Dhamma transmission and the ellipsis

From the oral transmission of the Pali Canon via Pali Text Society and Bhante Sujato we have :pray: :

They perceive earth as earth.
But then they conceive earth, they conceive regarding earth, they conceive as earth, they conceive that ‘earth is mine’, they take pleasure in earth.

And also (note the ellipses) we have:

They perceive the thought as the thought. But then they conceive the thought …

ASSERTION: Indeed, per Bhante Sujato, MN1 states:

They perceive the thought as the thought. But then they conceive the thought, they conceive regarding the thought, they conceive as the thought, they conceive that ‘the thought is mine’, they take pleasure in the thought.

Oddly, I feel wary typing the bold assertion, since it assumes a bullet-proof, incontestable application of ellipses (…). Yet the asserted statement itself rings true to me without qualm.

How very peculiar. I seem to be wary of the ellipsis. :thinking:

Am I suffering needlessly? Have we short-circuited the oral transmission?


That’s pretty much a given!

No, because we are not working from the oral transmission. We are working from the manuscript transmission, which already includes ellipses. These are even marked explicitly in Chinese translations, so they go back over a thousand years, and probably 2000.


Thank you, Bhante.

The context is that I have slowly been leaning towards the conviction that expanding ellipsis may not be so hard. I simply replaced “earth” with “the thought” in the above to arrive at the “expanded MN1”. This seems to be a fairly simple and mechanical substitution. We could, for example, offer a simple REST API for expanding suttas.

Unless you have concerns, I may end up wandering into implementing ellipsis expansion. I personally simply like listening to expanded suttas. I find the ellipses jarringly brief in recitation and would want the mind tracing out the entire sutta template for each individual replacement. The expansion would manifest as a user preference with absolutely no change to the core translation, which would retain all ellipses.


You are of course most welcome to try. But you can’t infer from one or two basic instances to a general principle. What you are trying is as hard as heck, and no scholar would attempt such a thing, or rely on the results. You’ll end up wasting a hell of a lot of time.

The only way to do this meaningfully is to take it on a case by case basis and lay it out by hand, which is what the readers are doing.

I would strongly encourage you to focus on getting the basic functionality down, get it active, get it working, get users using it, get feedback, improve it, and make it something that is actually going to be of value, and roll it out on multiple platforms. Then, when you have tie to sit back and relax, if you want, try figuring out an expansion algorithm as a hobby. But right now it will bog you down and cripple the whole project.


I have indeed been spoiled by the singular example of MN1. It keeps popping up in my daily study and practice. I do completely agree that guaranteeing exhaustive and correct expansion across all the suttas is a direction, and definitely not a near goal. Exhaustive and complete expansion is perhaps not practically or usefully achievable.

To support an expansion as a background project, we would need a way to indicate which suttas are expandable. For example, a simple whitelist of expandable suttas reviewed and approved by yourself over a lifetime would permit incremental progress. Each sutta would have its own “expansion unit test” so that approval need only be given once. These unit tests would preserve the approved integrity of all existing expansions if the expansion algorithm changes. In this manner, we could tackle one sutta at a time according to need or interest.

The main concern of this post was to establish whether such expansion would be a misstatement of the Dhamma or not. With your review and approval of each expansion, they would indeed be approved expansions. The whitelist would ensure that expansion could only be provided with the express approval of each translator for their individual sutta translations.

The other reason for this post was to gauge current overall interest in sutta expansion itself. Judging from the overall lack of hearts in the thread, I’d say that my passion for expansion is peculiar to me alone, so probably not worth any effort. The ellipses win! :smiley:

Thank you , Bhante. :pray:

1 Like

Not quite so! I often like to fill in ellipses when I read, and as for listening I would vastly prefer expanded versions.


My solitude is broken. :open_mouth:

@Gillian, given the projected cost of implementing expansion, please tell us:

  1. How many suttas would you expect to see expanded in the short term?
  2. What are your top five suttas requiring expansion?

My own answer is: 3 suttas: MN1, MN4, MN10 (all Sujato)


I also would like to see expanded versions- or have the option to expand and contract as required with a toggle.

1 Like

I think that’s a good suggestion: if the scope is limited to simple cases, and the programming outcomes are hand-checked, it could work.

But I would still strongly urge you to focus on getting an audio interface working first. That is something really valuable!


The AWS Polly SlowAmy adapter is done and unit tests are in place. The adapter converts text to voice one segment at a time. The results are cached on disk (in local/sounds) to reduce AWS service costs. Paragraphs are not composed using the service. Instead, they are concatenated on the sc-voice server using ffmpeg. Indeed, the existing sc-voice library is now capable of generating a sound file for a sutta.

