In a previous post we discussed issues that have been raised in an essay by Tse-fu Kuan, Rethinking Non-self: A New Perspective from the Ekottarika Āgama. One of the points they raise deals with the well-known passage in the Anattalakkhana Sutta where the self is discussed in terms of control. Essentially they argue that the Pali text has the idea that each of the 5 aggregates is not subject to control, but this contradicts the Upanishadic idea of a self, which that the self is the wielder of control.
In my earlier remarks, I pointed out that there is no real philosophical problem here: a controller requires a controlled, and the absence of a controlled shows there is no controller. Further, I identified a number of other weaknesses in the argument.
Nevertheless, it is an interesting point, and the differences between the Pali and Chinese texts in this are noteworthy. So let’s look a little closer. Of course, I know virtually no Chinese, so any corrections would be welcome. For this essay we just take the first case, that of rūpa, understanding that it applies to all five aggregates.
We are primarily concerned here whether there may be a simple textual reason for the difference between the versions. Now, it is possible that the difference is not textual but philosophical. But I find this very implausible. If the early schools really did diverge in this aspect of the qualities of a supposed self, this would be expressed in the abhidhammas, rather than an obscure wording of a sutta text. No-one really has any horses in this race, so it seems hard to explain the divergence as anything other than a transmission error.
The Pali text of SN 22.59 says this. I bold a few subtleties of the text, which I will discuss below.
Rūpaṃ, bhikkhave, anattā.
“Mendicants, form is not-self.
Rūpañca hidaṃ, bhikkhave, attā abhavissa, nayidaṃ rūpaṃ ābādhāya saṃvatteyya,
For if form was self, it wouldn’t lead to affliction.
labbhetha ca rūpe: ‘evaṃ me rūpaṃ hotu, evaṃ me rūpaṃ mā ahosī’ti.
And you could compel form: ‘May my form be like this! May it not be like that!’
Yasmā ca kho, bhikkhave, rūpaṃ anattā, tasmā rūpaṃ ābādhāya saṃvattati,
But because form is not-self, it leads to affliction.
na ca labbhati rūpe: ‘evaṃ me rūpaṃ hotu, evaṃ me rūpaṃ mā ahosī’ti.
And you can’t compel form: ‘May my form be like this! May it not be like that!’
The main purport of the bolded portions is to highlight the way that the Pali text establishes the logical connections between the ideas.
- The (partially elided ) hi flows the text on from the previous statement, and can usually be rendered “for”.
- abhavissa is the third person conditional for bhavati, i.e. “if it were”. This is a somewhat unusual form, although not hugely so.
- labbhetha is third person passive optative middle “it could be had of …”. And yes, this is pretty much as obscure in Pali as it sounds. The only other occurrence in the EBTs I can find is in a verse at Thag 7.2#3.
- the co-ordinated yasmā … tasmā … has the sense “Since … therefore …”.
- Note also the use of ca: here, it identifies two parallel clauses, showing that the sentence is to be divided up as shown.
So this passage includes are fairly large number of quite subtle logical connections and unusual forms. Thanks to the good work by Pali commentators, grammarians, and translators, we can sort this out without too much difficulty. But it is easy to see how these might get lost or misconstrued in the passage through dialects, still more so when translated to Chinese, a language that handles such things very differently.
Now, of course, we are not (I hope!) Pali fundamentalists here, and it is possible that these issues are an artifact of the Pali text. However, this seems to me unlikely. Understood properly, the text is very precise, clear, and idiomatic, and couches a subtle point in appropriately subtle language. It doesn’t convey to me any of the hallmarks of a corrupt or confused passage.
So let’s see what the Chinese translations made of this. Now, we have two renderings of this in the main parallel of the Samyutta, at SA 33 and SA 34. These have minor differences. I presume these are just a slight looseness in translation. In addition to Kuan’s translation in his essay, these are both translated on SC, so check those out. I won’t use any of those translations, but will line the Pali text next to the Chinese and see what happens. In addition, there is an independent translation at T102. I would guess there are likely to be other parallels, too, especially in the Vinayas. Anyway, this is what we have. The first couple of phrases are straightforward.
- Rūpaṃ, bhikkhave, anattā.
- Form is not self.
- SA 33: 色非是我
- SA 34: 色非有我
- T 102: 色不是我
- Rūpañca hidaṃ, bhikkhave, attā abhavissa, nayidaṃ rūpaṃ ābādhāya saṃvatteyya,
- For if form was self, it wouldn’t lead to affliction.
