Note to the Translation

The Diversity of Textual Versions and Editions

The translation of Engaging in Bodhisattva Behavior (sPyod-‘jug, Skt. Bodhisattvacarya-avatara) by Shantideva presents many textual problems. Written in Sanskrit during the first half of the eighth century C. E., several versions of the manuscript have passed down through the centuries. The Nepalese-German Manuscript Cataloguing Project, for example, has microfilmed forty-one handwritten manuscripts, of varying lengths. As far as I know, a comparative study has yet to be made on them.

A Tibetan translation of a Sanskrit recension, which may or may not be the same as any of the above-mentioned forty-one, has recently been discovered among the manuscripts buried at Dunhuang at the end of the tenth century. It contains 210 ½ less number of verses than the Tibetan canonical version.

According to the colophon of the Tibetan canonical version, the text was first translated into Tibetan in the early ninth century, during the Old Translation Period, based on a Kashmiri manuscript. The translators were the Indian master Sarvajnadeva and the Tibetan editor-translator monk Paltseg (dPal-brtsegs). Paltseg was one of the compilers of The Grand (Lexicon) for Understanding Specific (Terms) (Bye-brag-tu rtogs-par byed-pa chen-po, Skt. Mahavyutpatti), the first compendium of standardized Tibetan translation terms for Buddhist technical terms in Sanskrit.

The text was retranslated from a Magadha edition and commentary during the first half of eleventh century by the Indian master Dharmashribhadra and the Tibetan editor-translator monks Rinchen Zangpo (Rin-chen bzang-po) (958–1051) and Shakya Lodro (Shakya blo-gros). It is not clear which commentary this was. Rinchen Zangpo was the founder of the New Translation Period in Tibet.

The text was further corrected, retranslated, and finalized by the learned Indian master Sumati-kirti and the editor-translator monk Loden Sherab (Blo-ldan shes-rab) (1059–1109). This is the version preserved in the Tibetan canon, although various editions of the canon and of later publications of the Tibetan text have a number of textual discrepancies. The two earlier versions of the Tibetan translation have not survived, as far as I know.

According to the main compiler of the Tibetan canon, Buton (Bu-ston Rin-chen grub) (1290–1364), a hundred commentaries were written in Sanskrit to Engaging in Bodhisattva Behavior, but only eight were translated into Tibetan. The most well known one, perhaps because of the twentieth-century publication of the Sanskrit original, is Commentary on the Difficult Points of Engaging in Bodhisattva Behavior (sPyod-‘jug dka’-‘grel, Skt. Bodhisattvacarya-avatara-panjika). It was written by Prajnakaramati, in the eleventh century, and comments only on the first nine chapters of the root text.

Sumatikirti, the Indian pandit who helped with the Tibetan translation of the root text used for the canonical version, translated chapters 1, 2, 7, 8, and 9 of Prajnakaramati’s commentary into Tibetan together with the Tibetan translator Darma Drag (Dar-ma grags). The in-between chapters were translated by the Tibetan Lodro Zangdrag (Blo-gros bzang-grags). Thus, it is quite likely that the Sanskrit version used for the translation of the first nine chapters of the root text was the same as that which appears in Prajnakaramati’s commentary. Taking into consideration that there are several slightly different manuscript versions of this edition of the Sanskrit root text and commentary, there are still quite a number of discrepancies between the Sanskrit original and the Tibetan translation of the root text. The same is true concerning the published Sanskrit and canonical Tibetan versions of the tenth chapter, missing in Prajnakaramati’s work.

Many Tibetan masters from the four traditions of Tibetan Buddhism have written commentaries on the root text, based on the canonical version. Several of them were aware of the textual discrepancies and made occasional reference to different readings of some of the root verses as found in the Tibetan translations of the Sanskrit commentaries. Moreover, these Tibetan commentaries present a wide range of interpretation of the root verses.

The only conclusion that we can safely draw from the above sketch is that it is impossible, at present, to decide what is the authentic version of the text and what was its original or “actual” meaning. All of the versions and their commentaries make sense within the context of Buddha’s teachings. This accords with the principle that the enlightening words of a Buddha contain many levels of meaning and each disciple will understand them according to his or her stage of progress.

Faced with this situation, how best to translate the text into modern languages? Many English translations of Engaging in Bodhisattva Behavior have already been made, some from the Sanskrit version that appears with the Prajnakaramati commentary, plus the tenth chapter omitted there, and some from the Tibetan canonical translation. One of English works has even presented translations of both the Sanskrit and Tibetan versions for verses that significantly differ from each other. None, however, has attempted to reconcile discrepancies between the two that may have arisen simply due to scribe’s errors or differences in the structures of the two languages. This has been the challenge attempted here.

