associated with Mani. The Parthian texts seem to have arisen both as a result of translation from Aramaic and in the adaptation and development of Aramaic models. This resulted in two long Parthian litanies (āfrīwan) in two Parthian hymn-cycles Angad Rošnān and Huyadagmān; in sermon texts, possibly in a confessional text and, as a dominant feature of community life, in a great body of Manichaean hymns. The hymns follow an Aramaic pattern of alphabetically ordered strophes and, despite the damaged state of many of them, clearly represent hundreds of hymns used by the community. This extensive production will have taken some time to complete but may have been restricted to the late third and early fourth century. Evidence for later Parthian literary activity is sparse, but we know of at least one particular endeavor: hymns composed in memory of Mār Šād-Ohrmezd at the beginning of the seventh century¹⁴ Since there are no Parthian texts specifically associated with Uigur leaders (unlike the small number of Middle Persian ones) it seems that Parthian texts ceased to be produced, possibly but not certainly an indication that Parthian was dying out as a spoken language. Since Turfan was never a Parthian-speaking area this point is irrelevant and there is ample evidence for the use of Parthian as a church language by the Manichaean community. This evidence includes transcriptions (essentially transliterations) of Parthian hymns from Manichaean to Sogdian script, which show that the hymns needed to be accessed by people who did not know Manichaean script and probably did not understand Parthian but needed to sing the hymns or at least follow the sung or chanted text. The other piece of evidence is to be seen in Chinese and Old Turkish Manichaean texts that contain traces of Parthian rather than Middle Persian originals. The Old Turkish confession text uses Parthian terms and a recurring Parthian formula; the Chinese texts include translations of texts that otherwise survive in Parthian (and Old Turkish) but not in Middle Persian. One of these texts is the first part of the Huyadagmān; the other is a sermon, the sermon on the light-mind¹⁵ which, like the Parthian hymns, show that Parthian played a dominant role in community life. A further indication of this is that, though the Sogdian and Old Turkish-speaking Manichaean community or communities in Turfan (and a Chinese-speaking community in Dunhuang, but also in Toyuq bei Turfan) made translations of many prose texts and of some verse texts, there is no evidence for a Sogdian Manichaean hymnology but some for an Old Turkish Manichaean hymnology.

Nevertheless there are terminological differences between the Middle Persian and Parthian Manichaean texts which show slightly different routes taken by what are, after all, two different communities. Although the Middle Persian texts were transmitted by a community very much dominated by Parthian, they retain their own traits. Since it is unlikely that every text made by the Middle Persian communities in various parts of the Sasanian empire would have been channeled to the Parthian community as it arose on the basis of the first Manichaean mission to the Parthian-speaking north of the Sasanian empire, it seems more likely that at least some of the Middle Persian literature was transferred to the Parthian community later, through wandering elects from the Middle Persian communities or even through a regrouping of the communities because of persecution and flight to a different area. Mobility, whether missionary activity, the injunction on the elects to be itinerants, or flight from persecution, is a major factor in Manichaean activity.

The Sogdian and Old Turkish Manichaean literature has already been mentioned above in relation to Parthian. There is a definite possibility that the Parthian Manichaean mission soon led to a mission to Sogdiana but it is hard to find evidence for this. Al-Bīrūnī, writing in the tenth century, says that there were Manichaeans in Samarkand a generation previously but does not say from when this community dated. It could be old, but equally it could be quite recent. The question is important because it has consequences for the composition of Manichaean literature in Sogdian. It is surely highly significant that the Manichaean community in Turfan mainly used Parthian hymns as well as some Middle Persian ones. If the Manichaean community in Sogdiana was old, why did it not produce its own hymns? And if it did, why are these Sogdian hymns not present in Turfan? It may be the case that the quality and the quantity of Parthian hymns was such that any Manichaean community in Sogdiana was content to keep them (together with the Middle Persian texts) and be, initially and throughout its existence, a multi-lingual community in the same way that the community (or communities) in Turfan were, except that they added Old Turkish too. There is also the possibility that Sogdian Manichaeism was content to be in the second row, recognizing Middle Persian and Parthian texts as primary or hallowed—despite Mani’s insistence on ongoing translation of his works. Possibly the Manichaeans had an idea of a golden age of their religion in the third and fourth centuries, of which they were the guardians and to the literature of which they could add but not replace the older texts with more recent translations. Indeed, it would have been very difficult to replace the Parthian hymns composed in an intricate alphabetical scheme; perhaps the Sogdian-speaking community never had the resources or size to even contemplate the task. Sundermann’s suggestion that some Manichaean texts reflect local Sogdian features¹⁶ offers an interesting insight into Manichaean activity in Sogdiana but does not establish a date or, indeed, indicate the size of the community. It is equally possible that, through the mobility enjoined on the elect, the eastern Manichaean communities were linked by this mobility to itinerant Middle Persian and Parthian-speaking elects who thereby secured the dominant position of these two languages. There is evidence for speakers of Early Modern Persian in Turfan. There is also evidence for Syrian Manichaeans visiting Turfan, with an indication of a long line of such (consistently problematic) visits.¹⁷ The language of the Syrians is not specified but it may have been Syriac or, more likely, New Persian as suggested by some terms in the text.

