1. Sandhi kaṇḍa

02 Nov 2026 · 35 min read

The "Sandhi" chapter covers the representation of the language using letters and various rules for sandhi. sandhi is derived from saṃ + dhā meaning "putting together" and is used to refer to the transformation that result from the joining together of two words (or two parts of a word) for the sake of euphony. This chapter is translated by me.

Saññā (Representation)

Akkharāpādayo ekacattālīsaṃ (41 letters) ¶2

Akkharāpi akārādayo ekacattālīsaṃ suttantopakārā. Taṃ yathā-a ā-i ī-u ū-e o, ka kha ga gha ṅa, ca cha ja jha ña, ṭa ṭha ḍa ḍha ṇa, ta tha da dha na, pa pha ba bha ma, ya ra la va sa ha ḷa aṃ-iti.

The letters beginning with a are forty-one, useful for the suttas. They are as follows: a, ā, i, ī, u, ū, e, o, ka, kha, ga, gha, ṅa, ca, cha, ja, jha, ña, ṭa, ṭha, ḍa, ḍha, ṇa, ta, tha, da, dha, na, pa, pha, ba, bha, ma, ya, ra, la, va, sa, ha, ḷa, aṃ.

𑀅

/ə/

𑀆

/aː/

𑀇

/i/

𑀈

/iː/

𑀉

/u/

𑀊

/uː/

𑀏

/eː/

𑀑

/oː/

𑀓

/k/

𑀔

kha

/kʰ/

𑀕

/ɡ/

𑀖

gha

/ɡʱ/

𑀗

ṅa

/ŋ/

𑀘

/c/

𑀙

cha

/cʰ/

𑀚

/ɟ/

𑀛

jha

/ɟʱ/

𑀜

ña

/ɲ/

𑀝

ṭa

/ʈ/

𑀞

ṭha

/ʈʰ/

𑀟

ḍa

/ɖ/

𑀠

ḍha

/ɖʱ/

𑀡

ṇa

/ɳ/

𑀢

/t̪/

𑀣

tha

/t̪ʰ/

𑀤

/d̪/

𑀥

dha

/d̪ʱ/

𑀦

/n/

𑀧

/p/

𑀨

pha

/pʰ/

/b/

𑀪

bha

/bʱ/

𑀫

/m/

𑀬

/j/

𑀭

/r/

𑀮

/l/

𑀯

/w, ʋ/

𑀲

/s/

𑀳

/ɦ/

𑀴

ḷa

/l/

𑀁

aṃ

/ɲ/

Tatthodantāsarā aṭṭha (8 vowels ending with `o`) ¶3

Tattha akkharesu okārantā aṭṭha sarā nāma. Tattheti vattate.

The first 8 letters ending with o are called sarā (vowels).

𑀅

/ə/

𑀆

/aː/

𑀇

/i/

𑀈

/iː/

𑀉

/u/

𑀊

/uː/

𑀏

/eː/

𑀑

/oː/

Lahumattā tayo rassā (3 short vowels) ¶4

Tattha saresu lahumattā a, i, u iti tayo rassā.

The three metrically light vowels are called rassā (short): a, i, u.

𑀅

/ə/

𑀇

/i/

𑀉

/u/

Aññe dīghā (other vowels are long) ¶5

Tattha saresu rassehaññe dīghā.

Saṃyogato pubbe eo rassā ivoccante, anantarā byañjanā saṃyogo. Ettha, seyyo, oṭṭho, sotthi.

The other vowels are metrically heavy and called dīghā (long).

𑀆

/aː/

𑀈

/iː/

𑀊

/uː/

𑀏

/eː/

𑀑

/oː/

e and o before conjunct consonant clusters and double consonants are articulated short. The vutti gives two words for each vowel:

ettha
seyyo
oṭṭho
sotthi

Sesā byañjanā (The rest are consonants) ¶6

Sare ṭhapetvā sesā kādayo niggahītantā byañjanā.

The rest of the letters, beginning with ka, are byañjanā (consonants).

𑀓

/k/

𑀔

kha

/kʰ/

𑀕

/ɡ/

𑀖

gha

/ɡʱ/

𑀗

ṅa

/ŋ/

𑀘

/c/

𑀙

cha

/cʰ/

𑀚

/ɟ/

𑀛

jha

/ɟʱ/

𑀜

ña

/ɲ/

𑀝

ṭa

/ʈ/

𑀞

ṭha

/ʈʰ/

𑀟

ḍa

/ɖ/

𑀠

ḍha

/ɖʱ/

𑀡

ṇa

/ɳ/

𑀢

/t̪/

𑀣

tha

/t̪ʰ/

𑀤

/d̪/

𑀥

dha

/d̪ʱ/

𑀦

/n/

𑀧

/p/

𑀨

pha

/pʰ/

/b/

𑀪

bha

/bʱ/

𑀫

/m/

𑀬

/j/

𑀭

/r/

𑀮

/l/

𑀯

/w, ʋ/

𑀲

/s/

𑀳

/ɦ/

𑀴

ḷa

/l/

𑀁

aṃ

/ɲ/

Vaggā pañcapañcaso mantā (Five groups of 5 ending with `ma`) ¶7

Byañjanānaṃ kādayo makārantā pañcapañcaso akkharavanto vaggā.

There are 5 groups of 5 letters, beginning with ka and ending with ma.

ka-group

𑀓

/k/

𑀔

kha

/kʰ/

𑀕

/ɡ/

𑀖

gha

/ɡʱ/

𑀗

ṅa

/ŋ/

ca-group

𑀘

/c/

𑀙

cha

/cʰ/

𑀚

/ɟ/

𑀛

jha

/ɟʱ/

𑀜

ña

/ɲ/

ṭa-group

𑀝

ṭa

/ʈ/

𑀞

ṭha

/ʈʰ/

𑀟

ḍa

/ɖ/

𑀠

ḍha

/ɖʱ/

𑀡

ṇa

/ɳ/

ta-group

𑀢

/t̪/

𑀣

tha

/t̪ʰ/

𑀤

/d̪/

𑀥

dha

/d̪ʱ/

𑀦

/n/

pa-group

𑀧

/p/

𑀨

pha

/pʰ/

/b/

𑀪

bha

/bʱ/

𑀫

/m/

Vaggānaṃ paṭhamadutiyā so cāghoso. Ḷantāññe ghosā. (voiced and unvoiced consonants) ¶8

Ghosāghosasaññā ca ‘‘parasamaññā payoge’’ti saṅgahītā. Evaṃ liṅga, sabbanāma, pada, upasagga, nipāta, taddhita, ākhyāta, kammappavacanīyādisaññā ca.