Moreover, some translations have favored accuracy over poetics, and others have sacrificed accuracy for the sake of poetic beauty. Here, I have attempted to preserve the two.

Translation Method for Resolving Discrepancies between Tibetan and Sanskrit

I have received discourses on Shantideva’s text twice from His Holiness the Fourteenth Dalai Lama and twice from Geshe Ngawang Dhargyey. When His Holiness teaches the text, he often corrects the canonical Tibetan translation based on explanations he received from Khunu Lama Rinpoche Tenzin Gyaltsen, a great master well versed in Sanskrit. Moreover, His Holiness’s explanations always emphasize that the text is meant for meditation and everyday practice. Therefore, textual corrections need to be made based not only on what accords with Sanskrit grammar, but also on what accords more closely with practical advice on bodhisattva behavior. I have based the present translation on this precedent and this principle. Accordingly, I have followed primarily the canonical Tibetan version, but amended it, when necessary, according to the Sanskrit version that appears in the Prajnakaramati commentary, plus the published tenth chapter.

To say that the text is meant for use in meditation means that the whole or selected parts are to be read or recited from memory each day, aloud or silently, and reflected upon. This implies that the verses all connect with each other to form a flowing presentation of the various topics. The text does not consist of disjointed verses. Preservation of the flow of presentation or argument, then, is one of the main criteria I have used for establishing the context within which each verse needs to fit. I have therefore tried, as much as possible, to preserve the flow by adding conjunctions and so on, in parentheses, to help make the connections clearer.

Some of the differences in the Tibetan and Sanskrit versions may have arisen due to the Tibetan having been based on a slightly different Sanskrit manuscript version than the one that is currently available in published form. In such cases, I have followed the Tibetan version, so long as it fits in the context of the flow. It would be awkward to try to translate and incorporate both language versions into one text for lines that are totally different, and also there is no clear way to decide which version is more authentic.

Some of the discrepancies are due to a difference of one letter in a word, such as ākāra (appearance) in Sanskrit and the Tibetan translation obviously from āhāra (sustenance). A drop of water creating a smudge on a page of a handwritten manuscript, or a scribe’s error can be the cause of such discrepancies. In such cases, when both meanings make sense in the context of the flow of verses, I have translated both, with the Sanskrit in parentheses. When only one version has made sense in the context of the flow of verses, I have translated only that. Such cases have usually favored the Sanskrit version.

Sometimes, Tibetan nouns are followed by an instrumental particle when the Sanskrit grammar requires it to be genitive, or vice versa. Such discrepancies may also have arisen due to a scribe’s error or a smudge. In such cases, as well, I have followed the Sanskrit version when it makes more sense in the context.

One and a half verses from Sanskrit are missing from the Tibetan, and I have added them in parentheses. When words and phrases in the Sanskrit version have been omitted in the Tibetan translation, and they fit in well with the context, I have also added them in parentheses.

The greatest apparent source of discrepancies, however, is the difficulty of rendering the complexities of Sanskrit grammar into Tibetan. The two languages are extremely different in structure. Sanskrit is one of the most highly inflected Indo-European languages, while Tibetan belongs to the Sinitic family of languages and is far less inflected. Moreover, it is inflected according to different parameters. When the Tibetan is obviously trying to render the Sanskrit construction, and the verb or noun forms are ambiguous in Tibetan due to the limitations of Tibetan grammar, I have followed the Sanskrit grammar.

For example, Tibetan often renders both the dative and ablative cases of Sanskrit nouns with the postposition phyir, and does not distinguish the vocative from the nominative case. The Sanskrit present, past, and future participles, both active and passive, are often all represented by the past tense of a Tibetan verb together with the auxiliary byas. The Sanskrit optative, imperative, and future tenses are mostly all translated with the future of a Tibetan verb together with the auxiliary bya. Third person imperative constructions, as well as locative absolutes, present particular challenges. Tibetan also has difficulty in clearly distinguishing the Sanskrit active, middle, and passive voices, and often leaves out distinctions of singular, dual, and plural. Interrogative and relative pronouns are not easily distinguished from each other, and so on.

The only exception to this guideline concerns the person of the verbs. The Sanskrit version sometimes uses the first person, as when the meditator is addressing himself or herself. Sometimes it uses the second person, when the meditator is addressing his or her mind. And sometimes it uses the third person, as in making a general statement, or a participle construction, which avoids the issue of person. Tibetan verbs do not have personal endings, whereas Sanskrit verbs do. When the Sanskrit occasionally uses a pronoun, the Tibetan translates it. Otherwise, the Tibetan is not clear. I have rendered first and second person verbs in Sanskrit into first and second person in English. However, for the sake of making the text more obviously applicable to personal meditation practice, I have occasionally rendered the Sanskrit third person and participle constructions with the first person in English.

