Expanding the peyyālas: where to start

[arkiuat]

Continuing the discussion from How to find text in ellipses:

After reflecting some more on that that ten-year-old post, I took a closer look at DN 3 and DN 4.

The description of the gradual training in DN 2 begins at DN 2:40.1 and continues through to some point before DN2:99.1

What I completely missed on my first reading of DN 3 is that this whole thing is repeated from DN 3:2.2.2 to… some point. I only went and looked for it because I’d seen Bhante @sujato mention it in writing more than once or twice.

Likewise in DN 4 it begins from DN4:23.2. I missed this on first reading too, until I went back to seek it out. But even now I’m not sure where exactly the repeated section ends, in part because there is some slight variation in the exact wording of the ending, but mostly just because the abbreviation is so aggressive.

I’m sure I’ll find the same thing in DN 5 through, what, DN 13? But my uncertainty about where and how the expanded peyyala should end shows that I certainly don’t have the skills yet to start on this project, even for this low-hanging fruit of peyyala-expansion.

Editing to add: since I kept this in Q&A, I may as well make my question explicit: where exactly does the repeated text end, in each of these three suttas, DN 2 through 4?

[Snowbird]

@arkiuat

shows that I certainly don’t have the skills yet to start on this project

More likely it shows how ambiguous certain peyyalas can be. Some, of course, are clear. But others are not. This is probably a good explanation of why no one has ever seriously undertaken this project.

It becomes even trickier when we see that sometimes there are (small?) grammatical changes that need to happen to complete the elisions. Or if the interlocutor changes.

Perhaps the place to start would be to come up with a system of rating the levels or types of ambiguity. That way whoever uses or improves the data will have some systematic way to evaluate it. Or perhaps offering alternative ways of reconstructing the text.

[arkiuat]

@Snowbird

More likely it shows how ambiguous certain peyyalas can be. Some, of course, are clear. But others are not. This is probably a good explanation of why no one has ever seriously undertaken this project.

Well, if it’s really the case that no one else is interested, I guess I will have to start plugging away and see what I can come up with. I found it pretty disturbing when I realized that in my first reading of DN 2 and DN 3, I basically just read two framing stories with the main point elided. They were very interesting framing stories! But still.

I find it even more disturbing to think that probably lots of newbs have had this experience without ever realizing that’s what has happened.

@Snowbird

It becomes even trickier when we see that sometimes there are (small?) grammatical changes that need to happen to complete the elisions. Or if the interlocutor changes.

Yes, I noticed that in any expansion of DN 3 and DN 4, where the repeated section in DN 2 has vocative mahārāja, in DN 3 you need ambaṭṭha, and in DN 4 brāhmaṇa. That’s a substitution that needs to be made throughout.

@Snowbird

Perhaps the place to start would be to come up with a system of rating the levels or types of ambiguity. That way whoever uses or improves the data will have some systematic way to evaluate it. Or perhaps offering alternative ways of reconstructing the text.

Now there’s an idea for me to chew on for a while. Thanks!

I’m total novice at Pali, but at least I had four years of classical Latin in high school, so I know what an ablative is. (Latin has nothing like Pali’s multifarious ways of forming compound words though!)

One more thing I want to add: way back when I was a practicing Beanite Quaker in a bible study group, one of the striking experiences was when those of us who hadn’t already noticed it realized that almost a quarter of Exodus is entirely repeated with grammatical alteration: first god gives the instructions for constructing the tabernacle and its paraphernalia in extravagant detail, and then those instructions are carried out verbatim, except in the indicative instead of the imperative mood. Only talking about the English translation here.

Even a Quaker can acknowledge that the gradual training is a much bigger deal than the ancient tabernacle and its paraphernalia. But I’ve never seen that repetition elided, because the unabridged Tanakh is pretty short compared to the Tipitaka.

[Snowbird]

When I was mainly reading from hard/paper copies I had photocopied out the full sila section in Walsh’s translation and used it to read it “into” the suttas where it was missing. I think then for me it was mostly just for learning purposes.

It might be worth defining the purpose of reconstructing the peyyalas. I can think of a few purposes:

1. For recitation in Pali or translation, either alone or in a group. The SN 22.59 Anattalakkhaṇasutta and SN 35.28 Ādittasutta are good candidates for this.
2. For silent reading in translation to have the learning benefits of actually reading the repetitions
3. For determining a “more accurate” count of word usage.
4. For some other “deeper insights” that might come from experiencing the suttas “as they were intended”

One concern I have with reconstructing them in the Pali is that it will give a false sense of confidence in it’s accuracy. And as may be clear from my other posts, I don’t think that #3 above is a useful exercise.

