Stress and Intonation in Pali

Q1: Is this a reasonable brief statement?

It’s not possible to know what stress and intonation the original Pali speakers used, and the systems described today reflect the Sri Lankan, Burmese and Thai chanting traditions.

Q2: I believe (courtesy of the OCBS?) that there is also a 19th-20th European standard of Pali pronunciation, but where can I find this?

Thank you. :pray:


In addition to etymological clues, Pāli poetry contains regular and recognized meters which confirm the stresses and lengths of its vowels.

And tonal genesis significantly post-dates Pāli, so we know Pali had no tones like what modern Thai or Vietnamese has.


Very interesting questions.
There is a huge amount of Pali text not in metre, I’m not sure if there are specific instructions somewhere about stressed syllables?- perhaps in the Saddanīti?

I haven’t heard about what is mentioned in Q2.

1 Like

Besides vowel length, and therefore stress to a certain degree, (long vowels receive 2x the length of short, e and o usually long except when they precede a double consonant,)
Consonant strength is given (in descending order of strength):

-vagga consonants
-sibilant, spirant (s, h)
-semi vowels (y, r, l, v)

See Steven Collins’ Grammar p. 7.

Taking this into account, does this mean ‘pajahati’ is pronounced pa JA hati instead of pa ja HA ti due to the vagga consonant j being strongest?
Actually not sure!


While it is true that the metrical placement of a word gives strong clues to how it was pronounced (I believe Prof. Gombrich makes a big point about this somewhere…), there are certainly notable exceptions where syllable stress (and word pronunciation) is altered ‘metri causa’, e.g. samprasārana and svarabhakti.

The comments of Ven. Dhammanando is this thread may be helpful:

1 Like

Thank you Bhante @Khemarato.bhikkhu and @stephen for your comments and the useful links.

That I didn’t know, and is useful info. But beware: tone and intonation are different things.

Probably a throwaway remark that I over-interpreted. I’ll forget it.

I didn’t make it clear that I wasn’t thinking about poetry, but prose. I have been pondering why I find the IT generated voices on SC Voice very difficult to listen to. Phonemically they’re clear, I think they have syllable stress; so what is missing is intonation?

EDIT after 3 weeks: that wasn’t correct: The Bots on SCVoice have neither syllable stress nor intonation. If they had syllable stress, which is known, they would be tolerable to listen to without bothering about whether Pali did/din’t have intonation patterns – which is a good thing because modern speakers will be influenced by the tone/intonation features of their known langua ges.

1 Like

To make the point clear, I wasn’t (just) talking about poetry either.

Linguists use poetry to infer how words used to be pronounced. For example, most of what we know about old Chinese pronunciation comes from rhymes.

This, of course, assumes that the words were pronounced the same in the poems as in that language’s prose. As @stephen pointed out, there are some cases where the meter of the poem causes you to pronounce a word a bit differently (think, “That floats on high o’er vales and hills”) but such examples are the exception. For the most part, words are pronounced the same in prose as in poetry and thus knowing how a word was stressed in poetry tells us (in general) its pronunciation.

As far as the Amazon AI voices: yes. They are still lacking any real understanding of prosody.

Perhaps future AI language models could help text-to-speech algorithms shape their speech to better reflect the meaning of what they’re saying, but for now the algorithms are mostly stuck pronouncing one word one way (with a few heuristics like “pause between sentences” added in).


What would you say is the most commonly chanted prose sutta? Pref MN: With your help I could then find versions of the same sutta being chanted by Sri Lankan, Thai, Burmese and American monks. I should like to listen for the differences.

1 Like

Probably the Dhammacakkappavattana Sutta, I should think?


That was the one that came to my mind, but your opinion adds certainty. I’ll start collecting versions. Thank you.


You might want to add this version as well. It sounds like a very ancient style (5th century?):


Ven. Analayo in a recent book, Rebirth in Early Buddhism and Current Research, has analyzed those recorded chants in detail.
(See the section “Case Study in Pāli Xenoglossy”. )

Very interesting, and strange…


For our benefit could you summarise his findings?


Dhammaruwan’s story is related in detail, the recordings are described, 13 are transcribed, differences from the received text are related in detail.

One interesting thing noted is “at the time of recitation, the reciter did not have a proper knowledge of Pāli grammar””

Also interesting is that the recitations do not agree with any modern edition consulted.

There is also a discussion of “The Brain in the Listing of Anatomical Parts”, the brain being absent in the canonical listings.

The conclusion is that, “the evidence surveyed above suggests that Dhammaruwan’s chanting of these texts as a child is a genuine case of xenoglossy, in the sense of involving a recitation of material in Pāli that he did not learn and was not made to recite in this way in his present life in Sri Lanka. “

“…the chants contain a number of variants that are distinctly un-Sri Lankan. “

The recitations give an interesting snapshot of Pali recitation from a very distant time.


Not about Pali, but can we take the method used here and apply it to pali texts?

Based on what is put forward regarding Proto-indo-european languages in this video, my very simple mind thinks so.

1 Like

Which question are you answering please?

Good question! I don’t think that’s a reasonable statement. If we look at the way this topic is treated by Geiger, for example, we can see that stress impacts the quality of the vowels in related Prakrits. For example, in Jain Prakrit, arahant becomes arihant. This gives us a very good indication that the second vowel is unstressed, due to its weakening from a to i. (Like the unstressed Aus in Australia is dropped to give “Straya”.) The vowel in a syllable before a stressed syllable is normally unstressed, so this gives us hints as to where the stresses occur. While obviously stress can be different in different languages and change over time, what we see probably isn’t random change and findings from other Prakrits and Sanskrit could sensibly be used to infer historical Pali stress patterns. There are explicit rules for stress in Sanskrit.

Stress is not actually the same as heavy and light syllables. It is something else that tends to result in often sub-phonemic (argh is that the term?) lengthening of the stressed vowel, accompanied by the shortening of other vowels. As per Sanskrit, we would expect a heavy penultimate syllable to be stressed; if the penultimate syllable is light, the stress will move forward.

I don’t think the Pali on Sutta Central would actually have any kind of proper stress, the syllables would be equally stressed, which is characteristic only of “robot language”, hence why it might sound mechanical or difficult to listen to.


Actually, it was making this observation that set me off on this line of thought in the first place. I’m learning a lot. I need to study what you shared here carefully, and to apply it to what I am writing. I shall be back. Again, many th :pray:anks.

Btw, how would you have reacted if I had asked:

Q1: Is this a reasonable brief statement?
It’s not possible to know what intonation the original Pali speakers used, and the systems used today have come to us through the Sri Lankan, Burmese and Thai [?chanting] traditions.


Question 1 (adding some more characters to keep our robot overlords happy)


Not really because he is talking about phonemic features of the language and I am talking about nonphonemic features.