Gāndhāri and the Early Chinese Buddhist Translations Reconsidered: The Case of the Saddharmapuṇḍarīkasūtra

Daniel Boucher

Journal of the American Oriental Society, Vol. 118, No. 4 (Oct. – Dec., 1998), pp. 471 – 506 GANDHARI AND THE EARLY CHINESE BUDDHIST TRANSLATIONS RECONSIDERED:THE CASE OF THE SADDHARMAPUNDARIKASUTRA

DANIEL BOUCHER CORNELLUNIVERSITY

Scholars have for several decades now assumed that most if not virtually all of the Indic texts transmittedto China in the first few centuries of the Common Era were written in a Northwest Middle Indic language widely known as Gfndhari. Much of the data for this hypothesis has derived from the reconstructedpronunciation of Chinese transcriptionsof Indian propernames and Buddhist technical terms contained in the early Chinese Buddhist translations. This paper, inspired by the recent brilliant work of Seishi Karashima,attempts to reexamine this assumption from another angle. A closer look at problems in the translation process itself reveals that the collaboration of the Chinese members of the early translationteams may have been instrumentalin formulatingthe final shape these render- ings assumed. Such a realization will require us to reassess our use of these documents for the history of Indian Buddhist languages and texts.

I. THE GANDHARI HYPOTHESIS such a process into account and to raise some caveats with regard to our understanding of the underlying IT HAS FOR SOMETIME NOW been assumed that many Indian language of these translations. if not most of the early Chinese Buddhist translations Until quite recently, there were few thorough exami- derive from originals written in Northwest Middle Indic. nations of the early Chinese Buddhist translations.With A number of scholars have attempted to show that the the exception of a few brave Japanese souls, scholars of reconstructedpronunciation of many of the Chinese tran- both Indian and Chinese have generally been scriptions of Indian proper names and Buddhist tech- put off by the difficult if not at times impenetrable lan- nical terms in these translations reflect a Prakrit source guage of these texts. Moreover, there has been little to text that has much in common with, and perhaps is even attract scholars to these abstruse texts. While the trans- identical to, a language now widely known as Gandhari. lations of the first few centuries of the Common Era While there can be little doubt that the Chinese trans- had considerable impact on the gentry Buddhism that lators often heard recitations of Indic texts that were emerged after the collapse of the Han dynasty, they were heavily Prakritized,containing a numberof features that subsequently eclipsed by the translationsof Kumarajiva coincide with what we know of the Gandharilanguage, and his successors. It was these later translationsthat had it is not as certain that they saw such texts. This is to a greater impact on the development of the indigenous say, what has not been sufficiently taken into consider- schools of . ation is the fundamentallyoral/aural nature of the trans- From the other side of the Himalayas, Indologists have lation process in China. This paper is an attempt to take generally questioned-with good reason-the reliability of these first attemptedtranslations as documents for the study of Indian Buddhism. The majorityof our historical I have been fortunateto receive the kind advice and sug- data-prefaces, colophons, early bibliographies, etc.- gestionsof severalscholars who readan earlierversion of this paint a rather dismal picture of the earliest translation I paper. wouldlike at thispoint to extendmy profoundgratitude teams in China. The Indian or Central Asian missionary to Victor H. Mair and Seishi Karashimafor comments on things is frequently described as having little or no skill in Chinese;to KlausWille and Jens-Uwe Hartmann on variousIn- Chinese; it is virtually certain that practically no Chinese dianmatters; to RichardSalomon and G6rard Fussman for very of this early period commanded any Indian literary lan- usefulsuggestions on Gandharimatters; and to JanNattier and guage; and it is not at all clear how these texts were PaulHarrsion for miscellaneoussuggestions throughout. All of copied, transmitted,or preserved. As a result, it has been thesescholars contributed greatly in helpingme to avoida num- universally accepted that the translationsof later Indian- ber of mistakes; those that remain are where I strayed alone. trained specialists such as , as well as the very

471 472 Journal of the American Oriental Society 118.4 (1998) literal renderings in Tibetan, are far more trustworthyin transcriptionsand the language of the Dutreuil de Rhins absence of an Indic original. manuscript of the Dharmapada that had been discov- Be that as it may, the early translations are currently ered in the late nineteenth century.5Nevertheless, there enjoying an upsurge of scholarly attention. This new- were unresolved problems that kept Waldschmidt from found interest has come from two camps. Sinologists, led drawing firm conclusions concerning the nature of the in the West by Erik Ziircher, have sought to mine these underlying Prakrit. texts as repositories of early Chinese vernacular lan- The first attempt to identify and describe the features guage. The fundamentally oral/auralnature of the trans- of the Middle Indic idiom that appears in some of these lation process in China-a process that will be discussed early Chinese transcriptions as well as in a number of in detail below-has left remnantsof what appearsto be Central Asian languages is the groundbreaking article the spoken idiom of during the first few cen- by H. W. Bailey entitled "GGndhari," by which name turies C.E.1Indologists, on the other hand, have been scholars have continued to identify this Northwest drawn to these texts as early representatives of Maha- Prakrit.6 For Bailey, this Middle Indic language encom- yana Buddhist sitras drafted at a time thought to be passed the Asokan kharosthi edicts from Shahbazgarhi ratherclose, by Indian standards,to that of their compo- and Mansehra,7the various donative inscriptions from sition. In fact, these early translationspredate our oldest northwest India,8 the Dharmapada found near Khotan Sanskritmanuscripts by as many as four or five centuries (the Dutreuil de Rhins manuscript),9the documents from and may well reveal an earlier redaction of the Indian the ancient Shanshan kingdom found at Niya and Lou- textual tradition.In addition, it is also believed that these lan,10 and the miscellaneous traces preserved in Central early translationsmay contain clues concerning the Indic Asian and Chinese sources. language of transmission. Given the fact that almost all Since the publication of Bailey's article, attention paid of our extant Indic language materialsdate from a period to this language has steadily increased. In 1962 John when Sanskritization had already profoundly reshaped Brough published a masterful study of the Gandhari their idiom, these early Chinese sources may be one of Dharmapada which thoroughly discussed all aspects of our few windows into their earlier Middle Indic stage. the discovery, publication, and language of the manu- script as well as its relationship to other versions of the in 1914 Paul Pelliot had surveyed the tran- Already text. In discussing the broader role of Gandhari Prakrit of names in the Chinese translationsof scriptions proper in the transmission of , Brough also ad- the Milindapaiha in order to reconstruct their under- vanced the growing consensus that some early Chinese Indic forms.2 While Pelliot had noted similarities lying translations may have been translated from originals writ- between some of the names in the Chinese texts and forms originating in Northwest India, as well as the pos- sibility of Iranianinfluence, this was, in his own words, 5 Waldschmidt 1932, esp. pp. 231ff. "une etude provisoire." 6 1946. In the early 1930s Friedrich Weller and Ernst Wald- Bailey 7 Prior to article, the of the Asokan edicts schmidt turned their attention to the early fifth-century Bailey's language had received extensive such scholars as Johansson, Chinese translationof the Dirghagama.3Weller examined analysis by Senart, Biihler, and Woolner. For a systematic description of thirty-six transcriptions from the fifteenth of the the language of the kharosthi edicts, see Hultzsch 1925, Dirghdgama, noting that their reconstructedpronuncia- lxxxiv-xcix. The of Asokan studies that has since accu- tion showed many features closer to Prakritthan to San- corpus mulated is now quite large, constituting something of a sub- skrit, though he hesitated to label the specific idiom. field in its own right. Waldschmidt investigated an even larger body of tran- 8 On the language of these inscriptions, see Konow 1929, scriptions from the nineteenth sutra (the Mahdsamdja- xcv-cxv. important contributions have since been made sutra).4 He was perhaps the first to notice similarities Many toward some of the problems posed by these epi- between the reconstructed language of these Chinese clarifying graphs,particularly by H. W. Bailey, GerardFussman, and Rich- ard Salomon; see the bibliography in Fussman 1989, 488-98. 1See Ziircher 1977 and 1991. 9 For a list of the early studies on the linguistic problems of 2 Pelliot 1914. this text, see Brough 1962, viii-x. 3 Weller 1930 and Waldschmidt 1932, esp. pp. 226-49. 10 Boyer et al. 1920-29 and Burrow 1937. See also the rather 4 A revised edition of this text and a discussion of its lan- comprehensive list of kharosthi text/Gandhari Prakrit related guage in light of fifty more years of research can be found in publications focusing on finds from Chinese Turkestan (Xin- Waldschmidt 1980. jiang) in Lin 1996. BOUCHER:Gandhdri and the Early Chinese Buddhist Translations Reconsidered 473 ten in Gandhari.1 Brough was prudently cautious in his described as the impact of a lingua franca, a common remarks, recognizing that very few texts had been sys- language shared by speakers of diverse language groups tematically studied with this problem in mind. However, for the purposes of commerce, administration, or reli- within three years-and with no further studies under- gious intercourse, is far more uncertain.'7 taken to my knowledge-he was able to state: "Suffi- Bernhard would like to see the Dharmaguptaka school cient evidence, however, has now accumulated to estab- as primarily responsible for this spread of Gandhari in lish that the originals of these early Chinese translations .18 Some of Bernhard'sevidence indicating were mostly, even if not exclusively, texts written in the such a role for the Dharmaguptakas,however, has re- Northwestern (Gandhari) Prakrit."'2'While Brough's new- cently been shown to be problematic.19Furthermore, it found certainty is indeed curious, it is noteworthy that is well known that the Sarvastivadins had the most sub- his conclusions concerning the role of Gandhari Prakrit stantial presence in Central Asia, at least as discernible have been regularly repeated by subsequent scholars, from the preserved remains of Buddhist literaturein this generating what I call the "Gandharihypothesis." region and from the reports of Chinese pilgrims pass- Franz Bernhard, in an oft-cited article published in ing through. And, not insignificantly, the Sarvastivadins 1970, reiterated the now firmly established Gandhari are specifically connected with the Sanskritization of hypothesis:

Phonetic in Chinese translations of transcriptions early probablylasted only until midway throughthe reign of Huviska, Buddhist texts make it clear that Gandhari was the me- or approximately thirty-five years, when Chinese campaigns dium in which Buddhism was first in Central propagated reasserted themselves in the western regions (Hitch 1988, 185- the medium which Indianculture was trans- Asia, through 86). Hitch's thesis, however, depends upon the often repeated mitted from the northwestacross CentralAsia to China.13 but never substantiated supposition that Kushan expansion beyond the Pamirs could only have occurred under Kaniska. Bernhard describes Gandhari as "the Buddhist mission- Moreover, the evidence of Kushan control of the Tarim Basin dialect kind of com- ary par excellence," a lingua franca has consisted of little hard data: "The paucity of Kushan coins parable to ecclesiastical Latin of the European Middle in the area and the absense of other substantial evidence, liter- Ages. ary or archeological, make it likely that Kushan interests were It is difficult to know what would constitute evidence strategic or commercial and that they did not rule directly over for a franca in Central Asia on the basis of the lingua much of the region for any considerable time" (Rosenfield rather scant extant records.'4 There can be no doubt that 1967, 43). Gandhari had a noticeable on other it impact languages 17 The only clear case to draw from on this issue is the corpus encountered in Central and most scholars have Asia,15 of texts from the ancient kingdom of Shanshan. While these assumed that it had been most widely influential during administrative documents are written in a kind of Gandhari the of the Kushan in the first few height empire centu- Prakrit,it is also clear from internal linguistic evidence that the ries of the Common Whether this can be Era.'6 impact local spoken language of this region was a Tokhariandialect, albeit one that differs from that of either Agni or Kucha; see Burrow 1935. In addition, we know that Chinese became used 11 Brough 1962, 50-54. at least for business purposes from the time of Emperor Wu's 12 Brough 1965, 587. conquest of Kroraina(Chin. Loulan) in 263. Like much of Cen- 3 Bernhard 1970, 57. tral Asia, Shanshan was clearly a multilingual society. For a 14 bn the problem of categories of language in Central Asia, recent description of what these documents reveal about social, see Nattier 1990. political, and religious life in this region, see Atwood 1991. 15 18 Bailey 1946 discussed this influence on Khotanese and Bernhard 1970, 59-61. 19 Tokharianamong other Central Asian languages; for a survey For example, Bernhard claimed that an early Chinese of the impact of Gandhari on Parthianand Sogdian, see Sims- translationof the Karmavdcandbelongs to the Dharmaguptaka Williams 1983. school. However, as Hisashi Matsumurahas recently pointed out, 16 Douglas Hitch has attempted to pinpoint this influence the text in question is a mere extract from the Dharmaguptaka- more precisely (Hitch 1988). He argues that Kushan control of : "Once it has become clear that the extant two Chinese the southern silk route and the northwest TarimBasin coincided Karmavacana texts of the Dharmaguptakaswere compiled in with the rise of Kaniska-taking his ascension as the tradi- China, it is entirely meaningless to discuss what the original tional 78 C.E. On the basis of Chinese historical accounts and language of the Karmavdcandof this school was" (Matsumura numismatic evidence, Hitch hypothesized that this domination 1990, 69). 474 Journal of the American Oriental Society 118.4 (1998) canonical literature.20Nevertheless, some connection with crease of data. Sinologists have generally sought to use the Dharmaguptakas is not entirely without basis. The the transcriptional data to aid in the reconstruction of Chinese translation of the Dharmaguptaka-vinaya refers Ancient Chinese. Indologists have, conversely, used the to the recitation of the arapacana formulary21 and this reconstructed pronunciation of Chinese to determine the formulary has now been convincingly shown to be the underlying Indian language of the translation. The cir- syllabic order of Gandhari Prakrit in kharosthi script.22 cularity of this process becomes immediately evident Moreover, as mentioned above, the Chinese translation of and has not gone unnoticed by some of the principal the Dirghdgama, widely believed to belong to the Dhar- investigators: maguptaka school, has been repeatedly cited as derived from a Gandhari original.23 Since a good deal is known about the sound systems of Since Bernhard's article, the Gandhari hypothesis has various Middle Indic dialects and the ways they differed been repeated, more or less intact, by Indologists24 and from that of , the Chinese forms sometimes Sinologists25 alike, usually without any substantial in- allow us to guess whether the original language of a particulartext had a certain feature in common with San- skrit or was more similar to one or more of the Prakrits. 20 When care is taken to avoid information On the problem of school affiliation in relation to the pre- circularity, served Sanskrit remains from Central Asia, see von Simson obtained in this way can, I believe, be safely used in the reconstruction of BTD 1985, esp. pp. 84-85 on the evidence of the Dharmaguptakas. [Buddhist Transcriptional As von Simson points out, the only extant vinaya fragment of Dialect(s)].26 the Dharmaguptakasis in hybrid Sanskrit and a sutra fragment This brief overview of the of the attributedto this sect is in pure Sanskrit. Dr. Klaus Wille has development "Gandhari should make clear that the evi- informed me (personal communication, June 1995) that there hypothesis" dence marshalled to date the role of this may be some additional fragments of the Dharmaguptakavi- concerning Northwest Middle Indic in the transmission of naya in the Pelliot Sanskrit collection; they too are written in language Sanskrit. Buddhism to China is rather meager. It has in general 21 Levi 1915, 440. been founded upon a small body of transcriptions, prin- 22 Salomon 1990. cipally from a few in the Dirghagama only. And 23 the Indic Weller and Waldschmidtexamined only a relatively small the conjectures concerning underlying language of these have been to portion of the entire text in their early studies. The underlying transcriptions repeated sufficiently now as "facts." language of the Chinese Dirghagama will now have to be re- qualify But there are other From the Indian considered in light of the thorough study by Karashima(1994). problems. side, this has so much as to Karashimamakes it clear that the situation is more complicated hypothesis gained credibility inhibit the consideration of other Prakrits or mixtures than generally stated: "As we have seen above, the original of Prakrits as source It of language of the Chang ahan jing is not something that can be possible languages. is, course, even that texts in simply decided upon as Gandhari. When one looks at the possible, perhaps probable, composed Central Indian Prakrits were funneled the North- particulars,complex aspects emerge in which elements of San- through a transmission skritization, Prakrits,and local dialects were harmonized in ad- west language on route to China. Such could have these texts a number of dition to specific features of the Northwest dialect. We may still imprinted upon be able to call this dialect Gandhari in a broad sense, with the orthographic and dialectical features of the Gandhari But at the least this would have resulted necessary proviso that it differs considerably from the Gandhari language. very language as reflected in the Northwest inscriptions" (Karash- ima 1994, 51-52). 24 See, among others, von Hiniiber 1982, esp. p. 250: "If ten in Gandhari ... seems to make good sense in terms of the there has been a Gandharitext of the Upaligathas, it does not historical situation and has been supported by linguistic argu- seem to be too far fetched an assumption that the whole text ments by Bailey and Brough" (Pulleyblank 1983, 84). of the Madhyamagamapassed through a stage of development 26 Coblin 1983, 34-35. Coblin's study does in fact add a when it was written in this language once widely used in Cen- considerable amount of data to the transcriptionalcorpus from tral Asia" (von Hiniiber follows this remark by citing Brough some of the earliest translationsof Buddhist texts into Chinese, 1965). See also von Hiniiber 1983 and Nishimura 1987. though much more work remains to be done. Moreover, Coblin 25 See Pulleyblank 1983. Pulleyblank's adherence to the has suggested a more cautious approachto the underlying Indic Gandhari hypothesis is clear: "The hypothesis that the texts languages vis-a-vis the Chinese transcriptions in his more re- brought by the first Buddhist missionaries to China were writ- cent study, Coblin 1993, 871-72. BOUCHER:Gdndhari and the Early Chinese Buddhist TranslationsReconsidered 475 in texts that were linguistically mixed in some very com- variety of Indian and Central Asian locales. In this way plicated and difficult-to-discernways. I will returnto this I hope to avoid generic statements about "the" linguistic issue again at the end of this paper. medium of transmission. On the Chinese side, scholars have typically assumed Furthermore,rather than focusing upon the Chinese that the transcriptionalevidence accurately reflects the transcriptions of Indian names and terms, which, as I Indian source language. This takes for granted that the have suggested, raise a number of problems not all of Chinese scribes-and it was almost always Chinese which can be controlled, we shall look instead at mis- scribes who took down the final text-were able accu- takes in translation that were due in all probability to rately and consistently to distinguishthe Indianphonemes phonological confusions caused by a Prakritic or Cen- and find suitable equivalents for them with sinographs- tral Asian pronunciation of the text. It is my conten- all with no real knowledge of Sanskrit or Prakrit.Some tion-to be fleshed out below-that the fundamentally of the evidence gathered below will call this into ques- oral/aural nature of the translation process in China led tion, at least with regard to one of the early translation to a number of problems of interpretationfor Chinese teams. More importantly,however, even if the Chinese assistants on these teams who had limited skills in did for the most part accurately record the sound of an Indian languages. Indic word, that does not demonstratethat the word was For this purpose we are very fortunate to have the re- written in the Indian manuscript as they heard it. This cent and brilliant study by Seishi Karashima,29whose problem has been summarized by Heinz Bechert: work has broken entirely new ground in the study of these early translations. He has meticulously combed [W]e can only view with the greatestscepticism any through the earliest Chinese translation of the Sad- attemptsto come to conclusionsabout pronunciation on dharmapundarikasutra-that of Dharmaraksa,30whose the basisof orthography,since we mustnever lose sight translation is dated to 286 C.E.-and has provided a of the broadspectrum of possibledivergences between point-by-point analysis of the agreements and disagree- orthographyand pronunciation that we arefamiliar with ments of Dharmaraksa'stranslation with all of the extant from our knowledgeof the developmentof otherlan- Sanskrit manuscript remains as well as with the fifth- guagesand from examination of laterstages in the evo- century version of Kumarajiva.In so doing he has also lutionof the Indiclanguages themselves.27 offered ingenious explanations of some of the discrepan- cies between Dharmaraksa'stext and those of the various Thus on the Chinese side we have to consider the prob- Sanskrit manuscripts which may stem from confusions lem in reverse: evidence for a particular pronunciation caused by a more Prakritic-and, as I will argue, oral/ of an Indic locution does not ipso facto indicate the lan- aural-transmission of the text. guage in which that text was written. It is this problem The advantages of concentrating our attention on the that I will attempt to explore in more detail in this paper. Saddharmapundarikasutra(hereafter SP) then are mani- fold. We possess extensive manuscript finds with con- II. AN ALTERNATIVE APPROACH siderable divergences among them that often allow us to

