15 November 2013

The use of Negation in Vajracchedikā

Paul Harrison
This essay will reproduce and, to some extent, critique an argument put forward by my countryman, Paul Harrison, in his 2006 English translation of the Vajracchedikā Prajñāpāramitā or Diamond Sutra. My thanks to David Welsh for bringing this article to my attention and providing me with a copy of it. 

The Vajracchedikā was first translated into Chinese by Kumārajīva in ca. 402 CE (one also sees the date given as 401 and 403). The text that Kumārajīva translated was somewhat shorter than the one edited by Max Müller in 1881, suggesting that the text continued to change after it was first composed. The dating of the initial composition of Vajracchedikā is now disputed. Conze had argued that it belonged, with the Heart Sutra, to the period ca. 300-500 CE which follows the expansion of the basic text from 8000 to 100,000 lines. We now know that the Heart Sutra is a special case and was composed in the 7th century in China. Some scholars now argue that Vajracchedikā belongs to the earliest strata (see Schopen 1975: 153, n.16; Williams 1989: 42). According to one source cited by Schopen "...the latest date of establishment of the Diamond Sutra will be 200 AD or probably 150 AD" (153, n.16). It may be that, contra Conze, Vajracchedikā predates Aṣṭasāhasrikā (Schopen n.17). Jan Nattier has proposed that the Vajracchedikā was composed in a very different milieu (2003: 180, n. 18) "one of many reasons" is the difference in terminology: where Aṣṭa prefers experiential terms like na saṁvidyate 'is not found' and nopalabhyate 'is not obtained, Vajracchedikā is more confident using the verb 'to be' (asti, nāsti). For Nattier, this suggests that the Vajracchedikā is more at ease with ontology. Another of the quirks of Vajracchedikā is that it never mentions śūnyatā or svabhāva, which is odd for a supposedly late Prajñāpāramitā text.

In his notes to the revised edition of the (partial) Gilgit ms., Greg Schopen (1989) has listed the many problems in Conze's Sanskrit edition of Vajracchedikā (1957), along with previous editions. Conze has been too eclectic with his source materials and paid insufficient attention to chronology. His notes also leave much to be desired. We have reason to be suspicious of Conze's edition and thus of his translation of, and commentary upon, this text.

In 2006, Harrison & Watanabe published a new (partial) edition of the Vajracchedikā based on a manuscript (probably) from Bamiyan, Afghanistan, and held in the Schøyen Collection.  Like the edition by Schopen, this new edition is considerably shorter than previously published editions and close in content to the Chinese translation of Kumārajīva. In the same publication Harrison combined the Gilgit and the Schøyen partial manuscripts to create a single hybrid which represents the text as it circulated in Greater Gandhāra (roughly Northern Pakistan and Eastern Afghanistan) in the 6th or 7th centuries. This hybrid manuscript was then the basis of a new translation (Harrison 2006). The text and translation (sans Harrison's extensive notes and comments) are available online via Biblioteca Polyglotta at the University of Oslo.

In his notes, Paul Harrison tackles the vexed subject of paradox. His very interesting contribution is to provide a detailed argument for reassessing the idiom which is so particular to the Vajracchedikā, which he sums up as "X is non-X; hence it is called X." In particular, Harrison cautions against reading "some kind of mystical subversion of language" into this idiom. Richard H. Jones also resists the conclusion that the text was illogical or not meant to be understood, an idea which he says is "...frankly baffling and insulting to the ingenuity of the authors of this and other Perfection of Wisdom texts" (190, 220-3). I go along with this.

The mystification, even obfuscation, of Perfection of Wisdom texts has been actively pursued in some quarters. Correcting such misreadings will no doubt take time and will probably be resisted by those who enjoy the status quo. I put Harrison's case here, slightly modified with insights from Jones, in the hope that it will cause the more thoughtful amongst us to reconsider the Perfection of Wisdom. My own agenda is to try to demonstrate that the Vajracchedikā is another text which makes more sense when read using what I have called the hermeneutic of experience.


Buddhists will be familiar with the idea that prefixing a- (or an- for words beginning with vowels) to a noun or adjective negates it. (However, I have argued against the popular perception that this is the function of the syllable 'a' that gives it a central place in Prajñāpāramitā. See The Essence of All Mantras). Harrison begins by pointing out that such negated words can be treated as compounds of either of two types: as a karmadhāraya (not-X, no X, non-X) or as a bahuvrīhi (X-less, Lacking X, having no X). In his example the word aputra can be read as describing a person who is 'no son', with the possible implication of being unworthy of his parents; or the person might be 'sonless' or 'have no children'. Either reading is possible and only context can tell us which reading applies in any given case.

