Learning Pali Noun Declensions - Any Suggestions for How to Store All the Endings in your mind

I thought learning noun Declensions in Classical Greek was a nightmare. But compared to Pali it now seems so straightforward.

At first I thought it was the number of cases and/or the number of patterns (masc -a, neut -a, etc.) But more and more I think I struggle most with the fact that - unlike Greek - there are multiple endings even within a case and pattern. For instance, the Locative case for masc -a stems can be e, asmiṃ, amhi, or asi.

How do you store that? Do you just remember the most common ending? Do you remember all the possible endings for each Case and Pattern? Do you keep a Noun Declension sheet next to you and just look up lots of words?

I know how to study, but I feel I’m not sure I’m clear how to construct the storage buckets in my mind.

Any thoughts or suggestions about how to store this info in your mind is greatly appreciated. Thanks!


One possible approach and alternative to impressing declension tables into one’s mind is to instead (or concurrently) work with small bits of actual sutta text, and know the role of each word of that bit.
In other words, to learn the declensions through context.
(keeping a noun declension sheet near by can sure come in handy though!)


This is the right way. Most declensions are rare, you don’t have to learn them all at once.

The thing with language is, you don’t have to know everything about the grammar. Especially in Pali, the grammatical information is often redundant. The verb and the noun are both in plural, for example. Sometimes that’s helpful, you can tell which verb applies to which noun. But most of the time it’s obvious just from the context.

So as long as you can parse enough grammatical knowledge to make sense of the sentence, you’re good. Then pick up the more obscure forms along the way. Don’t neglect the degree to which we just absorb grammar through usage rather than formal study; after all, it’s how we learned language as kids.

When I was translating at Qimei, I had a big poster made of Nyanatusita’s grammatical tables, and stuck it on the wall. I still check it sometimes!

Good example: -e is common; -asmiṁ fairly common, while -amhi and -asi are less so. But still, they’re not super-obscure.

You already know -e. And you probably remember -asmiṁ after this discussion, if not before. So just keep the others in the background and look them up if need be.


Thank you, Bhante @sujato and @stephen! That’s very helpful.


Here a snippet to get you started, a famous statement…

Api cāhaṃ, āvuso, imasmiṃyeva byāmamatte kaḷevare sasaññimhi samanake lokañca paññāpemi lokasamudayañca lokanirodhañca lokanirodhagāminiñca paṭipadan’ti.


I did a rather unscientific search and found that many of the endings in those mega charts cannot be found in the first four nikayas. As I recall they could only be found in the commentaries.

Those mega charts have their place, however they are not a guide for what ought to be memorized if you decide to go that route.

One thing that was helpful to me was an index of those charts where it listed all the possible endings alphabetically indicating where they could be found on the chart. So if you come across an odd one you can just look it up that way.


Hi @JimInBC,

Pali isn’t really an “active” language for me because I am only working in Sanskrit now & these days I have been a bit of a hermit. @stephen and Bhante @sujato have excellent points, and famous modern linguists of the “extensive reading” movement like Stephen Krashen would probably agree. My style is slightly different and just reflects my own teacher’s preference.

I normally just tell my class outright that I am there to teach grammar. I wouldn’t pretend that it’s the same thing as teaching reading. To learn reading, you actually need to read a lot of stuff. It’s a different skill.

Anyway…because my original Sanskrit education was somewhat traditionally oriented, I was taught to memorise paradigms. It’s ok, you can do it. It’s just something like the national anthem that if you repeat it enough, you can’t forget it even if you want to, and then it’s there with you for life and you don’t need to look up anything (in theory).

The traditional memorisation method is to go across the column, devodevādevaṃdeve etc. Just make sure you have a textbook which lists the cases in the traditional order otherwise it will mess with your brain. And there is no breath if you can help it…there might be some Vedic science of timing the breath correctly, but I don’t know it. It should be a muscle memory thing, where it’s the same every time & you don’t need to think. If the row is getting confusing because there are multiple items, they can be split by naming the case and no…e.g. pathamā vibhatti ekavacanaṁ devo, bahuvacanaṁ devā…

If you are struggling with the multiple locatives, I would learn -e first and possibly come back to -smiṁ and -mhi later when you learn the pronominals, for reasons I will explain below. Noting that -smiṁ is given as the “base” form in the traditional account.

