Why the recent change from ṃ to ṁ?

Recently words in Pali that ended in ṃ now end in ṁ. What is the reason for that?



SuttaCentral uses the ISO 15919 standard:

As for the rationale, if you look at the glyphs used for Pali/Sanskrit, several of them have a dot underneath, and that dot always means the same thing—retroflex. That means they are pronounced with the tip of the tongue curled so that the back touches the roof of the mouth.

There is also the ‘n overdot’ ṅ, which means ‘nasal’. So when it comes to the other nasal, which sounds identical to ṅ and only formally differs, it’s logical to preserve the pattern by using ‘m overdot’ ṁ, which is what we find in ISO 15919.


M overdot (or rather, just the overdot) is also how the bilabial nasal is represented in Devanagari…which is a nice bonus, as an over dot is a very logical way to represent the nasalisation of a preceding vowel.

Next step: just putting the dot directly over the vowel haha.


I thought that this (dot 0n top) was to distinguish it from anushvara (dot below) which is a phoneme in its own right (the ng in English sing), but not retroflex like the other dots below; rather than a phonological process in response to the place of articulation of the following phoneme, which is more akin to sandhi.


Yeah…AFAIK m underdot and m overdot have always been exactly the same thing, exactly the same anushvara/anussara, with m overdot being a little more modern, and m underdot being old school IAST.

There are only a few relevant key points about the anussara sound “ng”.

—>The first is that it is a nasalisation of the preceding vowel, not a stand alone sound. If it is being read aloud in isolation, it should be called “ang”, with inherent vowel a.
—> it is also termed the “bilabial nasal” due to being made with the lips closed without oral air release.
—>The second is that its allowable vowel combinations are ang, ing and ung.
---->The third is that it assimilates progressively by place of articulation in proximity to another consonant, I.e. by becoming the nasal of that row. [This may not be fully represented in all scripts].

------->This last point may have been the sandhi you mentioned @Gillian, but it’s not related to m overdot vs underdot. The sandhi would be represented by the substitution of a different sound from the nasal row, either guttural n overdot, palatal n tilde, cerebral n underdot, dental n or labial m. These are consonants in their own right, they are different sounds.


When ṁ = ம் helps me with pronunciation :slightly_smiling_face: always got confused with the ṃ


Tamil should actually have another “anuvara” character that represents this precise sound with a little top circle?

1 Like

For the ing, ung, ang sounds , the letter ங் = ṅ is used.( goes with the vowel)

For example the word for fraction/part’ is ‘paṅkuu’ is spelt பங்கு

Lion is singham - சிங்கம்

The ñ sound has its own letter too ஞ்


Is the dot above in Tamil is for the vowel removal? like the al-kirima in Sinhala?

1 Like

Interesting. In Thai, that dot goes below (สฺ). ṁ is an upper circle just to the right of the vowel (าํ)


I’m not sure how much it relates to Sinhala, but yes, adding the dot on top removes the vowel and you just have the consonant.

ம் Sounds like ‘m’ with closed lips
ம sounds like ‘ma’ with them open


Thank you. I should have remembered this & thanks for taking the time to add more info. There still seems to be a difference, tho, between nasalisation of a preceding vowel and assimilation by place of articulation to a following sound. I have had advice to use the under and over dots to distinguish these.

But I can see that this might have been a useful pedagogic technique rather than the common use age I assumed it to be.


While we’re on the subject of diacritics, can someone explain ṅ and ñ? (i.e., taṅhā vs. tañhā)

1 Like

Yes, like the ṅ in saṅkhāra. I was just always puzzled by why some people romanized with the ñ: it seems impossible to pronounce. Funny thing, it seemed like all my first exposure to the word was printed “tañhā.” I never even tried to do ñ; I sort of instinctively guessed it should be ṅ (or, some reasonable facsimile thereof) instead.


Hi, the word for craving is often spelled ‘taṇhā’, the nasal corresponding to the cerebral, or retroflex group of consonants.

The letter n (nasal) with a dot above, ṅ, is part of the guttural, or velar group of consonants, pronounced in the throat. The nasal ñ is part of the palatal group, pronounced in the palate or mid-mouth.


Thank you, Bhante @sujato. That answers my question. And thanks to all for your interesting points.

1 Like

Thank you so much for pointing that out!

The retroflex n would therefore be a historical relic here from Sanskrit t.r.s.naa, due to historically retroflex environment.

How interesting! It’s cool how Pali sometimes has two spellings of a word, one from Sanskrit (eg pa.ti-from pra.ti-) and one more genuinely Pali (eg pati-).

Not sure if this ever happened with ta.nhaa though, as @stephen has pointed out, ta.nhaa was the only attested spelling I could find quickly.

Maybe tanhaa and tan~haa (if they are attested) result from people not being aware of the historical retroflexion and just applying Pali rules, as per the word “pan~ha” (a question).


Although the Sanskrit tṛṣṇā does become the Pāli taṇhā, retaining the retroflexed nasal, Pāli’s pañha comes from Skt. praśna I believe, so a different type of assimilation takes place there.

1 Like

Re: pan~ha…yes, the historical relationship to Sanskrit prashna is more complex, but I don’t think the historical relationship is what’s governing the assimilation here. N becoming palatal before h seems like just a physiological constraint thing that’s unique to Pali?

But yes, a different case in that the n in pan~ha was a historically dental, as opposed to historically retroflex sound, as in the case of ta.nhaa.


Wait! Found it! Was this what you were talking about?

The sequence śn becomes ñh, due to assimilation of the n to the preceding palatal sibilant


praśna → praśña → pañha