The first and second letters in each group, and sa, are āghosa (unvoiced, or surds). The remaining letters until ḷa are ghosa (voiced, or sonants).

These terms (āghosa and ghosa) are adopted from Sanskrit grammar in accordance with Kaccāyana’s “parasamaññā payoge” rule, as are other terms such as:

liṅga (grammatical gender)
sabbanāma (pronoun)
pada (word)
upasagga (prefix)
nipāta (particle)
taddhita (derived word)
ākhyāta (verb)
kammappavacanīyādisaññā (other words representing actions)

Aṃiti niggahītaṃ (`aṃ` is niggahīta or “nasal consonant”) ¶9

Aṃiti akārato paraṃ yo bindu sūyate, taṃ niggahītaṃ nāma.

Bindu cūḷāmaṇākāro, niggahītanti vuccate.

Kevalassā ppayogattā, akāro sannidhīyate.

In aṃ, the “dot” that is heard after the letter 𑀅 (a) - written as 𑀅𑀁 (aṃ) - is called the niggahīta or “nasal consonant”.

The verse: the dot, shaped like a crest-jewel, is called the niggahīta; because it is not used on its own (kevalassa appayogattā), the letter a is placed beside it.

A, kavagga, hā kaṇḍajā, i, cavagga, yā tālujā, u, pavaggā oṭṭhajā, ṭavagga, ra, ḷā muddhajā, tavagga, la, sā dantajā, e kaṇṭhatālujo, o kaṇṭhoṭṭhajo, vo dantoṭṭhajo (Pronunciation) ¶10

Letters	Type	Pronunciation
`a`, `ā`, `ka`-group, `ha`	kaṇḍajā	guttural
`i`, `ī`, `ca`-group, `ya`	tālujā	palatal
`u`, `ū`, `pa`-group	oṭṭhajā	labial
`ṭa`-group, `ra`, `ḷa`	muddhajā	lingual
`ta`-group, `la`, `sa`	dantajā	dental
`e`	kaṇṭhatālujo	guttural-palatal
`o`	kaṇṭhoṭṭhajo	guttural-labial
`va`	dantoṭṭhajo	dento-labial

Sarasandhi (Vowel sandhi)

sandhi (junction) is the meeting of two letters with no other letter and no pause between them (¶46 sannidhi-abyavadhānato). When a word ending in a vowel meets a word beginning with a vowel, the rules of this section offer three routes. The usual one elides the first vowel; the others keep both vowels and change the first, or put a consonant between them.

flowchart LR
  A(["a vowel meets a vowel"]) --> B{"is a vowel elided?"}
  B -- "usually the first ¶11" --> C["the second vowel stands<br/>sometimes i, u → e, o ¶13<br/>sometimes lengthened ¶14"]
  B -- "or the second, when unlike ¶12" --> D["the first vowel stands<br/>sometimes lengthened ¶15"]
  B -- "neither" --> E["the first vowel changes<br/>e → y ¶16 · o, u → v ¶17<br/>i, ī → y, occasionally ¶19<br/>eva → riva, short vowel before ¶20<br/>abhi → abbh ¶22"]
  B -- "neither" --> F["a consonant is inserted ¶21<br/>y v m d n t r ḷ"]

Loka aggoityasmiṃ ‘‘pubbamadhoṭhita massaraṃ sarena viyojaye’’ti pubbabyañjanaṃ sarato puthakkātabbaṃ.

For example, loka + aggo (best in the world):

Before succeeding word, detach vowel from (vowel-less) consonant, in accordance to Kaccāyana’s “pubbaṃ adhoṭhitaṃ assaraṃ sarena viyojaye” rule: lo + (k + a) + aggo
Vowel should be separated before consonant: lok + a + (a + g) + go

Sarā sare lopaṃ (elide vowel before vowel) ¶11

Anantare sare pare sarā lopaṃ papponti.

‘‘Naye paraṃ yutte’’ti assaro byañjano parakkharaṃ netabbo lokaggo.

Elide vowel before another vowel: lok + (a + a) + ggo
The vowel-less consonant should be joined to the next vowel, in accordance to Kaccāyana’s “naye paraṃ yutte” rule: lo + (k + a) + ggo
Final result: lokaggo

Saretyasmiṃ opasilesiko kāsasattamī, tato vaṇṇakālabyavadhāne kāriyaṃ na hoti. Yathā-maṃ ahā sīti, ‘‘pamādamanuyuñjantī’’tyādigāthāyaṃ ‘janā appamāda’nti ca. Evaṃ sabbasandhīsu.

sare (“before a vowel”) is a locative of close contact (opasilesika): the rule operates only where the two vowels touch, so where a letter or a lapse of time intervenes (vaṇṇakālabyavadhāne) it does not operate.

Examples:

maṃ + ahāsi (no elision: the letter ṃ stands between the vowels)
in the verse starting with pamādamanuyuñjanti (18Dh 2.4: Bālanakkhattasaṃghuṭṭhavatthu):

pamādamanuyuñjanti bālā dummedhino janā |
appamādañca medhāvi dhanaṃ seṭṭhaṃ’va rakkhati ||

No vowel is elided between janā + appamādañca: the pause at the end of the second foot intervenes. So in all sandhis.

Anantaraṃ parassa sarassa lopaṃ vakkhati, tasmānena pubbassa lopo ñāyati, teneva sattamīniddiṭṭhassa paratāpi gamyate.

Regarding the elision following another vowel, the elision of the previous vowel is understood because of the use of the 7th (locative) inflection form for sare.

Saretyadhikāro. Pana ime pana imetīha-sarā lopaṃ itveva.

For the next rule concerning vowels, consider elision of vowel for the example of pana + ime

Vā paro asarūpo (or, next dissimilar vowel is elided) ¶12

Asamānarūpāsaramhā paro saro vā lupyate, paname, panime.

As the vowels are dissimilar, either vowel can be elided:

pana + ime = paname
pana + ime = panime

Bandhussa iva, na upetītīdha –

And now let’s consider the example of bandhussa + iva.

Kvacāsavaṇṇaṃ lutte (optionally, substitution after elision) ¶13

Sare lutte parasarassa kvaci asavaṇṇo hotīti i u iccetesaṃ ṭhānāsannā e o. Bandhusseva. Nopeti.