In light of the problems discussed above, I shall at- tempt a somewhat different approach to examining the what we should call refugees than proselytizers. See Forte influence of Middle Indic-particularly Gandhari-on 1995, 65-70 for some tentative suggestions regarding the mo- the early Chinese Buddhist translations. I will, first of tives of An Shigao, the first translatorin China. This issue is all, restrict this investigation to one text. We know all tangential to this paper, though a more careful consideration of too well that Indian Buddhist texts were not transmitted the possible motives of these first Buddhist teachers in China to China in a single installment. They were brought over may reveal some interestingfacts about the homelands they left. a period of several centuries by an ethnically diverse 29 Karashima 1992. 30 group of missionaries28who themselves hailed from a Dharmaraksa,Chin. Zhu Fahu iff- (ca. 233-311), was born at and studied under an Indian teacher there. He was the most prolific of the early translators; his career 27 Bechert 1991, 17. spans over forty years and the earliest bibliography of Chinese 28 We call these early translators"missionaries" by conven- Buddhist translationscredits him with 154 translations,approx- tion; while it is likely that their endeavors included activities imately half of which are extant. The best overview of his life that we would typically label as missionizing, there is increas- and translation career can be found in Tsukamotoand Hurvitz ing evidence that suggests some of them may have been more 1985, 193-230. 476 Journal of the American Oriental Society 118.4 (1998) differentiatetranslation mistakes from redactional varia- translationthat may have been due to the misinterpreta- tions, something that can seldom be done with most tion of words or phrases whose forms, though distinct in Indian Buddhist texts.31Nevertheless, despite this quan- Sanskrit, would have coalesced in Prakrit,making them tity of manuscript material, it cannot be assumed that more difficult to distinguish for the translation team. I we can always proceed with full knowledge of the Indic will begin by giving examples that could be construed "original"underlying Dharmaraksa's translation. We will as providing evidence for a transmission of this text returnto this problem throughoutthe paper. specifically in GandhariPrakrit.32 The fact that the earliest translation of the SP is by Dharmaraksais also helpful for this examination. Be- Confusions Related to Vowels sides the fact that he was one of the most prolific of Dharmaraksa'stranslation exhibits frequentconfusions the early translatorsduring the formative period of Bud- between long and short vowels. This would be espe- dhism in China, we have a fair amount of information cially understandableif his Indic text were written in concerning his life and translation procedures that will kharosthi script, which does not ordinarily mark vowel bear upon our consideration of the range of forces oper- length.33 ating in this translation. He is, for example, one of the first of the translatorswho is Chinese foreign reportedby a/a biographersto have been fluent in both Sanskritand Chi- of Asian nese as well as the full range Central languages. KN 13.8: bdldn sahdyan parivarjayitvd dryesu sam- Our evidence for mistakes in the translation, then, will sargaratdn samahitdn us to reexamine these from the native provoke reports having avoided foolish company, they take as well as clues the hagiographies provide concerning pleasure in association among the Aryans actual dynamic among the participantson the translation teams. Dh 65a. 10: WW-t W;:-EiW In the evidence amassed below, I have in general fol- in the company of strong and close friends lowed Karashima'slead in the analysis of the philolog- (Krsh, 32) ical problems presented by Dharmaraksa'stranslation. Nevertheless, there are a number of places where I can- Dharmaraksahas confused bala (childish, foolish) for not accept Karashima'sreadings-places where I believe bala (strong); other examples include KN 48.7: baldh; he may have pushed the Prakriticexplanation further than Dh 70c.18: ~hJ (Krsh, 53); KN 99.4: bdldna etddrsa is warranted.I have, therefore, despite Karashima'shuge bhonti (they are fit for fools); Dh 79c.26: it7EJ-J7 (in- This body of evidence, cited only what I view to be valid ex- tent upon the faculties and powers) (Krsh, 81).34 as well: amples of confusions based upon a more heavily Prak- confusion occurs in the opposite direction ritic transmission of the text. Then, having looked at 32 such phonological problems, I will turn to an examina- The following abbreviations are used throughout the rest tion of two colophons to the translationthat reveal much of the paper: about the process by which it was renderedinto Chinese Dh: Dharmaraksa'stranslation of SP (references to the trans- as well as some of its early life in China. I will follow lation are to Taisho 263, vol. 9, by page, register, and line this with a look at other kinds of evidence from the trans- number). lation that expose in differentways the complexity of the KN: Kern/Nanjio 1908-12 (references are to page and line data for evaluating the underlying Indic language. It is numbers). hoped that such a problematizing of an early Chinese Kash: Chandra 1976; unless otherwise stated, this manu- translation will provide some important caveats for the script has been cited from the transcriptionof Toda 1981 use of these texts by both Sinologists and Indologists. by folio, side, and line number. BHS G and D: Edgerton 1953. Evidence for Gdndhdrl Prakrit Underlying Krsh: Karashima 1992. Dharmaraksa's Translation the SP of Translationsthroughout are mine unless otherwise indicated. 33 For a recent explanation of this convention in kharosthi In this section I will draw upon Karashima'sstudy in script and its implications for understandingthe phonology of order to highlight specific mistakes in Dharmaraksa's Gandhari, see Fussman 1989, ??33-34. 34 Note also that Dharmaraksa has confused etadrsa and 31 in A survey of the extant manuscripts can be found in (faculties), a confusion that is not easily explained Karashima 1992, 16-19. phonological terms. BOUCHER:Gandhari and the Early Chinese Buddhist Translations Reconsidered 477

KN 54.12: kdrunya mahyam balavantu tesu KN 3.5: Aniksiptadhurena I have great compassion for them (name of a : "whose burden is not abandoned") Dh 72a.29-b.l : :7;J [t; Dh 63a.28: pt I manifest great compassion and take pity on these fools (Krsh, 59)35 "not put down far away" (Krsh, 27).

i/i There appears to be a confusion here between -dhura (burden) and -dura (long distance); note also the confu- KN 120.5: pratipatti darsenti bahuprakaramsattvana sion between aspirateand non-aspirateconsonants which [the Buddha] teaches good conduct to be- will be discussed below. We should also mention that ings in multifarious ways this rendering of this bodhisattva's name occurs in the works of previous translators, for example, Zhi Qian's Dh 83a.24: &' )lm ,;iAte5 early third-century translation of the Vimalakirtinir- thus the brilliance of a great lamp illumi- desasatra (Taisho 474, vol. 14, 519b.15). Thus we must nates the innumerable masses (Krsh, 91)36 always allow for the possibility that such a name could have been drawn from an established lexicon of trans- Karashima has a confusion between proposed pra- lation equivalents and would not therefore represent and Note also tipa(tti) (good conduct) prapa (lamp). evidence for the underlying language of this Indic the confusion between voiced and unvoiced intervocalic original.38 stops, a widespread Prakriticphenomenon. Besides these there are also a numberof other vocalic confusions, but many are confusions of quality rather u/a than length and are either common in many Prakrits or of a different nature. Thus they can- KN 54.6: samharsayami vividhair upayaih represent problems not be used to indicate a Gandhari source. I gladdenthrough various means

Dh 72a.20-21: jLb,fi:f-~ tT[fz Confusions Related to Consonants what [sentient beings] love in their hearts There are quite a number of mistakes in - has many forms. raksa's translation that appear to be due to confusions between aspirated and unaspiratedvoiced consonants in Karashima has that Dharmaraksa proposed incorrectly both initial and intervocalic position. Weakness of as- divided these two words, a confusion between causing piration-discerned from occasional interchange of as- -dhair upa- and (Krsh, 58). This is not ripa proposal pirated and unaspirated stops in Gandhari texts and certain since Dharmaraksa's could be an at- rendering inscriptions-is frequently cited as a defining feature of to translate vividhair with itf, tempt upayaih sexiang Gandhari among the Prakrits and is especially common this would be an unusual for though extremely rendering among the consonants g/gh and d/dh (Burrow 1937, ?? in his of translations.37 upaya corpus 24-27; Brough 1962, ? 49; Fussman 1989, ? 35).

38 35 We should note that this example is a bit ambiguous. Thus Chinese renderings (translations or transcriptions) Whileit is likely thatDharmaraksa mistook balavantu as from established in the early period under possible Gandhari influ- balavatas Karashimahas suggested,he also translatedbala- ence cannot be cited from later texts as evincing the continued vantuin his renderingda ~t. We will see otherinstances of this influence of the Northwest Prakrit. Once these terms became kind of "double translation"below. partof the indigenous Chinese Buddhist vocabulary,translators 36 Karashima has suggested an alternative rendering: "(The often defaulted to them even if their Indic text may not have Buddha) burns [ran , = ran J,] a great candle." reflected the same phonology or exact meaning. A common 37 Furthermore,sexiang occurs elsewhere in Dharmaraksa's example that could be cited is shamen (Early Middle Chinese: SP for rupa: KN 76.3/Dh 75c.2, KN 290.12/Dh 109c.10, KN sa-man) i?Jr, which transcribessramana but appears to reflect 295.10/Dh 110b.11, etc. I would like to thank Prof. Karashima the particular Gandhari development of sr > s, (samana). On for calling these additional examples to my attention. this issue, see de Jong 1981, 111-12 and Nishimura 1987, 51-52. 478 Journal of the American Oriental Society 118.4 (1998)

g/gh KN 84.7: kumbhandakadarunaraudracittdh kumbhdndaswhose minds are cruel and KN 15.1: ghantasamahai hostile with multitudesof bells39 Dh 76c.17: Dh 65b. 12: fi:: W the intentions of the kumbhdndasand there being a large quantityof incense gandharvasremain hostile and harmful (Krsh,34) (Krsh,70)

Here a confusion appearsto occur between ghanta (bell) Karashimahas proposed that Dharmaraksadivided this and gandha (incense); there is also the interchange of phrase incorrectly, taking it as kumbhdnda-kddaruna- voiced and unvoiced stops as well as dental and retroflex and interpreting it as gandharva. Despite the ingenuity consonants. of this proposal, there are still unresolved problems with this interpretation. d/dh KN 4.13: drumena ca kimnarardjena(proper name) KN 56.8: aham pi samksobhi imasmi darune utpanna sattvdna kasayamadhye Dh 63b. 12: 'I42 ~P~ F: I too have arisen in this dreadfulcom- "pay heed to the dharma" kinnararaja motion[i.e. the world],in the midstof the (Krsh,28) impuritiesof beings40 The confusion between druma and dharma has two pos- Dh 72c.3: ~:L_-: _fiAttA FM.FUJj sible explanations:we could hypothesize with Karashima At that time I was a bhiksu,41and I too that Dharmaraksa'sIndic text-or his pronunciation of came forth amongthe masses of men in it-was subject to the usual metathesis of liquids char- orderto upholdthis dharma.(Krsh, 60) acteristic of Gandhari and the later Dardic languages, hence dharma > dhrama or drama (with concomitant The relationship between the Sanskrit and the Chinese confusion of u and a); or that druma was pronounced is not entirely clear. It may be that Dharmaraksacon- with an epenthetic -a- [daruma] in which the unaccented fused darune (dreadful)with dharana (preserve, uphold) -u- was heard only weakly, thereby making its pronun- or perhaps even with dharma. The word daruna seems ciation nearly indistinguishablefrom that of dharma. We to have given Dharmaraksaparticular problems as he should also note that the following kimnara king in the often made mistakes in its interpretation: list presented here is Mahadharma,accurately rendered by Dharmaraksaas 5/T5[,,:. It is telling that two KN 253.11: suddrune completely different Indic words in such close proximity extremelycruel could be translatedwith the same sinographs. Unable to hear a difference between druma and dharma, Dharma- Dh 104c.15: tL5 raksa'stranslation assistants may have logically assumed able to receive(< dharana?) (Krsh,152) that Dharma- would precede Mahadharma-.

b/bh and -t-/-(d)dh- 39 Note that Kash 21a.6-7 reads: ghanthasamudgebhi. The of illustrate several 40 Cf. Kash 62a.l: ... ddruni satvesu kaSatthamadye. following groups examples utpamna different that occurred There- With regard to our examination of the confusions related to as- problems simultaneously. fore I will discuss them while also pirationin Dharmaraksa'stranslation, we shouldalso note that together attempting to the various developments at in this one line the KashgarMS itself has made two errorsof this distinguish phonological work. To with, there are many examples of a con- kind: kaSattha presumably stands for kasatta (cf. BHSD 174; begin fusion between a form of the verb xbhu (to be, become) note also Kash 53a.2: kasatrra) and madye here is a mistake for and bodhi madhye. The manuscriptis quite clear in both cases. (enlightenment). 41 Karashimaproposes that biqiu th:kEhere represents an in- stance of metathesis in Dharmaraksa'stranslation: (sam)-ksobhi/ 42 bhiksu (Krsh, 60). variant: shun [Ig. BOUCHER:Gandhdri and the Early Chinese Buddhist Translations Reconsidered 479

bhoti/bodhi KN 99.4: baldna etdTdrsabhonti gocaras such arethe domain of fools KN 283.6: sukhasthito bhoti sada vicaksanah [worldlybooks] the wise one is at ease always Dh79c.26: TTA Etti Dh 108b.27: Mt g |_ and they practice toward enlightenment the wise alwaysdwell at ease in enlighten- focusing upon the faculties and powers ment(Krsh, 167) (Krsh,80)

KN 57.15: loke utpadu bhoti purusarsabhanam In these examples Dharmaraksa'stranslation also ig- thereis the appearancein the worldof the nores the nasal present in the third person plural form; bulls of men,i.e. the buddhas mistakes regarding nasals will be discussed in more detail below. In the latter he has also inter- Dh 72c.26-27: THfAr ;k{lM? SA example, preted gocaras from its etymological root Icar rather thereis a buddhain the world,a greatsaint than in its more standard Buddhist sense of "range, and sage, who manifestsnoble enlighten- domain, association." ment(Krsh, 60) sphere, KN 336.5: ca bhonti caritah Dharmaraksahas confused bh- and b- as well as -t- and ye - kotiyah -dh- in these examples. We might expect that the latter confusion was heard as no more of a difference than that andwhich bodhisattvas who havepracticed for kotisof aeons between -t- and -d-, which are interchanged in other contexts as well.43 While weakness of in aspiration Dh 116c.12: ~ ~~_ $tM t Gandharicould be cited in both cases, it is nevertheless if bodhisattvasseek astounding that the translator(s) would have produced a enlightenment,they [should]practice for kotis of aeons (Krsh, text that so completely departs from the Indic version.44 191) bhonti/bodhi Karshima records an important Sanskrit variant here: KN 45.9: ye bhonti hindbhirata bodhi caritva (instead of bhonti caritah). This variant comes from a Central Asian in the Otani Col- thosewho areengaged in lowly pursuits fragment lection that was transcribed by N. D. Mironov and Dh 70a.23: APT EW whose readings are preserved in the notes to N. Dutt's those who do not delightin full enlighten- 1953 edition of the SP. Since Dharmaraksa,as we have ment(Krsh, 50) seen, has a propensity to confuse bho(n)ti with bodhi,45 it is difficult to draw conclusions about his conformity to one or another manuscript tradition in this example. In fact, this is a very good illustration of a problem one is 43 The situationin Gandhari is actually more complicated regularly faced with in these early Chinese testimonies thanthis. Broughhas astutelyhypothesized (1962, ?43a) that to Indian redactional histories. the appearancein the GandhariDharmapada of -dh-in placeof -t- resultsfrom a furtherweakening of the intervocalicstop to the pointat whichit wouldhave soundedlike -dh- to at least some scribes.Such a shift wouldhave been facilitatedby the 117.4/Dh82c.2 (Krsh,89); KN 177.6/Dh91b.10 (Krsh113); fact that-dh- had alreadytaken on the valueof a fricative[6]; KN 287.8/Dh109b.5 (Krsh, 170); KN 287.10/Dh109b.8 (Krsh, as -t- and -d- both weakenedover time towardsthe spirant,a 170);Kash 342a.5 [KN 355.10: bhavet]lDh 119b. 10 (Krsh,198); differencebetween the aspiratedand unaspiratedstop was no KN 394.3/Dh125a.2 (Krsh, 215). longer felt. But this confusioncould also have occurredin 45 Furtherexamples of this confusion between bhonti/ Dharmaraksa'stranslation under the influenceof an Iranian bho(n)di and bodhi can be found: Kash 54a.l[KN 43.3: bha- pronunciation,without specific connection to GandhariPrakrit. vanti]/Dh70b.4 (Krsh,51); Kash 224b.51Dh102a.21 (Krsh, Cf. also note 47 below. 142);KN 236.5/Dh102a.24 (Krsh, 143); KN 296.1/Dh1 l1b.13 44 Otherexamples of a confusionbetween bhoti and bodhi (Krsh,174); KN 326.10/Dh115b.3 (Krsh, 188); KN 355.1/Dh includethe following:KN 63.2/Dh73c.25-26 (Krsh,63); KN 119b.2(Krsh, 198). 480 Journal of the American Oriental Society 118.4 (1998)