After memorization, another pedagogical approach would then be to point out some interesting features of the cases. For example, that the forms devasmiṃ / devamhi are related, with -mhi also being a way of representing -smiṃ. Because I am this kind of teacher, I would ask my students how you get from -smiṃ to -mhi? Which ending do they think is older and why? Do you see any features here that we talked about as being common in Prakrits (hint: begins with “m”.) Do you remember seeing something similar happen in the ablative endings -smā and -mhā? How can sa become ha? Did you notice the nasalisation has also been lost? Why might a final nasal disappear in Prakrit? What terms can we use to describe these letters? Where are they made in the mouth etc etc etc (it’s possible to go on for a while this way…).

Have you ever thought about how our English word “hall” is related to “sālā”? Or why the Sinhala word sāvā (hare) is also hāvā (hare)? And why do we have these additional two case endings, given that we only have -e in Sanskrit? (hint: -smiṁ and -mhi are actually the pronominal locative endings [c.f. Skt -smin] which have migrated into this paradigm…c.f. tasmiṁ samaye).

So once you have thought about it a bit. There is only really one original masc. loc. sing -a stem ending, which is -e. The rest have bled over from the pronominals, which you will have to learn later anyway.

And by the time they have thought about it that much, the class should have also picked up several important historical phonetic shifts like s<–>h and metathesis, have been introduced to the concept of a pronominal, and understood some of these things in a logical historical manner instead of just having them presented as inherently meaningless variation.

Also if the class was advanced I would note Prakrit -mmi for -mhi & Buddhist Hybrid Sanskrit -i and -ṁhi and -ṁse. I have never taught and advanced class though.

P.S. not sure if seeing a paradigm presented like below would help (I did this up just now) or whether it would make things worse…I’ve put things back in their original places historically and added them as notes to help the table look less cluttered?


Calling Bhante @Dhammanando: do you know how they deal with the multiple case endings to do paradigm recitation for Pāli in Asia? Is the whole paradigm just recited straight through with all the variant case endings, & do you know if there is a “correct” place to take a breath when reciting a paradigm?


Hi Suvira,

Yes, that’s how it’s done in most Pali schools in Thailand. Prince Vajirañanavarorasa, the late 19th century creator of modern Thai monastic education, basically modelled his approach to teaching Pali on the way that he’d been taught Latin by his British tutors (who were reportedly very hard taskmasters). And so as in Victorian Latin teaching, all the tables of declensions, conjugations, prefixes, suffixes, indeclinables, etc. have to be systematically memorised.

This, by the way, is actually considered the easy way of learning Pali grammar, and is regarded very sniffily by visiting Burmese monks and by the handful of Thai Pali schools that have adopted the Burmese pariyatti system.

In the Burmese approach you start by memorising the 673 suttas (in the sense of “aphoristic rules”) that make up Kaccayana’s Pali grammar. As you’re memorising these (which is mostly done as homework and takes about six months) your teacher will spend the class time going through one of the commentaries to Kaccāyana, usually Buddhappiya’s Padarupasiddhi. (The commentary is essential since the aphorisms are even more succinct and compressed than those in the Abhidhammatthsangaha).

So, let’s take noun and adjective declensions, for example.

You start by memorising aphorism 55:

55. Si yo, aṃ yo, nā hi, sa naṃ, smā hi, sa naṃ, smiṃ su.

Kā ca pana tāyo vibhattiyo? Si, yo iti paṭhamā, aṃ, yoiti dutiyā, nā hi iti tatiyā, sa, naṃiti catutthī, smā, hi iti pañcamī, sa, naṃ iti chaṭṭhī, smiṃ, su iti sattamī.

This gives you the names of the seven cases along with fourteen placeholder forms (the text in bold) representing their inflectional endings.

Then you have to memorise all the rules for converting the placeholder form into the form that’s actually required, depending on the gender, number and case of the noun in question. Each rule is derived from some later aphorism. And so you sit in class reciting like this (I’m translating from Thai):

Masculine nouns in -a. Purisa, “man”.

Nominative: Change si to o: puriso. Change yo to ā, purisā.