After elision of previous vowel, the next vowel is sometimes changed to a different vowel, i.e. i and u becomes e and o, the vowels nearest in place of articulation (ṭhānāsannā):¹

bandhussa + iva = bandhuss + (a + i) + va = bandhuss + (i→e) + va = bandhusseva
na + upeti = n + (a + u) + peti = n + (u→o) + peti = nopeti

Tatra ayaṃ, yāni idha, bahu upakāraṃ, saddhā idha, tathā upamantyetasmiṃ –

Consider the following examples:

tatra + ayaṃ
yāni + idha
bahu + upakāraṃ
saddhā + idha
tathā + upamaṃ

Dīghaṃ (next vowel becomes long) ¶14

Sare lutte paro saro kvaci ṭhānāsannaṃ dīghaṃ yāti. Tatrāyaṃ, yānīdha, bahūpakāraṃ, saddhīdha, tathūpamaṃ.

After elision of previous vowel, the next vowel is sometimes lengthened:

Examples:

tatra + ayaṃ = tatr + (a + a) + yaṃ = tatr + (a→ā) + yaṃ = tatrāyaṃ
yāni + idha = yān + (i + i) + dha = yān + (i→ī) + dha = yānīdha
bahu + upakāraṃ = bah + (u + u) + pakāraṃ = bah + (u→ū) + pakāraṃ = bahūpakāraṃ
saddhā + idha = saddh + (ā + i) + dha = saddh + (i→ī) + dha = saddhīdha
tathā + upamaṃ = tath + (ā + u) + pamaṃ = tath + (u→ū) + pamaṃ = tathūpamaṃ

Kiṃsu idhetyatra –

Consider example kiṃsu + idha.

Pubbo ca (also, the previous vowel becomes long) ¶15

Sare lutte pubbo ca kvaci dīghaṃ yāti. Kiṃ sūdha.

After (optional) elision of (next) vowel, the previous vowel is sometimes lengthened:

kiṃsu + idha = kiṃs + (u + i) + dha = kiṃs + (u→ū) + dha = kiṃsūdha

Te ajja, te ahaṃtettha –

Consider examples te + ajja and te + ahaṃ.

Yamedantassādeso (ending `e` is replaced with `y`) ¶16

Sare pare antassa ekārassa kvaci yo ādeso hoti, tyajja, ‘‘dīgha’’nti byañjane pare kvaci dīgho, tyāhaṃ.

e ending followed by another vowel is sometimes replaced by y; the vowel that then stands before a consonant is sometimes also lengthened (dīgho), in accordance with Kaccāyana’s consonant rule “Dīghaṃ”.²

te + ajja = t + (e→y + a) + jja = tyajja
te + ahaṃ = t + (e→y + a) + haṃ = ty + (a→ā) + haṃ = tyāhaṃ

Kvacīti kiṃ. Nettha.

What does “sometimes” mean? There are exceptions to the rule.

ne + ettha = n + (e + e) + ttha = nettha (¶16 not applied; the final e is elided by ¶11 instead — ne = te, “those”. In practice e becomes y only before an unlike vowel, as in tyajja and tyāhaṃ, never before another e; Mitra (1935) writes this restriction into the rule as “a dissimilar vowel”)

So assa, anu etityettha –

Consider examples so + assa and anu + eti.

Vamodudantānaṃ (ending `o` and `u` is replaced with `v`) ¶17

Sare pare anto kārukārānaṃ kvaci vo ādeso hoti. Svassa, anveti.

o and u ending followed by another vowel is sometimes replaced with v.

so + assa = s + (o→v + a) + ssa = svassa
anu + eti = an + (u→v + e) + ti = anveti

Kvacīti kiṃ. Tayassu, sametāyasmā.

Exceptions:

tayo + assu = tay + (o + a) + ssu = tayassu (rule not applied)
sametu + āyasmā = samet + (u + ā) + yasmā = sametāyasmā (rule not applied)

Idha ahaṃ tīdha –

Consider example idha + ahaṃ.

Do dhassa ca. (also, `d` from `dh`) ¶18

Sare pare dhassa kvaci do hoti. Dīghe – idāhaṃ. Kvacīti kiṃ. Idheva. Cakārena byañjanepi, idha bhikkhave.

dh ending followed by another vowel is sometimes replaced with d.

idha + ahaṃ = idh + (a + a) + haṃ = i + (dh→d + a) + haṃ = id + (a→ā) + haṃ = idāhaṃ (dīghe, “with lengthening”: the a of ahaṃ is lengthened by ¶14 “Dīghaṃ” because the vowel before it was elided)

Exceptions:

idha + eva = idh + (a + e) + va = idheva (rule not applied)

The replacement may occur even if dha is followed by a consonant letter.

idha + bhikkhave = i + (dha→da) + bhikkhave = ida bhikkhave (two words, as the source and Mitra (1935) print it)

Pati antaṃ, vutti assetīha –

Consider examples pati + antaṃ, and vutti + assa.

Ivaṇṇo yannavā (occasionally, `i`-vaṇṇa is replaced with `y`) ¶19

Sare pare ivaṇṇassa yo navā hoti. Kata yakārassa tissa ‘‘sabbo cantī’’ti kvaci cādese ‘‘paradvebhāvo ṭhāne’’ti sarato parabyañjanassa ṭhānāsannavasā dvittaṃ. Paccantaṃ, vutyassa.

i-vaṇṇa (i and ī) followed by another vowel is occasionally replaced with y. Sometimes, after ti is changed to y, ty becomes c (based on Kaccāyana’s “sabbo canti” rule), next consonant following vowel can be doubled (based on Kaccāyana’s “paradvebhāvo ṭhāne” rule).³

pati + antaṃ = pa + (ti→ty→c→cc + a) + ntaṃ = paccantaṃ
vutti + assa = vut + (ti→y + a) + ssa = vutyassa

Navāti kiṃ. Paṭaggi,

Exception:

pati + aggi = pat + (i + a) + ggi (rule not applied) = pa + (t→ṭ) + aggi = paṭaggi

Ettha ‘‘kvaci paṭi patisse’’ti patissa paṭi, vaṇṇaggahaṇaṃ sabbattha rassadīgha saṅgahaṇatthaṃ.

Here, following Kaccāyana’s “kvaci paṭi patisse” rule, t becomes ṭ. Using the term vaṇṇa refers to both short and long versions of a phoneme.

Yathā evetīha –

Consider example yathā + eva.

Evādissa ri pubbo ca rasso (start of `eva` replaced with `ri`, and shortening of previous vowel) ¶20

Sarato parassa evassādiekāro rittaṃ navā yāti. Pubbo ca ṭhānāsannaṃ rassaṃ. Yathariva. Yatheva.