bhuta/buddha that such a reading could not have been represented in writing in the underlying Indic text.47 However, we are There are instances in which bhuta appears to have been ahead of ourselves here and should continue with an confused for buddha: examination of the linguistic data before setting forth hypotheses about how this translation acquired its cur- KN 45.14: vadami yeneha ca bhutaniscayam rent form. by which I will speak here about true resolve -th-/-d- Dh 70b.l: &W;i A fP T for which reason can on what the [I] speak Sanskrit -th- and -dh- are both generally represented Buddha has decided (Krsh, 50) by -dh- in Gandhari, as well as occasionally by -d- (e.g., yada < yathd) through weak articulation of aspiration. KN 200.3: bahubhis ca bhitair gunair abhistuto Of course, the voicing of -th- to -dh- is a common Prakritic Pischel and it is lauded for his many genuine qualities development (cf. 1955, ?203) not unlikely that the difference between the aspirated and Dh 95c.1:1: ~~, ~ unaspirated voiced dental stops would have been lost as brilliantly glorified and praised the virtues discussed above. of the buddhas (Krsh, 124) KN 166.6: natho 'si sastasi guru 'si jdto There also seem to be instances in which Dharmaraksa's you are a protector, a teacher, a guru translation team misinterpreted -(d)dh- as having been derived from -t-: Dh 90a.12: tqi) [0--al L what is taught by the Blessed One is like Kash 47a.4-5: ... evdham s'aradvatiputra buddhajnina- the roar of a lion (Krsh, 109) (darsana)samdarsaka it is I, Saradvatiputra, who display the KN 176.8: sadhu darsana buddhanam lokanathina exhibition of buddha-knowledge46 tdyinam Dh 69c.8: excellent is the sight of the buddhas, the lords of the the ones [I] manifest the knowledge of truth (Krsh, world, holy 47) Dh 91a.27: - :. t lAh4 U] Excellent! May the buddhas, roars [sic!] of buddha- KN 330.13: maharsind prakasayanten'ima the world, ones who have obtained bril- bhumim liance . . .(Krsh, 112) by the great seer who reveals this buddha- hood Dharmaraksa has confused -ndtha (lord, protector) with -nada While this - ([lion's] roar). phonological develop- Dh 116a.4: {Lj... . t A___ Jt ment is not impossible to explain, it is considerably more the great saint ... makes a detailed revela- tion and establishes this true stage (Krsh, 190) 47 As noted above, Brough has already well explained the use of -dh- [= 6] in place of original -t- or -d- in the Dharma- In these two examples, Dharmaraksa took buddha- in pada. This convention could have been known to Dharmaraksa the beginning of compounds as bhuta-. Since the normal and/or his assistants. We could also speculate that if Dharma- Prakritic development is clearly from unvoiced to voiced raksa's pronunciation habits were influenced by a Tokharian stops, we might speculate that Dharmaraksa's translation idiom as, for example, the Gandharitexts from Niya were (cf. assistants, hearing an intervocalic voiced dental stop, Burrow 1937, ??14-15, 19), intervocalic consonants could have perhaps pronounced with considerable friction, deduced been orally represented by Dharmaraksaas devoiced. In such it to be derived from an unvoiced stop, despite the fact circumstances his assistants would still have had to deduce the derivation of the word from a pronunciation in which voiced and unvoiced consonants collapsed together, but they might 46 KN 40.11 reads: tathagatajianadarsanasamdarsaka eva- have been more likely to choose the devoiced equivalent under ham sdriputra. such conditions. BOUCHER:Gandhdrl and the Early Chinese Buddhist Translations Reconsidered 481 troubling that Dharmaraksa'stranslation team would not KN 3.4: avalokitesvarena have been guided more by context-not to mention a Dh 63a.27: :E ("the sound of the broaderknowledge of Buddhistidiom-than such a trans- 't illuminating world"?) lation suggests. Moreover, this is especially perplexing when we note that Dharmaraksahas made such a mis- Dharmaraksaappears to have interpreted the name as take in close proximity with a more accurate rendering dbhd-loka-svara, with the confusion of -v- and -bh- of the same expression. For example, at KN 301.7/Dh (perhaps through pronunciationas [3]). For other occur- 11 la.7, Dharmaraksaonce again confused lokandthawith rences in Dharmaraksa'sSP, see 128c.22-23, 26, 27, 29; *lokanada. Yet in the preceding verse (KN 301.4/Dh 129a.1, 7, 8-9, 10, etc. llla.4) lokandtha is found again and is translated by Dharmaraksaas shixiong tt1S (world hero). Though not Evidence Consistent with Gandhdri Prakrit but not an entirely acceptable rendering of lokandtha, it does Restricted to this Language indicate a serious attempt to render that Indic locution. The close proximity of a plausible translation and an There are a number of confusions in Dharmaraksa's entirely mistaken one for the same word will be another translationsthat appearto be caused by a Prakritictrans- piece of data that will lead us to speculate on the actual mission whose phonological features are represented in division of labor within the translation team. Gandhari, but are also known to other Prakrits as well. Thus, to avoid prejudicingthe analysis of the underlying -bh-/-v- language of Dharmaraksa'stranslation, I will present this evidence separately, noting where appropriatealter- As Brough has observed (1962, ??12 and 44), -bh- in native explanations for such occurrences. all probability had the phonetic value of a voiced bila- Confusions Between Labials bial fricative [,] in the northwest and could be repre- sented in kharosthi as or even -h- script -bh-, -vh-, -v-, p/v (e.g., lavhu < labha, uvha'i < ubheya, silaprava < slla- prabha). Such a phonetic value for -bh- would have Though frequently cited as a feature of Gandhari made it difficult to distinguish from -v-, especially in an phonology, the development of p > v is a widespread oral/aural context among non-native speakers. Prakritic phenomenon (cf. Pischel 1955, ?199; von Perhaps the most noticeable occurrence of this pho- Hinuber 1986, ?181). netic confusion is found in the name of the famous bodhisattva Avalokitesvara. There have been a number Kash 121a.3: bahuprakarampravadanti dharmam of scholarly discussions already on the variations and they declarethe dharmain manyways50 interpretationsof the name of this bodhisattva among our extant Sanskrit manuscripts and Chinese transla- Dh 83a.28: Wl it 4 tions.48Without retracing all of the scholarship here, it they bring about decline to the manifold is reasonably safe to assume that the original form of dharma(Krsh, 91) this appellation is that found in some of the Sanskrit re- mains of the SP from Central Asia (i.e., Avalokitasvara) Karashima has suggested that pravadanti was miscon- strued as and approximatedin Kumarajiva'sfifth-century transla- prapatanti (lit., they fall down), though we tion (guanshiyin t E] = avalokita[loka]svara ?).49 over [i.e., hears]the sounds"?)in his translationof the Ugra- pariprccha (Taisho 322, vol. 12, 15b.7) as did Zhi Qian (ca. 48 For a survey of scholarly opinions up to 1948, see Mall- 220-52) in his translationof the Vimalakirtinirdesasutra(Taisho mann 1948, 59-82; the few studies that have appeared since 474, vol. 14, 519b.16).The translationof the Sukhavativyaha this work have contributedlittle to the discussion. attributedto Lokaksema(ca. 168-88) transcribedthe name: 49 See Mironov 1927, 243. The renderingguanshiyin Et? helougeng ffiHf (Taisho 361, vol. 12, 290a.27); see von has been said to originate with the translator Kang Sengkai Stael-Holstein 1936, 351-52, n. 3 and Brough 1970, 83 and (Safighavarman ?) (mid-third cent.) in a translation of the nn. 13-16. Once again, however, this attribution is quite im- Sukhdvativyaha(see von Stael-Holstein 1936, 352, n. 3), but probable. Brough's attempt to link this transcription with a the attributionof this translationto him is highly questionable. name that appears in a second-century Gandharaninscription is Before Kang Sengkai, the Parthian translator An Xuan (ca. also not without problems (cf. Brough 1982). 50 180) renderedthe name as kuiyin Fi: ("[the one who] watches KN 120.8 reads: bahuprakdramhi braviti dharmam. 482 Journal of the American Oriental Society 118.4 (1998)

might expect the verb here to have been understood as a esized to have the phonetic value [ts],51 we would not causative (prapatenti). Thus both -v- and -d- were taken expect it to have been confused with Sanskrit c or ch by as derived from unvoiced originals (-p- and -t-). a person who read a kharosthi manuscript. Such a mis- take would suggest either that this aksara was miscon- KN 398.4: adavati (a word within a dharanimantra- strued in its aural reception or that it was assimilated in pada) a more typically Prakriticfashion. Dh 130b.3: sentencefor H- ("a wealth")(Krsh, 237) KN 35.13: rtjdna ye mahipati cakravartino ye dgatdh ksetrasahasrakotibhih Ada is confused with rich, we iadhya(wealthy, opulent); those kings, great lords, universal monarchs would assume a derivation from addha through assimi- who have come fromthousands of kotisof lation of the consonant conjunct along with weakened as- lands piration. -vati appearsto have been confused with -pada; an original -p- would have been assumed for the -v- and Dh 69a.13-14: ]3:E F-IV {EfE52 fL, the -t- may have been voiced. In both of the cases cited lordsof the world,supreme kings, wheel- here, as well as others cited elsewhere, an existing -v- turningsage-kings, united in thought,they was interpretedincorrectly as deriving from -p-, despite numberhundreds of thousandsof kotis the fact that it is unlikely to have been so representedin (Krsh,46) writing. Karashima has made the ingenious suggestion that xin -m-/-v- jL, (heart, mind) here is probably derived from a devel- opment in which ksetra was pronounced chetta, which was then confused with citta. Dharmaraksa's The alternationof -m- and -v- is quite common in the translation Prakrits (Pischel 1955, ?? 248, 250) "but is rare in assistant presumably would have understood this clause as "who have fallen Gandhari sources other than the Dharmapada"(Brough ye dgatdh citta, to, obtained (a par- ticular state of This confusion is one of 1962, ?36). The most probable explanation is that -m- mind)." perhaps the clearest of a mistake that serves as a notation for an allophone of /vl in nasalized examples could only have occurred in an oral/aural translation context or from a contexts (see Pischel 1955, ??251, 261; Brough 1962, ?36; von Hiniiber 1986 ??209-11). Thus we find words non-GandhariPrakrit source text. in the GandhariDharmapada such as bhamana'i (< bha- vand-), sramana (< sravanam), and nama (< ndvam). Confusions Related to Nasals and anusvdra There are several instances of an apparentconfusion KN 343.13: esa mulam gamisyate drumasya by Dharmaraksaof dental and retroflex nasals, which he will go to the foot of the tree show a clear pattern in the Gandhari Dharmapada: ini- tial n- remains n and both intervocalic -n- and -n- are Dh 117c.19: ff .* -I _T written -n-. (cf. Brough 1962, ?45; von Hinuber 1986, will the of the andhe seek out foot kingof ?93). These nasals appear, however, to have been regu- trees (Krsh,193) larly interchangedin the Niya documents (Burrow 1937, ?34) and kharosthi inscriptions from the northwest, not We presumably have here a confusion between gami- to mention the fact that their differentiationoften became (he will and (he will seek). syate go) gavesyate Despite weak in the Prakritsbroadly (cf. Pischel 1955, ?224).53 Brough'scomment above, the interchangebetween these two roots is common in the Gandhari documents from Niya (cf. Burrow 1937, ?50). 51 See Konow 1936, 610; Bailey 1946, ?4; Brough 1962, ?16; for a fullerdiscussion of the problemsrelated to this con- Confusion Between ks and c/ch junctin Middleand New Indo-Aryan,see Turner1936. 52 I haveread here with the variant. One of the distinguishing features of GandhariPrakrit 53 Konowremarks with regardto these two nasals:"Here in kharosthi script is the preservation of a distinct sign thereis an apparentdifference between the systemof Dhp.and for the conjunct ks, a conjunct that is assimilated in all that of Doc. [Niya Documents] and, so far as we can see, Indian other Prakrits (see Burrow 1937, ?48; Bailey 1946, ?4; Kharosthi inscriptions. It is, however, remarkablethat the Kur- Brough 1962, ?52). Although this sign has been hypoth- ram casket inscription, which contains a quotation of a canon- BOUCHER:Gdndhari and the Early Chinese Buddhist Translations Reconsidered 483

KN 113.11: kanaka Although Karashima's explanations are ingenious and one-eyed (occurs in a list of adjectives surely possible, we might also expect that once a sig- describing lowly servants in the parable of nificant confusion was made in the understandingof a ch. 4) passage, as in the case of kanaka/kanaka here, consid- erable reworking of the whole verse would have been Dh 82a.5: . necessary to make some sense of one's reading. In this possessedof highestquality gold (Krsh, 87) regard we should note that Dharmaraksa'stranslation is made sensible in light of the story to come. That is to say, There appears to be a confusion here between kanaka Dharmaraksa'srendering of v. 21 here anticipates the (one-eyed) and kanaka (gold). Such a confusion, as one narrative development that is revealed some ten verses might imagine, has altered this verse beyond recognition later, where the father states that all his wealth and prop- in translation. The fuller Sanskrit context in KN (ch. 4, erty will be given over to his son, who was being pre- v. 21) reads: pared gradually for his inheritance. If this is the case, then Dharmaraksa,knowing the whole text before the purusams ca so tatra vankds ca kanaka prayojayeta ye actual translation, backwards from knowl- kunthakasca interpolated edge of what was to come. On the other hand, it is also kucailakd krsnaka hinasattvah paryesatha tam naru possible that a revision was made-by Dharmaraksaor karmakdrakam by another proofreader-after the first draft of the trans- lation was completed. A more systematic gathering of Thereforehe directedhis who men, lowly beings were such occurrences may make clearer the range of activi- vagabonds,one-eyed, dull-witted,poorly clothed, of ties at work in the production of the final text. blackcomplexion: "you will seek out this manwho (is There are also a number of instances in Dharma- to be servant." my) raksa's translation in which his interpretationof a word suggests that existing nasalization was omitted or that Dharmaraksarenders this as follows (82a.4-6): non-existing nasalization was assumed:

>A54; :gj ; 1 t KN 272.3: (aranyavrttakds caiva) kanthd pravari- yana ca [thosewho dwell in the wilderness],being This man (= the father)consoled him, explainingmat- dressedin rags tersto him fully: "therewill be the highestquality gold gatheredhere, and it will be providedfor you as uten- Dh 107a.4: SPJW,K sils for eatingand drinking; you will overseeall of the how shall [we?]reply? (Krsh, 160) accounts,tasks, and servants." The confusion here seems to be from taking kantha as katham in what My rendering is by no means certain, and given the (rags) (how? manner?). Karashima considerable disparity between Dharmaraksa'stransla- also proposes a mistake between pravariydna (< prda/vr) and Vvad tion and the parallel Sanskrit, we should expect certain pra (to answer) to account for the verb here, but I am more inclined difficulties. Karashimahas attemptedto account for sev- to see this as a free rendering by eral more of the differences between the Sanskrit and Dharmaraksaof pra/vr (to choose), molded here to fit the Chinese versions as phonological confusions (Krsh, 87). incorrect perception of kanthd as an adverb. However such an explanation appears less certain here. KN 350.6: na durgandhi will not have an offensivesmell ical passagewritten in practicallythe same languageas Dhp., has no trace of the distinction between n Dhp. and n. We are Dh 118b.20-21:T _g;b_ left with the impression that in this a Dhp. respect represents will not fall into unfortunatedestinies normalizationwhich may be due to the influenceof anotherlit- (Krsh, 195-196) eraryPrakrit, or belongsto a limitedterritory within the area covered this where the by dialect, treatmentof n was different" Durgandhi (foul smelling) is here confused with durgati (Konow1936, 607). (unfortunate destinies). While I agree in principle with 54 Variantreads fu Z. Karashima'sreading of this passage, I should also note 484 Journal of the American Oriental Society 118.4 (1998) that Karashima fails to mention the previous clause in The numbers fifty and five hundreddiffer only in a long Dharmaraksa'stext: JH9: (his body will always be "a" and are confused even in Buddhist Sanskrit litera- pure in odor). This could either be an attempt to render ture.56In the case of the numbers 20 and 200 and 30 and the previous clause of the Sanskrit: na tasya ... puti- 300, the loss of anusvdra either in the manuscriptor in mukhambhavati (he will not have a putridmouth), or, as pronunciation could account for the confusion.57 Only I think more likely, it is meant to render both clauses in 40 and 400 are difficult to explain, though it would be a telescoped fashion. If this latter proposal is correct, easy to imagine this numberbeing brought into line with then the inclusion of ~I* in Dharmaraksa'strans- the other numbers of the same magnitude. Karashima lation would constitute another instance-to be exam- would also like to see an instance of the unnecessary ined in more detail below-of what I have called a addition of anusvdra in qi xian PF[ (its limit), where "double translation." Faced with an orally delivered ato would have been confused with amto (antah). While phrase that had at least two possible interpretationsas this does help account for the ratherout-of-place xian P pronounced, the Chinese assistants, unable to decide be- at the head of this line, it also ignores the fact that in tween the two, presented both options. both verses ato ("than this/that")is well translated:)/t i "surpassing that" (v. 11) and i:fl1 "also surpasses KN 304.1-4: ato bahutaradcanye parivarair anantakaih this" (v. 12). It seems somewhat more plausible to me to paicdaatiya gangaya catvdrimsac ca trim- read 'PR here as "their scope, compass," indicating the (v. 11) magnitude of the retinues ratherthan their delimitation. samo samantatah vimsatiganigya parivarah It is difficult to draw entirely clear conclusions con- ato bahutarad dasa ca cdnye yesam panca the evidence of Dharmaraksa'sinconsistent rec- ca (v. 12) cerning ognition of nasalization in his underlying Indic text. The Others, even more numerousthan that, with best discussion of this problem vis-a-vis Gandhari is by unlimited retinues, [equal to the sands] of Gerard Fussman.58 Fussman convincingly demonstrated fifty Ganges rivers, and to forty, and to thirty, that in early Gandhari inscriptions, heavy open syllables Others, even more numerousthan that, who (i.e., a syllable with an etymologically long vowel fol- will everywhere have retinues equal to [the lowed by only a single consonant) in initial or medial sands] of twenty Ganges rivers, and to ten, would have been with nasalization and to five. position pronounced by the law of morae whether such nasalization was Dh lllb. 11-14: marked or not. Thus inconsistencies in marking anu- svdra-either its absence or its inclusion where it never that its notation in had al- er mer ra tat; ter existed-suggest writing XAwk% AHWY+ ready become superfluous to at least some scribes in the Their numbers will surpass that; their reti- northwest: nues will be inconceivably [large, equal to] the sands of five hundred [Ganges] rivers, Si le scribe n'a pas jug6 bon d'indiquer la nasalisation, or to four [hundred]or to three [hundred], c'est qu'elle allait de soi a partirdu moment oh la syllabe Or to two hundred [Ganges] rivers; their 6tait lourde et ouverte: en d'autrestermes, toute syllabe attendants will be like this. Their measure ouverte lourde comporte des vibrations nasales, que sa will also surpassthis, be it five or ten [riv- voyelle soit une voyelle 6tymologiquement breve et ers]. (Krsh, 177-178) nasalis6e ou que sa voyelle soit une voyelle 6tymolo- giquement longue, 6tymologiquement non nasalis6e, et There are multiple problems with these two verses from nasalis6e la suite conform6ment a une tendance fourteen. First of all, Dharmaraksahas clearly par chapter des indo-aryennes.59 mistaken several of the numbers: spontan6e caract6ristique langues Sanskrit (KN) text Dharmaraksa'stext 56 pahcdaatiya (50)55 i5. (500 < *pafcasatl) See Yuyama 1992. catvdrimsac (40) (1 (T) (400 < *catuhsati) 57 In the case of 200, we could hypothesize that it was read trimsati (30) -~ (300 < *trisatf) as duvisati and that the aksara du- was perhaps mistaken as a virmati (20) _~ (200 < *dvisatT) particle (= tu). 58 Fussman 1989, ?33 ff. 59 Fussman 1989, 478; for remarks on nasalization in the 55 On these BHS forms, see BHSG, ??19.29-30. Niya documents, see Burrow 1937, ?47. BOUCHER:Gdndhari and the Early Chinese Buddhist Translations Reconsidered 485

