Where can i get Pali Devnagari text data
If you just want to see the Pali texts in the Devanagri script, then you can set that when viewing a sutta.
But the actual Pali data is all kept in the international Romanized system. I believe that the tools at Aksharamukha are used to convert between scripts.
Is that what you are after? The Pali is all on the Github repo: suttacentral/bilara-data
5 Likes
I am new in the field of Pali research , and i tried your suggestion but somehow not getting that script in pali , instead sutta is shown in plain english
To see the root, Pali, text, you can select show root text from the view menu, or tap the V key on your keyboard until it shows in âside-by-sideâ mode.
The english will still show in Roman script. If you want to just have the root text without the English, then click on the âParallelsâ and scroll down to the âMahasanghitiâŠâ version.
The reason we have so many scripts available is because there is no âcorrectâ script to display Pali in. If you find devanagari easier to read than Roman script then itâs a great choice. However, choosing a special script isnât important to learning Pali.
3 Likes
Welcome!
What Venerable Pasanna said is correct, there is no script connected exclusively with the Pali language. It has been written down in various scripts over the centuries, probably starting with Brahmi.
There are, however, various manuscript traditions that would have been written in different scripts. So, for example, manuscripts in Sri Lanka were written in the Sinhala script.
As far as I know there are no existing Pali manuscript traditions that use Devanagri.
On SuttaCentral, we have a single manuscript for Pali, namely the Mahasangit edition (abbreviated as ms.) See here:
The Pali text on SuttaCentral is the MahÄsaáč gÄ«ti edition, sometimes called the âWorld Tipitakaâ. This is a version of the Sixth Council text that was edited and published by the Dhamma Society of Bangkok, based on the digital text supplied by the Vipassana Research Insititute. The Dhamma Society undertook an extensive review of the text, correcting it against multiple printed editions. They took as their mainline source the first edition of the Sixth Council text, which represents the Burmese manuscript tradition. We believe this is the most accurate, consistent, and well-edited digital edition of the Pali canon available.
3 Likes
Thank you for clarifying , i will look into it
1 Like
Thank you !!
I am currently working on this Pali language for building a better Tokenizer for Large Language Models , Cause existing onces are unable to understand the meaning of the words or like they fail to understand grammer of it as they are totally built in considering english language and trained on English Text .
As Pali is a severely low-resource language compared to English
Be sure to take a careful look at the license page to see how the SuttaCentral.net content can be re-used. There is a note about using SuttaCentral created content for AI training at the very bottom.
The license page indicates that the original Pali text is not under any copyright. But lots of folks around here are uncomfortable with even using those for training LLMs. If you want to avoid using any content at all from SuttaCentral.net, you can find an edition of the World Tipitaka here.
If you are interested in learning how to read Pali, there are Events > Pali Course offered from time to time.
You may also be interested in this discussion: Building a Pali language analyzer
Iâm curious how you got interested in this. If you like there is a thread to introduce yourself
4 Likes
PS: If youâd like to know why Bhante Sujato is opposed to all things AI, one of our mods has kindly collected links to his essays in this post. Itâs a lot.
I believe at the moment mods are trying to consolidate all of the AI related discussion to this thread, so you could comment there. Just be aware that there have been lots of discussions already and we generally try not to re-hash things.
4 Likes
Thank you so much for this detailed and welcoming response, and especially for the links! I completely understand and respect the communityâs stance on generative AI, and I appreciate you taking the time to explain it to a newcomer.
To give you some background on my interest: I am an engineering student currently working on my Masterâs thesis. My research sits at the intersection of Natural Language Processing (NLP) and ancient languages.
I am actually not trying to build a generative AI or a âchatbot.â My thesis is focused on Computational Linguistics - specifically, exploring how modern AI tokenizers (like the ones used by OpenAI or Google) fundamentally fail to understand morphologically rich, continuous-script languages like Pali. I am building a custom tokenizer to prove that these models destroy Pali grammar by failing to understand things like Sandhi splitting and morpheme boundaries.
The âBuilding a Pali language analyzerâ thread you linked is exactly the kind of structural linguistic work I am researching!
To ensure I completely respect SuttaCentralâs ethical requests and Bhante Sujatoâs stance, I will absolutely take your advice and use the World Tipitaka edition for my tokenization experiments rather than scraping SuttaCentral data.
Thank you again for pointing me in the right direction and for keeping the community boundaries so clear. Iâm looking forward to reading through those threads!
6 Likes