The immediate goal is to build up the words/en.json data file. This is a critical file for sc-voice, and it serves several functions:

  1. en.json stores custom IPA pronunciations that we will be adding with the help of volunteers listening to suttas. For example, you will see the custom IPA for “earth”, which folks had difficulty understanding.
  2. en.json maps word variants to a canonical pronunciation. For example, SuttaCentral uses three different spellings of bhikkhu, and they will all be pronounced the same. The variants will also be used to auto-generate regular expressions with alternates for search (i.e., all bhikkhus will be found).
  3. en.json stores the punctuation semantics, which are needed to establish word boundaries for looking up pronunciations.
  4. en.json stores the romanization rules that will allow us to enter search terms without diacritics. The search terms will be replaced with their proper Pali spellings for the actual search.
  5. en.json will allow us to enter the short form of a sutta (e.g., “Mulapariyaya” vs. “Mulapariyayasutta”) for search.
  6. en.json will hopefully enable a custom vocabulary for voice search–right now my Chromebook fails on simple things like saying “SuttaCentral”. I get “ceuta Central”. Voice search is really a V2 feature, but it’s good to anticipate enough to inform current designs.
  7. en.json is used for English. We can extend this customization capability to other languages in the future.

The sc-voice service requires Node.js running on a Linux server. Ideally this would be a SuttaCentral server. But if the environments are not compatible, we can just integrate at the REST level.

The library API is fully unit-tested and is stabilizing quickly. This means that the curious folks out there can feel free to download and play using their own AWS Polly accounts. I’ll be happy to add documentation as needed. The prototype will use the translation repository content for searches. This should give the prototype enough content to present information in a way that SuttaCentral search does today. The UI will be necessarily minimalist and will be easily navigable with screenreaders. I’ll be testing with ChromeVox, for example. SIghted users will miss the richness of existing SuttaCentral UI, but that is intentional, since the sighted user will have to be second priority to the assisted user on the voice assisted “read me a sutta” page. We will no doubt iterate for quite some time on the prototype as we all explore various assisted use cases.


Mat, thanks for chiming in. Please let us know which are your top 5 suttas for expansion. Bhante Sujato is right about the difficulty of expansion. Today I saw some suttas I do not know how to expand. By providing a list of top suttas you would like expanded we can hopefully prioritize our expansion efforts to provide the broadest benefit.

1 Like

I regularly expand ellipses, often copying from SuttaCentral or a Kindle version of a Wisdom Publication into my own doc and then working on it there. It is particularly valuable for suttas I wan to use in chanting and for contemplation practices.


I am very much impressed by your work, @karl_lew! And I’d be interested in helping make this work for German.

I am not a programmer but have started learning little things in order to help coding a few texts for SC and might be able to learn some more… however don’t have much spare time at the moment.

Anyway, I see the application for other languages than English rather as a future project, not for now yet—so maybe I can have a chance to help. But maybe saying this is already a step too much, and I’ll be just drowned and drowned in work… :scream:


As far as I know, the German version of MN (Mettiko Bhikkhu) is already fully expanded.

1 Like

Ohh… I’m not talking of the expansion, but the text-to-voice-conversion.

Oops - sorry.

1 Like

My experience of expanding elisions is that it is not something that could be automated. The extension is often not that difficult but does require care and often words need to be changed.

1 Like

Sehr gut! Leider habe ich seit langem kein Deutsch gesprochen oder gelesen. Aber vielleicth @sabbamitta könnte uns helfen?


With so many helping hands emerging to help expand the suttas, perhaps it is best that others coordinate the expansions with Bhante Sujato. All I need for voice assistance is access to these expansions as JSON segments per the overall plan to adopt a JSON standard for suttas. I will need at least one expansion (I’ll probably do MN1 myself) to test the voice assistance code, but I am quite excited about the possibility of others helping, especially with the difficult ellipses.:heart::pray:

Vielen dank! Gerade jetzt sind wir nicht ganz bereit. Aber hoffentlich bald!


Hi Karl

My experience was MN 1 was one of the most problematic to deal with.

The sutta has 24 descriptions which repeat once for the untaught ordinary person, once for the disciple in higher training, 4 times for the arahant and twice for the Tathagata.

We decided when expanding the elisions to group together the expansions for the arahant and the Tathagata - below is an example of one of the 24 descriptions expanded for the arahant.

“He directly knows water as water. Having directly known water as water, he does not conceive himself as water, he does not conceive himself in water, he does not conceive himself apart from water, he does not conceive water to be ‘mine,’ he does not delight in water. Why is that?
Because he has fully understood it, I say.
Because he is free from lust through the destruction of lust.
Because he is free from hate through the destruction of hate.
Because he is free from delusion through the destruction of delusion.

If you don’t do this the reading would be over 2 hours long.

I would choose a different sutta to start with.

I’m happy to share the word files we have used which have expanded elisions - although they have not been checked or approved by anyone.