English translations of Vajracchedikā and other Prajñāpāramitā texts almost always opt for the karmadhāraya reading. Thus in Conze's (1975) translation we find:
And this world-system the Tathāgata has taught as no-system. Therefore it is called a 'world system'. (52; §13c)
I think everyone agrees that this is nonsense, even if we disagree on the significance of such nonsense. See, for example, Shigenori Nagatomo (2000) for an attempt to make "make intelligible the logic that is used in this Sutra in which a seemingly contradictory assertion is made to articulate the Buddhist understanding of (human) reality" (213).

There exist at least four Sanskrit variations on this sentence, 8 Chinese translations, and one Tibetan. The Gilgit Sanskrit manuscript reads:
yo 'py asau lokadhātur adhātuḥ sa tathāgatena bhāṣitas tenocyate lokadhātur iti | (Harrison 137)
We've mentioned that the English practice is to treat the compounds as karmadhāryas ("no-system"). Harrison points out that the Chinese also read karmadhārayas, because they negate the terms using 非 fēi rather than 無  (cf the negations in the Heart Sutra which all use 無). What interests Harrison is that the Tibetans treat the compounds as bahuvrīhi (X med pa or X ma mchis pa) and that this seems to be the better reading. In the case of lokadhātu/adhātu we might, with the Chinese, construe this as 'the world system (or realm, sphere, element, etc.) is not a system'; or '...is a non-system'; or '...is no system at all.' However, we may also read it as saying that lokadhātu lacks a system or that there is no system in it (138). Harrison translates:
"Any world system there is has been preached by the Realized One as systemless. Thus it is called a world-system" 
The obvious question is whether we can chose either option arbitrarily? Harrison thinks not. He thinks we must chose to read these compounds as bahuvrīhis, i.e., as adjectives of the unnegated term. In this case, adhātu is an adjective describing the substantive lokadhātu. In order to show this, he first lists all thirty of the terms that are negated in the text. In each case what is negated is the second part of the compound: where we have a compound of the form XY, the negative is almost always aY, though sometimes aXY with an implied negation of Y. For example, lokadhātu > adhātu; puṇyaskandha > askandha; ātmabhāva > bhāva and so on.

Reading the Negations in Vaj

The key to understanding this idiom, according to Harrison, is in the phrase nirātmāno dharmā 'dharmas are selfless', which is found in the Vajracchedikā (17h) but echoes an Āgama phrase. It's also expressed anātmakāḥ sarve dharmāḥ, 'all dharmas are selfless'. The Pāli counterpart of this phrase, sabbe dhammā anattā, is more ambiguous and is frequently read as 'all dharmas are not-self'. However, Harrison argues that the ambiguity is not present in Sanskrit, so that a term like nirātma must be read as a bahuvrīhi (i.e., as self-less, rather than non-self).
"Thus nirātmāno dharmā means that all dharmas lack a self or essence, or to put it in other words, they have no core ontologically, they only appear to exist separately and independently by the power of conventional language, even though they are in fact dependently originated" (139)
I'm largely in agreement with Harrison. One of the key features of Prajñāpāramitā thought is a trenchant critique of substance ontologies which became a feature of Buddhist Abhidharma thought in North India around the beginning of the common era.

However, I differ from Harrison in attributing the problem to "appearance" and "conventional language". Clearly, we do have experiences. This is not a matter of appearance or convention. The use of the term "illusions" (indicating that experience is not real) is itself an aspect of the shift of attention from dharmas qua experience to dharmas qua reality. The latter leads to the necessity to split reality into relative and absolute (or conventional and real). If we do not go down the path of real dharmas the issue of conventional vs real language does not arise. In other words, the split of language into conventional and real is dependent on framing the discussion in terms of real or unreal. The early Buddhists deny the validity of this dichotomy with respect to experience (e.g., Kaccānagotta Sutta SN 12.15) and thus imply that the conventional/real split is not valid, either. Thus, if we accept the early Buddhist argument, and I do, then we need not invoke conventional language and illusions. 

