Like David, I was going to avoid getting into the crosshairs on this, despite agreeing with the Jean-Luc and the beginning of David’s response. But David's “myth” of the alleged disparity between the Chinese and Tibetan translations and even the supposed disparity in social consequences, a myth that has been in circulation for too long in Buddhist studies, requires some comment.

First, as to the agreement. For exactly the reason that David mentions, brackets (and parentheses) should separate what can legitimately be attributed to the original underlying text and what is being added for clarity by the translator. Over the years, many undergraduate students have written papers or essays spinning out detailed and clever interpretations of texts they read in translation, but primarily extrapolating and unpacking implications of a word that they are unaware has no counterpart in the original text.

Sanskrit in many genres puts a premium on conciseness, and can sometimes be so condensed as to require a bit of extra padding in translation to make explicit what is tacit. On such occasions brackets and parentheses may be optional at the discretion of a translator. And there is a principle by which things already stated needn’t be repeated, but remain implicit in later passages by context. Then, too, filling in the gaps for clarity in a translation is often necessary, and whether to use brackets or parentheses can be optional (perhaps determined by how much extra is being supplied). But there can be entire arguments that are presented in cryptic, ellitical shorthand; then, depending on the degree of confidence a translator has at identifying the fully developed argument, brackets, and or parentheses, and or additional annotations, may be necessary.

It is true that a certain stylistic vogue took hold during the 20th c. that way overdid such interpolations, so that frequently, not just occasionally, more than 50% of a translation resided in brackets that added little clarity but made reading more difficut rather than easier. So the push-back against overuse of brackets and parentheses is understandable and necessary. As Buddha might recommend, however, a middle way — a more efficient and effective use of interpolative graphics rather than banning them altogether, should be the goal.

As for the myths:

First, the Chinese translations are, in general, no less literal or accurate than the Tibetan. Both take liberties, and one reason another 20th century vogue — back-translating texts from Tibetan and/or Chinese into Sanskrit — has gone out of style is the continual recovery of lost Indic texts (not always in pure Sanskrit) which highlighted how inaccurate the back-translations often were, whether constructed from Chinese, Tibetan or both.

The idea that TIbetan translations are more literal or accurate is a myth that doesn’t hold up to scrutiny. Matt Kapstein’s recent comments on this list about the calques of Tibetan titles for Indian texts is only the tip of the iceberg. Sometimes the Chinese texts are more accurate than the Tibetan. One example should suffice. In Dignāga’s Ālambana-parīkṣā for which only a few passages survive in Sanskrit, but three Chinese translations (by Paramārtha, Xuanzang, and Yijing — three of the most famous translators, the Yijing translation embedded within his translation of Dharmapāla’s commentary which also doesn’t survive in Sanskrit) and a pair in Tibetan (the verses alone, and another with the verses inserted into the svavṛtti) plus a translation of Vinītadeva’s commentary which, as chance would have it, quotes and relies heavily on Dharmapāla’s commentary. The most “literal” (and therefore difficult to decipher) is the Dharmapāla; one can recognize many of the parallels between it and the Tibetan version of the Vinītadeva, but the latter is often equally cryptic in places (a group working on it frequently asked a group working on the Chinese for help clarifying passages). And here is perhaps the most significant example of where the accuracy of the Chinese easily eclipses the Tibetan. Dignāga devotes five of the eight verses to refuting three different theories of how atoms could serve as ālambana: 

1. that the ālambana is a paramāṇu, 

2. that it is a samūha of paramāṇus, and 

3. that it is a saṃcita (or saṃcita-ākāra) of paramāṇus (Dharmapāla differentiates this from a paramāṇu by calling it an aṇu, which is considered the minimum amount of paramāṇus necessary to be perceived - not all discussions in India adhered to that terminological distinction) 

The main difference between two and three is that, in the jargon of some Buddhists of the day, a samūha is simply a prajñapti, and thus cannot cause anything, including being unable to cause a perception, and thus cannot be an ālambana, while a saṃcita is, its advocates claim, a dravya and thus can cause a perception as well as convey its own svarūpa to the perception — the two requirements Dignāga stipulates for an ālambana to be an ālambana (based on the abhidharmic assumptions of his pūrvapakṣas).

All three Chinese versions distinguish the three theories by using different terms for each; the Tibetan uses only ‘dus pa for both the second and third theories, the result being that when a respected Tibetan geshe was asked to give instruction on the text for the group preparing a translation from Tibetan, the geshe was surprised to learn there was a third position (and the book the group published makes serious equivocations about whether there is or isn’t a third position — despite the fact that the Vinītadeva text preserved in Tibetan explicitly announces the third position as “the third” when coming to it (but the anonymous Tibetan translator, perhaps the same translator[s] that translated the verses and verse+svavṛtti versions, also used ‘dus pa for samūha and saṃcita in that text as well). On the other hand, the easiest of the translations to read—and least literal—is Xuanzang’s translation which takes a number of interesting liberties — some of which, like the practice of the aforementioned undergraduates, became the fodder for extensive doctrinal exegesis in East Asian circles, unrecognized as Xuanzang’s own contributions. He didn’t use brackets or parentheses.

To further easily test whether Tibetan is as isomorphic to Sanskrit as the propaganda insists, start with the Tibetan term don. Usually presumed to be the Tibetan for artha, in fact, according to the online Tibetan translation tool at http://www.thlib.org/reference/dictionaries/tibetan-dictionary/translate.php it is used as an attested equivalent for kārya, kāryatva, parakārya, artha, arthatas, bhāva, vastu [and I would add that yul, the main Tibetan equivalent for viṣaya has a similar spread and overlap - the translation tool lists deṣa.diś,  pradeṣa, *-antara, viṣaya, gamya as equivalents, but I would suggest in practice there are more]. Not only is there no one-to-one literal correspondence, which can complicate discussions of epistemology when terminological precision is required, but inflections can also get murky. Chinese translations, incidentally, can be fastidious about indicating Sanskrit inflections (though many modern readers overlook them).

As for Buddhism being successful in Tibet but not China — there are more East Asian Buddhists, all of whom rely on the Chinese canon, than there have ever been Tibetan Buddhists. When the PRC relaxed its suppression of all religions starting in the 1980s, Buddhism, which had been underground, reemerged into plain sight in vast numbers, and continues to flourish. Despite being persecuted for centuries by Confucians in Korea, Buddhism remains the largest religion there, and one needn’t point out its continued cultural significance in Japan.

Dan Lusthaus