[arkiuat]

@Snowbird

It might be worth defining the purpose of reconstructing the peyyalas. I can think of a few purposes:

1. For recitation in Pali or translation, either alone or in a group. The SN 22.59 Anattalakkhaṇasutta and SN 35.28 Ādittasutta are good candidates for this.
2. For silent reading in translation to have the learning benefits of actually reading the repetitions
3. For determining a “more accurate” count of word usage.
4. For some other “deeper insights” that might come from experiencing the suttas “as they were intended”

I certainly had the first two in mind (and sparing the uninformed newb the experience of missing the main point on their first reading, which I suppose falls under the second purpose).

I would qualify the third: I agree with you that counts of individual words aren’t useful for much, but I am interested in something like the research program expressed in Mark Allon’s Style and Function (1997), which is about repeated pericopes and long formulaic phrases, not word counts, and I’d substitute this concern as the third purpose. Any future effort to carry that research forward would benefit from expansion, as there are many such shorter pericopes and formulaic components hidden inside this very long pericope of the gradual training. The four-jhānas pericope in one of its forms and a whole bunch of precepts are in there, for instance.

I take your fourth point to refer to the reasons why the original reciters repeated these long texts in different suttas (other than the obvious mnemonic and checksum value of sheer repetition), but the reasons that applied for eliding these texts in the first place no longer apply to the medium in which we are currently interacting with them (for the most part: I still read them on paper too), and the elisions have introduced disadvantages that are no longer necessary. This would be reason enough alone, for me.

@Snowbird

One concern I have with reconstructing them in the Pali is that it will give a false sense of confidence in it’s accuracy.

I share this concern, which is why I was reluctant at first to do anything more than offer help. It’s also why your earlier suggestion about coming up with “a system of rating the levels or types of ambiguity” is so excellent! I might limit my project to just my private English versions: it could be extended to those translations released with permissions to do such things, without touching the Pali at all.

A halfway stage would be to adopt a less aggressive elision scheme than the one we have inherited from the ancient texts, one that would give

• a clear description of exactly what repeated text we’re skipping over,
• a clear pointer to where to find the full spelling-out of the elided text (which sections of DN 2 exactly, in this case) and
• a less ambiguous indication of where exactly the repeated text ends (both in the abridged text and in the unelided source).

That would go a very long way to addressing my concerns without introducing as much of a false sense of confidence.

[josephzizys]

@arkiuat

What I completely missed on my first reading of DN 3 is that this whole thing is repeated from DN3.2.2.2 to… some point. I only went and looked for it because I’d seen Bhante @sujato mention it in writing more than once or twice.

This is a hugely important observation for many reasons.

In particular it means we have to realize that to the reciters of DN, the patipada as at DN2:40.1 is the core teaching of the buddha at ALL of DN2, DN3, DN4, DN5, DN8, DN9, DN10, DN11 and DN12. (DN13 uses the same trope, but removes everything form jhana on and replaces it with the Brahmaviharas.

Secondly that fully ten suttas, in both the Pali and the Chinese, in the first collection of both canons, agree that this trope is fundamental to the teaching, is manifestly significant.

Third we should see that because of the peyyālas we must accept that the four jahna teaching, the five hiunderence teaching, the minds of others, the mind made body, past lives, etc etc are ALL repeated ten times, universally, at the start of both the Pali and Chinese canons.

@Snowbird

More likely it shows how ambiguous certain peyyalas can be. Some, of course, are clear. But others are not. This is probably a good explanation of why no one has ever seriously undertaken this project.

Someone really should. One starting point would be to simply locate them, which would at least indicate which cases we have to check, presumably by tracing the pe back to a previous occurrence of the full trope, so in DN3’s case to DN2, and so on.

@arkiuat

I found it pretty disturbing when I realized that in my first reading of DN 2 and DN 3, I basically just read two framing stories with the main point elided. They were very interesting framing stories! But still.

It is disturbing! and it does distort peoples understandings of exactly what is said and how often in the suttas.

@Snowbird

It might be worth defining the purpose of reconstructing the peyyalas. I can think of a few purposes:

One purpose might be to get a more accurate count of certain tropes, for example jhana where it disappears because of a pe referring to DN2.

Anyway, I will continue to follow this thread with interest and wish you all the best of luck with the investigation.

One way to start, just reflecting my own bias and interests, would be to take the long suttas first, as then you have a case limited to the hundreds as opposed to the SN AN cases where you have thousands and thousands of documents.

Even better you could start with the DN MN documents that have direct parallels in the same volumes in the agamas, this reduces the set of documents to 120 and is probably doable by hand.

in each case ape appears you could replace it with the original occurrence i.e at DN3 “pe…” you could write (DN2:40.1 to DN2:98.7) thus directly referring to the original text in a link.