The Buddhist model of cognition itself shows why this is so: cognition (vijñāna) is always sense-object, sense-faculty and sense-cognition working together. "Direct knowledge of objects" is simply not possible in this model as sense-cognition is always part of Buddhist knowledge production. "Reality" gets crowbarred into Modern Buddhism, but in early Buddhism there seems to be no concept that corresponds to our concept of reality. Early Buddhists accepted no noumena behind phenomena. The Buddha was always concerned with experience and understanding the nature of experience.  

Thus, the problem is not one of "appearance", but one of interpretation. The naive realist feels themself to be in direct contact with reality. And, as part of this interpretive framework, the sense of self is also interpreted as real (giving rise to a number of false notions such as disembodied consciousness; pure subjectivity; a true self; i.e., a substantial entity behind what feels like subjective experience; and persistence of consciousness after death). When we understand that these ideas were meant to be applied to the domain of experience, rather than the broader domain of reality, then we eliminate a great deal of confusion both in language and in metaphysics. 

This difference aside, Harrison's proposal to read negated compounds like adhātu in the Vajracchedikā as bahuvṛhis is very interesting and useful. In the case of a term like prajñāpāramitā we get the negated term apāramitā. As he says, prajñāpāramitā "does not contain any perfection [pāramitā] within itself, it is devoid of perfectionhood, so to speak, which would constitute its essence."

So the form of the first part of the argument is now: Any X kind of Y is Y-less according to the Buddha. In other words, just because we can talk about various kinds of dhātu (loka-dhātudharma-dhātumano-dhātu etc) does not make dhātu a thing; does not make dhātu real; does not imply a substantial entity. There is no dhātutva or dhātu-ness, no noumena lurking in the background to give our experiences the qualities of permanence, satisfactoriness or substance.

Final Affirmation

Harrison's explanation of the final affirmation ("thus is it called Perfection of Wisdom") I find less convincing, precisely because I'm unconvinced by the arguments about so-called conventional language.
"If there was perfection in the perfection of insight, then perfection would exist apart from the perfection of insight, and we would have two things, not one, and we could no longer speak about anything as the perfection of insight... However, there is no perfection existing as an entity in and of itself apart from the perfection of insight..." (139-40).
This approach echoes discussions in Nāgārjuna's Mūlamadhyamakakārikā, and does draw out an important aspect of the critique in the sense that it is critical of essence or noumena. However, I think Jones does better. The form of the affirmation in Sanskrit is:
tenocyate lokadhātur iti; i.e., (without sandhi) tena ucyate "lokadhātuḥ" iti.
Jones points out that the tendency of previous translators to render tena as 'hence, thus, that is why' introduces a paradox. He argues that in fact no paradox is implied (222). What translators, including Harrison, it seems, are doing here is making the final statement a direct logical consequence of the previous: XY is Y-less, therefore it is called XY. But as we have already observed, this inference is not logical. The fact that a world-system is not a system does not logically infer that we should call it a world-system. In this case the word "therefore" seems out of place, to say the least, and this raises the question of how we translate tena.

Conventionally, tena in this position can be translated as "therefore". Apte's dictionary sv. tad has "...tena the instrumental of tad is often used with adverbial force in the sense of 'therefore', 'on that account', 'in that case', 'for that reason.'" Jones is arguing that tena does not have adverbial force here. Jones construes the sentence as being in the form: "XY is not a (real) Y. The word 'XY' is used this way." 

The iti following lokadhātuḥ is equivalent to putting the word in quotes. The Sanskrit: lokadhātur iti corresponds to English [the word] 'World System'. The verb ucyate is a third person singular passive from √vac, 'to say, to speak'. Here, the passive requires the weakest grade of the root vowel and √vac undergoes samprasaraṇa to become uc; and ucyate means 'it is said, it is spoken'. And here we understand that it is the word lokadhātu (in the nominative) that is being spoken. With a passive verb the agent will be in the instrumental case and tena is the only word in the instrumental case. Thus, here we can take tena 'by him' to be the agent of the verb rather than an adverb. The use of the pronoun tena would usually refer to a previously mentioned agent, in the previous phrase i.e. tathāgatena 'by the Realised'.