We might hypothesize, then, that Dharmaraksa,faced nese translations-an effect that clearly must be taken with a kharosthi manuscript in which nasalization was into account in the search for the underlying language of at best irregularlymarked, made mistakes both of omis- the source text. Given the rathersizeable body of docu- sion and addition in his recitation of the text. However, mentationof this process in China-colophons, prefaces, while this is certainly possible, we must also consider and bibliographers'notices-we can often reconstructin the fact that this problem is, in Fussman's words, "car- broad terms the roles of the Indian or Central Asian act6ristique des langues indo-aryennes."That is to say, missionary, his assistants, and his Chinese scribes in tra- we often find in Prakritbroadly a tendency to lengthen versing the huge linguistic and cultural divide separat- an open syllable through nasalization (cf. Pischel 1955, ing the Indian and Chinese worlds. The general steps of ?74), or, conversely, to lengthen a vowel with omission this process have been conveniently summarizedby Erik of anusvara (e.g., visati < vimsati; von Hintiber 1986, Ziircher: ?112). This phenomenon occurs regularly in Pali60and even occasionally in Buddhist Hybrid Sanskrit.61And it The mastereither had a manuscriptof the originaltext is important that we heed Brough's caveats concerning at his disposalor he recitedit frommemory. If he had Nepalese scribal tendencies-including a general care- enoughknowledge of Chinese(which was seldomthe lessness in the marking of anusvara-in assessing the case)he gave anoral translation (k'ou-shou IlR), other- likely "original" form underlying our witnesses to BHS wise thepreliminary translation was made, "transmitted", lexical items.62We have, in short, no exact means to as- by a bilingualintermediary (ch'uan-i 4iW). Chineseas- sess the quality of the manuscript tradition from which sistants-monks as well as laymen-noted down the Dharmaraksamade his translation. translation(pi-shou !), after which the text was Taken by themselves, then, confusions related to na- submittedto a finalrevision (cheng-i IE-, chiao-ting1 salization are inconclusive with regard to the underlying t). Duringthe workof translation,and perhaps also on language or script of Dharmaraksa'ssource text. But otheroccasions, the mastergave oral explanations (k'ou- taken with the growing body of other problems in this chieh 11) concerningthe contentsof the scriptures translation, we might be just as inclined to see here mis- translated.63 takes in the interpretationof Dharmaraksa'srecitation of the Indic text by assistants with just enough knowledge More specifically for our investigation, we have two of Sanskritto confuse themselves in situations for which extant colophons to Dharmaraksa'stranslation of the SP context did not clearly decide among the alternatives for (Zhengfahuajing IEi ,k) preserved within Sengyou's them. This would be all the more likely if Dharmaraksa's early sixth-centurybibliography, the Chu sanzangjiji ij recitation of the source text was influenced by pronun- 'AgiB [Compilation of Notices on the Translationof ciation habits in which heavy syllables-especially in the Tripitika,hereafter CSZJJ], which preserve explicit verse-were representedhaphazardly. details of the translation process and the roles of the participants.The first colophon is translatedas follows:64 III. THE CHINESE TRANSLATION PROCESS Onthe tenth day of theeighth month of the seventhyear In the above survey of the linguistic data for a possible of the Taikangreign period [= September15, 286 C.E.], Gandhariinfluence on the transmission of the Indic text the Yuezhibodhisattva sramana from Dunhuang, Dhar- of the SP, I have on a number of occasions referred to maraksa,holding the foreign (hu 1) scripturein his Dharmaraksa'srecitation of the Indic text and to his as- hand, orally deliveredand issued65the twenty-seven sistants'possible misunderstandingsof specific locutions. chaptersof the Zhengfahuajing, conferring(shou t) it Before proceeding with an evaluation of the presented uponthe updsakaNie Chengyuan,Zhang Shiming, and data or additional evidence, I should pause to ZhangZhongzheng, who togethertook it downin writ- consider in more detail the process by which Dharma- ing.66Zhu Decheng,Zhu Wensheng,Yan Weibo,Xu raksa'stranslation attained its final form. In fact, it is my contention that this process had a significant and hitherto 63 underestimated effect on the shape of these early Chi- Zurcher1959, 31. For otherscholarly discussions of the translationprocess in China,see Fuchs 1930, van Gulik 1956, Hrdlickovi1958, Ch'en 1960, Tso 1963. 60 64 Geiger 1994, ?6. Taisho 2145, vol. 55, 56c.16-24. 61 65 BHSG, ?3.1-4. kouxuan (var. adds chuan)chu []_i(L) f. 62 66 Brough 1954, 355. bishou :1, literally "received with the brush." 486 Journal of the American Oriental Society 118.4 (1998)

Wencheng, Zhao Shuchu, Zhang Wenlong, Chen Chang- Dharmaraksa.70 We can only imagine what a xuan, and others all took pleasure in encouraging and hodge-podge of linguistic backgrounds such a variety of assisting. It was finished on the second day of the ninth assistants would have brought to the translation process. month [= October 6]. The Indian sramana Zhu Li and the Kuchean Bo Yuanxin both collated (can- il the translation.On the sixth of the second jiao ) day times, and members of this clan are known to have been partic- month of the first it was reexamined. year67 Furthermore, ularly active in the productionof Buddhist texts at Dunhuangin on the fifteenth of the fourth month of the first day year later periods (see Teiser 1994, 146, n. 26). With regard to Zhu of the 291 Sun Yuankangreign period [= May 29, C.E.], Decheng and Zhu Wensheng, who "took pleasure in encourag- Bohu of it with Chang'ancopied simple glosses.68 ing and assisting" the work on the Saddharmapundarikasutra, Hurvitz states: "These two Chinese lay brethren with the sur- A number of points from this colophon stand out name Chu [Zhu] must have been very devout indeed, since, al- immediately. First, this translation, like many in Dhar- though still laymen, they had left the secular community, an act maraksa's corpus, was very much an international affair. symbolized by abandoning their clan name and taking instead No fewer than thirteen participants in the translation the name Chu, which, as indicated above, is short for 'T'ien- process are mentioned by name, and these include Chi- chu,' i.e., India" (Tsukamotoand Hurvitz 1985, 486, note "ad"). nese,69 an Indian, a Tokharian, as well as, of course, the Hurvitz'sspeculation-and that is all this is-is dubious for two reasons. For one, despite the Chinese-looking personal names, 67 The question, of course, is first year of which reign period. it is not impossible that they were both naturalized Indians Tsukamoto and Hurvitz 1985, 551, note 3 assume the reign living in China. Secondly, if they were Chinese, it is likely that period to be Yongkang [= 291], but that is unlikely given the they were monks, given that they had adopted the ethnikon of a fact that that reign period only begins in the third month. The foreign master, perhapseven Dharmaraksahimself (cf. Ziircher first new year after the Taikang period is Taixi, which would 1959, 68). Among the assistants on Dharmaraksa'svarious trans- make this date equivalent to March 3, 290. This problem is lation committees with the ethnikon zhu, only two, Zhu Li and exacerbated by the fact that there are four rapidly succeeding Zhu Fashou, are clearly of Indian descent and both are de- changes of reign titles in the years 290-291; whether the anon- scribed as sramanas. ymous colophon writer was in touch with such changes at court 70 The ethnic identity and linguistic affiliation of the Yuezhi is impossible to determine. Tang 1938, 112 and Okabe 1983, is one of the most vexed subjects in Central Asian history. De- 21 read yuan nian 7dC here as a mistake for [Taikang]jiu nian spite decades of studies drawing upon Greco-Roman, Chinese, AfL [= March 25, 288]. This reading has the advantage of ex- Tibetan, and Central Asian sources, there has yet to be a con- plaining why a new reign title was not specifically mentioned sensus on many of the most fundamental issues. Much of the in the notice. problem lies in the great difficulty-and probable impossibil- 68 Exactly what the Chang'an devotee Sun Bohu did is not ity-of pinpointing the identity of the Yuezhi before their entirely clear. The colophon states that he xie sujie ,I*, expulsion by the Xiongnu out of Gansu in the second century "copied [the translation, making] a simple exegesis." Okabe B.C.E. Maenchen-Helfen 1945 is almost certainly correct in sug- 1983, 21 proposes to read xie sujuan ,1,,P, "copied it onto gesting that the ethnikon Yuezhi in Chinese sources ceased as a pure silk." Though perhaps a clearer reading, there is no obvi- sociological-ethnic term after the migration of the Great Yuezhi ous reason to adopt such a emendation. Interestingly, Sun Bohu to the west. From that point, this designation represented a is mentioned in Dharmaraksa'sbiography in the Gaoseng zhuan composite people: one group (the Dayuezhi tkA 5) settled in (Taisho 2059, vol. 50, 327a. 6-7) as one of the several people the western Tarim Basin and eventually conquered Bactria, who regularly "held the brush and collated [the translation] in where they adopted an Iranianlanguage and culture; others (the detail at the request of Dharmaraksa."It is not unreasonableto Xiaoyuezhi /I\ J 5) remainedin the Nanshan region (in modern hypothesize that if Sun Bohu did in fact play a significant role Gansu) among the Qiang tribes and probably spoke a Tokharian on Dharmaraksa'stranslation committees as the Gaoseng zhuan language. The problem of Dharmaraksa'sethnic identity is not suggests, then he very well may have produced a series of without significance for this investigation. As noted several exegetical notes to the Saddharmapundarikasutrafor the faith- times already, Dharmaraksa'sown pronunciation habits could ful in Chang'anas he copied down the text, perhaps even at the have been responsible for some of the translationconfusions we request of Dharmaraksahimself. have considered and will consider below. It would be of some 69 Among the Chinese on this translationcommittee are three interest then to know if his native pronunciationwas affected by members of the Zhang X clan: two scribes and one of the Iranianhabits, perhapsto a greaterdegree than Gandharispeak- patrons. Wolfram Eberhard(1956, 213-14) has listed this clan ers in northwest India, or by a Tokhariandialect, as the inhab- name among the prominent families at Dunhuang from early itants of the Shanshan kingdom appear to have been. BOUCHER:Gandhdri and the Early Chinese Buddhist Translations Reconsidered 487

They almost certainly would have had a diverse range of Exactly what transpiredbetween the recitation of the skills in Indian languages, and, perhaps more impor- Indic text and the creation of the literary Chinese trans- tantly, they would have had an equally diverse range of lation obviously cannot be known with certainty.But it is pronunciationhabits. clear that we cannot take for granted the polyglot skills Secondly, we might also notice that the translation attributed to Dharmaraksa by the Chinese hagiogra- was completed in about three weeks, quite rapid for such phers. Other preserved colophons suggest something of a large text, even by ancient standards.7'It is difficult to an evolution in his Chinese ability. In the colophon to judge what effect this pace would have had on the fin- his earliest recorded translation, the Suvikrdntacinti- ished product, but it is worth noting at the very least that devaputrapariprcchd (Taisho 588), Dharmaraksa is once this translation was carried out under circumstances not again described as "orally conferring and issuing" the altogether typical of Dharmaraksa'sforty-year career. text, after which the Parthian An Wenhui and the Ku- What is most important about this colophon, how- chean Bo Yuanxin "transferredthe words" (chuanyan ever, are the actual steps of the translation process that 1f#). Their transmission was then taken down with it reveals. First, it mentions that Dharmaraksaheld a the brush by Nie Chengyuan, Zhang Xuanbo, and Sun manuscript in his hands. Dharmaraksais the first trans- Xiuda.74It is clear here that Dharmaraksawas not able lator in China who is reported to have held an actual independently to translate the Indian text into Chinese. manuscript during the translation, though we should be Later in his career, however, the colophons state explic- cautious in attaching too much significance to what may itly that he "held the foreign text in his hand and orally be a simple omission of detail. What follows then is of delivered it into Chinese" (kouxuan jinyan Fi_F ).75 crucial significance. Dharmaraksais said to have "orally delivered and issued" the whole of the SP, conferring has argued that chu refers to an oral translation as opposed to the text the Nie and two other upon updsaka Chengyuan yi i, a written one. Since all translationsby Indian and Central scribal assistants.72 the is Though colophon ambiguous Asian missionaries were carried out orally, there appears little at this I take it to mean that Dharmaraksacon- point, point to such a contrast. Richard Robinson (1967, 298, n. 28) veyed a recitation of the Indic text to Nie Chengyuan contends that chu at least sometimes refers to the recitation of with a series of along preliminary glosses and exegeses, the Indic text, not its translation into Chinese; he cites several and that he, not Dharmaraksa,then converted the oral examples. Robert Shih seems, in part, to support this position: draft translationinto Chinese, influenced as well literary "Dans les prefaces, la difference entre 'publier' et 'traduire'ap- his of the Indic recitation.73 by apprehension parait clairement. Celui qui tient en mains le texte indien joue un role plus importantque celui qui traduitl'indien en chinois" 71 The year Taikang7 was an especiallyactive periodfor (Shih 1968, 168). While the greater importance of the foreign Dharmaraksa.Besides the SP, he also translatedthe Panca- master was certainly acknowledged by the Chinese bibliogra- vimsatisahasrikaprajfiapramita, the Visesacintibrahmapari- phers, we have looked at data that calls this into question-at prccha, and the Ajatagatrukaukrtyavinodand,all of which are least without substantial qualification. Arthur Link has gone sizabletexts. furtherto suggest that chu is "an abbreviation for the technical 72 Nie Chengyuanwas without a doubtDharmaraksa's closest Buddhist compound i-ch'u [t_] .... That is i-ch'u means disciple.He is mentionedin a numberof colophonsto Dharma- 'translated [with the result that a book] is issued,' or more raksa'stranslations, including the earliest,the Suvikrantacin- simply, 'translate"'(Link 1960, 30). None of these positions is tidevaputrapariprccha,translated in 267 C.E. Thus he had over fully satisfying. To "issue" an Indian text is to bring it out of its twentyyears of experienceworking on Dharmaraksa'stransla- native guise, to make it available. That process, however, re- tionteams by the timeof the renderingof the SP.Furthermore, quired at least two steps that were not necessarily performedby he is eulogizedin Dharmaraksa'sbiography as follows:"[Nie] the same person. The Indian text had to be recited aloud, its es- Chengyuanwas wise and experienced,talented and princi- oteric script being otherwise impenetrable to native assistants. pled-devout in the work of the dharma.When MasterHu It also had to be glossed in Chinese, since the Indic sounds [Dharmaraksa]issued scriptures,he [Nie Chengyuan]would were no less befuddling than the manuscript.While we can rea- frequently examine and revise them" (CSZJJ, vol. 55, 98a.26- sonably hypothesize that Dharmaraksaboth recited the Indian I will return 27). to Nie Chengyuan and his possible influence text of the SP and explained it in at least general terms for his on the translationof the SP below. Chinese assistants, it is unlikely that chu can be thought of as 73 The crucialword here is chu fi, a very common verb, yet "to translate"in the way that we now use the word. difficult to pin down, describing translation procedures in 74 CSZJJ, vol. 55, 48b.22-26. 75 China. It is often translatedas "to publish," but that does noth- See, for example,the colophonto his Lalitavistara,trans- to ing clarify the designated activity. ArthurWaley (1957, 196) lated in 308 C.E. (CSZJJ, vol. 55, 48b.27-c.l). 488 Journal of the American Oriental Society 118.4 (1998)

The "translator" of the SP is by comparison more dif- bhiksu Kang Nalii copied the whole of the Zhengfahua ficult to discern and this very well may have been inten- at Luoyang. Then, together with the upasakas of pure tional. It is easy to imagine that Dharmaraksa, despite conduct and restraint Zhang Jibo, Dong Yingxuan, Liu having attained considerably improved skill in Chinese Changwu, Chang Wen, and others, he took the sitra over the course of twenty years of translation work, original in hand and went to the White Horse Monastery would still have been unable to translate the text on his (Baima si El ,)77 to face Dharmaraksa.He [Dharma- own. He would in all likelihood have contributed con- raksa] orally collated the ancient teaching [with the siderably to his assistants' understanding of the text, add- translation] and explained its profound meaning. On the ing exegetical comments as well as his own suggestions uposadha day78-the fourteenth-in the ninth month about appropriate renderings. Nevertheless, much of the [= November 3, 290], he gave79 to a great assembly at translation was shaped, I suspect, by a Chinese who was the EasternOx Monastery (Dongniu si t-:-) a lecture almost certainly not fully qualified in the Indic lan- on this sutra. All day and all night everyone was com- guages.76 Obviously the significance of this for an un- pletely enraptured(by his talk). [The translation] was derstanding of the underlying language of the Indic text again revised (chongyi jiaoding MEr, lit., "collated is considerable. and established").80 The colophon also tells us that quite a number of individuals "encouraged and assisted" in the translation. Once again a few points deserve to be highlighted. First We have no way to know what "encouragement" and "as- of all, besides the copying of the translation recorded in sistance" consisted of, though we would expect that those the first colophon of the year 291, this colophon records named were the most generous patrons of the translation yet another copy being made in the year 290, this time committee. The translation was proofread after comple- at Luoyang. Moreover, the international constituency of tion by an Indian monk and Kuchean layman-a Ku- Dharmaraksa's associates continues, here involving a chean who, as we saw above, had previous experience Sogdian81 monk, and interestingly, a number of Chinese with Dharmaraksa. This combination of disparate na- upasakas who probably were under his tutelage. Dhar- tionalities and Buddhist "ranks" in what should have maraksa himself is also on the move, travelling to the been an important conclusion to the translation is per- famous White Horse temple of Luoyang to spread the haps not as unusual as it might seem, given the generally teaching of this newly translated scripture.82 international character of Dharmaraksa's cohorts at Dun- huang and Chang'an. But it should again be borne in mind that whereas the translation of the Indic text may 77 Read si X~ with the variant. have been substantially shaped by a Chinese assistant, 78 I read benzhai T4 (lit., "original fast") as referringto the the Chinese translation is here checked by an Indian and monastic holy day, the uposadha, at which time monks often a Tokharian. recited the pratimoksa and laymen took special vows. This des- This short but rich record of the translation process ignation occurs again in the colophon to Dharmaraksa'strans- is not the last we hear of the fortunes of this text. The lation of the Lalitavistara (CSZJJ, vol. 55, 48b.28). anonymous colophon is also preserved in the following 79 Shitan l1tfW appears to be a translation-transcriptionof Chu sanzang ji ji: dana, "giving." 80 CSZJJ, vol. 55, 56c.25-57a.2. On the twenty-eighth day of the eighth month of the first 81 The ethnikon kang f is generally taken to represent Sog- year of Yongxi reign period [= October 18, 290], the dian, but WolframEberhard has shown that there is some reason to believe that early use of this ethnikon may have designated 76 Aside from the pilgrims who studied extensively in India, two different clans: one that was native to Kangguo (present- native it is unlikely that any Chinese in traditional times truly com- day Samarkand)and another, the old Kangju, who were these manded any Indian literary language. Cf. the remarksby R. H. to Gansu before being forced to emigrate to Transoxiana; van Gulik: "... [T]he average Chinese scholar considered a latter may have been Yuezhi (Eberhard 1955, 150). It is also monk after knowledge of the Indian script alone tantamount to a knowl- possible that this ethnikon was adopted by a Chinese which became com- edge of the Sanskrit language. Chinese terms like fan-hsiieh- ordinationby a Sogdian preceptor,a practice fourth centuries. seng 'a monk who has studied Sanskrit'as a rule means nothing mon among Chinese clerics in the third and more than 'a monk who has mastered the Indian script'" (van 82 This appears to be one of the few recorded instances of his Gulik 1956, 13). For a fascinating discussion of how even a very Dharmaraksatravelling this far east. The vast majority of learned Chinese Buddhist scholastic fundamentally misunder- translations were carried out in Dunhuang and Chang'an. that stood the nature of Indian languages, see Link 1961, 281-99. Though not explicitly stated, there are several indications BOUCHER:Gdndhari and the Early Chinese Buddhist TranslationsReconsidered 489