[arkiuat]

@josephzizys

One starting point would be to simply locate them,

Definitely a worthwhile project all on its own!

[josephzizys]

just as one example of how the pe presents problems;

the first pe in SN is SN2.17 at the end of the sutta it says

Idamavoca …pe…

this is translated by @sujato as “That is what the Buddha said. …”

but the Buddha isn’t mentioned, just the “thus it was said”.

So we might think, lets go back through SN up to this point to find the actual passage being shortened.

There are only 4 suttas prior to SN2.17 that use the word “Idamavoca”

SN1.1 “Idamavoca sā devatā.”
SN1.31 “Idamavoca bhagavā.”
SN2.1 “Idamavoca kassapo devaputto;”
SN2.8 “Idamavoca tāyano devaputto;”

So we take SN1.31 as the source of the pe, and the ONLY word missing is “bhagavā”.

A few points.

First, this is a wierd use of pe, it uses one word to indicate one other word. why bother? why not just say “bhagava”?

Second, this is the only pe before SN3.2, which then uses the same pe, i.e just as a replacement for bhagava.

we don’t get an actual pe representing more than one word until SN3.3, where it refers to the previous line in the same sutta, i.e

aḍḍhā mahaddhanā mahābhogā pahūtajātarūparajatā pahūtavittūpakaraṇā pahūtadhanadhaññā, tesampi jātānaṁ natthi aññatra jarāmaraṇā.

SN3.4 is the same as SN2.17

SN3.5 refers to SN3.4

that only leaves about 2292 cases in SN.

[arkiuat]

Yes, this sort of thing is why I refer to Sīlakhandhavagga in DN as low-hanging fruit: most benefit for least effort, relatively speaking.

I’m not even going to start looking at the peyyālas in SN and AN until after I’ve worked up DN2-12. However, if anyone wants to catalog them while I’m working on Sīlakhandhavagga, that would be great!

[Snowbird]

@josephzizys

So we might think, lets go back through SN up to this point to find the actual passage being shortened.

Indeed. In the Daily Sutta Emails I feel obligated to try and reconstruct them for SN1 and SN2 especially. But they are much trickier than one would think. Obviously it’s low stakes. That’s why I just go ahead and do it. ¯\_(ツ)_/¯ It’s a problem throughout the SN in places one wouldn’t expect.

[Dogen]

Pāli Audio website has (afaik) all four Nikayas recited out with the peyyālas expanded. Their recitation is based on (mostly) Bhante Sujato’s translations.

@Snowbird

• For recitation in Pali or translation, either alone or in a group. The SN 22.59 Anattalakkhaṇasutta and SN 35.28 Ādittasutta are good candidates for this.
• For silent reading in translation to have the learning benefits of actually reading the repetitions
• For some other “deeper insights” that might come from experiencing the suttas “as they were intended”

Someone could ask them if they’d be willing to share the text files (which I just did). Also, they might even be useful additions to the website for the reasons above.

[Snowbird]

@Paliaudio.com’s owner is here, too. It would be good to get their thoughts.

[arkiuat]

@Dogen

Someone could ask them if they’d be willing to share the text files (which I just did).

@Snowbird

@Paliaudio.com’s owner is here, too. It would be good to get their thoughts.

Thanks to both! This reminds me that I’ve already expanded some peyyāla before, in a private copy of Dhammacak SN 56.11, and of course I relied on chanting texts.

Incidentally, progressing a little further with DN 3, I ran into the issue of section numbering for the “restored” text. The text that is repeated begins at DN 2:40.1, which corresponds to DN 3:2.2.2.

The text from DN 2 is actually repeated up through DN 3:2.2.7, and then the first peyyāla is between DN 3:2.2.7 and DN 3:2.2.8.

My first thought is to number sections of text within the expansion as DN 3:2.2.7.1, 2.2.7.2, and so forth. Is this going to break any part of the existing system, and if so, what alternatives are there?

And that brings us to the next issue: in DN 2, the section numbering resets at the beginning of the Shorter Section on Ethics. And there are differences in the depth of sublevels that show up before then (sections in DN 2 that only have 2-level numbering correspond to sections in DN 3 that have 3-level numbering).

And the third issue: at what I’m tentatively labelling DN 3:2.7.6 (corresponding to DN 2:43.1) there is not only an interlocutor substitution (“Ambaṭṭha” instead of “great king”), there’s also a content substitution: where in DN 2 the refrain is “This reflects on their ethics,” in DN 3 it’s “This reflects on their conduct,” and this line is the first place where that makes a difference. (And I checked: Bhante is translating two different Pali words here.)

I’m going to keep working on this for a while and we’ll see what I come up with.

