Pronouncing m in Pāli

Are Pali ṃ and ṁ both anusvara?

Do they share all phonetic features? Are they pronounced the same or differently?

Thanks :pray:


Yes. You’ll find it represented by various characters in different transliteration systems, but no difference in pronunciation is intended. Thus:

Early Pali Text Society:

International Alphabet of Sanskrit Transliteration (IAST); Thesaurus Indogermanischer Text- und Sprachmaterialien (TITUS); and later Pali Text Society: aṃ

ISO 15919: aṁ

Harvard-Kyoto; and Indian languages TRANSliteration (ITRANS): aM

Velthuis: a.m

IPA: ə̃


Thank you so much. I saw both ṃ and ṁ in the same printed text and thought I must be missing some fine distinction!!

Thank you for that; I was assuming ṃ was velar.

:pray: May you be well and happy, Bhante.


It would indeed be pronounced as a velar (i.e., as ṅ) if it were followed by k, kh, g or kh. There is an inconsistency in different editions as to whether sangha, for example, is written as saṅgha or as saṃgha.

In fact even though /˜/ is supposed to be the sound, the places where aṃ, iṃ and uṃ would actually be realized as /ə̃/, /ɪ̃/ and /ũ/ are greatly outnumbered by those where it’s euphonically modified to accord with what comes after it. Sometimes the said modifications get reflected in the spelling (e.g. when the Thais write saṃyojana as saññojana), and sometimes not.


? Perhaps I’m not understanding the IPA correctly, but this appears to me to be a vowel without any consonant. I assume you mean something like /əŋ/?

1 Like

No. A niggahīta is a nasalization of either a, i or u, but differs from other Pali nasals in that there is no other place of articulation but the nose. The IPA symbol for this is a tilde on top of the vowel.

We don’t have this sound in English, except in loanwords, but here are some French examples:

quinze – /kɛ̃ːz/ (fifteen)
deux ans – /dø.zɑ̃/ (two years)
intelligence - /ɛ̃.tɛʒɑ̃s/

And Portuguese:

limão – /liˈmɐ̃w̃/ (lemon)
presidente /pɾeziˈdẽtɨ/
São Paulo – /sɐ̃w̃ ˈpawlu/


! I’ve literally never heard anyone pronounce Pāli like this. Do you have a citation, so I can learn what else I’ve been pronouncing incorrectly?

1 Like

That’s not surprising. I mean you would only be likely to hear, say, saṃsāra pronounced as /sə̃sɑːɾə/ if the speaker’s mother tongue were one that had an /˜/ in it. If it doesn’t then we can expect the ṃ to be realized as something else, usually as n (e.g. in Sri Lanka) or ṅ (e.g., in Thailand). I gather from the link below that /˜/ is present in a lot of modern Indian languages but not in many SE Asian ones.

Well, okay, but it might not be a good idea to try developing a style of pronunciation different from that of the monks you’re living with.

There are a few web pages where people have attempted to use IPA to represent the pronunciation of Pali as it’s described in the old grammars. I’m not 100% in agreement with any of them, but you could make a start with these three:

Wikipedia - Pali phonology


Prof. Vichin Phanupong

And then Eisel Mazzard has a great page outlining the regional pronunciation of Pali in Cambodia, Myanmar, Laos and Thailand. (Though you’d need to learn the IPA first to make any sense of it).

Reading Pali in its Indigenous Systems of Orthography


I wrote this on a different forum a while ago:

Before we begin, although I have a passion for linguistics, I have no special training in it. Consider these the words of a fellow amateur.

Another user said:
You mean I’ve been saying it wrong for 20 years?! That’s it, my Buddhism has gone up in smoke!
Damn, I must still be catholic!

Speaking of Catholics, this is a good segue.

Pāli is a dead language. Pāli is a living language. Sanskrit is a dead language. Sanskrit is a living language. Latin is a dead language. Latin is a living language.

There’s a whole (tiny) country that uses Latin as their official language, and there used to be an entire denomination of Christianity making daily use of it before ~1965. There’s an entire priesthood (several sets of priesthoods, actually) that still makes daily use of Sanskrit. Pāli chanting is still done by monks of the Theravāda tradition to this day. In Scarborough, here in Canada, you can take Pāli lessons at the Scarborough Mahāvihāra. Latin was offered at my high school as an elective. Dead, living, dead, living, etc. Latin is a lot like Pāli, it’s just missing a small country using it as a state-sanctioned official language.