The most importantobservation from this colophon is Sanskrit or Prakrit vocabulary and little knowledge of the last line: "[the translation] was again revised." The grammarmisunderstanding phonemes that may not have colophon is not specific as to who revised the translation been easily distinguished in pronunciation. We would or for what reason. It seems reasonable in the context to expect that the general purportof a given passage would suppose that Dharmaraksahad revised the translation as have been explained to him by Dharmaraksa; at the a result of his oral collation and lecture in Luoyang. same time, his own understandingsand misunderstand- Although we do not know what alterations he may have ings of what he perceived to be key lexical items may made, for our purposes it is important to note that sev- have intruded into the translation, producing at times a eral years after the original work of the translationcom- quite different if not nonsensical rendition of the pas- " mittee-the committee upon whom the recitation of the sage. That Dharmaraksafelt the need to revise again" Indic text was conferred and by whom it was translated, the translation during his preaching at Luoyang is then written down, and proofread-Dharmaraksa indepen- not surprising. That he did not find every mistake con- dently changed its shape. It is this revised text that we firms our sense of his limited Chinese skills. presumably have received. Thus the very phonological features that point to a Having discussed in some detail what is known about Gandhari source text-mistakes with regard to vowel the method by which the SP was translated in 286, we length, aspirated and unaspiratedstops, and distinctions can begin to reconsider our evidence for the connection between the nasals-are precisely the sounds that we between the underlying language of Dharmaraksa'sIn- might expect a Chinese assistant to have difficulty dis- dic text and Gandhari Prakrit. Dharmaraksa'srecitation tinguishing. In addition, this situation would have been of the SP was almost certainly influenced by a recension complicated no doubt by the influence of Central Asian of the text containing many more Prakritic forms than pronunciation habits in Dharmaraksa'srecitation of the any of the manuscripts that have come down to us, not Indic text. I have noted above the possibility of voiced to mention the likelihood of interference from his na- intervocalic stops taking on spirancy in Iranian fashion tive language and the habits of his Indian teacher. We (e.g., avalokitesvara > abhd-loka-svara:-v- > [[] > -bh-) may seriously question then whether his Chinese assis- or being devoiced as in Tokharian(e.g., durgandhiunder- tant or assistants would have been able consistently to stood as durgati: -(n)dh- > -t-). It should be clear by now distinguish Indic locutions whose pronunciationin Prak- that the oral/auralnature of the translationprocess must rit-whichever Prakrit-would have presentednumerous be treated with as much consideration as the linguistic opportunities for confusion. data itself. Furthermore,there is a considerable body of Thus mistakes such as the bho(n)ti/bodhi confusion other kinds of evidence that may provide even more de- can hardly be attributedto Dharmaraksa.If it is possible tails about the underlying language of the Indic text and that Dharmaraksa could have read a manuscript that the roles of the translationparticipants. interchanged bh and b (e.g., a kharosthi manuscript in Gandhari Prakrit), it is nearly impossible that the same IV. ADDITIONAL DATA manuscript would have also added aspiration to -(n)ti.83 The fact that most instances of both bho(n)ti and bodhi Double Translations in the text are translated correctly makes it all the more unlikely that the person with the greatest knowledge of One of the most unusual features of Dharmaraksa's Indic languages would have confused these words only translation idiom and one to which I have alluded al- occasionally. On the contrary, it is not difficult to imag- ready is the occurrence of what I call double transla- ine a CJinese assistant with an incomplete knowledge of tions. These are cases in which an Indic term is rendered twice in close proximity, presumablybecause two differ- Dharmaraksa'stranslation of the SP was carried out at ent words had collapsed together in pronunciation, at least as recited Dharmaraksa.84His translation Chang'an. by assis- unable to decide between two or more 83 The dubiousnessof a Gandhariinfluence in this example tants, possible offered both the fact that has been furtheremphasized by G6rardFussman (personal options, possibilities despite such a almost in communication,June 1995):bhavati is almostalways attested rendering always resulted nonsense. We will look at several of as hoti (pronounced hoti, hodi, or ho'i) in northwest kharosthi examples this phenomenonbelow. inscriptions;bodhi is oftenwritten bosi (pronounced[bozi]), at least from the first centuryof the commonera. A phonemic overlapbetween the two words is thus highly unlikelyin a 84 For an interestingparallel to this phenomenonin the Gandhfiritext datingfrom the thirdcentury. Uighurtranslations of ChineseBuddhist texts, see Zieme1992. 490 Journal of the American Oriental Society 118.4 (1998)

KN 162.5: lokavidu decision was not made between the two possibilities, one who understandsthe world(epithet of resulting in an incoherent translation. a buddha) KN 231.3-4: tathagatapaniparimarjitamiurdhanasca te Dh 89b.13: tZ,5_ bhavisyanti sagelyfather of the world(Krsh, 108-9) they will have their heads strokedby the hand of the tathagata

KN 193.1: yatha vayam lokavidu bhavema Dh lOlb.16: alS17kJ, ,#' Tt75t? _ ,i1 just as we will become knowersof the to seekthe water of thetathdgata and aspire world to be in theBuddha's palm-this is theresult of the practiceof formervows (Krsh,140) Dh 93b.23-24: 42 tZ Dharmaraksa did not this we will becomewise fathersof the world Clearly adequately convey line to his assistant, if he even understood it himself. (Krsh,119) It appears that there was a confusion between -pdni- Dharmaraksaappears to have renderedboth -vida (wise) (hand) and pdni(ya) (lit., "drinkable,"hence "water"), and -pitu (father). While there are a numberof instances exhibiting the interchangebetween n/n that we discussed of an interchange between p and v in kharosthi doc- above. As one can plainly see, by not adequately differ- the uments and inscriptions-if that were the script of Dhar- entiating possiblities presented by the recitation of maraksa's manuscript-it is obvious that both words the text, Dharmaraksa'stranslation assistants produced an could not have been representedin the same place. Such utterly nonsensical rendering of the passage. Even if a mistake suggests that the pronunciation of these two Dharmaraksahimself were responsible for some of the words (-vidu and -pitu) had coalesced, and therefore, confusions, for example, by having indicated alternative Dharmaraksa'stranslation assistants, unable to deter- "possibilities" of interpretation,it is clear that he could mine the proper reading, deduced that two voiced con- not have fully understood and accepted such a rendering sonants here (-v-,-d-) could have been derived from two in Chinese. In fact, data such as these make it especially difficult to believe unvoiced consonants (-p-, -t-). It is also possible, as I that a single person with adequate have mentioned several times now, that Dharmaraksa's knowledge of both Indic languages and Chinese could have left such translations pronunciation habits were influenced by a Tokharian intact. Much of our evidence idiom in which -v- and -d- were devoiced, which would suggests to the contrarythat Dharmaraksa'srecitation of the text was also account for the uncertaintyof interpretation. Indic mediated by someone with a modest command of Sanskrit/Prakritvocabulary and a rather KN 301.6: sviakrds caiva te sattvdh poor grasp of Sanskrit grammar.This points to the semi- andthese beings of good disposition bilingual intermediariesthat our colophons speak of and in the case of the SP, Nie Chengyuan, in particular. Dh 11 la.6: x :U N beingswho havegood causes/rooms (Krsh, pratyaya/pratyeka-buddha 176) It has been known for some time that there are two versions of the title of the buddha of It appears here that Dharmaraksaand/or his assistants widely occurring the "second vehicle": from understood both akara (ground, reason, cause, disposi- pratyaya-buddha (awakened and on tion; cf. BHSD, 86) and dgdra (dwelling, house, room). [external?] causes) pratyeka-buddha (awakened one's The alternationof these two terms has led to Of course we have already seen several examples of own). confusions between voiced and unvoiced intervocalic a number of folk etymologies in Buddhist literature, as well as in modern The best discussion consonants.8 What is astounding here though is that a scholarship.86 Dharmaraksahimself would have beenunclear as to the actual 85 We might also hypothesizethat such a confusioncould wordintended by the Indic manuscript.At the very least we have resulted from a kharosthi manuscript in which the nota- are remindedof the complexityof decidingamong multiple tion -k- couldstand for -g-, as in the GandhariDharmapada (cf. indeterminablefactors in the transmissionand receptionof Brough 1962, ??30-31) or, conversely, the notation -g- [= y] thesetexts. could stand for either -k- or -g- as in the Niya documents (cf. 86 Therehave been two ratherunsatisfactory monographs on Burrow 1937, ?16). If this were the case, it is possible that the pratyekabuddhafigure: Kloppenborg 1974 and, more re- BOUCHER:Gandhari and the Early Chinese Buddhist TranslationsReconsidered 491 of the philological problems related to this figure is by KN 49.2: ekam idam ydna89dvitiya nasti K. R. Norman.87 Norman convincingly demonstrates that this is the only vehicle; there is no second the available , Prakrit, and Sanskrit evidence in Bud- Dh 71a.2: i, - I _- dhist and Jain texts and inscriptions points to pratyaya- buddha as the original form of the word, and that as for wisdom/vehicle, there is one, never pratyeka-buddha represents an incorrect back-formation two (Krsh 1993, 139) (as would the Jain patteya-buddha). Translations reflecting one or the other form of the KN 49.7-8: sarve ca te darsayi ekayanam ekam ca term occur throughout the , yanam avatdrayanti including those of Dharmaraksa as well. Yet there are ekasmi ydne paripacayanti acintiyd prani- also instances in which Dharmaraksa (or perhaps his sahasrakotiyah assistants) was unable to decide between the two: All [buddhas] have manifested but a single vehicle, and introduceone vehicle KN 10.4: pratyekayanamca vadanti tesam they only. With this one vehicle they bring to matura- to them about the they speak solitary vehicle tion inconceivably numerous thousands of kotis of living beings Dh 64b.4: X ,- - furthermore they are able to obtain the Dh 71a.8-10: At:,t -~g5 i _ 1t :t vehicle of the conditioned-solitarybuddhas For the sake of [the bud- Dharmaraksa's rendering reflects an underlying pratyaya- beings everywhere, dhas] manifest one vehicle; therefore they eka-[buddha-]ydna, clearly nonsensical in any context, teach this path to liberate the unliberated. but cognizant, interestingly, of the two words possible They always teach for the sake of men the that could have collapsed in Prakritic pronunciation. equanimous path/knowledge, converting We should also note that this particular double trans- hundredsof thousands of millions of kotis lation predates Dharmaraksa. In his translation of the of beings (Krsh 1993, 143) Vimalakirtinirdesasatra, Zhi Qian (ca. 220-52) has the Note that besides the double translation here following: ~xz.,;-, fl (furthermore [I] will (daohui establish others in the practice of the sravakas and of the tEi._), this verse also clearly establishes the semantic of and a pratyaya-eka-buddhas) (Taisho 474, vol. 14, 522a.26). equivalence sheng T (vehicle) dao (path). Thus, as in previous examples, we must consider the pos- KN 189.1-2: md khalv ima ekam eva buddhajndnam90 sibility that Dharmaraksa and his team borrowed well- srutvd dravenaiva pratinivartayeyurnaivo- known locutions from translations. previous pasamkrameyuh ydna/jidna bahupariklesam idam buddhaiidnarm9 samuddnayitavyamiti Besides a number of alternations between these two These [beings], having heard this one and words-cases where Dharmaraksa reads ydna when one only buddha-knowledge, should not casu- or more of the Sanskrit manuscripts reads jndna and ally turn back and not go all the way [think- vice versa-there are several instances in which a Chi- ing]: "To acquire this buddha-knowledgeis nese rendering of both terms was provided for one or the fraught with too many difficulties." other Indic word.88 Dh 92c.14-15: X Z5iS ;-3- riti- .T _ cently, Wiltshire 1990. On the latter see the review by Collins Furthermore,the Buddha taught from the (1992). beginning that there is one vehicle; having 87 Norman 1983. heard the Buddha teach the dharma, [these beings] do not the 88 I will in this section draw upon an article that Karashima accept path/knowledge (Krsh 1993, 140) published in 1993. He there makes the provocative claim that the very conception of "vehicle" as a central motif of identifi- cation for the 89 Mahayanamay very well be founded on an incor- Karashima (1993, 139) notes that one Sanskrit MS (Add rect back-formationof the Middle Indic word for "knowledge" 1682 housed at the Cambridge University Library)reads jndna in the process of Sanskritization. It is my intention to produce here. an translationof English this very interesting article in the near 90 Kash 183a.3 reads: buddhaydnam. future. 91 Kash 183a.4 reads: buddhaydnam. 492 Journal of the American Oriental Society 118.4 (1998)

What is especially striking about this example is that in Dh 66c.2-3: ~ iS:--'Elt. -p,iE {i~a-f both lines of this verse we have buddhajiinanm(or in [the buddha]will encourage and develop in- the case of the Kashgar MS,buddhayanam) represented numerable bodhisattvas and inconceivable in the Sanskrit, but two different renderings in the hundreds of thousands of kotis of men Chinese, the latter a double translation. While Dharma- (Krsh 1993, 148)94 raksa's strict adherence to four-character prosody cer- To complicate matters further,there are also instances in motivated the use of a two-character equivalent tainly which Dharmaraksaconfused dhyana with dana (giving): here, this example would suggest either a certain amount of indecision on the part of the scribe, or perhaps an in- KN 13.10: dhydnena te prasthita agrabodhim tentional attemptto indicate the ambiguity of a Prakritic they set out for highest enlightenment locution. through meditation _ jniina/dhyana Dh 65a.13: ;f L t setting their thoughts on giving, they seek There are several instances in which jiina is rendered the noble enlightenment (Krsh, 32)95 as dhydna and vice versa in Dharmaraksa'stranslation. Such an interchange presumably would have taken place, as Karashimarightly suggests, through a Prakritic KN 24.13: sarvasvaddndniparityajantah development jiana > jana (or jana) < jhana < dhyana, forsaking donations of their whole property them at least in some contexts making indistinguishable, Dh 67a.3-4: -J:Jt TEN, especially to the ear.92There is also at least one case no meditation rise to causes [sic!] where a rendering of both terms appears together: gives (Krsh, 40)96 KN 53.2: yend vinesyanti 'ha pranakotyo baud- These would to a confusion dhasmi jiinasmi andsravasmin examples appear represent between dhy and d. We should expect the assimilation of by which means [the buddhas] will train dental with to palatal stops, though kotis of beings here in the buddha- stops y conjuncts there are the Prakrits (cf. Pischel knowledge which is without depravities exceptions among 1955, ?281). Moreover, we can never rule out the pos- Dh 71c.23-24: WfilJt]{L%tO t I'Y WAAM sibility of redactional differences playing a role in the by which they spur on beings numbering in disparity between the Sanskrit and Chinese versions. To the kotis in the meditation/wisdom with give one relevant example, KN 14.4 reads: desenti te which they dry up the "outflows" (Krsh pranasahasrakotinamjninena te prasthita agrabodhim 1993, 149) (they [the buddhas] teach thousands of kotis of beings, and these set out for highest enlightenment by The confusion between jiana and dhyana provides ad- (beings) means of As Karashima points out (Krsh, ditional informationconcerning pronunciation.We know knowledge). Asian MS from the Stein Col- that the consonant clusters in both words would have 33), the Central fragment lection (H3 Kha. i 24, fol. 4a.8; Toda 1981, 265) also been assimilated to nearly homophonous sounds in at reads here. But the Kashgar MS (Kash 20a.6) least some contexts (jni > jj > j and dhy > jh with loss of jnanena and the MS97 both read danena. Dharmaraksa aspiration). This is further confirmed by the Gilgit perceived to follow the latter. fact that jiina is also confused with jana (people): appears

Kash 31a.2-3: samddapeti bahu bodhisatvd(m) acintikd kotisahasra jinne93 94 There are other instances in which jnina is confused with [the buddha] inspires inconceivably many jana, prajana, and jina; see Karashima 1993, 147-48. thousands of kotis of bodhisattvas toward 95 But note that in the first pdda of this same line Dharma- knowledge raksa renders the word dhydna correctly: dhydyanta varsdna sahasrako.ty ("being in concentration for thousands of kotis 92 for hundredsof For example, in the Sanskrit kharosthi document no. 511 of years . . ."); ,J,'{tt,: ("meditating from Niya we find dhyana representedas jana: te jdna parami thousands of kotis of years ..."). 96 Karashima also gata (they attain mastery in meditation); see Boyer et al. 1927, Besides the dhydna/dina confusion, pro- as 186 (reverse, 1. 6). poses that Dharmaraksamistook parityajantah pratyaya x/jan, 93 KN 23.6 reads: samadapeti bahubodhisattvin acintiydn leading to his rather bizarre rendering. 97 uttami buddhajiine. Watanabe 1975, 2: 10.8. BOUCHER:Gandhdri and the Early Chinese Buddhist Translations Reconsidered 493

The examples we have examined in this section pro- of the examples we have considered in this section are vide us with some information concerning Dharma- from verse. Unless I have seriously misinterpreted the raksa's pronunciation of his Indic text. Confusions data, there appearsto be evidence that Dharmaraksamay between jndna andjana suggest that the jn- conjunct had in fact have pronounced the same word (jiina) in vari- been assimilated to j- (< jj). The confusion with ydna is ous contexts in different ways, though this data also sug- possible "since in the North-Indian dialects ya and ja gests that his manuscriptwould have at least sometimes have in many cases coalesced together" (Pischel 1955, written this word as jana. We would be hard pressed to ?236). The confusion with dhydna, pronouncedjana (or account for the range of mistakes otherwise.'1 jhdna), would also seem to confirm this, as would its It is difficult to know what to make of this phenome- rendering by the transcription chan a (Early Middle non of "double translation."One is tempted to specu- Chinese dzian). And such a hypothesis will also help us late-and, of course, only that-that given their obvious with the dhyana/dana confusion. If dental stops became limitations in Indian linguistic matters, the Chinese as- increasingly voiced, tending toward the fricative [6] as sistants may have believed that both meanings of simi- in Gandharipronunciation (under Iranianinfluence), it is larly pronounced expressions inhered in the Indic term not at all difficult to imagine that a word with initial or before them. In some cases this stretches our imagina- intervocalicj (dhydna > jana), pronounced in Northwest tion to rathersevere limits, and we are left to explain the fashion as [z], would have been confusable with [6].98 often extreme semantic disparity of words in such trans- But even if this hypothesis be accepted-and it is lation "binomes." We might, at the very least, hypothe- certainly not clear that it should be-the underlying lan- size that whoever was responsible for these double guage of the text is still not determined.As Edgertonhas translations in all probability viewed both meanings as convincingly demonstrated,Buddhist Hybrid Sanskritor- somehow compatible with the reading of the text. It is thographycan be quite misleading as an index of actual difficult, it seems to me, to attributethese double trans- pronunciation.99It is clear from an examination of the lations to the sole agency of Dharmaraksahimself. Surely verse portions of BHS texts that these sitras were origi- he could not have understoodsuch Chinese renderingsas nally pronounced with far more Prakritic features than sensible. On the contrary, such a hypothesis highlights are now preserved in the manuscripts. For example, the not only the linguistic shortcomingsof his Chinese assis- most widely occurring meter in the SP is the tristubh- tants-in this case almost certainly Nie Chengyuan-but jagati, which requires that the third, sixth, seventh, and also a ratherunderdeveloped knowledge of fundamental ninth syllables be light. Thus a consonant conjunct occur- Buddhist terms and ideas. Nevertheless, these examples ring initially in a following syllable would have to be may a broader and more systematic investigation, pronounced as assimilated even if it were not resolved or- as they could illuminate more precisely the burgeoning thographically.In one of the examples just cited, acintika attempts of Chinese Buddhist literati to deal with a thor- kotisahasrajndne, the ninth syllable, -sra must be metri- oughly linguistic and religious "other"for the first time cally light, though orthographically it is heavy by posi- in Chinese history. tion, being followed by the conjunct jfi-which must In the following few sections I would like to discuss thereforehave been assimilated in actual pronunciation. other kinds of data that do not primarily depend upon What is not as clear from the Indic texts, however, is phonetic confusions caused by the pronunciationof the exactly how such conjuncts would have been assimi- lated. In the case of jni- there are a number of possibili- ties: j,(< jj), n (e.g., ndna), n (e.g., and < djfn), and n 100 It is important to recognize, however, that these transla- (e.g., ndna) (cf. Pischel 1955, ?276). It is entirely pos- tions of and do not a text written in sible that such conjuncts could have been assimilated in jndna dhydna require Gandhari Prakrit, but a Buddhist Sanskrit manu- different fashions by speakers of diverse Prakrits and only Hybrid script read aloud under the influence of a Prakrit dialect in certainly by foreign translators of these texts. It is in which both jn- and dhy- were assimilated to - We cannot this regard that the evidence for phonetic confusions in (jh-). assume that Indian texts were Central Asians in Dharmaraksa'stranslation become particularlytelling. All pronounced by Indian fashion. Even in the Buddhist Sanskrit texts preserved at we find evidence for 98 Gilgit, pronunciationsdifferentiated under In a Taxilaseal inscription(Konow 1929, 100),for exam- the influence of Iranianhabits; see von Hinuber 1989, 357-58. ple, we find a case in which mahajana almost certainly stands Giventhe overwhelmingimportance of oral/auralinteraction to for = mahddhana,where -j- [z] was interchangedwith -dh- = [5]; the Chinese translation process, such a consideration must al- cf. Brough 1962, ?6b. ways be central to our examination of data for the underlying 99 See Edgerton 1935 and 1946. Indic text. 494 Journal of the American Oriental Society 118.4 (1998)

