17 November 2017

All of them Arahants. Notes on Aṣṭasāhasrikā and Speech Acts.

I'm doing some preparation for reading Chapter One of the Aṣṭasāhasrikā-prajñāpāramitā-sūtra (Aṣṭa) with a friend and have ended up making a load of notes. I'm mainly looking at the edition by Vaidya, but comparing it where possible with the Gāndhārī, and two versions in Chinese, one by Lokakṣema, translated in 179 CE (the earliest), and one by Kumārajīva, translated in 404 (the most popular). This will be too laborious to do for the whole text, but might help shed light on particular passages (we may well get a publication out of it at some point).

The first sentence is:
evaṃ mayā śrutam ekasmin samaye bhagavān rājagṛhe viharati sma gṛdhakūṭe parvate mahatā bhikṣusaṃghena sārdham ardha-trayodaśabhir bhikṣuśataiḥ, sarvair arhadbhiḥ kṣīṇāsravair niḥkleśair vaśībhūtaiḥ suvimuktacittaiḥ suvimuktaprajñair ājñair ājāneyair mahānāgaiḥ kṛta-kṛtyaiḥ kṛta-karaṇīyair apahṛta-bhārair anuprāpta svakārthaiḥ  parikṣīṇabhava-saṃyojanaiḥ samyag-ājñā-suvimuktacittaiḥ sarvaceto vaśiparamapārami-prāptair ekaṃ pudgalaṃ sthāpayitvā yaduta āyuṣmantam ānandam ||
In this batch of notes, I will make some miscellaneous comments about numbers, dhāraṇī, and the absence of bodhisatvas. I then look at how speech act theory can inform translation, using one of these adjectives (in red) as my example.


In english we say that there were "twelve hundred and fifty bhikṣus". However, Sanskrit Buddhists texts say this differently, using the form "x hundreds-of-bhikṣus" (where hundreds-of-bhikṣus is a compound, bhikṣuśatāḥ). In this case the number of hundreds is ardha-trayodaśa or literally "half-thirteen". This means thirteen-less-a-half, or twelve-and-a-half. And "twelve and a half hundreds" = 1250. The significance of this number is unclear, but it crops up in other texts as well.


One of the overall things that strikes me about the string of adjectives (in red) is how much it looks like a dhāraṇī. There is the same kind of iteration and alliteration, e.g. suvimuktacittaiḥ suvimuktaprajñair,  ājñair ājāneyair, and kṛta-kṛtyaiḥ kṛta-karaṇīyair. If change the instrumental plural to the standard eastern Prakrit nominative singular ending, it emphasises the similarity e.g. 
kṣīṇāsrave niḥkleśe vaśībhūte suvimuktacitte suvimuktaprajñe ājñe ājāneye mahānāge kṛtakṛtye kṛtakaraṇīye apahṛtabhāre anuprāpte svakārthe  parikṣīṇabhavasaṃyojane samyagājñāsuvimukte
Tack a svāhā onto the end of this and it could be a dhāraṇī as found in most Mahāyāna texts after about the 4th Century. It seems we could say that dhāraṇī make use of literary techniques already in use, such as the tendency to iterate adjectives, to double up (or higher multiples) in order to create emphasis. 

The form of this statement, using the instrumental plural is rare in Pāḷi, occurring only in the Samaya Sutta, recorded twice: in the Dīgha Nikāya (DN ii.252) and the Saṃyutta Nikāya (SN i.25). 
Evaṃ me sutaṃ – ekaṃ samayaṃ bhagavā sakkesu viharati kapilavatthusmiṃ mahāvane mahatā bhikkhusaṅghena saddhiṃ pañcamattehi bhikkhusatehi sabbeheva arahantehi; 
Thus have I heard: one time the Bhagavan was dwelling with the Sakyas in a large grove in Kapilavattu, together with a large congregation of five hundred bhikkhus, all of them arahants. 
The Pāḷi number idiom is slightly different. Pāḷi says "five measures (pañcamatta) of one hundred bhikkhus (bhikkhusata)." However, most of the other adjectives are familiar in one way or another. 

Note that Lokakṣema's translation doesn't have a list of adjectives and thus looks a lot more like the opening of the Pāḷi Samaya Sutta. It suggests that the adjectives were added after the original composition. This highlights that the Mahāyāna texts were not part of a Canon (a collection of texts in fixed form), even when written but, instead, they were continuously added to over the centuries. 