- SA 33: 若色是我者,不應於色病、苦生
- SA 34: 若色有我者,於色不應病、苦生
- T 102: 若是我者,色不應病及受苦惱
It’s in the next bit that differences appear, which I highlight. Since T 102 is quite different I will treat that separately.
- labbhetha ca rūpe: ‘evaṃ me rūpaṃ hotu, evaṃ me rūpaṃ mā ahosī’ti.
- And you could compel form: ‘May my form be like this! May it not be like that!’
- SA 33: 亦不應於 色欲令如是、不令如是
- Also one should not regarding form wish to make it be like this or like that.
- SA 34: 亦不得於 色欲令如是、不令如是
- Also one does not have regarding form the wish to make it be like this or like that.
So there are some interesting variations here. Notably, while all three translations handle the opening statements, which are more grammatically obvious, the same, they diverge here, both from the Pali and from each other.
SA 33 and SA 34 introduce an extra 不, a negative which is absent from the Pali.
SA 33 uses 應 “should”, where SA 34 has 得 “have”. Here, 得 clearly represents the Pali labbhetha and is the preferred reading. Indeed, SA 33 uses 得 below. But the difference, I think, hints at the struggle with an oblique idiom.
Both SA 33 and SA 34 use 欲, a term for “want, desire”, which does not have a literal parallel in the Pali. Probably this also arises from the same need to express the idiom.
From here they continue as expected, reversing the negative in the final statement, and thus proving that the difference from the Pali is not incidental.
- Yasmā ca kho, bhikkhave, rūpaṃ anattā, tasmā rūpaṃ ābādhāya saṃvattati,
- But because form is not-self, it leads to affliction.
- SA 33: 以色無我故,於色有病、有苦生
- SA 34: 以色無我故,於色有病、有苦生
- na ca labbhati rūpe: ‘evaṃ me rūpaṃ hotu, evaṃ me rūpaṃ mā ahosī’ti.
- And you can’t compel form: ‘May my form be like this! May it not be like that!’
- SA 33: 亦得於色欲令如是、不令如是
- SA 34: 亦得於色欲令如是、不令如是
- Also [one] does have regarding form the wish to make it be like this or be like that.
The real question here is the handling of the negatives. Now, notice one thing. In the Pali text, the complex sentence begins with a negative (nayidaṁ). In some cases, such a negative construction might be distributed across both clauses, and this may in certain instances create ambiguities. Negatives are hard! However, the repeated use of ca shows clearly that the clauses are distinct, and the negative should be applied to the first clause. So one possible source of the confusion would that the Chinese translation, or its source, improperly distributed this negative, letting it leak outside its scope.
This explanation, however, cannot apply when the argument is reversed. There, the Pali clearly does not have a negative in the first clause, and does have it in the second: there is no ambiguity. So we have two possible situations:
- The Pali removed the negative from the first clause and added it to the second.
- The Chinese improperly distributed the first negative and removed the second to avoid inconsistency.
So it’s easier to see how the Chinese could have arisen by a simple textual mistake. This is, of course, in addition to the fact that the Chinese is at a further remove from the original, and as already seen, struggles to represent the grammar precisely and consistently.
T 102 takes a very different approach, and appears to dispense with the grammatical problem altogether. I suspect the translator simply chose a more direct and simpler way of expression. Rather than speaking of having or wanting form to be like this or that, it has a direct speech passage, conveying the same sentiment. In any case, this does not support the use of the extra negative. This is made explicit in the next section, which again does not closely parallel the other texts, but rather appears to paraphrase the text:
- 我欲如是色,我不欲如是色
- I want form like this, I do not want form like this
- 既不如是,隨情所欲,是故當知,色不是我
- Since it does not accord with one’s wishes, for this reason it should be understood that form is not self.
Thus it seems on close inspection that there is no compelling case to assume a corruption in the Pali tradition in this passage. To recap:
- There is no philosophical problem with the problem of controller/controlled, as they are two sides of the same coin, i.e. control.
- The Pali passage is perfectly coherent.
- The subtleties of the Indic passage appear to have created a number of problems for the Chinese translators.
- It’s easier to explain the difference as arising in the Chinese (or its source) rather than the Pali.
- The independent Chinese translation supports the Pali.
Lacking a Sanskrit parallel for direct comparison, it may not be possible to draw any further conclusions from these passages. Of course, I have not considered all the Chinese passages here, and these conclusions, tenuous as they are, might well be refuted by other contexts.