Indic text. Instead we will look at evidence in the trans- It is difficult to determine to which of these two Sanskrit lation that points to the translators'understanding of the recensions (among others: cf. Krsh, 91) Dharmaraksa's text-its grammar,idiom, and ideas. translation corresponds. Dharmaraksa'sxing xing 'ttf could reflect either carya or adhydsaya; his taren If1,2 Grammatical Mistakes seems to render Kashgar'spa(re)sa satvana but his yiqie qunmeng -3JiJ could also be an attempt to translate There are several instances in which Dharmaraksa's sarvesa sattvdna of KN. Regardless of the redactional translationreflects a seeming confusion between -mdna, differences, it appears that Dharmaraksahas mistakenly the inflectional ending of the present middle participle, taken janamandh (or prajdnamdnah) as consisting of and manas (mind, thought): two words: jdna- (men) and -mdnd.h(thoughts) rather than as a participle derived from \Ijia. KN 24.15: aninjamdnas avedhamdndh [the bodhisattvas] are immovable and with- Abstract suffix -tva out stirring KN 255.3-4: kotisahasran bahavah arhatve yo 'pi Dh 67a.6: , L, , <,4 sthapayet their minds contemplate impermanenceand sadabhijiin mahdbhdign yathi gahgdya they do not become unrestrained(Krsh, 40) valikdh Even should one establish in Dharmaraksa's to reflect two confu- -ship rendering appears many thousands of kotis of the greatly vir- sions. he seems to have taken as derived First, aniiija- tuous, endowedwith the six supernatural from anitya (impermanent), probably through the com- faculties and [numerous]as the sands of the mon Prakritic pronunciation of the latter as anicca Ganges... (Gandharianica), assuming loss of nasal and confusion of voiced and unvoiced palatal stops.?01Secondly, Dhar- Dh 105a.6-7: ,It,,l,-f NA-,t maraksaappears to have taken the present middle parti- ciple ending in -mana as manas (mind, thought). If innumerablethousands of kotis [of people?] erect arhat- [sic!]-the most KN 120.7: tathagatas carya prajanamanah sarvesa excellent sages [possessed] of the six super- sattvana 'tha pudgaldntim naturalfaculties, [numerous]as the sands on the banks of the . . . (Krsh, 153) the tathagata, knowing the actions of all Ganges beings and persons ... Karashima proposed that Dharmaraksa's ta it ()- Kash 121a.2-3:tathdgatadhyasaya jdnamdndh pa(re)sa completely out of place here-derives from a confusion satvana ca pudgalana ca between the abstract suffix -tva in arhatve and one of the Middle Indic words for one that is at- the tathdgatas, knowing the dispositions of stupa (thuva), other beings and persons ... tested in inscriptions from Mathura and Taxila.'02 This would assume a confusion between the aspirated and Dh 83a.26-27: 0 Bf- ,Xti 7T ftfl,.L -J, ATMs unaspirated dental consonants, noted already above, as the tathcgatas all observe the traits and ac- well as the regular development of p > v. We would also tions of people, the thoughts of others, and have to presume the insertion of an epenthetic -u-, here all mankind (Krsh, 91) under the influence of the labial semivowel, again a fairly common Prakritic development (cf. von Hinuber 1986, ?155). While this explanation may seem to stretch credibility, it is difficult to discern an alternative. In ad- 101 With to Gandhari,Fussman ?18 and n. 32) regard (1989, dition, the syntax of Dharmaraksa's translation, gener- an of the of < c. If this gives early example development j ously strained in my own rendering, suggests that he did is authentic-and that is not clear-it is unusual. development not perceive both kotisahasran bahavah and sadabhijndn We should both and c to to before expect j develop ya-sruti as referring to mahdbhdgdn (literally, "those possessed final disappearance.A good example of the j > ya-sruti shift is exhibited in one of Dharmaraksa'sfew transcriptions:the name Ajita is renderedas ayi M{1 (66a.17), Early Middle Chinese ?a 102 jit (j hereis IPAhigh frontglide). Elsewhere he translatedthis See Konow 1929, 48 (Mathura Lion Capital) and 87 name as moneng sheng Afl ("cannot be surpassed"). (Taxila Vase inscription). BOUCHER:Gdndhari and the Early Chinese Buddhist Translations Reconsidered 495 of a great share," thus the highly fortunate, illustrious, noted above that the Gandhari sources differ in their and in religious contexts, the virtuous and holy).'03This treatmentof these two nasals. Obviously such a mistake verse then provides yet anotherpiece of evidence for the plays havoc with the understanding of the verse and erratic-to put it charitably-knowledge of Sanskrit cannot be attributedmerely to phonological confusions. grammarof Dharmaraksa'stranslation team. KN 27.12: pujam ca tesadmvipuldm akarsit Mistaken Division of Words he performedextensive homage to them Two Nepalese MSS (one [K'] brought from Tibet by We have noted several above in already examples E. Kawaguchi and preserved in the Toyo Bunko in which Dharmaraksaor his assistants misconstrueda pas- Tokyo105and MS no. 3/672 [?678] preserved in the the words in the sentence In sage by dividing improperly. National Archives of Kathmandu,Nepal) read vipulam one case, for Dharmaraksatook vividhair example, upd- ahdrsit (< Ilhr "to offer") instead. This latter reading yaih as vividhai rupai(yaih). I will note two other appar- seems to account for Dharmaraksa'stranslation: ent cases of such a mistake. Dh 67c.8: - t KN 120.3-4: anuvartamdnas tatha nimitta- nityakalam universal (Krsh,42) carina braviti dharmam greatsage dharmesvaro igvaru sarvaloke mahesvaro Karashimahas proposed that Dharmaraksaread the end lokavindyakendrah of this line as vipula-mahdrsi(t), taking the final verb as The lord of the dharma,lord over the whole maharsi (great seer). world,great lord, chief of the leadersof the world,always preaches the dharmain Semantic Misinterpretations conformitywith those who follow [mere] appearances. There are a number of instances in which Dharmar- aksa's rendering of a word or phrase indicates that he or Dh 83a.21-23: ~;tM,-J 'W.J R4 p , V * his assistant had correctly perceived the respective Indic locution but had construed it with an incorrect connota- In and on consoling urging [others]always tion. I will cite only two examples. atjust the righttime, he has neverengaged in acts out of hope for merit;in the whole KN 14.4: desenti te prdnasahasra kotindm worldhe is the venerableof the dharma, they [the bodhisattvas]teach thousands of and is considered all as the by greatlord, kotisof living beings the supremetathdgata. (Krsh, 91) Dh 65a.29: While there are several in this interesting problems [the bodhisattvas]induce to make the one that people verse, principally concerns us here is the theirconfessions (Krsh, 33) fact that Dharmaraksaor a member of his translation team has mistakenly interpretedthe first line in the neg- Dharmaraksaseems to have interchanged two senses of ative, presumablyby taking the -na of the gen. pl. nimi- the verb desenti/desayati (< /dis): "to teach, impart" ttacdrina as the negative markerna.04 We have already and "to confess" (cf. BHSD, 272).

KN 16.4: lokanatho atha vydkarisyaty ayu bodhi- 103 sattvan In all fairnessto Dharmaraksa,it would appear that Kern thenthe of alsomistranslated this verse in his Englishrendering (Kern 1884, protector the worldwill makea for thesebodhisattvas 242, v. 32); Iwamoto'sJapanese translation is to be preferred prediction (Iwamoto1964, 2: 199, v. 32). Dh 65c.3-4: mW, - 104 Karashimawould also like to see a mistakehere between may [theworld hero] explicate for oursake braviti and bhaveti ("speaks, teaches") (< bhavayati, "culti- [matters concerning] these bodhisattvas vates")through the confusionof b(r) andbh andan unnoticed (Krsh,35) distinctionbetween weakly articulatedpalatal vowels. While this againis certainlypossible, I am moreinclined to see an emendationby Dharmaraksa(or, of course,his assistants)as the passagethat focused upon nimittacdri, albeit in a ratherstrained source.Having interpreted the -na as a negativeparticle, they fashion. 105 would likely have been inspiredto continuea sense for this Toda 1980, lb.1. 496 Journal of the American Oriental Society 118.4 (1998)

Dharmaraksahas confused here two meanings of the What we find then is considerable inconsistency- verb vydv/kr: "to elucidate, explain" and "to predict, redactional differences aside-in the way Dharmaraksa prophesize" (cf. BHSD, 517); the syntax of this clause handles the term lokavid(u) when it occurs outside the is seriously strained by this misconstrual. Clearly the context of the standardlist of the ten epithets of a bud- person responsible for these renderings was not able dha. How shall we account for this? Should we hypoth- consistently to appreciate the precise nuances of certain esize that Dharmaraksahimself misconstrued the word Buddhist technical terms. when it was isolated from the standard list? This is One of the issues raised by these various misunder- certainly a possibility. However, it would seem at least standingswithin Dharmaraksa'stranslation-be they pho- curious that the very translator who adequately and netic, semantic, or grammatical-is to understand why sometimes expertly translatedmany more technical pas- Dharmaraksaand/or his translationassistants interpreted sages elsewhere in the text would have sometimes un- certain words or phrases correctly in some places but not derstood a standardtitle of a buddha and sometimes not. in others. While a systematic survey of all the possibil- I would contend on the contrary that this is yet another ities is well beyond the scope of this paper, I would like piece of a growing body of evidence that points toward to examine at least one clear case of the role that seems Dharmaraksa'sprincipal assistant, Nie Chengyuan, as the to have been played by the context in which certain words source of such problems. We can easily imagine that occur within the text. Dharmaraksa'srecitation of the Indic text-replete with We examined two instances above in which the epi- Prakritisms together with possible Central Asian influ- thet lokavidu (one who understandsthe world) was con- ences-would have presented numerous difficulties to strued as *lokapitu (father of the world). This epithet of a Chinese assistant, however learned, for understanding a buddha occurs twenty-eight times in the Sanskrit text key terms in the text and renderingthem adequately into (KN edition) of the SP. Twenty-two of these occurrences Chinese. Deprived of an illuminating context, not to are found within the standard list of ten epithets of a mention a thorough Sanskrit education, Nie Chengyuan buddha, and in every case, Dharmaraksahas rendered would almost certainly have been at a loss to separateout the term correctly: shijian jie HtIF.106 Of the six various possible meanings of certain Middle Indic words remaining instances of lokavidu, all of which occur pronounced by a Yuezhi monk. Presumably Dharma- outside the standardlist, two of these are not represented raksaoften providedglosses, which accounts for the large in the Kashgarmanuscript of the Sanskrittext, and these number of correct renderings of most terms. But it also are also cases where they are not found in Dharma- appears that on many occasions Nie Chengyuan would raksa's translation, suggesting the likelihood of a re- have trusted his instincts, unwisely at times, and trans- dactional difference between the KN edition and the lated what he heard, or what he deduced from what he Indic text underlying his translation.107In two other heard. Thus in the case of the two renderingsof lokavidu instances Dharmaraksa'stranslation offers a double ren- as "wise father of the world,"Nie Chengyuan, unable to dering, loka-vidu/pitu, as already discussed. In another decide between two possible meanings of a word as pro- case Dharmaraksarenders the epithet as shixiong tHfI nounced, arrived at a compromise solution. It is virtually (world-hero < *lokavira ?), and in the sixth instance he certain that Dharmaraksa'smanuscript could not have leaves the term untranslated.108 read *lokapitu or any variation thereof in these places. It is not difficult to appreciate the appeal of this in- terpretation to a Chinese assistant. And it should be 106 In five of these twenty-twooccurrences within the stan- pointed out that the term lokapitr, "father of the world," dardlist Dharmaraksa an abbreviatedlist of epithetsand gives does occur in the SP itself (see KN 77.8, 80.4, 326.7).109 thus does not actually translate the term lokavid in those cases. Interestingly, at KN 80.4, which reads te tathdgatasya Some of these instances appear to reflect redactional differ- lokapitur abhisraddadhanti (they have faith in the ences in the Sanskrit text where Dharmaraksahas, for example, the father of the world), Dharmaraksatrans- followed the abbreviatedlist of the MSrather than the tathagata, Kashgar lates as taZ5~' jj i~ (76a.15, those who have full list of the KN edition (e.g., Kash 24b.2-3/Dh 66a.3). faith and take pleasure in the tathdgata who has re- 107 At Kash 38b.3 we find sugatana instead of KN's lokavi- nounced the world). Though not noticed by Karashima, dusya; Dharmaraksaprovides zhufo Hit (buddhas). The other there appears to be a confusion here between lokapitur missing occurrence is at Kash 55a.3-4, though the reading here is less clear. Dharmaraksaseems to reflect neither this reading nor that of the KN version. 109 108 Dh 90a.18; KN 166.10: lokavidu; Kash 160a.2: lokadhi- On conceiving of the Buddha as one's father, see also SP patir (lord of the world). ch. 13 (KN 286-87). BOUCHER:Gandhart and the Early Chinese Buddhist TranslationsReconsidered 497 and *loka-vidhu(ta) (one who has abandonedthe world), generated by the great Peng bird as it flies toward the exhibiting yet anotherconfusion between p/v and voiced/ southernocean. The Zhuangzienjoyed a great resurgence unvoiced dentals with assumption of aspiration (or, of interest among the literati during the third century, as more likely, spirancy). This kind of inconsistency, the evidenced most notably by the commentaryof Guo Xiang fact that identical terms received such varied interpre- (d. ca. 312?), a contemporaryof Dharmaraksa.10 tations, reveals, I suspect, something of the dynamics Twenty years before his translation of the SP, Dhar- of the translation process itself. It would be difficult, it maraksa had already used this metaphor to render the seems to me, to suppose that a single bilingual agent Sanskrit term maya (illusion) in his translation of the was fundamentally responsible for the final version of Suvikrantacintidevaputrapariprccha.Since the Sanskrit this translation. In essence we would have to believe text is not extant, I cite the Tibetan translationhere (Stog that such a translator'sskills in Indic languages would Palace MS, vol. 69 [tsha], 368b.4): nyon-mongs sprin have fluctuated so wildly as to inspire absolute nonsense 'dra sgyu-ma rmi-lam Ita bur Ita ([the bodhisattvas] see in some places, despite well-translated passages else- defilements [klesa] as like a cloud, an illusion [maya], where. But before drawing final conclusions, let us look like a dream). Dharmaraksarenders this as follows: Pfi:f at one more type of evidence of a ratherdifferent nature. ;F7 ,%$D[il ~/~g--?: (What is produced is like a dream or an illusion of shimmering air. [The Chinese Literary Borrowings bodhisattvas] view all as foolishly distinguished like this [by deluded people?]).l" In speculating upon the roles of the participantsin the It may be worth noting that the Suvikrantacinti is translationprocess I have thus far exclusively examined Dharmaraksa'sfirst recorded translation (267 C.E.) and the translators'attempts to come to terms with the Indic that two of his assistants from the SP translation-the or presumed Indic text. But we also find some evidence upasaka Nie Chengyuan and the Kuchean layman Bo that one or more of the members of the translationcom- Yuanxin-also participated in this work twenty years mittee supplied specific terms or passages that can only earlier."2 The colophon to the Suvikrdntacinti spe- be understood as borrowings from Chinese literary and cifically states that Dharmaraksa'soral recitation was religious traditions. One such passage occurs very early "linguistically transferred" (chuanyan fi` ) by two as- in the first chapter of Dharmaraksa'sSP and stands out sistants, suggesting a limited level of ability in Chinese as having no parallel in any extant Sanskrit manuscript, on the part of Dharmaraksa at this early date. Assuming nor is it found in Kumarajiva'sfifth-century translation. that the Kuchean and Parthian who "transferred the I render it as follows: words" would have been unlikely to have been versed in the Chinese classics, we can reasonably take such lit- [The bodhisattvas] roam the trailoka like the through erary allusions as yema to be the doing of the Chinese rays of the sun. They understandall dharmasas illusory, updsaka Nie Chengyuan.13 as conjured, like shimmering air (yema j,i) or reflec- tions-all without real existence, in abiding non-abiding. 110 Although [these bodhisattvas] experience samsara, they My renderingof this locution follows Guo Xiang's neither go nor come. When they see color and form, commentary, which glosses the metaphor as youqi J, (Sibu these appear as original non-being (benwu S:,). Al- beiyao edition, 2a). For a discussion of the impact of this though they emerge in multiple rebirths,they are perpet- renewed interest in the Zhuangzi on third-century Chinese ually without arising or cessation. They lead the masses Buddhism, see Tsukamotoand Hurvitz 1985, 173-77. 111 /without being attached to the three abodes. They elu- Taisho 588, vol. 15, 97c.16-17. See also 107a.16 for cidate the wisdom of emptiness but are without concep- another instance of this term. tualizations or inclinations. They give rise to the three 112 See the colophon in CSZJJ, vol. 55, 48b and the discus- gates of liberation and reach the three penetrating in- sion of this colophon in section III of this paper. 113 sights, but are without the thought of past, future, or Early records by native bibliographers indicate that Nie present. They convert the masses and cause them to Chengyuan's work on the translation committees extended be- understandoriginal non-being. (Dh 63a.20-26) yond the usual scribal duties. From Dharmaraksa'sbiography we learn: "First Dharmaraksaobtained the foreign text (huben One of the first lexical items to note in this passage $) of the Chaoriming [sanmei] jing ELHB[3H ], from is what I rendered as "shimmering air," yema f,% in the Western Regions and translatedit. But [his translation]was Chinese, which literally means "wild horses." This is prolix and full of repetitions. Then the upasaka Nie Chengyuan without a doubt a metaphor drawn from the first chap- altered it, thoroughly correcting the prose and verse sections, ter of the Daoist classic Zhuangzi, describing the wind reducing it to two rolls. The suitrawhich is transmittedtoday is 498 Journal of the American Oriental Society 118.4 (1998)