So the phrase reads: 
tena ucyate lokadhātuḥ iti
by him / is said / "lokadhātu"
The word 'lokadhātu' is said by him [i.e., the Realised].
Jones argues that this means that the Tathāgata uses a word like lokadhātu always keeping in mind that there is no substantial, really existent kind of 'dhātu'. This is emphasised later in the text: a bodhisattva perceives no ātma, satva, jīva, or pudgala, which here translate roughly as 'substance, essence, soul or homunculus' (Vaj §6 ). Or, to put it another way, the names we give to persistent and repetitious experiences cannot hide the truly ephemeral, unsatisfactory and insubstantial nature of experience. I've argued before that this is only true (or only straightforwardly true) when the domain under consideration is experience. Hence, I see a continuity here with one of the most important threads of early Buddhist thought that is epitomised by the Kaccānagotta Sutta (SN 12.15) and the Sabba Sutta (SN 35.23).

The Role of Translator

I think what both Harrison and Jones are getting at is that the translator is also an interpreter. A translator assumes that the text made sense to the author and tries to understand the sense and communicate it in another language. Conze's interpretations unconsciously colour every line of his translations. However, he frequently choses unclarity precisely because it suits his interpretation (this was, for example, how I understood his misreading of the first sentence in the Sanskrit Heart Sutra resulting in a simple grammatical error). Faced with a sentence that presents a difficult reading, the translator's job is to try to get across what they think the author was getting at. Slavishly sticking to a literal rendering of the words is seldom helpful, especially if one also uses the foreign syntax. Buddhist authors use both the Sanskrit language and sectarian Buddhist idiom to convey ideas; the text is embedded in the context of a worldview. It's not sufficient to translate an idiom literally because idioms are not used literally. Consider English idioms like "I'm going to see a man about a dog", or "he'd lose his head, if it wasn't screwed on". Literalism only leads us astray and yet Conze admits he has translated as literally as possible to the point of reproducing the Sanskrit syntax. This approach has marked Conze as a leading exponent of what has been called Buddhist Hybrid English.

Of course, for Conze, perhaps influenced by Suzuki and late Buddhist commentators, the quotient of nonsense in his texts seemed to give him a certain amount of pleasure. It gave him scope to play the gnostic and insinuate that he understood this text in a nonconceptual way through (deep) meditation, whereas his academic readers, approaching Buddhism intellectually, had to be content with illogical nonsense. There are constant digs at the plodding intellectual non-meditator in Conze's commentaries. As his memoirs make plain, Conze felt, with some justification, deeply aggrieved at his treatment by the academic establishment of the UK and USA and was contemptuous of most of his colleagues. Conze is thus a complex figure and his work is complicated by such factors as well.


Thanks to Harrison and Jones, we now have two possible interpretations of this and similar passages: one which conveys nonsense and implies occult profundity; and one which conveys some sense and is no less profound but in a more obvious way.

Taking the whole sentence again:
yo 'py asau lokadhātur adhātuḥ sa tathāgatena bhāṣitas tenocyate lokadhātur iti |  
The Tathāgata taught a world-system that is without a [noumenal] system, the word 'world-system' is used this way by him.
I think it's worth repeating that this statement is not less profound than Conze's gnostic interpretation or more mystical readings. That fact that we can understand the statement does not make it less valuable.

Of course, whether we can experience a "world-system" is moot. With cosmological terms like lokadhātu we are in an abstract realm. Even in modern cosmology everything we know about the universe is inferred rather than experienced. The object of the senses here is an abstraction formed in the mind on the basis of sense data; i.e., it is an object of the mind-sense (manas). The Vajracchedikā is, in fact, equivocating on this element of Buddhist cosmology. "OK," it is saying "even if you believe in a world-system (or any other cosmological or metaphysical entity), your experience of it is still subject to the laws of experience: impermanence, disappointment and insubstantiality." Whatever categories, abstractions, ideas, entities you can come up with, your experience of them is subject to these constraints because it is only through experience that you know anything at all. Here the relatively unsophisticated Buddhist approach to psychology has distinct advantages. By lumping all mental experience under the heading of manas we are less likely to get caught up in finessing the details of the aspects of humanity that make us feel special. Our ability to think abstractly is remarkable, but to Buddhists it's just another kind of experience about which we make epistemological mistakes. 