Occasionally, the e of eva following vowel is changed to ri. The previous vowel is shortened.

yathā + eva = yath + (ā→a + e→ri) + va = yathariva
yathā + eva = yath + (ā + e) + va = yatheva (rule not applied)

Na imassa, ti aṅgikaṃ, lahu essati, atta atthaṃ⁴, ito āyati, tasmā iha, sabbhi eva, cha abhiññā, putha eva, pā evatīha vā tveva –

Consider examples:

na + imassa
ti + aṅgikaṃ
lahu + essati
atta + atthaṃ
ito + āyati
tasmā + iha
sabbhi + eva
cha + abhiññā
putha + eva
pā + eva

Ya va ma da na ta ra ḷā cāgamā. (insertion of `y`, `v`, `m`, `d`, `n`, `t`, `r`, `ḷ`, `c`, `g`) ¶21

Sare pare yādayo āgamā vā honti, cakārena go ca. Nayimassa, tivaṅgikaṃ, lahumessati, attadatthaṃ, itonāyati, tasmātiha, sabbhireva, chaḷabhiññā, puthageva, ‘‘rassa’’nti byañjane pare kvaci rasso. Pageva.

y, v, m, d, n, t, r, ḷ can be inserted following a vowel. g also, by the sutta’s ca (cakārena go ca; the g of puthageva and pageva is Kaccāyana’s “Gosare puthassāgamo kvaci” and “Pāssa canto rasso”). Before a consonant the vowel is sometimes shortened (‘‘rassa’’nti, Kaccāyana’s “Rassaṃ” rule): pageva.

na + imassa = n + (a + y + i) + massa = nayimassa
ti + aṅgikaṃ = t + (i + v + a) + ṅgikaṃ = tivaṅgikaṃ
lahu + essati = lah + (u + m + e) + ssati = lahumessati
atta + atthaṃ = att + (a + d + a) + tthaṃ = attadatthaṃ (“one’s own good”)
ito + āyati = it + (o + n + ā) + yati = itonāyati
tasmā + iha = tasm + (ā + t + i) + ha = tasmātiha
sabbhi + eva = sabbh + (i + r + e) + va = sabbhireva
cha + abhiññā = ch + (a + ḷ + a) + bhiññā = chaḷabhiññā
putha + eva = puth + (a + g + e) + va = puthageva
pā + eva = p + (ā→a + g + e) + va = pageva

Vāti kiṃ. Cha abhiññā, putha eva, pā eva.

Ettha ‘‘sare kvacī’’ti sarānaṃ pakati hoti, sassarūpameva, na vikārotyattho.

Exceptions:

cha abhiññā (no sandhi insertion)
putha eva (no sandhi insertion)
pā eva (no sandhi insertion)

According to Kaccāyana “sare kvaci” rule, sometimes there is no morphological transformation of original vowels.

Abhi uggatotyatra

Consider example of abhi + uggato.

Abbho abhi (`abbh` from `abhi`) ¶22

‘‘Abbho abhī’’ti abhissa abbho. Abbhuggato.

According to Kaccāyana “Abbho abhi” rule, abhi can become abbh

abhi + uggato = (abhi→abbh) + uggato = abbhuggato

Byañjanasandhi (Consonant sandhi)

Byañjanetyadhikāro. Kvacītveva. So bhikkhu, kacci nu tvaṃ, jānema tantīha –

In the rules that follow, byañjane (“before a consonant”) and kvaci (“sometimes”) are understood (adhikāra, a heading word carried forward into each rule).

Consider examples:

so + bhikkhu
kacci + nu + tvaṃ
jānema + taṃ

Lopañca tatrākāro (elision, then insertion of `a` in place) ¶23

Byañjane pare sarānaṃ kvaci lopohoti, tatra lutte ṭhāne akārāgamo, cakārena okārukārāpi. Sabhikkhu kaccino tvaṃ, jānemu taṃ.

Vowel is sometimes elided if followed by consonant, and in place of the elision a is inserted. Instead of a, o and u can also be inserted.

so + bhikkhu = s + (o→a) + bhikkhu = sabhikkhu
kacci + nu + tvaṃ = kacci + n + (u→o) + tvaṃ = kaccino tvaṃ
jānema + taṃ = jānem + (a→u) + taṃ = jānemu taṃ

Kvacīti kiṃ, somuni.

Exception:

so + muni = somuni (no rule applied)

Ughoso , ākhātantīha – dvebhāve ṭhāne itveva.

Consider examples, still under “doubling in place” (dvebhāve ṭhāne, the “Paradvebhāvo ṭhāne” quoted at ¶19):

u + ghoso
ākhātaṃ

Vagge ghosāghosānaṃ tatiyapaṭhamā. (in `vagga` `ghosa/āghosa` 3rd and 1st) ¶24

Vagge ghosāghosānaṃ catutthadutiyānaṃ tabbagge tatiyapaṭhamā honti yathāsaṅkhyaṃ yutte ṭhāne, ugghoso, rasse akkhātaṃ.

Within a vagga (consonant group), when the fourth or the second letter — the voiced and the unvoiced aspirate — is doubled after a vowel, the first member of the pair is the third or the first letter of that group respectively (ggh for gh, kkh for kh), where this is fitting (yutte ṭhāne). In akkhātaṃ the ā is also shortened (rasse), by Kaccāyana’s “Rassaṃ” rule.

u + ghoso = u + (gh→ghgh) + oso = u + (ghgh→ggh) + oso = ugghoso
ā + khātaṃ = (ā→a) + khātaṃ = a + (kh→khkh) + ātaṃ = a + (khkh→kkh) + ātaṃ = akkhātaṃ

Para sahassaṃ, atippakhotīha

Consider examples:

para + sahassaṃ
atippa + kho

kvaci o byañjane (sometimes `o` before consonant) ¶25

‘‘kvaci o byañjane’’ti okārāgamo. Parosahassaṃ. Gāgame ca, atippagokho.

In accordance to Kaccāyana rule “kvaci o byañjane”, in some cases, the vowel o is inserted before a consonant.

Examples:

para + sahassaṃ = par + (a + o) + sahassaṃ = parosahassaṃ (“more than a thousand”)
atippa + kho = atippa + (g + o) + kho = atippagokho (after insertion of g in accordance with ¶21 (“Ya va ma da na ta ra ḷā cāgamā”))

Ava naddhātyatra

Consider transformation of ava.

o avasse (`o` from `ava`) ¶26

‘‘o avasse’’ti kvaci avassa o. Onaddhā.