The most remarkable fragments are the Bactrian and Tocharian B ones because of the implications that derive from their existence. There is one Bactrian fragment in Manichaean script. Some further Bactrian fragments in Bactrian script (a derivative of Greek cursive script) are also in the Turfan Collection but their damaged state has not allowed any satisfactory interpretation of them and it is not clear if they could be Manichaean. So the Bactrian fragment in Manichaean script is unique.¹⁸ It is a page from a book. It was folded down to a small square, possibly for use as an amulet but it is clear from the text that the page was never complete on its own and that the text contains a discussion (in a sermon or a treatise) on Manichaean themes couched in part in the Buddhist language. This is not unusual for Manichaeism and would of course not be unusual for Bactria, where the book from which this page was taken was probably made. It is unlikely to have been made in Turfan, though we cannot entirely exclude that possibility. The fact that the page is made of parchment supports its more westerly origin, because the Manichaean fragments in Turfan are all written on paper, with the exception of one fragment on parchment and a few on cloth. A parchment page is in keeping with what we expect for a Manichaean book that was not produced in China. The page therefore throws a light on a Manichaean community in Bactria who were using a highly-developed Manichaean literature and Manichaean script. This agrees with a text which tells us about a Manichaean mission to the eastern border.¹⁹ In particular, Bactria occupied a central position on the trade-routes that made up the Silk Road, along which Manichaeism also travelled. However, we do not know the date of the Bactrian page. It could be from the fourth century onwards. For lack of evidence we also do not know what role Manichaean Bactrian may have played in the Manichaean mission eastwards and whether Bactrian-speaking Manichaeans were present in Turfan in any number.

There is only one Tocharian B Manichaean text.²⁰ It is written, together with an Old Turkish Manichaean text, in a carefully prepared Indian style book. Though there is only one text, part of it is also preserved on two fragments of paper that were either part of a scroll or belonged to a codex book. The date of the text is not clear. Whatever the origin of the known copy, which, accompanied by an Old Turkish version, cannot be from before the late eighth century and is probably later, the question is whether the Tocharian B text is evidence for a Manichaean mission in a Tocharian B-speaking area before or after the establishment of Manichaeism at Turfan. The choice is essentially between a Manichaean mission from Bactria before the sixth century or after the rise of Manichaean texts in Old Turkish. Note that Tocharian B as a language has a special status and is well attested as a language both of Buddhist and of secular texts, monastic accounts etc., whereas Tocharian A, the sister language, is nearly exclusively the language of Buddhist texts. Tocharian B is generally regarded as a spoken language whereas Tocharian A is apparently a literary language. Additionally, texts in Tocharian A were recovered further east than texts in Tocharian B. A Manichaean mission coming from Bactria would have reached the area of Tocharian B first. On the other hand, some significant and extensive Buddhist texts in Old Turkish were translated from versions in Tocharian A, demonstrating the important role placed by Tocharian A in the development of Old Turkish Buddhism and showing that Tocharian A was a prestigious language. It seems possible that the Manichaean text in Tocharian B might have been composed on the basis of the Old Turkish text copied with it with the aim of projecting Manichaeism into the prestigious realm of Tocharian, though, curiously not Tocharian A. The lack of any other text makes it impossible to choose between the two options.

Digital images of the Berlin Turfan Collection of texts from Eastern Central Asia (Turfan and other sites on the northern branch of the Silk Road visited by the four German Turfan-Expeditions between 1902 and 1914) can be found at http://turfan.bbaw.de and there Digital Turfanarchiv I and II. Direct access to the database is also possible at http://idp.bbaw.de for the German version or http://idp.bl.uk for the English version, but the language can be chosen anew in any version of the database.

References

Footnotes