Another locution that stands out in this passage from In order to reevaluate this hypothesis, I examined the SP is benwu ;,, "original non-being." This term evidence from the earliest translation of the Sad- was a centerpiece of the negative ontology of the dharmapundarikasutra,that of the third-centuryYuezhi Xuanxue (Mysterious Learning) movement, embodied translator Dharmaraksa, which indicated phonological most notably in the third-centuryfigure Wang Bi.114 For confusions due to a transmission of this text in a lan- Wang Bi, drawing tangentially on the language of ab- guage having similarities to GandhariPrakrit. sence in the Laozi, benwu functions as the ontological This evidence, culled for the most part from the study ground of all reality. But even before Wang Bi, trans- by Seishi Karashima,reveals possible confusions related lators such as Lokaksema (ca. 168-88) and Zhi Qian to vowel length, aspirated and unaspirated consonants, (ca. 220-52) had used this term to express the principles voicing of intervocalic stops, and confusions related to of sunyata thought. We cannot engage here nasalization. I also noted that despite the similarity be- in the debate of what influence-if any-Mahayana tween the possible source language influencing these emptiness doctrine exercised on early Xuanxue thought. phonological confusions and Gandhari Prakrit, several Regardless, it is clear that terms like benwu and yema of them are unlikely to have been representedin the In- were already current among literati by the time Dhar- dic text underlying Dharmaraksa'stranslation. Instances, maraksabegan his translationcareer.115 And these terms for example, where Dharmaraksa'stranslation assumes certainly would have been consciously chosen to ex- aspirationwhere none could have existed run counter not press facets of Mahayanametaphysics so as to appeal to only to what we know of Gandhari, which has a predi- third-centuryChinese intellectuals. lection for dropping aspiration, but also of the Prakrits For our purposes in this paper, such indigenous ex- generally. pressions within clearly interpolated passages point to We have also looked closely at the translationprocess a decidedly Chinese source. And once again, the most itself as attested by the two extant colophons. These likely candidate for this influence is Nie Chengyuan, colophons present detailed but not entirely unambiguous providing yet another datum for what appears to have information of the specific roles played by each member been a significant if not overwhelming native Chinese of this internationaltranslation committee. What stands influence on the translationprocess. out from this information is the fundamentallyoral/aural nature of the translation process. Dharmaraksarecited V. CONCLUSION the Indic text aloud to his Chinese assistant Nie Cheng- yuan who, in conjunction with two other Chinese scribes, I began this investigation by surveying the rather converted Dharmaraksa'soral glosses into literary Chi- meager evidence marshalled to argue that the majority nese. I have supposed that Nie Chengyuan, who almost of early Chinese Buddhist texts were translated from certainly would not have fully commanded any Indian Gandhari originals. And we have also seen that the literarylanguage, would have been dependentupon Dhar- repetition of this claim has led to its widespread accep- maraksa for a reading of the Indic manuscript. Never- tance among scholars, generally without significantly theless, his rendering into literary Chinese would have augmented data. been simultaneously informed by his apprehensions and misapprehensionsof specific Indic locutions-locutions that Dharmaraksamay well have expected his pupil of thatone" (CSZJJ, vol. 55, 98a.23-26);see also the separateno- twenty years to have understood. Despite the fact that tice to the translationof the Chaoriming sanmeijing appended this translationwas subsequently proofreadby an Indian to the list of Dharmaraksa'scorpus, CSZJJ,vol. 55, 9c.5-8. Nie monk and Kuchean layman and was later revised by Chengyuan's role in redacting the Chao riming sanmei jing is Dharmaraksahimself, many of the mistakes remained especially noteworthy in light of the fact that this scripture unnoticed. This confirms our sense that Dharmaraksa, contains a passage very much like the one cited above from like the other foreigners on the committee, was not the Suvikrantacinti: --Jz)$a

raksa's translation. Such an unusual translation practice phonetic alternations, there are apparentconfusions that highlights the lexical uncertainty that the translation may lend themselves to a graphic explanation: assistant faced in determining an appropriaterendering for an Indic term that was neither clear nor made clear to KN 18.3-6: iti hy ajitaitena paramparoddhdrena can- him. If Dharmaraksa'sability in Indic languages must to drasuryapradipandmakdndimtathdgatdndm arhatdm ekandma- a large degree be assumed in order to account for the samyaksambuddhandm dheydndm ekakulagotranadm yad idam large number of correct the transla- readings throughout bharadvdjasagotrdnidm vims'atitathagata then such limitations to Nie tion, again point Cheng- sahasrany abhivan I tatrdjita The confusions exhibited in to yuan. relationship yina/ therewere are of the Thus,Ajita, successively20,000 jidna/dhydna particularlyrevealing overlap- tathagataswho hadthe samename and the of that Nie almost cer- ping pronunciation Chengyuan same family-namely, Bharadvaja-as the tainly must have faced. If Dharmaraksa'spronunciation Tathagata,Arhat, Samyaksambuddha Can- of these words was affected, for example, by their po- drasuryapradipa.Then, Ajita.... sition (verse or prose) or graphic irregularities in the then we should not be to find SP 65c.29-66a.1: R]--Ht tnJ' f ' manuscript, surprised - problems of interpretationby a Chinese assistant with ,M tft "MZkL=XZ* %1riffAR only limited ability in discerning the correct form from Thus there were 80 tathagatas all having the among various Prakritic possibilities. While some of same name Candrasuryapradipa,117all in- these confusions also suggest a pronunciationinfluenced heritingthe samefamily name. If we were to line them there would by CentralAsian habits, none of them requires an under- up, be 20,000 The Buddhasaid to .... lying text written in Gandhfri Prakrit. tathdgatas. Ajita This also brings us to the problem of and orthography We have two instances of the name Ajita in the Sanskrit its relationship to pronunciation. It be might supposed passage but only one in the Chinese. Moreover, we have that, regardless of the actual of underlying language a very strange state of affairs in the Chinese: Dharma- Dharmaraksa'sIndic manuscript,some of these mistakes raksa describes the number of tathagataswho have suc- could be accounted for by supposing a text written in cessively appeared as Candrasuryapradipato be eighty, kharosthi script.116 And indeed, besides the above-noted and then immediately following, to be twenty thousand. There is, of course, no mention of "eighty" in our Indic 116 text. But if we suppose Dharmaraksato have been work- Karashimahimself has hypothesized as much: ".. judg- ing from a kharosthi manuscript that read *ayita, ex- ing fromthe confusionrelating to anusvaraor the vowel length, hibiting the widespread Prakritic development j > y we mayassume that the originaltext was probablywritten in a (Pischel 1955, ? 236), then we could speculate that he kind of the Kharosthiscript, in which these differentiations misread the kharosthi y as s -two of the most werenot denoted"(Karashima 1992, 275). Of coursewe have graphi- cally similar aksaras in this script-and understoodasiti no reasonto doubtthe presence of thekharosthi script in China. ("eighty"). What is curious in this case is that this name John Broughhas publisheda kharosthiinscription found at was read correctly, both in transcription (cf. note 101) Luoyang(Brough 1961) and Lin Meicun has similarlyreported and in translation, several times in nearby passages. But on a kharosthiinscription found on the baseof a Buddhaimage discoverednear ancient Chang'an (Lin 1991). On this pointI mightalso tentativelysuggest that the desig- nationof Dharmaraksa'sIndian manuscript as huben 4S by states that Dharmaraksa"recited and issued [shuochu~Id3] the anonymousauthor of the colophonmight also supportthe the brahmitext [fanwen~,;] in Chang'an,conferring it upon proposalof a kharosthiscript text as contrastedwithfanshu X [Nie] Chengyuan"(CSZJJ, vol. 55, 57c.19-21). A close lin- i (brahmi MS) in other colophons. As possible supporting guisticanalysis of this translation,which was carriedout with evidence, I would note that the colophonto Dharmaraksa's the sameprincipal assistant and at nearlythe same time, may translationof the Lalitavistaraalso describeshis Indicmanu- inform us as to the respective difficulties of handling texts script as hubenand Broughhas arguedthat this translation transmitted in kharosthi vis-a-vis brahmi script for Dharmar- stems from a GandhariPrakrit original, as evidencedby its aksaand his committees. arapacanaformulary (Brough 1977). One way to test this 117The fact that Dharmaraksarenders this name as riyue hypothesismay be to examineDharmaraksa's translation of the dengmingBI J:f (*Suryacandra-pradipa)merely reflects a ViSesacintibrahmapariprcchasatra(Taisho 585). This text was Chinesepredilection for the order"sun and moon" rather than translatedless thansix monthsbefore the SP and its colophon the reverse. 500 Journal of the American Oriental Society 118.4 (1998) here Dharmaraksanot only misread the text, but pro- Of course, we cannot presume that all of these mistakes duced a translation that is transparently incoherent. are the result of Nie Chengyuan's misunderstandings.It Thus, when the colophon states that this translationwas is likely that Dharmaraksahimself would have some- proofreadby a Kuchean layman and an Indian sramana, times misread his Indian manuscript, which could have such mistakes remind us to take such information cum itself been fraught with scribal errors of indeterminable grano salis. types. Furthermore, Dharmaraksa may have provided Moreover, there is no reason to assume that Buddhist glosses to his assistants that would have been mislead- Hybrid Sanskrit manuscripts were not transmitted in ing. Nevertheless, the predominance of correctly trans- kharosthi script. Among the kharosthi documents dis- lated items in much of the text (e.g., lokavid when it covered at Niya are two that are written in Sanskrit: occurs within the standardlist of epithets) side-by-side document no. 511 is composed in a mixed Buddhist San- with occasional mistakes, even when context demanded skrit with a numberof Prakritismsand document no. 523 a narrower reading, suggests a rendering by someone is in pure classical Sanskrit, replete with long vowels, whose understandingof the Indic text was imperfectly visarga, virama, and proper sandhi."8 These documents mediated. Unless the Indic text contained unusually ir- were certainly composed by someone conversant with regular variants of the same words, or Dharmaraksa's the brahmi script as indicated by the fact that the verses understandingand recitation of the text fluctuated in er- are numberedin both documents with brahmi numerals. ratic ways, the most probable explanation, it seems, rests In all probability, the modifications to the kharosthi with the middlemen: the Chinese assistants who were script that made correct Sanskrit possible would have responsible for receiving the Indian text with a severely occurred under a brahmi influence."9 limited arsenal of linguistic tools and who transformed In addition to mistakes based upon phonological con- their understanding of it into a semi-literary Chinese fusions, we have also found evidence for grammatical translation.121 misunderstandings,mistaken division of words, and con- Lastly, we have taken notice of an interpolation in notative misrenderings-all of which again point to a Dharmaraksa'stranslation that perhaps more than any translatorwith limited skill in Indic languages. We dis- other piece of data points to the strong likelihood of a covered that context seems to have played a significant native Chinese source. In this passage we observed two role in Dharmaraksa'sor his translationassistant's arriv- locutions that were doubtless derived from the contem- ing at an accurate rendering of certain lexical items.'20 porary Chinese literary vocabulary,reflecting an attempt by the Chinese members of the translation committee to narrow the gulf between the Indian and Chinese reli- 118 See Boyeret al. 1927,185-87 and 191, respectively.On gious worlds. document no. 511, see the recent though problematic article by In short, what this rather sizable mass of data would Hasuike 1997. seem to indicate is that the evidence for the underlying 119 Such influence between scripts can work in the other direction as well. Douglas Hitch (1984) has examined the in- ing from such parts-with, as I have indicated, only sporadic fluence of kharosthi on brahmi script in Iranianenvironments. success. Furthermore,it has been suggested that brahmi script could be 121 The fact that the Chinese translations are nearly always used to write Gandhari Prakrit, though the evidence is by no attributedto one usually foreign translator,in our case Dharma- means conclusive: see Harmatta1967. raksa, and not his committee has more to do with concerns for 120 I should point out that while the investigation here has legitimation and orthodoxy in China than with historical accu- focused upon problems related to the reception of words on the racy. Antonino Forte has astutely observed: "The assignment of part of the translation team, we could also mention the high the responsibility for a translation was an extremely important incidence of syntactical problems that range from loose ren- matter as its purpose was to reassure the Buddhist establish- derings of the Sanskrit to nearly incomprehensible strings of ment and the government of the full authenticity and orthodoxy Chinese characters. Many of these loose renderings have the of a work. This need to make one person responsible often feel of paraphrases.Others appear to be word-by-word transla- meant that the actual contributionof other members of the team tions, often with considerable violation of Chinese syntax. Both tended to be unacknowledged. The paradox thus often arose of unable to types of renderings-the paraphrase and the literal-suggest the accredited translator, usually a foreigner, being that Dharmaraksatranslated the Indic text in a piecemeal fash- speak or write Chinese, while the actual translatorsreceived so of a num- ion, glossing words and phrases while providing some additional little attention that, but for the colophons at the end known their exegesis of the overall import. It would have fallen to Nie ber of translations, we would often not have even Chengyuan then to construct a coherent literary Chinese read- names" (Forte 1984, 316). BOUCHER:Gandhdri and the Early Chinese Buddhist Translations Reconsidered 501

Indic text of this translation is in fact evidence for the with them-that the transposition was no formalized Chinese reception of the Indic text. And this reception, translation. It was another kind of transformationfrom as we have seen, suffered at times from rather severe one dialect into another dialect, that took place in the limitations in expertise.122 Thus the attempt to see course of a tradition, which was still an oral tradition, GandhariPrakrit specifically beneath our extant Chinese but had already entered the process of being formalized translation must take into account the complex inter- linguistically ... 124 action between an orthographicallyindeterminable Indic text, its recitation by a Yuezhi monk trained by an However, these positions are not necessarily as sharply Indian master at Dunhuang, and its transmission to a opposed as they might first appear. Norman has shown linguistically underpreparedChinese updsaka. that these "translations" were often carried out by In addition, the linguistic complexity of the under- scribes who applied certain phonetic rules mechani- lying Indic text cannot be underestimated. Even if we cally.125 Nevertheless, some of these transpositions led want to suppose the existence of a considerable number to hypercorrections and mistaken interpretations, sug- of Buddhist texts written in the Gandhari language, gesting that the movement between these dialects was most canonical texts used in the northwest would have not always clear even to learned scribes.126 This problem originated from central Indian Prakrits.And the process was especially acute in Gandhari, as G6rard Fussman of turning such Prakrits into Gandhari would have de- has recently indicated: cidedly shaped and perhaps significantly altered the final text. K. R. Norman, for example, has argued: "It can- Il ne faut pas surestimer la gene qu'apporte a l'usager not be emphasized too much that all the versions of ca- l'existence d'une orthographevieillie assez eloignee de nonical Buddhist texts which we possess are la prononciationr6elle.... Dans ces conditions les textes translations, and even the earliest we possess are trans- bouddhiques gandh. s'ecartaient tellement de la norme lations of some still earlier version, now lost."'23Heinz parlee qu'ils n'6taient parfois plus comprehensibles, Bechert, on the other hand, has suggested that transla- meme a leur r6dacteur.127 tion-a linguistic transfer between mutually unintelligi- ble languages or dialects-is too strong a characterization For our purposes then it is important to realize that of this process: before an Indian sutra arrived in China, it may have undergone one or more stages of transference between Some scholarsbelieved that this transformationwas a Middle Indic languages. This process almost certainly real "translation"of texts whichat thattime already ex- would have resulted in a very mixed and layered text.128 isted as writtenliterary texts. Others think-and I agree

124 Bechert 1980, 12. 122 125 There is anotherwell-known case, from a much later Norman 1993, esp. pp. 95 ff. on translation techniques. period,of a translationgoing awrythrough the inadequacyin Cf. Brough's remarks on the Gandhari Dharmapada: "We Sanskrit lexical and syntactical knowledge of the Chinese can, however, see immediately that the translation involved- "translators." This is the pseudo-translationof the Jatakamdla whether it was done in one or more deliberate stages, or simply from the Northern Song dynasty (960-1127), discussed most happened through imperceptible gradations in different lines of cogently in Brough 1964. Brough demonstratedthat while this descent-is scarcely more than a mechanical transposition be- supposed translation certainly did no justice to AryagSra's tween the sound systems of the dialects" (Brough 1962, 113). poetic masterpiece, it was related to it in a particularway. The 126 For an amusing story from the Malasarvistivddavinaya Chinese who worked on this text clearly did not understandthe showing the effects of an uncomprehended "translation,"see Sanskrit, but they did recognize-sometimes erroneously- Brough 1962, 45-48. 127 certain Sanskrit words in an order parallel enough to the San- Fussman 1989, 485. skrit text to 128 rule out chance association. For a recent attempt to On this issue I have been reminded by Professor Jens- explain the reason behind this stark decline in translationcom- Uwe Hartmann (personal communication, June 1995) that it see petence, Bowring 1992. Of course Dharmaraksa'sSP trans- may be necessary to separate the redactional history of dgama lation is not nearly so incompetent as this pathetic attempt;but texts from that of . It is highly likely that al- it is to illuminating observe that Chinese understandings of most all dgama texts were composed in Middle Indic languages Sanskrit texts were perhaps always focused at the lexical level, that were liable to phonological shifts as the bearers of these those excluding few pilgrims who had studied in India. texts migrated across India. It is certainly possible that some of 123 Norman 34. 1990, these texts, such as the Dirghdgama, could have undergone a 502 Journal of the American Oriental Society 118.4 (1998)

Moreover, it is precisely this predicament, Fussman tinued uncertainty as to the location(s) of the early Ma- suggests, that led Buddhists in the northwest to adopt hayana, these philological discussions are likely to have the use of Sanskrit as their linguistic norm: ramifications beyond any particular text. It must be emphasized at this point that I have not Surtout il n'existait a ma connaissance aucun texte proven-nor have I attempted to prove-that Dhar- gandh. dont le prestige fit tel qu'il pfit servir de norme: maraksa's underlying Indic manuscript was not written on sait bien que le bouddhisme n'est pas originaire de under the influence of Gandhari Prakrit. If, despite some et les grands sutra bouddhiques, s'ils exis- qualifications, there is sufficient evidence that points to taient en gandh., n'y existaient qu'en traductionfaite ou this manuscript as having been written in kharosthi refondue sur un original en m-i gang6tique. La seule script, we would expect a fair number of Gandhari fea- norme possible 6tait le skt., dont le prestige est bien tures to be represented even in a Buddhist Hybrid San- attest6 aux environs de n.e....129 skrit text.132 But what I have attempted to show is that these early Chinese translations are imperfect testimo- We would expect then that the Indic text of the SP was nies to the Indian source texts. There is much that is not shaped by the burgeoning role of Sanskrit in north India well understood about these early translations and much beginning from approximately the first century B.C.E.130 of this will have to be solved within sinology. As should Edgerton has in fact already shown that the idiom he be abundantly clear by now, the Chinese-ness of these called Buddhist Hybrid Sanskrit "was not a pure Prakrit but a hybrid dialect, based on a Prakrit, but partially Sanskritized from the start."'31 Though the Indic text 132 underlying Dharmaraksa's translation would have cer- On this matter it would be appropriatethen to ask what tainly contained many more Prakritic forms that were we mean by a "Gandharitext." Richard Salomon has suggested increasingly disguised with an orthographically San- (personal communication,July 1995) that what we should mean skritic veneer, there can be little doubt that the original is a text written in kharosthi script. The real Gandhari-ness of composition was already in the hybrid language. And such a text would consist largely in its graphically derived this language, as Edgerton repeatedly emphasized, was phonemic features through which an original Middle Indic text an artificial language, in no way identical to any living was channelled. There is considerable merit in this proposal, vernacular or otherwise literary Prakrit. Given the de- especially in that it allows us to focus our investigation some- bate that has surrounded the linguistic status of BHS what. However, to the extent possible, I would tentatively like since Edgerton's monumental study, as well as the con- to differentiate the language of composition from the language of transposition. That is to say, a Central Indian Prakrittext- especially one still transmittedorally-would in all probability series of phonetic transpositions that coincided in many ways have been more linguistically fluid, more prone to "translation" with features of the Northwest Prakrit.On the other hand, Ma- across Prakrits, than a text originally composed in Buddhist hayana sutras were in all probability composed in a language Hybrid Sanskrit. Both are hybrid works and both may have already undergoing Sanskritization and may have been more taken on Gandhariphonemic features by a transmissionthrough likely to have been circulated in written form at a date close to northwest India (and transcription in kharosthi). But the lin- their composition, causing a somewhat greater fixity in their guistic nature of a text composed in BHS, insured in many linguistic shape. This supposition, however, is difficult to prove cases by an analysis of the verse sections, may well representa on the basis of our extant sources. differentgenre, a different set of literary,political, and religious 129 Fussman 1989, 486. forces that distinguishes it from the use of Prakrit. There is 130 Th. Damsteegt, in his monographic study of epigraphical much about the beginnings of Sanskritization that we still do hybrid Sanskrit, attempts to show that the Sanskritization evi- not understand.It would be premature,it seems to me, to prej- denced in Buddhist inscriptions began in Mathura with the udice our investigation strictly at the level of script. Again, coming of the Saka satrapsand radiatedout from there, while at document no. 511 from Niya may be a good example of a BHS the same time absorbing linguistic features from the northwest. text filtered through a Gandhari-using environment. And the He hypothesizes that BHS would have developed in such an language of this document is clearly distinguishable from the environment, from which it too would have spread to Buddhist others found at this site though nearly all are in kharosthi communities elsewhere (Damsteegt 1979, 238-66). Despite the script. To label this document as aberrant,to dismiss the possi- mass of data presented, several aspects of his thesis are in seri- bility of more texts of this kind, would be to beg the question ous doubt: see the review by Fussman 1980. our investigation of the Chinese translationsposes: "Werethere 131 Edgerton 1936; see also his BHSG, ?1.33ff. more?" BOUCHER:Gandhdri and the Early Chinese Buddhist Translations Reconsidered 503 texts intrudes throughout and must be taken seriously in Despite all the uncertainties, I hope to have shown that any assessment of the source language. these early Chinese translations hold tremendous poten- The gist of this long digression is that any proposal tial for advancing our knowledge about the language of that a Chinese Buddhist translation derives from Gan- the Buddhist texts transmitted from India in the first half dhari must also take into account the complex history of of the first millennium. Above all else it should be evi- Indian Buddhist texts, generally, and the process of their dent that we need fewer generic statements that merely translation into Chinese, specifically. Given the impor- repeat the scholarly assumptions of our predecessors and tance of such philological discussions for Buddhist tex- more focused studies-one text at a time-that unpack tual history, we obviously must proceed carefully.133 the philological clues contained in these mongrel docu- ments. Karashima's study is but the first serious attempt in this regard. Obviously we are in need of many more. 133 The recent acquisition of several new kharosthi script/ GandhariPrakrit Buddhist texts by the British Library will, we hope, add importantnew data to our very partial understanding to date belong to a mainstream-probably Dharmaguptaka- of Gandharan textual history. See the preliminary report on order. Thus there remains no Indian textual evidence to link the these manuscripts by Salomon (1997). It is interesting to note Gandhiri language and the early Mahayana, though, of course, that all of the texts identified among these new manuscripts we cannot exclude the possibility of still more discoveries.