According to Kaccāyana rule “o avasse”, ava is sometimes changed to o

Example:

ava + naddhā = (ava→o) + naddhā = onaddhā

Kvacīti kiṃ. Avasussatu.

Exception:

ava + sussatu = avasussatu (no rule applied)

Niggahītasandhi (Niggahita Sandhi)

Niggahītantyadhikāro . Kiṃ kato, saṃ jāto, saṃ ṭhito, taṃ dhanaṃ, taṃ mittantiha –

In the rules of this section, niggahīta is the governing rule (adhikāra): each rule concerns a word-final ṃ.

Consider examples:

kiṃ + kato
saṃ + jāto
saṃ + ṭhito
taṃ + dhanaṃ
taṃ + mittaṃ

Vaggantaṃ vā vagge. (sometimes, end of `vagga` from ṃ before `vagga`) ¶27

Vaggabyañjane pare bindussa tabbagganto vā hoti. Kiṅkato, sañjāto, saṇṭhito, tandhanaṃ, tammittaṃ.

When a word ending in niggahīta (ṃ) is followed by a consonant of one of the five consonant groups (vagga), the niggahīta may optionally be changed to the nasal consonant of that group (the letter with the “dot” in that group).

Examples:

kiṃ + kato = ki + (ṃ→ṅ + k) + ato = kiṅkato
saṃ + jāto = sa + (ṃ→ñ + j) + āto = sañjāto
saṃ + ṭhito = sa + (ṃ→ṇ + ṭ) + hito = saṇṭhito
taṃ + dhanaṃ = ta + (ṃ→n + d) + hanaṃ = tandhanaṃ
taṃ + mittaṃ = ta + (ṃ→m + m) + ittaṃ = tammittaṃ

Vāti kiṃ. Na taṃ kammaṃ.

Exception:

na + taṃ + kammaṃ = na taṃ kammaṃ (rule is not applied)

Vākāreneva le lo ca. Pulliṅgaṃ.

Similarly, ṃ can become l before another l

Example:

puṃ + liṅgaṃ = pu + (ṃ→l + l) + iṅgaṃ = pulliṅgaṃ

Vātyadhikāro . Evaṃ assa, etaṃ avocetīha –

vā (optionally) is understood in the rules that follow. Consider examples:

evaṃ + assa
etaṃ + avoca

Madā sare. (`m` and `d` before vowels) ¶28

Sare pare binduno ma dā vā honti. Evamassa, etadavoca.

When a vowel follows a niggahita, sometimes the niggahita can be changed to m or d

Examples:

evaṃ + assa = eva + (ṃ→m + a) + ssa = evamassa
etaṃ + avoca = eta + (ṃ→d + a) + voca = etadavoca

Vāti kiṃ. Maṃ ajini.

Exception:

maṃ + ajini (rule not applied)

Taṃ eva, taṃ hītīha –

Examples:

taṃ + eva
taṃ + hi

Eheñaṃ. (from `e`, `h` to `ñ`) ¶29

Ekāre, he ca pare binduno ño vā hoti. Dvitte – taññeva, tameva. Tañhi, taṃ hi.

When e or h follows the niggahīta (ṃ), it may optionally be changed to ñ; before e this is with doubling (dvitte, by Kaccāyana’s “Paradvebhāvo ṭhāne”).

Examples:

taṃ + eva = ta + (ṃ→ñ + e) + va = ta + (ñ→ññ + e) + va = taññeva
taṃ + hi = ta + (ṃ→ñ + h) + i = tañhi

Exceptions:

taṃ + eva = ta + (ṃ→m + e) + va = tameva
taṃ + hi = taṃhi (rule not applied)

Saṃyogotīha –

Consider example saṃ + yogo.

Saye ca. (and with `y`) ¶30

Yakāre pare tena saha binduno ño vā hoti. Dvitte – saññogo, saṃyogo.

And for y: when the niggahīta (ṃ) is before y, the two together may optionally be changed to ñ, with doubling (dvitte, by Kaccāyana’s “Paradvebhāvo ṭhāne”).

Example:

saṃ + yogo = sa + (ṃy→ñ) + ogo = sa + (ñ→ññ) + ogo = saññogo

Exception:

saṃ + yogo = saṃyogo (rule not applied)

Cakkhu aniccaṃ, ava sirotīha - āgamo, kvacitveva.

Consider examples - optional insertion of ṃ:

cakkhu aniccaṃ
ava siro

Niggahītañca. (and niggahitaṃ) ¶31

Sare, byañjane vā pare kvaci bindvāgamo hoti. Cakkhuṃaniccaṃ, avaṃsiro.

Sometimes, a niggahita (ṃ) is inserted before a following vowel or consonant.

cakkhu + aniccaṃ = cakkh + (u + ṃ + a) + niccaṃ = cakkhuṃaniccaṃ
ava + siro = av + (a + ṃ + s) + iro = avaṃsiro

Vidūnaṃ aggaṃ, tāsaṃ ahaṃtīha –

Consider examples:

vidūnaṃ + aggaṃ
tāsaṃ + ahaṃ

Kvaci lopaṃ (sometimes elision) ¶32

‘‘Kvaci lopaṃ’’ti sare bindulopo, vidūnaggaṃ. Dīghetāsāhaṃ.

According to Kaccāyana’s rule “Kvaci lopaṃ”, the niggahīta (ṃ) before a vowel is sometimes elided. In tāsāhaṃ this is with lengthening (dīghe): once the ṃ has gone, tāsa + ahaṃ has two as, one is elided by ¶11 and the other lengthened by ¶14 “Dīghaṃ”.

Examples:

vidūnaṃ + aggaṃ = vidūna + (ṃ + a) + ggaṃ = vidūn + (a + a) + ggaṃ (by ¶11) = vidūnaggaṃ
tāsaṃ + ahaṃ = tāsa + (ṃ + a) + haṃ = tās + (a + a) + haṃ (by ¶11) = tās + (a→ā) + haṃ (by ¶14 “Dīghaṃ”) = tāsāhaṃ

Buddhānaṃ sāsanaṃ, saṃ rāgotīha –

Consider examples:

buddhānaṃ + sāsanaṃ
saṃ + rāgo

Byañjane ca (and before a consonant) ¶33

‘‘Byañjane ce’’ti bindulopo, buddhānasāsanaṃ. Dīghesārāgo.