Whatever ontology we might subscribe to, there are always these epistemological constraints that leave us off balance. Though we might make apparently valid ontological inferences, commitment to any particular ontology as an individual is always premature because knowledge proceeds from experience, not from reality, and experience is always a co-creation (pratītya-samutpāda) of objects, our sensory apparatus and our mind. This is directly contradicted by Buddhist mystics, who sometimes claim that direct knowledge of reality is the goal of Buddhism, but it is what the Buddhist texts say over and over; and it is also what my friends who go deep into meditation also say. In this view the problems of human existence are due, in effect, to epistemological errors which can be corrected by careful observation of experience under controlled conditions and the guidance of an experienced mentor. The problems addressed by Buddhist practice are, on the whole, not caused by ontological errors. With some caveats, what we experience is not an illusion, but we do have illusions about what we experience. Of course, we do make ontological errors, but the Buddhist texts do not seem overly concerned with this type of error, which is relatively easily corrected.

I've said that ontological commitments based on individual experience are always premature. However, we can get around this by pooling our resources. One of my frustrations with philosophy and, in particular, Buddhist philosophy, is that it always seems stuck in the point of view of an absolutely isolated individual and takes no account of our collective endeavours. In fact, we social primates almost always work in teams. We can have reliable knowledge about the world around us through comparing experiences, though even when this collaborative effort is coordinated and formalised (as in the sciences) most knowledge is still considered provisional, because there is always the possibility of a "black swan event". Most importantly, by communicating with others we do know that objects exist apart from our perception of them, even though we might be slightly fuzzy about the details of that object. One only has to watch the heads at a tennis match - turning this way and that as they follow the action - to know that the ball is not something that we alone are perceiving. The only way to maintain that objects do not exist, is to artificially disallow the evidence of others.

But the whole focus of the Buddha's teachings is away from the objects and on the experience itself. And in experience neither 'real' nor 'unreal' apply. Even if the object were permanent, the experience itself would still change because it relies partly on us - our sense organs and sense cognition. These simple facts can be used to direct our practice of Buddhist techniques in an effective direction. The Prajñāpāramitā teachings continue this focus.

As a final aside aside, Jan Nattier has an interesting take on this type of negation: "[In the Aṣṭa and the Vajracchedikā] the initial negations are directed not at 'dharmas' or at things in general, but at the bodhisattva and the practices in which he is engaged. It is my strong suspicion that this 'rhetoric of negation' first emerged as a tactical attempt to undercut the potential for bodhisattva's arrogance, and was only later generalized to what came to be considered to be a new (anti-abhidharma) ontology" (2003 135-6, n.62). Hopefully in the future Nattier will collect her thoughts on the Vajracchedikā and publish them.



Conze, Edward. (1957) Vajracchedikā Prajñāpāramitā. Serie Orientale Roma XIII. Roma. 
Conze, Edward (1975) Buddhist Wisdom Books: The Diamond Sutra and the Heart Sutra. 2nd Ed. George Allen & Unwin.
Harrison, Paul. (2006) 'Vajracchedikā Prajñāpāramitā: A New English Translation of the Sanskrit Text Based on Two Manuscripts from Greater Gandhāra', in Buddhist Manuscripts in the Schøyen Collection (Vol. III). Hermes Publishing, Oslo, p.133-159.
    Harrison, Paul & Shōgo WATANABE (2006) 'Vajracchedikā Prajñāpāramitā.' in Buddhist Manuscripts in the Schøyen Collection (Vol. III). Hermes Publishing, Oslo, p. 89-132.
    Jones, Richard H. (2012) The Heart of Buddhist Wisdom: Plain English Translations of the Heart Sutra, the Diamond-Cutter Sutra, and Other Perfection of Wisdom Texts. New York: Jackson Square Books.
      Shigenori Nagatomo. (2000) 'The Logic of the Diamond Sutra: A is not A, therefore it is A.' Asian Philosophy, 10(3): 212-244
        Nattier, Jan. (2003) A few good men : The Bodhisattva path according to the Inquiry of Ugra (Ugraparipṛcchā). University of Hawai'i Press.
            Schopen, Gregory. (1975) 'The phrase ‘sa pṛthivīpradeśaś caityabhūto bhavet’ in the Vajracchedikā: Notes on the Cult of the Book in Mahāyāna.' Indo-Iranian Journal. 17(3-4): 147-181.
              Schopen, Gregory. (1989) 'The Manuscript of the Vajracchedikā Found at Gilgit,' in Studies in the Literature of the Great Vehicle: Three Mahāyāna Buddhist Texts, ed. by L. O. Gómez and J. A. Silk, Ann Arbor, pp. 89-139.
                Williams, Paul. (1989) Mahāyāna Buddhism: the Doctrinal Foundations. London, UK: Routledge.
                  Related Posts with Thumbnails