Neither Sanskrit nor Tibetan differentiates gender in third person singular verbs. For the sake of simplicity, I have translated them all as masculine, based on the fact that Shantideva was a monk and was writing primarily for monks. Shantideva’s presentation of meditation on the uncleanliness of the body, as an antidote to longing desire, attachment, and distraction in meditation, is gender neutral in both Sanskrit and Tibetan. Although some of the commentaries specify that the discussion refers to the body of a woman, since that would be the most relevant for a monk audience, I have left the discussion gender neutral as in the original.

Sometimes the Tibetan terms chosen for translating a Sanskrit term have several meanings. When a secondary meaning of the Tibetan corresponds more closely to the Sanskrit term, and the primary meaning does not, I have chosen the secondary meaning, since this was obviously the meaning intended by the translators. When both the Sanskrit and Tibetan terms have several disparate meanings, I have chosen the meaning that is common to both.

Further, sometimes there is a discrepancy in the two versions concerning which words modify or go together with which other words in the verse. In the Sanskrit version, the case and number endings indicate the connections quite clearly, whereas Tibetan is not inflected in the same way. When the Sanskrit makes more sense from the context, I have followed the Sanskrit. When the differences seem insignificant, I have followed the Tibetan.

Occasionally, the order of the phrases in the verses do not correspond with each other, and change the emphasis in the verse. When the Sanskrit ordering gives an emphasis that fits better into the context of the flow of the verses, or is more poetic, I have followed the Sanskrit. When it has not made much difference, I have followed the Tibetan.

Another complication concerns poetic devices. The Sanskrit frequently uses alliteration, puns, and plays on words, which are not conveyed in the Tibetan. I have attempted to convey this device by using them sporadically throughout the English translation, although not necessarily where they appear in Sanskrit. The Sanskrit often repeats a word several times within a verse, or uses it in different forms and inflections, while the Tibetan often uses several terms. Although English style may frown upon such repetition, I have followed the Sanskrit style as much as possible, to convey some of the flavor of Sanskrit poetics. When Tibetan has repeated a word several times within a verse, but the Sanskrit has employed different terms, I have generally followed the Sanskrit, especially when the Tibetan repetition may have been due to a scarcity of synonyms in Tibetan.

Moreover, both the Sanskrit and the Tibetan are in metered verse. Although I have not used a strict meter for the English text, I have tried as much as possible to render it into a loose English meter, so that it flows easily from the mouth. Hopefully, this will make the text more conducive for recitation, meditation, and memorization. Because of the rules of meter, both Sanskrit and Tibetan frequently need to add filler conjunctions, particles, and words of emphasis, to fill out the meter. I have translated them when they also contribute meaning within the context, and have occasionally added filler conjunctions, such as “and,” in English, also for the sake of meter.

Many passages, especially in the ninth chapter on far-reaching discriminating awareness (the perfection of wisdom) have several interpretations, as evidenced by the wide variety of commentaries. As a root text, the verses therefore need to be as neutral as possible in meaning, so that they can act as a root from which the various explanations and levels of understanding can grow. On the other hand, the verses also need to make sense on their own, even without commentary. I have attempted to meet both needs, by adding as few words in parentheses as possible to such passages. When I have added “true” in parentheses before “existence,” for example, this accords with the fact that all the commentaries interpret the text from the Madhyamaka viewpoint, despite differences in presentation of Madhyamaka.


I wish to thank Renate and Rainer Noack, who requested me to teach Engaging in Bodhisattva Behavior weekly at the Buddhistische Gesellschaft Berlin, and to the students there who requested me to teach it slowly, thoroughly, and deeply, regardless of how long this might take. The course began in November 2000, took one year to complete the first eight chapters, and has been on the ninth chapter for the last three years, with still a little less than half of that chapter covered. This has afforded the opportunity to delve into each word of the text and its background, and to prepare for the class a rough running translation of the verses that would accord more closely with the explanations.

I wish to thank Christian Dräger and Christian Steinert, who have been translating these rough English verses into German as we have been going along, and for encouraging me to continue the process.

I wish also to thank Albrecht Seeger, who requested me to go through the Tibetan text with him over the last half year. This became the circumstance for revising my entire translation, in accord with the Sanskrit, so that I could explain each grammatical construction and the translators’ choice of Tibetan words. Going through the text, word for word, with him has helped to fine-tune the accuracy of the translation, so that, as much as possible, each word in the Tibetan version is accounted for in the English.

Finally, I wish to thank His Holiness the Dalai Lama, who will be teaching the entire text in Zürich, Switzerland, in August 2005. His planned teaching has inspired me to prepare this translation in timely fashion, so that it may be of benefit for that occasion and serve His Holiness’s aims. May it be of benefit to all.