Accusative: aṃ remains aṃ, -a in purisa is elided: purisaṃ. Change yo to e: purise.

Instrumental: rassa the ā in nā, a is gunated to e: purisena. Hi remains hi, a is gunated to e: purisehi.

Dative: change sa to ya, dīgha the ā: purisāya. Naṃ remains naṃ, dīgha the ā: purisānaṃ.


Later, when you get to aphorism 99, you learn that the placeholders smā, hi and smiṃ can change to mhā, bhi and mhi in all genders:

99. Smāhismiṃnaṃ-mhābhimhivā.

Sabbato liṅgato smāhismiṃ iccetesaṃ mhābhimhiiccete ādesā honti vā yathāsaṅkhyaṃ.

Purisamhā, purisasmā, purisebhi, purisehi, purisamhi, purisasmiṃ.

But it’s not until aphorism 108 that you learn that locative smiṃ can become e and ablative smā can become ā:

108. Smāsmiṃnaṃvā.

Tasmā akārato sabbesaṃ smāsmiṃiccetesaṃ ā e ādesā honti vā yathāsaṅkhyaṃ.

Purisā , purisasmā, purise, purisasmiṃ.

So, as you can see, it’s a slow, cumbrous and dull-as-ditchwater way of learning the grammar, even if it does enjoy a certain advantage when it comes to thoroughness.

When I was at the Pali school at Wat Benchamabophitr in Bangkok we used Vajirañanavarorasa’s Gradgrindian Latin-teaching system. I don’t recall needing to take a breath when reciting the noun declensions. One inhalation sufficed to get me from puriso to purisesu.

With verb conjugations, we’d recite each verb in its parassapada and attanopada forms for each of the eight tenses and take a breath after every second tense. Like this:

vattamānā and pancamī.
sattami and parokkhā
hiyyattani and ajjattanī.
bhavissantī and kālātipatti.

Well, that gave me a nice excuse to use the lungs emoji for the first time ever.

Best wishes,


I like to think of well known texts that feature a particular form, for example
bhagavato is the gentivie/dative because: Namo tassa bhagavato
bhagavatā is the instrumental because: svākhāto bhagavatā dhammo


What worked for me (for Sanskrit, Avestan etc) was rote learning, first memorising the masc. a-stem declension, then ntr. a-stem etc etc. I set aside a period of time each morning for recitation and would run through them again when sitting on the bus/washing up/etc. I actually came to really enjoy it!
I think you’ll find that the process becomes much quicker and easier once you’ve a few declensions under your belt.


What an interesting thread.

Hi @JimInBC, nice to see you!

This fact continues to stand out to me as the weirdest feature of Pali. “Wait, the locative plural can be any one of arehi, arebhi, ārehi, ārebhi, ūhi, uhi, ūbhi or ubhi?! Obviously those kind of overlap and you can recognize groups (roughly it’s arehi or uhi with variations in length and bh sometimes showing up for h…). But still… it’s not like this is some sort of outlier on the nominal inflection chart — almost every inflectional combination has multiple forms. Good grief!

I even made a web version of Nyanatusita’s chart to try to look things up:


This is sort of along those lines but I’m afraid the data is still a bit buggy:


If yuo search for āya (type aaya) you will get various inflections that take that form.


I am relying on Anki for now: Pali Made Easy - AnkiWeb

They include declension for some of them, I got 5 of the Pali ankis decks now, I think 2 or 3 of them tests me on that.

I find it easier to read the Pali if I know the words, then I can deal with the grammar at my own pace. I learnt the grammar long time ago, but didn’t got much exposure to daily chanting and Pali back then, didn’t memorize much.


@Snowbird - Happen to have this around somewhere? Bhante @Suddhaso has this “reverse” chart he (used to?) use(¿s) - https://archive.org/download/PaliDeclensionChartV2/declension%20chart%20v2.pdf in case that’s helpful for anyone here :slight_smile:


I would be careful to avoid missing the forest for the trees.
The -ar nouns in Pāli are fairly rare, and many of these endings are super rare.
I wonder if some them ever occur in the Sutta Pitaka!

Once the masculine -a case endings are clear in one’s mind, family resemblances can be seen with the others. (e.g. -ehi)
PS Don’t you mean the instrumental/ablative plural?