If you know something about Latin, you know it comes in a few varieties: ecclesiastic, classical, and old. Similarly, with Sanskrit, we have classical and Vedic. With Pāli, we have less standardized names, but we might call them “how Pāli was pronounced” and “how Pāli is pronounced” for our purposes.

Ecclesiastic Latin is generally understood to be Latin pronounced with more-or-less an Italian accent. It’s how the Popes generally speak/spoke Latin, and how priests would have been trained to pronounce Latin before its liturgical use fell out of favour. It is still Latin, it is just pronounced differently. Back when the entire Roman Catholic world used Latin (and even further back when the entire Western Christian world used Latin), Latin was pronounced regionally according to local phonologies influenced by local languages.

In my 3rd year of university studies, during the acquisition of my rather pointless ethnomusicology degree (I don’t regret it), I was in a male vocal ensemble that participated in a performance of Carl Orff’s Carmina Burana. The choral conductor was keen on us pronouncing the Latin in the Medieval German manner, as it might have been pronounced ~1200. So when we got to the lines:

In taberna quando sumus,
non curamus quid sit humus,
sed ad ludum properamus,
cui semper insudamus.

“Quando sumus,” became “Kvando zumus,” the U being a V, certain S’s being Z’s. There were a few more (minor) phonological quirks to the Latin we used, as reconstructed by whoever our conductor consulted, but this one stuck out the most. Certain “C’s” were also pronounced “ts,” instead of “ch.” Back in the day, there was a Polish way to pronounce Latin, a French way, an Italian way, a German way, and even a regional English way (!!!) to pronounce it, along with various prescriptivist standards for people to endlessly complain about. All these regional pronunciations were influenced by the languages these Christians used in their daily lives alongside this liturgical/sacred language.

In addition to this, we have Classical Latin (and Old Latin that is so old we don’t know much about it). You know Caesar? He’s “see·zr” in Canadian English. He’s “che·sar” in Ecclesiastic Latin (forgive my imprecise phonetics, I’ll start using IPA shortly), and “kai·sar” in Classical Latin. Now the German loanword “Kaiser” makes a lot more sense, right? It’s “Caesar” with a more classical pronunciation, essentially.

Pāli is pronounced regionally in Southeast Asia much like Latin used to be in Europe. AFAIK, it was originally either a Burmese or Thai convention to pronounce “ṁ” and “ṇ” as “ng” in the sense of ŋ. I think the Sri Lankan convention of pronouncing it as “ŋ” also comes from the Thai, who re-established Buddhism in Sri Lanka after it had died out there in the early 1800s (hence why there is a “Siyam Nikaya” in Sri Lanka, from the word “Siam”). Past revivals of Buddhism in Sri Lanka after periods of decline were on the part of the Burmese rather than the Thai, AFAIK.

Some romanization schemata from Sri Lanka actually render ṁ and ṇ as the character “ŋ,” from IPA.

Another user said:
Am I right in thinking that the ṁ is pronounced like “ng”? So it sounds more like, “Buddhang saranang”?

It is a minor point of controversy in Sanskrit (and therefore by extension Pāli) linguistics, namely the manner in which the anusvāra is meant to be pronounced generally and when it acts as a syllable coda (i.e. ends a syllable).

Common practice in Theravāda-influenced contexts is to pronounce it “ŋ,” like in the material [another user] posted, which seems like it might be from a Thai-produced textbook. You could try to be “proper” and pronounce it like a Prākrit from ~200BC-200AD, a sibling/descendant of the dialect clusters that gave rise to Vedic Sanskrit, but that begs the question as to why you would want to stand out necessarily.

In my experience, if everyone in the room says “buddhaŋ saraṇaŋ,” and you are there going “buddhã saraṇã […],” (nasalizing the previous vowel, with no “m,” “n,” or “ŋ” sound, which we will discuss later) people are more likely to think you are just wrong than thinking you are a phonological pedant who looked up the “right” way to do it.

Generally speaking:
If ṁ comes before a velar consonant (k, g), it is ŋ, like in “sāṁkhyaṁ” (sə͂ːŋkʰjə͂ː).
If ṁ comes before a palatal consonant (c, j), it is ñ, like in “saṁjñaḥ” (sə͂ñdʑñəh).
If ṁ comes before a dental consonant (t, d), it is n, like in “cittasaṁtatiḥ” (tɕɪtːəsəntətih).
If ṁ comes before a labial consonant (p, b), it is m, like in “saṁbuddhaḥ” (səmbʊdʰːəh).
If ṁ comes before a spirant, it is n like above with dentals (saṁskṛtam – sənskɹtəm)

“Proper” Pāli, depending on how pedantic you want to be, will loosely observe these anusvāra rules to a certain extent. Do remember though, that complicatedly, Pāli did not descend from “proper” Classical or Vedic Sanskrit, but rather descends from languages in the same “dialect cluster” as Vedic Sanskrit, which then became regional Prākrits.