According to Kaccāyana’s rule “Byañjane ca” (the vutti’s ce’ti is ca + iti in sandhi), the niggahīta (ṃ) is sometimes elided even before a consonant.⁵ In sārāgo this is with lengthening (dīghe): the a left standing before r is lengthened by Kaccāyana’s consonant rule “Dīghaṃ”.

buddhānaṃ + sāsanaṃ = buddhāna + (ṃ + s) + āsanaṃ = buddhānasāsanaṃ
saṃ + rāgo = sa + (ṃ + r) + āgo = s + (a→ā) + rāgo (by Kaccāyana’s “Dīghaṃ”, §25 (37)) = sārāgo

Bījaṃ ivetīha –

Consider example bījaṃ iva.

Paro vā saro. (or next vowel) ¶34

Binduto paro saro vā lupyate, bījaṃva.

The next vowel following a niggahita (ṃ) is optionally elided.

bījaṃ + iva = bīja + (ṃ + i) + va = bījaṃva

Evaṃ assetīha –

Consider example evaṃ assa.

Byañjano ca visaññogo. (and detached consonant) ¶35

Binduto pare sare lutte saṃyogo byañjano vinaṭṭhasaṃyogo hotīti pubbasalopo. Evaṃsa.

When the vowel after the niggahīta (ṃ) is elided (by ¶34), the conjunct consonant that followed it loses its conjunction (vinaṭṭhasaṃyogo), so the first s is dropped (pubbasalopo): ṃss becomes ṃs.

evaṃ + assa = eva + (ṃ + a) + ssa (by ¶34 “Paro vā saro”) = evaṃ + (s) + sa = evaṃsa

Vomissaka sandhī (Combination sandhi)

Anupadiṭṭhānaṃ vuttayogato. (besides those seen, apply combinations) ¶36

Idhāniddiṭṭhā sandhayo vuttānusārena ñe yyā, yathā – yadi evaṃ, bodhi aṅgātīha – yādese iminā suttena dayakārasaṃyogassa jo, dhayakārasaṃyogassa jho, dvitte – yajjevaṃ, bojjhaṅgā.

For unexplained sandhi conjunctions not previously seen, eg.:

yadi + evaṃ
bodhi + aṅgā

Apply whichever previously stated rules are appropriate. Here, once the i has become y (yādese, by ¶19 “Ivaṇṇo yannavā”), by this rule:

the conjunct dy becomes j
the conjunct dhy becomes jh

together with doubling (dvitte):

yadi + evaṃ = yad + (i→y + e) + vaṃ (by ¶19) = ya + (dy→j + e) + vaṃ = ya + (j→jj + e) + vaṃ = yajjevaṃ
bodhi + aṅgā = bodh + (i→y + a) + ṅgā (by ¶19) = bo + (dhy→jh + a) + ṅgā = bo + (jh→jjh + a) + ṅgā = bojjhaṅgā

Asadisasaṃyoge ekasarūpatā ca. (and dissimilar conjunctions becomes similar) ¶37

Pari esanātīha – yādese rakārassa yo, payyesanā.

Example: in pari + esanā, once the i has become y (yādese, by ¶19 “Ivaṇṇo yannavā”), the r of the unlike conjunct ry is assimilated to the y:

pari + esanā = par + (i→y + e) + sanā (by ¶19) = pa + (ry→yy + e) + sanā = payyesanā

Vaṇṇānaṃ bahuttaṃ, viparītatā ca. (and various substitutions, reversals) ¶38

Sarati, iti eva, sā itthī, busaṃ eva, bahu ābādho, adhi abhavi, sukhaṃ, dukkhaṃ, jīvotīha –

Consider examples:

√sar + a + ti
iti + eva
sā + itthī
busaṃ + eva
bahu + ābādho
adhi abhavi
sukhaṃ
dukkhaṃ
jīvo

Māgamo sakāre akārassa u ca, sumarati.

After ma is inserted after sa (an insertion of the kind of ¶21 “Ya va ma da na ta ra ḷā cāgamā”), the preceding a becomes u (the u of ¶23 “Lopañca tatrākāro”, as in jānemu taṃ):

√sar + a + ti = s + (a + ma) + rati = s + (a→u + ma) + rati = sumarati

Issa vo, itveva.

i becomes v (an extension of ¶17 “Vamodudantānaṃ”, which gives v for o and u):

iti + eva = it + (i→v + e) + va = itveva

Paralope ākārassa o, sotthī.

When the next vowel is elided (paralope), ā becomes o:

sā + itthī = s + (ā + i) + tthī = s + (ā→o) + tthī = sotthī

Mādese , pubbadīghe ca ekārassa i. Busāmiva.

When m is substituted for the niggahīta (mādese, by ¶28 “Madā sare”) and the preceding vowel is lengthened (pubbadīghe), e becomes i:⁶

busaṃ + eva = busa + (ṃ→m + e) + va (by ¶28) = bus + (a→ā) + meva = busām + (e→i) + va = busāmiva

Vādese havakāravipariyayo. Bavhābādho.

When v is substituted (vādese, by ¶17 “Vamodudantānaṃ”), h and v are reversed:

bahu + ābādho = bah + (u→v) + ābādho (by ¶17) = ba + (hv→vh) + ābādho = bavhābādho

Adhissa kvaci addho, dīghe-addhābhavi.

Sometimes adhi becomes addh, with lengthening of the following vowel (dīghe, as by ¶14 “Dīghaṃ”):

adhi + abhavi = (adhi→addh) + abhavi = addh + (a→ā) + bhavi = addhābhavi

Binduno, okārassa ca e. Sukhe, dukkhe, jīve.

aṃ and o becomes e:⁷

sukhaṃ = sukh + (aṃ→e) = sukhe
dukkhaṃ = dukkh + (aṃ→e) = dukkhe
jīvo = jīv + (o→e) = jīve

Radānaṃ ḷo (`r` and `d` replaced by `ḷ`) ¶39

paḷibodho, pariḷāho.

Examples:

pari + bodho = pa + (r→ḷ) + i + bodho = paḷibodho
pari + dāho = pari + (d→ḷ) + āho = pariḷāho

Sare, byañjane vā pare binduno kvaci mo. (Sometimes `ṃ` can become `ma` before a vowel or consonant.) ¶40

Mama abhāsi, buddhama saraṇaṃ, pubbe mo paraṃ na netabbo ayuttattā.

Sometimes m replaces the niggahīta (ṃ) before a vowel or a consonant.⁸

Examples:

maṃ + abhāsi = ma + (ṃ→m) + abhāsi = mam abhāsi
buddhaṃ + saraṇaṃ = buddha + (ṃ→m) + saraṇaṃ = buddham saraṇaṃ

pubbe mo paraṃ na netabbo ayuttattā

The m of the preceding word is not to be carried over to the following word, because that would be unfitting (ayuttattā, the converse of ¶11’s “naye paraṃ yutte”): the results stay two words, mam abhāsi, not mamabhāsi.