It might not be so weird if you look at it from a linguistic or scholastic point of view. Before codified and standardized languages, dialectal variation was the norm (and still is in non-standardized languages of today). My understanding is that it is not even known for sure what the area of the original speakers of Pali was. And scholars seem to think it might be some sort of mixture of various Prakrit languages of India. And with so many centuries of copying, we can’t rule out some errors here and there either.


I know that people like to make a lot of the variation potentially indicating dialect, but the some of the variations that are found in Pali declension forms could equally be attributed to the passage of time.

For example, starting from Sanskrit -smin, -smiṁ is basically the same form as Sanskrit (the older form). Of these two, -mhi is the newer Prakritic form. But that too becomes archaic by the time of the Prakrits of 0-1000CE, where you find -mmi.

Maybe this is why in the 13th century, Buton said that the Sthavīras speak Paiśācī AKA “ghost”…because as a Central Indian Prakrit with pre-CE features, Pāli was recognizable as the archaic ancestor of the court language of the Deccan.

Unless I am imagining things. But see table of Prakrit declensions below which is (a) clearly not Pāli despite looking a LOT like Pāli and (b) also features a number of the same multiple forms for the ablative and locative, maybe hinting that the pronominal overlap phenomena is a feature of Prakrit in general and not just a Pāli thing. And not a result of the mixing of languages.

Masculine: halia- “ploughman, farmer”

ēkavacanam bahuvacanam
prathamā haliō haliā
dvitīyā haliaṁ haliē
tr̥tīyā haliēṇa haliēhiṁ
pañcamī haliāō, haliāhi, haliā haliēhiṁto
ṣaṣṭhī haliassa haliāṇaṁ
saptamī haliammi, haliē haliēsu
saṁbuddhiḥ halia haliā

Thank you Suvira for your insight. It might in some way be similar to variation in other languages. For example Spanish has two different subjunctive past conjugations that are equivalent. But then one of them is, more or less, more used in speech, the other more in writing. So you could say “estuviera” or “estuviese” and it is pretty much the same thing, except that one sounds just slightly more archaic. And it is not a dialectal difference.


Hi @stephen!

Indeed yes, I will fix it above, thanks. :man_facepalming:

Your points about the rareness of many of the forms is well-taken; I just digitized what was in Bhikkhu Nyanatusita’s table, from here:

(I also failed to do the cell merging that Nyanatusita did, because it’s kind of a project in HTML, which contributes to adding trees to the forest!) He mentions that the italicized forms are rare, so for example of the eight I mentioned as a “worst-case scenario”, he distinguishes half of them as rare:

arehi arebhi
ārehi ārebhi
ūhi uhi
ūbhi ubhi

So indeed, as you point out, that already removes half of the trees from that corner of forest. And even among the remaining forms, pairs that vary just in vowel length such as ūhi/uhi or clearly related consonants like h/bh in arehi vs arebhi (and here there are echoes of the kinds of historical changes @suvira points out). And again as you point out, the whole column of -ar nouns is quite rare.

I guess there are two ways to look at this sort of thing: with “learner eyes” or “linguist eyes”. Nyanatusita’s chart is pretty terrifying at first glance to a learner, simply because there are so many forms listed (840!). But as you point out, that is misleading — Pali is very learnable due the to the fact that in the vast majority of these cells only one or two inflections are really used.

Nevertheless, linguist eyes (which I am afraid I have a case of) look at such a table and it looks pretty unusual from a cross-linguistic perspective; not because it’s weird to have some variation, but for variation to be so widespread throughout the table. My linguist eyes lead me to say “ooh, neat. that’s complicated”, whereas my learner eyes look at that table and :scream_cat: like anyone else! Thank goodness we have Pali teachers like you, @johnk , and Bhikkhu Bodhi to guide us through the forests.

It would be great to have a frequency count of all these inflections across the corpus but of course that would be very hard to do, because a program won’t know which inflection a given affix represents.
It occurs to me that one useful thing to do would be to tag all these forms by which grammars they occur in, as a sort of proxy for a frequency count — if Warder and De Silva and Gair & Karunatillake all introduce a form early on, we can be pretty sure it’s common.