[srkris]

They are nearly all artificial expansions. There is no need to say something in 1000 words when 200 words would suffice to give the same idea. Even the suttas as they currently stand contain plenty of redundancies.

These redundancies served a real purpose in the first few generations of Buddhism (i.e. when writing was a novel art, and copyists - who were not necessarily all monks - made many mistakes) - which is to ensure that if there was textual corruption in the process of copying manuscripts over and over, they at least had the same text written out in full elsewhere that they could refer to and fix mistakes. Pretty soon the writers of the manuscripts must have seen that to write the same thing over and over was extremely time consuming and pointless so they began to take more care in writing repetitive texts just once correctly, and shorten the repetitions with the letter “pe(yyāla)” i.e. “repetition” in later occurences. However it was not consistently enforced so some writers still left behind repetitions while others removed repeated phrases/paragraphs.

However a lot of common templated phrases and paragraphs are repeated across suttas (which is the way these suttas were constructed by their original writers in the first place), so even if they are abbreviated, there are still plenty of redundancies left behind.

I dont see it as a worthwhile task to expand the redundant texts whether for reading/reciting or for anything else.

[arkiuat]

@srkris

I dont see it as a worthwhile task to expand the redundant texts whether for reading/reciting or for anything else.

That’s as may be, but there was my other suggestion:

@arkiuat

A halfway stage would be to adopt a less aggressive elision scheme than the one we have inherited from the ancient texts, one that would give

• a clear description of exactly what repeated text we’re skipping over,

• a clear pointer to where to find the full spelling-out of the elided text (which sections of DN 2 exactly, in this case) and

• a less ambiguous indication of where exactly the repeated text ends (both in the abridged text and in the unelided source).

That would go a very long way to addressing my concerns

What do you think about this idea?

[srkris]

@arkiuat

a clear pointer to where to find the full spelling-out of the elided text (which sections of DN 2 exactly, in this case) and

It is sufficiently clear for those who can read the Pali.

For example in DN2, the first instance of “pe…” in [3.2. Makkhaligosālavāda] begins with “‘yathā nu kho imāni, bho gosāla, puthusippāyatanāni …” then has “pe…” and then “sakkā nu kho, bho gosāla, evameva diṭṭheva dhamme sandiṭṭhikaṁ sāmaññaphalaṁ paññapetun’ti?”

So if you look in the immediately prior paragraph where the words puthusippāyatanāni and sakkā nu kho — the pe… in 3.2 abbreviates everything between those words ( puthusippāyatanāni and sakkā nu kho) that are present in 3.1.

The abbreviated text in bold below:

‘yathā nu kho imāni, bho kassapa, puthusippāyatanāni,
seyyathidaṁ—hatthārohā assārohā rathikā dhanuggahā celakā calakā piṇḍadāyakā uggā rājaputtā pakkhandino mahānāgā sūrā cammayodhino dāsikaputtā āḷārikā kappakā nhāpakā sūdā mālakārā rajakā pesakārā naḷakārā kumbhakārā gaṇakā muddikā, yāni vā panaññānipi evaṅgatāni puthusippāyatanāni, te diṭṭheva dhamme sandiṭṭhikaṁ sippaphalaṁ upajīvanti;
te tena attānaṁ sukhenti pīṇenti, mātāpitaro sukhenti pīṇenti, puttadāraṁ sukhenti pīṇenti, mittāmacce sukhenti pīṇenti, samaṇabrāhmaṇesu uddhaggikaṁ dakkhiṇaṁ patiṭṭhapenti sovaggikaṁ sukhavipākaṁ saggasaṁvattanikaṁ.
Sakkā nu kho, bho kassapa, evameva diṭṭheva dhamme sandiṭṭhikaṁ sāmaññaphalaṁ paññapetun’ti?

[arkiuat]

@srkris

It is sufficiently clear for those who can read the Pali.

I’m not sure what point you are trying to make here.

Are you saying that you have no compassion for the uninformed newb who reads DN 3 in translation, for the first time, and tries to understand it, only to find out later that they only read the framing story and missed the main point?

Also, you didn’t answer my question, but I guess your reply means that you don’t like my halfway-stage suggestion either.

[stephen]

@arkiuat

reads DN 3 in translation, for the first time, and tries to understand it, only to find out later that they only read the framing story and missed the main point

I think he’s saying that if the entire sutta is read, one can deduce the missing bits, as they are present in a preceding section.

[srkris]

@stephen

one can deduce the missing bits, as they are present in a preceding section

That’s right, and to know that, one doesn’t even need to know Pali, one just needs to take the few words before and after the “pe…” and compare those words with the preceding non-abbreviated sections to know what’s abbreviated.

Perhaps the OP wants a different (unabbreviated) format for the translation rather than the original.