Binduto parasarāna maññassaratāpi. (after an `ṃ` change to another letter, the following vowel can also change to another vowel) ¶41

Taṃ iminā, evaṃ imaṃ, kiṃ ahaṃ tīha-issa a. Tadaminā.

Issa u, akārassa ca e, bindulopādo. Evumaṃ, kehaṃ.

The vowels after a niggahīta (ṃ) may also become other vowels (aññassaratā): i becomes a in tadaminā; i becomes u and a becomes e in evumaṃ and kehaṃ, with elision of the niggahīta and so on (bindulopādo: ¶32, then ¶11).

Examples:

taṃ + iminā = ta + (ṃ→d + i) + minā (by ¶28 “Madā sare”) = tad + (i→a) + minā = tadaminā
evaṃ + imaṃ = eva + (ṃ + i) + maṃ (by ¶32 “Kvaci lopaṃ”) = ev + (a + i) + maṃ (by ¶11) = ev + (i→u) + maṃ = evumaṃ
kiṃ + ahaṃ = ki + (ṃ + a) + haṃ (by ¶32) = k + (i + a) + haṃ (by ¶11) = k + (a→e) + haṃ = kehaṃ⁹

Vākyasukhuccāraṇatthaṃ, chandahānitthañca vaṇṇalopopi. (For ease of pronunciation in sentences and to maintain meter in verses, phonemes can be dropped) ¶42

Paṭisaṅkhāya yonisotīha – pubbayalopo, paṭisaṅkhāyoniso.

Example:

paṭisaṅkhāya + yoniso = paṭisaṅkhā + (ya) + yoniso = paṭisaṅkhāyoniso

Alābūnityādo akāralopo. (eg. `alābūni`, the initial `a` can be dropped) ¶43

Lābūni sīdanti, silā plavanti.

Example: the initial a of alābūni is dropped for the metre (chandahānitthaṃ, ¶42):

alābūni + sīdanti = a + lābūni + sīdanti = lābūni sīdanti

Vutyabhedāya vikāropi. (Changes can also occur to avoid breaking metrical rules) ¶44

Akaramhase tetyādo sakāre garuno ekārassa iminā lahuakāro, akaramhasa te kiccaṃ.

Example, the long e in se should become short a:

akaramhase te = akaramhas + (e→a) + te = akaramhasa te

Akkharaniyamo chandaṃ, garulahuniyamo bhave vutti. (The arrangement of letters forms meter, and the pattern of heavy and light syllables forms rhythm.) ¶45

Dīgho, saṃyogādipubbo rasso ca garu, lahu tu rasso. Yathā- ā, assa, aṃ, a.

A long vowel is heavy, and so is a short vowel followed by a conjunct consonant or a niggahīta (saṃyogādipubbo, “before a conjunct and the like”); a short vowel is light.

Examples:

ā (heavy)
assa (heavy + light)
aṃ (heavy)
a (light)

Evamaññāpi viññeyyā, saṃhitā tantiyā hitā. (Other sandhi rules should be understood as beneficial connections to the text.) ¶46

Saṃhitāti ca vaṇṇānaṃ, sannidhabyavadhānato.

And saṃhitā (connection) is so called from the letters’ contiguity, with nothing intervening (sannidhabyavadhānato).

Mitra (1935) adds in brackets that ī “is also changed into o” and ā “into o”. As printed these pairs contradict the vutti’s ṭhānāsannā, which admits only i → e and u → o; they were presumably meant as ī → e, ū → o (the vaṇṇa reading) and garbled in print. The vutti’s pairs stand here. ↩
The vutti’s ‘‘dīgha’’nti byañjane pare kvaci dīgho quotes Kaccāyana’s consonant rule §25 (37) Sarā kho byañjane pare kvaci dīghaṃ papponti, not the after-elision “Dīghaṃ” of ¶14 (§15 (17)): in te + ahaṃ → tyāhaṃ nothing is elided, e → y being a substitution. Mitra (1935) cites the same rule (his Kaccāyana 1.3.3). An earlier version of this page linked §15 (17); the Kaccāyana page’s own table for tyāhaṃ, under §17 (19), reaches the word by the other route, lengthening by §15 (17) before the y substitution. ↩
Mitra (1935) collapses the two rules into one statement, “ty is sometimes changed into cc”, and reuses it for yajjevaṃ (¶36); the vutti names both, “Sabbo canti” for ty → c and “Paradvebhāvo ṭhāne” for the doubling, and they are kept apart here. Both the CSCD and the 1935 edition print the input as Pati antuṃ, a shared misprint for antaṃ (the result is paccantaṃ), which the lead-in above silently corrects. ↩
The CSCD prints attha atthaṃ in the lead-in; the first word is atta “self”, as the result attadatthaṃ requires (attha + atthaṃ would give atthadatthaṃ) and as the canon has it (attadatthaṃ paratthena, “one’s own good”, (18Dh 12.10: Attadatthattheravatthu)). Mitra (1935) reads atta + atthaṁ, and the Rūpasiddhi page derives atta atthaṃ → attadatthaṃ in the same way; the lead-in and the derivation are corrected accordingly. ↩
The vutti’s ‘‘Byañjane ce’’ti is Byañjane ca + iti in sandhi (as ‘‘o avasse’’ti at ¶26 is O avassa + iti), that is Kaccāyana’s §39 (54) “Byañjane ca”, “and before a consonant”; an earlier version of this page read ce as “if”. Mitra (1935) renders the ca as “even when a consonant follows” and, like this page, attributes the ā of sārāgo to the consonant rule “Dīghaṃ” (§25 (37), his Kaccāyana 1.3.3). ↩
Mitra (1935) starts from busā + eva, inserts m by his II.11 (our ¶21), shortens the ā by Kaccāyana’s “Rassaṃ” and prints busamiva. The vutti’s busaṃ eva … mādese, pubbadīghe ca has the m substituted for the niggahīta and the preceding vowel lengthened, so busāmiva stands. ↩
sukhe, dukkhe, jīve are Pakudha Kaccāyana’s words for three of the seven bodies in the Sāmaññaphalasutta (6D 2.3.4: Pakudhakaccāyanavāda), sukhe dukkhe jīve sattame: ①⨀ forms ending in -e where the usual forms end in -aṃ and -o. Mitra (1935) has no entry for this last set of examples. ↩
The CSCD prints Mama abhāsi, buddhama saraṇaṃ, where the final ma is the graphic form of a word-final m; an earlier version of this page read the first input as maṃ + bhāsi, which is in neither edition, and joined the second into one word. Mitra (1935) reads the first as maṁ + ahāsi = mamahāsi (echoing ¶11’s maṃ ahāsi) and prints the second as buddham saraṇaṁ, two words; neither result is attested in the Tipiṭaka, so the first input cannot be settled from usage, and the CSCD’s abhāsi is kept. ↩
Mitra (1935) says the ṃ is elided and “i is changed into e”; the vutti’s akārassa ca e makes the a of ahaṃ the vowel that becomes e, after the i and ṃ of kiṃ are dropped (bindulopādo), which is what the derivation shows. For evumaṃ his account and this page agree: the ṃ is elided (his Kaccāyana “Kvaci lopaṁ”), then i → u; the v is the lemma’s own. ↩
Mitra (1935) agrees: he quotes only 'lāpūni sīdanti (a misprint for lābūni) and calls the elision metrical. An earlier version of this page listed silā plavanti as a separate example. ↩