A huge controversy is when the anusvāra comes at the end of a syllable, and academics are split 3 ways on how this would have been realized. I’ll use “a” as the vowel for these examples:

  1. ãm, 2) ãŋ, and 3) ãː.

In 1, the vowel is nasalized and a “m” sound follows. In 2, the vowel is nasalized and a “ŋ” sound follows. In 3, the vowel is nasalized and lengthened and there is no “m” sound left.

For various reasons, I fall on the side of supposing that the ṁ at the end of words generally was a lengthening and nasalization of the previous vowel (without an “M” sound) in some or most regional Sanskrit pronunciations and likely in Pāli.

We see from Sanskrit to Pāli:

cittam (Sanskrit) --> cittaṁ (Pāli)

The full-on Sanskritic “M” becomes a Pāli nasalization, which is a very common sound-change in languages, more common than “M” becoming “Ŋ” (i.e. ŋ/“ng”) in this case, according to my unlearned instincts and a handful of papers I skimmed in writing this.

But that is just my opinion, and as I said before I am not an expert. I wouldn’t even call myself a Sanskrit expert, let alone Pāli.

Common practice is to realize ṁ and ṇ as ŋ, and you will stick out and maybe seem unlearned if you don’t, since Pāli nowadays is pronounced regionally, like Latin once was.

If there’s wrong material in the above, I’d be grateful for some correction.

The ṁ should also be lengthening certain of the nasalized vowels, and it looks like I didn’t observe that systematically in the IPA above.


It’s the geopolitical material concerning Sri Lankan Buddhism and relations to Thailand and Burma that I suspect could use the most correction. Reflections with regards to this would be very welcome. My only source for making that was wikipedia. I’ll forward any corrections to the area where this was originally published.

It’s a nasalised mid short vowel.

You are, atm, hearing Pali pronounced with a Thai accent. I first heard the chants pronounced with an Aussie accent. Very different.

1 Like

:joy: Of course. Sadly that laudable goal of harmony is basically impossible: we already have many pronounciations. Some people here use a heavy, “rural” Thai accent (making the vowels sharp and whiney), some use very rounded vowels, there’s no agreement on the retroflex or double consonants and even the sounds that do exist in both English and Thai (kh, g, etc) are used on different letters! There’s even a (rather hilarious) meme here that ṃ means “wrinkle your nose” (while pronouncing it exactly the same as an ordinary ง) so there are a number of monks who chant with their noses atwitch :mouse: :joy:

Thankfully I have enough ear and time to learn more than one way of doing things :slightly_smiling_face:

Well then, I look forward to you publishing your pronounciation guide! :pray::grin::grin::grin::pray:

Thanks, Bhante :pray:

So much fun! I also had the pleasure in University of learning to sing Latin in a variety of styles: a skill that, to my surprise, is coming in handy!

So I tried doing some chanting making the nasalization (“evaṃ me suttaṃ” now sounds French :joy:) and I hit a bit of a stumbling block trying to do both retroflex consonants and nasal vowels. It seems to require a very different (more Indian?) accent than I’ve been using.

Does anyone have recordings of Pāli speakers / chanters doing this all “correctly” so I can get a feel for the accent? Or even just a video of people speaking a language that has both retroflex consonants and nazal vowels? [edit: it seems that contemporary Punjabi might fit the bill of having a similar phonemic range to Pāli?]


Interesting. Would love to see @Dhammanando’s full understanding.

Here is what I’ve written for my Pali classes. (Still kind of draft.) Hope the markup transfers to discourse properly. (Edit: it didn’t fully, sorry. It’s a bit messy, but good enough I hope. So .n means the n with dot below, and .m is ṁ, and aa is ā.)

The niggahīta (‘anusvaara’ in Sanskrit) is represented in Roman script by , , or ŋ. Nowadays it is often pronounced as ‘ng’, as in ‘thing’, identical to the consonant . This is an acceptable pronunciation, but is probably not fully true to the original.