Though this is a Mahāyāna text and the critical edition is based on relatively late manuscripts, with the oldest being from the 10th Century, there are no bodhisatvas present. This seems significant, because the presence of bodhisatvas seems to be an important feature of Mahāyāna.

However, when we look at the old translations we find a different story. Lokakṣema's translation from 179 AD, 《道行般若經》 (T224) says: 
[8.425.c06] 佛在羅閱祇 耆闍崛山 中,摩訶比丘 僧不可計,諸弟子 舍利弗 、須菩提等;摩訶薩菩薩無央數,彌勒菩薩 、文 殊師利菩薩  等。 
Once the Buddha was at Rājagṛha on the Vultures Peak with a huge congregation of monks, impossible to count, all of them disciples (弟子), including Śāriputra and Subhūti; and countless mahāsatva bodhisatvas, including Maitreya and Mañjuśrī. 
This kind of hyperbole is what we expect from a Mahāyāna sūtra. By the way, the word "disciples" (弟子) seems to reflect an underlying śrāvaka, though we expect arhat here, and is probably a mistake. It may reflect the idea that arhat was the goal of the śrāvakayāna, whereas the bodhisatva was the goal of the bodhisatvayāna

Unfortunately, the Gāndhārī manuscript (dated to 70 AD) is damaged and/or missing at this point. By the late 4th Century Kumārajīva's text (T227), while still considerably shorter than the later manuscripts, is completely conventional:
[537a25 - 26]  如是我聞。一時佛在王舍城耆闍崛山中。與大比丘僧千二百五十人倶皆是阿羅漢。
Thus have I heard: one time the Buddha was staying at Rāgagṛha on the Vulture's Peak, with a great congregation of 1250 bhikṣus, and all of them were arhats. 
Here the word arhat is transliterated as 阿羅漢, which in Middle Chinese was alahan. It may reflect a Prakrit original (cf Pāḷi arahant), but, by Kumārajīva's time, was fairly standard, though there were many variant "spellings" in Chinese, e.g.,  阿盧漢; 阿羅訶, 阿羅呵; 阿梨呵, 阿黎呵. This was later abbreviated to lohan or louhan (these Romanizations represent modern Mandarin pronunciations)

So the text is seemingly quite different at different times, assuming that the Chinese translations accurately reflect their source texts. Nor are the differences explicable as a linear evolution. Lokakṣema has bodhisatvas present, whereas others did not. So was Lokakṣema's text different or was he taking liberties? We don't know, because the Chinese did not preserve the Sanskrit originals of the Indic texts that they translated and, indeed, very few texts survive from that period, anywhere. The surviving Sanskrit manuscripts of Aṣṭa are on corypha palm leaves and date from about the 10th Century onwards. Note also that—especially in the earlier translations—the translators were working from single manuscripts that were most likely riddled with copying errors.

One of the things this brief comparison shows is that there is no single sūtra called Aṣṭasāhasrikā. I tried once before to bring out this with respect to the Vajracchedikā-prajñāpāramitā-sūtra (Is There Any Such Thing as 'a Text'? 20 December 2013). Similarly, Jonathan Silk has recently called into question the applicability of traditional philological techniques of identifying the "original" text, (Establishing / Interpreting / Translating: Is It Just That Easy?), which I used as the basis of my essay, The Heart Sutra and the Crisis in Buddhist Philology (30 June 2017).

Effectively, the text of any given sūtra is different for different people at different places and times. And this is an argument for the prominent position that translations play in Western Buddhism, despite the fact that, as far as the Prajñāpāramitā literature is concerned, there are no good translations, as yet. With the forthcoming book by Paul Harrison we may finally have a decent translation of the Vajracchedikā but, as my work has shown, we do not yet have a reliable source text from which to translate the Heart Sutra. An accurate translation of Aṣṭa would be a fine thing.

Speech Act Theory and Translation

In this short sub-essay I'll take a single example and examine how we should understand it in the light of speech act theory. 