BIBLIOGRAPHY

Atwood, Christopher. 1991. "Life in Third-fourth Century .1965. "Comments on Third-CenturyShan-Shan and Cadh'ota: A Survey of Information Gathered from the the ." BSOAS 28: 582-612. PrakritDocuments Found North of Minfeng (Niya)." Cen- .1970. "Nugae Indo-Sericae." In W. B. Henning Me- tral Asiatic Journal 35: 161-99. morial Volume, ed. M. Boyce and I. Gershevitch. Pp. 81- Bailey, Harold W. 1946. "Gandhari."Bulletin of the School of 88. London: Lund Humphries. Oriental and African Studies 11: 764-97. 1977. "The Arapacana Syllabary in the Old Lali- Bechert, Heinz, ed. 1980. Die Sprache der altesten buddhis- tavistara."BSOAS 40: 85-95. tischen Uberlieferung / The Language of the Earliest Bud- .1982. "Amitabhaand Avalokitesvara in an Inscribed dhist Tradition.Gottingen: Vandenhoeck & Ruprecht. GandharanSculpture." Indologica Taurinensia 10: 65-70. .1991. "Methodological Considerations Concerning Burrow, Thomas. 1935. "TokharianElements in the Kharosthi the Language of the Earliest Buddhist Tradition."Buddhist Documents from Chinese Turkestan."Journal of the Royal Studies Review 8.1-2: 3-19. Asiatic Society: 667-75. Bernhard, Franz. 1970. "Glandhariand the Buddhist Mission 1937. The Language of the Kharosthi Documents in Central Asia." In Anjali: Felicitation Volume Presented from Chinese Turkestan.Cambridge: Cambridge University to Oliver Hector de Alewis Wijesekera on his Sixtieth Press. Birthday, ed. J. Tilakasiri. Pp. 55-69. Peradeniya. Caillat, Colette, ed. 1989. Dialectes dans les litteratures indo- Bowring, Richard. 1992. "Buddhist Translations in the North- aryennes. Paris: College de France. ern Sung" Asia Major, 3rd ser., 5.2: 79-93. Chan, Alan K. L. 1991. Two Visions of the Way: A Study of Boyer, A. M., E. J. Rapson, and E. Senart, eds. 1920-29. the WangPi and the Ho-shang Kung Commentaries on the Kharosthi Inscriptions Discovered by Sir Aurel Stein in Lao-Tzu. Albany: SUNY. Chinese Turkestan,pts. I-III. Oxford: Clarendon Press. Chandra,Lokesh. 1976. Saddharma-Pundarika-Sutra:Kashgar Brough, John. 1954. "The Language of the Buddhist Sanskrit Manuscript. New Delhi: The International Academy of Texts." BSOAS 16: 351-75. Indian Culture; reprint, Tokyo: The Reiyukai, 1977. 1961. "A KharosthiInscription from China."BSOAS Char, S. V. R. 1991. "Methods and Principles Used in Trans- 24: 517-30. lating the Buddhist Tripitaka into Chinese." Chinese Cul- 1962. The Gandhari Dharmapada. London: Oxford ture 32.3: 37-47. University Press. Ch'en, Kenneth. 1960. "Some Problems in the Translation of .1964. "The Chinese Pseudo-Translation of Arya- the Chinese Buddhist Canon." Qinghua xuebao ~SH Sura's Jatakamald."Asia Major, n.s., 11: 27-53. [Tsing Hua Journal of Chinese Studies], n.s., 2.1: 178-86. 504 Journal of the American Oriental Society 118.4 (1998)

Coblin, W. South. 1983. A Handbook of Eastern Han Sound Ruins of Niya in Xinjiang]. Indogaku bukkyogakukenkya Glosses. Hong Kong: The Chinese University Press. [Journal of Indian and ] 44.2: (164)-(166). .1993. "BTD Revisited: A Reconsideration of Han von Hinuber, Oskar. 1982. "Upali's Verses in the Majjhimani- Buddhist Transcriptional Dialect." Zhongyang yanjiuyuan kaya and the Madhyamagama."In Indological and Bud- r Lishi yuyan yanjiusuo jikan ~ E: t f ! P fff5fi dhist Studies: Volumein Honour of Professor J. W. de Jong FIJ [Bulletin of the Institute of History and Philology, on his Sixtieth Birthday, ed. L. A. Hercus et al. Pp. 243-51. Academia Sinica] 63.4: 867-943. Canberra:Faculty of Asian Studies. Collins, Steven. 1992. "Problems with Pacceka-Buddhas." .1983. "Sanskrit und Gandhari in Zentralasien." In Religion 22: 271-78. Rohrborn and Veenker 1983, 27-34. Damsteegt, Th. 1979. Epigraphical Hybrid Sanskrit: Its Rise, . 1986. Das dltere Mittelindisch im Uberblick. Wien: Spread, Characteristicsand Relationship to BuddhistHybrid Osterreichischen Akademie der Wissenschaften. Sanskrit. Leiden: E. J. Brill. .1989. "Origin and Varieties of Buddhist Sanskrit." Eberhard, Wolfram. 1955. "The Origin of the Commoners in In Caillat 1989, 341-67. Ancient Tun-huang."Sinologica 4.3: 141-55. Hitch, Douglas. 1984. "Kharosthi Influences on the Saka .1956. "The Leading Families of Ancient Tun-huang." Brihmi Scripts." In Middle Iranian Studies: Proceedings Sinologica 4.4: 209-31. of the International Symposium organized by the Katho- Edgerton, Franklin. 1935. "The Meter of the Saddharma- lieke Universiteit Leuvenfrom the 17th to the 20th of May pundarika." In M.M. K. S. Shastri Commemoration Vol- 1982, ed. W. Skalmowski and A. Van Tongerloo. Pp. 187- ume. Pp. 39-45. Madras. 202. Louvain: Uitgeverij Peeters. .1936. "The Prakrit Underlying Buddhistic Hybrid 1988. "Kushan Tarim Domination." Central Asiatic Sanskrit." Bulletin of the School of Oriental Studies 8: Journal 32: 170-92. 501-16. Hrdlickova, V. 1958. "The First Translations of Buddhist .1946. "Meter, Phonology, and Orthographyin Bud- Sutras in Chinese Literature and Their Place in the Devel- dhist Hybrid Sanskrit."JAOS 66: 197-206. opment of Storytelling." Archiv Orientdlni 26: 114-44. .1953. Buddhist Hybrid Sanskrit Grammar and Dic- Hultzsch, E. 1925. The Inscriptions of Asoka. Corpus Inscrip- tionary, 2 vols. New Haven: Yale Univ. Press; reprint, tionum Indicarum, I. Oxford: Clarendon Press. Kyoto: Rinsen Book Co., 1985. IwamotoYutaka. W** 1962-67. Hokkekyo,i. Tokyo: Forte, Antonino. 1984. "The Activities in China of the Tantric Iwanami Bunko; reprint, 1992-93. Master Manicintana (Pao-ssu-wei ,E~'t: ?-721 A.D.) de Jong, J. W. 1981. "Fa-hsien and Buddhist Texts in Ceylon." from Kashmir and of his Northern Indian Collaborators." Journal of the Pali Text Society 9: 105-16. East and West, n.s., 34: 301-45. Karashima Seishi. -[F1,ij 1992. The Textual Study of the .1995. The Hostage An Shigao and his Offspring: Chinese Versions of the Saddharmapundarikasutrain the An Iranian Family in China. Italian School of East Asian Light of the Sanskrit and Tibetan Versions. Tokyo: The Studies, Occasional Papers 6. Kyoto. Sankibo Press. Fuchs, Walter. 1930. "Zur technischen Organisation der Uber- . 1993. "Hokkekydni okeruj6 (yana) to chie (jniina)- setzungen buddhistischer Schriften ins Chinesische." Asia daij6 bukkyo ni okeru yana no gainen no kigen ni tsuite" i Major 6: 84-103. |;l_:t3t'51 (yana) L (jnana)- f3f4Lr ~t Fussman, G6rard. 1980. "Review of Th. Damsteegt, Epi- 7 yana -0DS S -C_'li [Vehicle(yana) and Wisdom graphical Hybrid Sanskrit: Its Rise, Spread, Characteris- (jnana)in the :On the origin of the concept of Yana tics and Relationship to Buddhist Hybrid Sanskrit (Leiden: in MahayanaBuddhism]. In Hokkekyo no juyo to tenkai i E. J. Brill, 1978)." Journal asiatique 268: 420-26. iSOfg VCiR [The Acceptance and Development of .1989. "Gandhari 6crite, Gandhari parlie." In Cail- the Lotus Sutra], ed. Taga Ryiigen. Pp. 137-97. Kyoto. lat 1989, 433-501. .1994. Choagonkyo no gengo no kenkyu-onshago -- Geiger, Wilhelm. 1994. A Pali Grammar.Translated from the bunseki o chushin to shite :filN ODIia tff E- German by B. Ghosh. Revised and edited by K. R. Nor- AM"3Y:;F,t-: LrC [A Study of the Original Language man. Oxford: The Pali Text Society. of the Chang ahan jing: Focusing on an Analysis of the van Gulik, R. H. 1956. Siddham: An Essay on the History of Transcriptions].Tokyo: Hirakawa Shuppansha. Sanskrit Studies in China and Japan. Nagpur. Kern, Hendrik. 1884. Saddharma-pundarika or the Lotus of Harmatta, Janos. 1967. "The Oldest Brahmi Inscription in the TrueLaw. Oxford: Clarendon Press; reprint, New York: Innermost Asia." Acta Orientalia (Budapest) 20: 1-32. Dover Publications, 1963. Hasuike Toshitake. 1997. "Shinky6 Niya iseki shutsudo no Kern, Hendrik, and Bunyiu Nanjio, eds. 1908-12. Saddharma- bukky6 bunken ni tsuite (1)" l 'fi:k=t5&tf?l ) -C pundarika. Bibliotheca Buddhica, X. St. Petersburg:Impe- )lV'C(1) [On Buddhist Literature Excavated from the rial Academy; reprint, Osnabriick: Biblio Verlag, 1970. BOUCHER:Gandhdri and the Early Chinese Buddhist Translations Reconsidered 505

Kloppenborg, Ria. 1974. Paccekabuddha. Leiden: E. J. Brill. Pulleyblank, Edwin G. 1983. "Stages in the Transcription of Konow, Sten. 1929. Kharosthi Inscriptions with the Exception Indian Words in Chinese from Han to Tang." In R6hrborn of Those of Asoka. Corpus InscriptionumIndicarum, II, pt. 1. and Veenker 1983, 73-102. Oxford: Clarendon Press; reprint, Varanasi, 1969. Robinson, Richard H. 1967. Early in India and .1936. "Note on the Ancient North-Western Pra- China. Madison: Univ. of Wisconsin Press. krit." BSOS 8: 603-12. Rohrborn, K., and W. Veenker, eds. 1983. Sprachen des Bud- Levi, Sylvain. 1915. "Sur la recitation primitive des textes dhismus in Zentralasien. Wiesbaden: Otto Harrassowitz. bouddhiques."Journal asiatique, s6r. 11, vol. 5: 401-46. Rosenfield, John M. 1967. The Dynastic Arts of the Kushans. Lin Meicun. 1991. "A Kharosthi Inscription from Chang'an." Berkeley and Los Angeles: Univ. of California Press. In Papers in Honour of Prof Dr. Ji Xianlin on the Occa- Salomon, Richard. 1990. "New Evidence for a Gandhari Ori- sion of His 80th Birthday, eds. Li Zheng et al. Vol. II: pp. gin of the Arapacana Syllabary."JAOS 110: 255-73. 119-31. Nanchang: Jiangxi renmin chubanshe. .1997. "A PreliminarySurvey of Some Early Buddhist 1996. "KharosthiBibliography: The Collections from Manuscripts Recently Acquired by the British Library." China (1897-1993)." CentralAsiatic Journal 40.2: 188-220. JAOS 117: 353-58. Link, ArthurE. 1960. "Shih Seng-yu and His Writings."JAOS Shih, Robert. 1968. Biographies des moines eminents (Kao Seng 80: 17-43. Tchouan) de Houei-Kiao. Louvain: Institut orientaliste. .1961. "The Earliest Chinese Account of the Compi- von Simson, Georg. 1985. "Stil und Schulzugehorigkeit bud- lation of the Tripitaka."JAOS 81: 87-103; 281-99. dhistischer Sanskrittexte." In Zur Schulzugehorigkeit von Maenchen-Helfen, Otto. 1945. "The Yueh-chih Problem Re- Werkender Hinayana-Literatur,ed., Heinz Bechert. Vol. I: examined." JAOS 65: 71-81. pp. 76-93. Gottingen: Vandenhoeck & Ruprecht. de Mallmann, Marie-Th6rese. 1948. Introduction a l'e'tude Sims-Williams, Nicholas. 1983. "Indian Elements in Parthian d'Avalokitegvara.Paris: Mus6e Guimet. and Sogdian." In Rohrborn and Veenker 1983, 132-41. MatsumuraHisashi. 1990. "Miscellaneous Notes on the Upali- von Stael-Holstein, Baron A. 1936. "Avalokitaand Apalokita." pariprccha and Related Texts." Acta Orientalia (Copen- Harvard Journal of Asiatic Studies 1: 350-62. hagen) 51: 61-113. Tang Yongtong. tiJfY 1938. Han wei liangjin nanbeichao Mironov, N. D. 1927. "Buddhist Miscellanea, I: Avalokite- fojiaoshi ri ]iOg~!f b [A History of Buddhism gvara-Kuan-Yin." JRAS: 241-52. during the Han and Six Dynasties Period]. 2 vols. Changsha: Nattier, Jan. 1990. "Church Language and Vernacular Lan- ShangwuYinshu guan, reprint, Beijing: Zhonghua shuju, 1983. guage in Central Asian Buddhism."Numen 37.2: 195-219. Teiser, Stephen. 1994. The Scripture of the Ten Kings and the Nishimura Minori. N^tf 1IJl1987. "Gandaragobukky6ken to Making of Purgatory in Medieval Chinese Buddhism. kan'yaku butten j',,--i3r[I /L:," [Gan- Kuroda Institute Studies in . Hono- dhari and the Chinese Translations of Buddhist Texts]. lulu: Univ. of Hawaii Press. Sanko bunka kenkyijo nenpo _tl fifei--_ [An- Toda Hirofumi. PBffi; : 1980-85. "Saddharmapundarika- nual Bulletin of the Sanko Research Institute] 20: 49-125. sutra, Nepalese Manuscript (K')." Tokushima daigaku Norman, K. R. 1983. "The Pratyeka-Buddhain Buddhism and kyoyobu. Rinrigakka kiyo 0J - Wi4~Et Jainism." In Buddhist Studies, Ancient and Modern, ed. [Tokushima University General Education Section, Bulle- P. Denwood and A. Piatigorsky. Pp. 92-106. London: Cur- tin of the Ethics Dept.]: 8-11. zon Press. 1981. Saddharmapundarikasatra: Central Asian .1990. "Pali Philology and the Study of Buddhism." Manuscripts, Romanized Text. Tokushima:Kyoiku Shuppan The Buddhist Forum 1: 31-39. Center. .1993. "The Languages of Early Buddhism." In Pre- Tso Sze-bong. {?tt 1963. "Lun zhongguo fojiao yichang mier colloque Etienne Lamotte (Bruxelles et Liege, 24-27 zhi yijing fanshi yu chengxu P igt 4 ,,M t septembre 1989). Pp. 83-99. Louvain-la-Neuve: Universit6 jl~4J" [Essay on the Methods and Procedures for Trans- de Louvain, Institut orientaliste. lating Sutras at Buddhist TranslationCenters in China]. Xin Okabe Kazu6. FAZM 1983. "Yakky6 to shaky6 RJ &VA Yaxuebao :1i [The New Asia Journal] 5: 239-321. " [Translated Scriptures and Copied Satras]. Toyo gaku- Tsukamoto Zenryu. 1985. A History of Early Chinese Bud- jutsu kenkyda T ~*WffT [Journal of Oriental Studies] dhism: From Its Introduction to the Death of Hui-yiian. Tr. 22.2: 13-47. Leon Hurvitz. 2 vols. Tokyo: Kodansha International. Pelliot, Paul. 1914. "Les noms propres dans les traductionschi- Turner,Richard L. 1936. "Sanskrit a-kseti and Pali acchati in noises du Milindapanha."Journal asiatique, s6r 11, vol. 4: Modern Indo-aryan."BSOS 8: 795-812. 379-419. Waldschmidt, Ernst. 1932. Bruchstucke buddhistischer Sutras Richard. 1955. A Pischel, Grammarof the Prakrit Languages. aus dem zentralasiatischen Sanskritkanon, I. Leipzig: Tr. SubhadraJha. Delhi; reprint, 1981. Deutsche Morgenlandische Gesellschaft. 506 Journal of the American Oriental Society 118.4 (1998)

.1980. "Central Asian Sutra Fragments and their Re- Zieme, Peter. 1992. "Alternative Ubersetzungen in alttiirki- lation to the Chinese Agamas." In Bechert 1980, 136-74. schen buddhistischenWerken." In Festgabe an Josef Matuz: Waley, Arthur. 1957. "Review of Zenryu Tsukamoto, Choron Osmanistik-Turkologie-Diplomatik,ed. Christa Fragnerand kenkyu."BSOAS 19: 195-96. Klaus Schwarz. Pp. 343-53. Berlin. Watanabe, Shoko. 1975. Saddharmapundarika Manuscripts Zurcher, Erik. 1959. The Buddhist Conquest of China: The Found in Gilgit. 2 vols. Tokyo: The Reiyukai. Spread and Adaptation of Buddhism in Early Medieval Weller, Friedrich. 1930. "Uber den Aufbau des Patikasuttanta, China. 2 vols. Leiden: E. J. Brill; reprint, 1972. 2: Ubersetzung des chinesischen Textes." Asia Major 5: .1977. "Late Han Vernacular Elements in the Earli- 104-40. est Buddhist Translations." Journal of the Chinese Lan- Wiltshire, Martin G. 1990. Ascetic Figures before and in guage TeachersAssociation 12.3: 177-203. Early Buddhism: The Emergence of Gautama as the Bud- .1991. "A New Look at the Earliest Chinese Buddhist dha. Berlin: Mouton de Gruyter. Texts." In From Benares to Beijing: Essays on Buddhism Yuyama, Akira. 1992. "Paicasat1-, '500' or '50'? With Special and Chinese Religion in Honour of Prof Jan Yiin-hua, ed. Reference to the Lotus Sutra." In The Dating of the His- Koichi Shinohara and Gregory Schopen. Pp. 277-304. torical Buddha / Die Datierung des historischen Buddha, Oakville, Ontario: Mosaic Press. ed. Heinz Bechert. Vol. 2: pp. 208-33. Gottingen: Vanden- hoeck & Ruprecht.