In the Asian scripts the niggahīta is a diacritic, not a seperate character. Usually it’s a circle or dot. (E.g.
ku.m is කුං or ก,ุํ while ku is කු or กุ.) According to many grammarians this diacritic indicates nasalisation of the preceding vowel. Nasalised vowels are pronounced with air escaping from the nose. For example, saṁ is pronounced much like the French ‘sans’. Since the niggahiita follows either a, i or u, there are three nasal vowels in Pali. These are transcribed as ã, ĩ, and ũ in the International Phonetic Alphabet.

The niggahīta does not only indicate nasalisation of the vowel. It can also indicate a nasal consonant which naturally results from the vowel nasalisation. The (i.e. the common ‘ng’ pronunciation) is one these consonant, but not the only one.

When a nasal vowel (that is, aṁ, iṁ or uṁ) is followed by a consonant, it can produce a nasal consonant (, ñ, .n, n or m) with has the same place of pronunciation as that consonant. This can be represented in writing by changing the niggahiita to the nasal consonant. (E.g. taṁ ca --> tañca.) Even though the niggahīta is lost in writing, the implication is that the preceding vowel remains nasalised. This is indicated here using ã, ĩ and ũ, but ordinarily these are just written as a, i and u. Examples:

  • Gutteral: alaṁ-karoti --> alãṅkaroti
  • Palatal: kiṁ ca --> kĩñca
  • Retroflex: *susaṁ-.thitā --> susã.n.thitaa
  • Dental: saṁ-nipatati --> sãnnipatati
  • Labial: dutiyaṁ pi --> dutiyãmpi

(This is somewhat comparable to English ‘hangar’ (“haṅgar”), ‘bench’ (“beñch”), ‘hand’, ‘lamp’, and many other words, where the nasal consonant (n or m) also has the same place of articulation as the following consonant.)

When a nasal vowel (aṁ, iṁ or uṁ) is followed by another vowel, it can merge with the following vowel

  • idaṃ ahaṃ --> idā̃haṃ

Or it can produce a consonant—often m, but others are found as well.

  • ala.m ariyaa --> alãmariyaa
  • ki.m assa --> kĩmassa
  • ta.m idã.m --> tãyida.m
  • eta.m eva --> etãdeva

These rules for writing are not always followed. Often the niggahita remains written. For the sake of pronunciation the nasal consonant then has to be deduced.

  • saṁgho, pronounced ‘sãṅgho’ (saṅgho itself is also written)
  • saṁyutto, pronounced ‘sãñyutto’ (saññutto is also written)
  • ahiṁsaa, pronounced ‘ahĩnsaa’
  • saṁsaaro, pronounced ‘sãnsaaro’
  • saṁvaaro, pronounced ‘sãmvaaro’

This applies also to the niggahiita at the end of a word.

  • Ekãṁ samayãṁ bhagavaa sāvatthiyãṃ viharati., pronounced * 'Ekãn­samayãm­bhagavaa­sāvatthiyãm­viharati.’*

The niggahita at the end of a sentence is probably best pronounced however it comes naturally, either as the nasal vowal or any nasal consonant. In Sanskrit m is customary.

  • Katamesaṃ catunnaṃ?, pronounced 'Katamesãñcatunnãm?’

This all seems confusing at first, but actually follows quite naturally as soon as the vowels are nasalised.

However, to complicate things, there are the separate nasals consonants as well, which do not nasalise the vowel (e.g. ki.nāti, anveti). And some manuscripts use the niggahiita for these too! E.g. aṁña = añña, bhaṁte = bhante. Fortunately, this is rarely done in romanized editions.


glad to see this discussed. I am also a little confused by the two terms anusvara and niggahita in reference to the pure nasal m.

1 Like

Anusvara and niggahita are the same. The former is the Sanskrit term and the latter is the Pāli term.


thank you. Unusual in that Pali and Sanskrit terms are often similar. These two are quite dissimilar.


it’s also interesting to see where the pure nasal is placed in the alphabet depending on which of the learning materials you look at. For example, Lily de Silva simply places it among the miscellaneous letters grouped after the labials whereas James Gair and W S Karunatilleke place it at the end of the vowels and again, in brackets, following the gutturals, or velars.


Given that the alphabetical sequence is based primarily on the position sounds are produced in the mouth, the logical positions for the pure nasal aret the beginning, between vowels and consonants, or at the end (after gutterals); “among the miscellaneous letters grouped after the labials” doesn’t make sense, given that assortment of sounds is made in/around the centre of the mouth.


It might be helpful not to think of the niggahita as a ‘letter’, but rather an ‘effect’ applied to some vowels.


You probably told us before, bur now the penny drops. Thanks!
Is this the only case of a process getting it’s own orthographic symbol and thus place in the alphabet, and dictionary entry?

1 Like