Like most of the adjectives in our list, this is a compound. It combines kṣīṇā, 'cut off', and āsrava, 'inflow' ← ā√sru,* where √sru means "flow", but also "to flow out of, to gush forth". The addition of the ā- prefix to verbs involving direction, usually reverses the direction, i.e., suggests, as a first approximation, "inflow, influx". We could translate kṣīṇāsrava as "inflows cut off". We could play around with synonyms such as "influx", or more traditional attempts at interpretation such as "taints, corruptions" or the wildly interpretive "intoxicant biases" (Nyanatiloka's Buddhist Dictionary). But what does any of this mean? What precisely is "flowing in"? I've been a Buddhist for more than 20 years and I still draw a blank when I see these terms.
* the change from sru to srava is regular and expected. The root sru undergoes guṇasro, which is conceived as sr(ău). We add a to create an actions noun giving, sr(ău)a. This creates an internal sandhi which resolves as srava

Can we do better? I think we can. Many writers, not least Richard Gombrich, have pointed out that āsrava (Pāḷi assava) is a term taken from Jainism. In that context, actions (karma) cause an inflow (āsrava) of "matter" or "dust" (the Sanskrit word here is unclear) that sticks to the soul (jīva) and keeps it in saṃsāra. But the word āsrava might also be translated as "channel for acquisition of karma", i.e. the Jains see āsrava both as the flow and channel for flow of karma. Jains believe that suffering removes (nirjarā) karma from the jīva, thus liberates it from saṃsāra. Another way of thinking about it in Jain terms, is that āsrava is a way that the consequences of karma impinge on the person. An āsrava is a karma conduit.

So what might it mean to cut off the inflowing or channels for inflowing of karma? It means that the person concerned is not going to suffer the consequences of any past actions because the flow of karma has ceased. Nor will they create any new karma (conceived of as consequential actions that will result in rebirth) that might prevent from being finally liberated from rebirth. Someone who is kṣīṇāsrava does not create new karma and has no old karma waiting to manifest.

In other words, to be kṣīṇāsrava is to be free of karma: free in the sense of not subject to any consequences of past actions; and free in the sense of not having to worry about what they do because they can no longer do actions that result in rebirth. Of course, Aṅgulimālā might be considered an exception, since he still has to suffer from past karma but he is still not making any new karma and won't be reborn.

Dictionaries are helpful tools, but to really understand a language one has to think beyond the dictionary, to see words in their cultural context. This is particularly important for Sanskrit which is used in a wide range of distinct contexts which may use the same words very differently. Similarly, etymology can tell us what the parts of a compound originally meant, but not how the individual words are used at a particular place or time, let alone the meaning of a compound.

Speech Acts and Translation

The theory of speech acts was developed in the USA in the 20th Century, largely by two men, John L. Austin and John Searle. Their analysis was part of a movement away from seeing language in merely semantic terms by applying principles deriving from pragmatism. Semanticists ask "What does language mean?", while pragmatists ask "What does language do?" Austin and Searle mapped out the kinds of things we do with language. They treated spoken sentences as "speech acts". In this pragmatic view, semantics must be subordinated to pragmatics, if only because of irony, i.e., when we say one thing, but mean something else. If I say "I love your new haircut", a semanticist can only analyse the words themselves and conclude that I do love your new haircut. A pragmatists also listens to my tone of voice and watches my face as I say it, and they might realise I don't like your new haircut, at all, and that I am mocking you. Semantics cannot cope with sarcasm or irony, because the same words are used as if I was sincere. Pragmatics doesn't just add a dimension to semantics, but shows that "sense" occurs in the context that goes well beyond word choices.

This is one of the problems of working with written texts. Written texts have no eyebrows or tone; we cannot tell how the author intended us to read their words, whether as literal truth, informative myth, entertaining legend, or some other interpretation. To take a real example, one of us might read a Buddhist text such as the Pāḷi Tevijjā Sutta as a parody, which changes its meaning entirely; while another dismisses the idea that Buddhists could portray the Buddha as having a sense of humour as projection, and argues for a more literal reading. 

Speech act theory suggests that we can understand a communication in terms of what was said, what was meant, and what was understood. The technical terms for these are locution, illocution, and perlocution (and be aware that the technical definitions of all of these terms are a lot more sophisticated than how I have boiled it down here). The case of kṣīṇāsrava illustrates this very nicely. Obviously kṣīṇāsrava is a locution; i.e., it is a declaration about an arhat that helps to establish legitimacy and authority on several levels. It establishes the status of people present (who subsequently participate in the dialogues); it helps to establish the status of arhats as a class of people; and because the Buddha is surrounded by a large number of them, his authority and legitimacy is also established. Buddhists are obsessed with these political issues of status, legitimacy, and authority from the earliest records of their thinking.

Conze's attempt to translate kṣīṇāsrava is "their outflows dried up". This is a perlocution for Conze, it represents what he has understood, but it is also a new locution, something he is declaring. This is a feature of translation. The author composes a text and perhaps writes it down as a document. The translator reads the text, tries to understand it in the source language, then they compose a text in the target language which they hope will have the same illocutionary force. A translation is always a new locution. It's never the same locution.

Conze wants us to understand this thing about arhats: "their outflows dried up". This is similar to how Kumārajīva's translation team understood term, since they translate 諸漏已盡 "all leaks completely exhausted"

So, contrary to the dictionaries and Jain usage, which clearly suggest that ā√sru means "in-flow", both Conze and Kumārajīva understand "outflow". One of the things about borrowed terms is that they are thoroughly decontextualised, so the knowledge that this word āsrava originally came from a Jain context was lost and not recovered until after Conze was writing.

I'm not sure about other readers, but when I think of "their outflows dried up", I think of a leaky container, particularly a human body leaking fluid from various orifices (the Chinese 漏 "leak" only reinforces this!). I have a cold at present with a runny nose and sore throat. I have a lot of extraneous outflows that I wish would dry up. So, what Conze seems to be saying, on face value, is that the leaking body fluids from the arhats have dried up. It certainly does not conjure any sense of what the term means in practice, or convey anything to me that I intuitively find meaningful.

By looking at how the word was used in its original context we have deduced that the illocutionary force of kṣīṇāsrava is that arhats are free of karma. And we can use this conclusion "free of karma" as our translation. To my mind, as a Buddhist who has explored Buddhist karma doctrines in some depth, this makes a great deal of sense; whereas, "their outflows dried up" doesn't communicate anything relevant to me (and produces a load of irrelevant associations). 

What I do not control is how the reader will understand this - the perlocution. For example there are many different ways of thinking about karma and I can't be sure that all of them will fit my conception. Some might take this to mean that the arhats are free from moral restraint, for example, and able to act immorally with impunity. Though this would not be what I intended to say, not my illocution, it might be a perlocution for the reader.


This "essay" is really just a collection of notes with no overall theme except that they arise out of reading Aṣṭa and thinking about how to translate it. However, one of the major themes I've explored over the years is just how difficult translating really is. I've tried to convey how little confidence we should have in translated documents as representative of the author's intentions. The very idea of "the text" is much more fluid in our Buddhist milieu that it is for, say, Christians.

On one hand, this ought to legitimate translations. We know that in most Asian countries, Indic texts were abandoned quickly once translations became widely available. Indic texts were not generally preserved in the long term, but remained theoretically important. On the other hand, the whole point of the story of Xuanzang going to India (and he was only one of many such pilgrims from China and Tibet) was that he felt the translations of Kumārajīva and others were not sufficiently clear or comprehensible. It is a little ironic then, that while his translations were generally considered superior by scholars, none of Xuanzang's translations ever become popular or replaced those of Kumārajīva in the popular imagination.

Translations are seldom really about translating individual words. The basic unit of meaning is the sentence. That is to say, it is how words are used in sentences that convey the authors' intentions. A list of adjectives is a special case. But the single word example of kṣīṇāsrava does seem to highlight many of the problems with English translations of Buddhist texts. We are not there yet in terms of fully migrating to English as a medium for communicating the Dharma. We are still struggling with Buddhist Hybrid English and with incomprehensible word for word translations.

One of the problems we seem to have is that few scholars are going over the ground and bringing the light of new discoveries to familiar texts. I think this is partly a problem of how such work is funded now. Everyone is busy working on "new" areas and previously untranslated texts (which seem to become more obscure with each passing year).

Another text I've been looking at recently is Lewis Lancaster's unpublished dissertation on the Chinese translations of Aṣṭa which compares the versions - and delineates three periods of the text. I hope to write up some notes on this as well, because it is apparent that this 50 year old document has not had the kind of influence on the popular imagination of Prajñāpāramitā that it should have. It is  a great pity that in the 50 years since Lancaster's doctoral dissertation no one seems to have followed up on the doors that it opened. Certainly, no new translation of Aṣṭa has appeared to replace the faulty one produced by Conze. One bright spot is Seishi Karashima's glossary of Lokakṣema's Aṣṭa, which ought to make a new comparative translation much easier. 


Related Posts with Thumbnails