1. Sandhi kaṇḍa

Saññā (Representation)

Akkharāpādayo ekacattālīsaṃ (41 letters) ¶2

Tatthodantāsarā aṭṭha (8 vowels ending with o) ¶3

Lahumattā tayo rassā (3 short vowels) ¶4

Aññe dīghā (other vowels are long) ¶5

Sesā byañjanā (The rest are consonants) ¶6

Vaggā pañcapañcaso mantā (Five groups of 5 ending with ma) ¶7

Vaggānaṃ paṭhamadutiyā so cāghoso. Ḷantāññe ghosā. (voiced and unvoiced consonants) ¶8

Aṃiti niggahītaṃ (aṃ is niggahīta or “nasal consonant”) ¶9

A, kavagga, hā kaṇḍajā, i, cavagga, yā tālujā, u, pavaggā oṭṭhajā, ṭavagga, ra, ḷā muddhajā, tavagga, la, sā dantajā, e kaṇṭhatālujo, o kaṇṭhoṭṭhajo, vo dantoṭṭhajo (Pronunciation) ¶10

Sarasandhi (Vowel sandhi)

Sarā sare lopaṃ (elide vowel before vowel) ¶11

Vā paro asarūpo (or, next dissimilar vowel is elided) ¶12

Kvacāsavaṇṇaṃ lutte (optionally, substitution after elision) ¶13

Dīghaṃ (next vowel becomes long) ¶14

Pubbo ca (also, the previous vowel becomes long) ¶15

Yamedantassādeso (ending e is replaced with y) ¶16

Vamodudantānaṃ (ending o and u is replaced with v) ¶17

Do dhassa ca. (also, d from dh) ¶18

Ivaṇṇo yannavā (occasionally, i-vaṇṇa is replaced with y) ¶19

Evādissa ri pubbo ca rasso (start of eva replaced with ri, and shortening of previous vowel) ¶20

Ya va ma da na ta ra ḷā cāgamā. (insertion of y, v, m, d, n, t, r, ḷ, c, g) ¶21

Abbho abhi (abbh from abhi) ¶22

Byañjanasandhi (Consonant sandhi)

Lopañca tatrākāro (elision, then insertion of a in place) ¶23

Vagge ghosāghosānaṃ tatiyapaṭhamā. (in vagga ghosa/āghosa 3rd and 1st) ¶24

kvaci o byañjane (sometimes o before consonant) ¶25

o avasse (o from ava) ¶26

Niggahītasandhi (Niggahita Sandhi)

Vaggantaṃ vā vagge. (sometimes, end of vagga from ṃ before vagga) ¶27

Madā sare. (m and d before vowels) ¶28

Eheñaṃ. (from e, h to ñ) ¶29

Saye ca. (and with y) ¶30

Niggahītañca. (and niggahitaṃ) ¶31

Kvaci lopaṃ (sometimes elision) ¶32

Byañjane ca (and before a consonant) ¶33

Paro vā saro. (or next vowel) ¶34

Byañjano ca visaññogo. (and detached consonant) ¶35

Vomissaka sandhī (Combination sandhi)

Anupadiṭṭhānaṃ vuttayogato. (besides those seen, apply combinations) ¶36

Asadisasaṃyoge ekasarūpatā ca. (and dissimilar conjunctions becomes similar) ¶37

Vaṇṇānaṃ bahuttaṃ, viparītatā ca. (and various substitutions, reversals) ¶38

Radānaṃ ḷo (r and d replaced by ḷ) ¶39

Sare, byañjane vā pare binduno kvaci mo. (Sometimes ṃ can become ma before a vowel or consonant.) ¶40

Binduto parasarāna maññassaratāpi. (after an ṃ change to another letter, the following vowel can also change to another vowel) ¶41

Vākyasukhuccāraṇatthaṃ, chandahānitthañca vaṇṇalopopi. (For ease of pronunciation in sentences and to maintain meter in verses, phonemes can be dropped) ¶42

Alābūnityādo akāralopo. (eg. alābūni, the initial a can be dropped) ¶43

Vutyabhedāya vikāropi. (Changes can also occur to avoid breaking metrical rules) ¶44

Akkharaniyamo chandaṃ, garulahuniyamo bhave vutti. (The arrangement of letters forms meter, and the pattern of heavy and light syllables forms rhythm.) ¶45

Evamaññāpi viññeyyā, saṃhitā tantiyā hitā. (Other sandhi rules should be understood as beneficial connections to the text.) ¶46

Footnotes

Tatthodantāsarā aṭṭha (8 vowels ending with `o`) ¶3

Vaggā pañcapañcaso mantā (Five groups of 5 ending with `ma`) ¶7

Aṃiti niggahītaṃ (`aṃ` is niggahīta or “nasal consonant”) ¶9

Yamedantassādeso (ending `e` is replaced with `y`) ¶16

Vamodudantānaṃ (ending `o` and `u` is replaced with `v`) ¶17

Do dhassa ca. (also, `d` from `dh`) ¶18

Ivaṇṇo yannavā (occasionally, `i`-vaṇṇa is replaced with `y`) ¶19

Evādissa ri pubbo ca rasso (start of `eva` replaced with `ri`, and shortening of previous vowel) ¶20

Ya va ma da na ta ra ḷā cāgamā. (insertion of `y`, `v`, `m`, `d`, `n`, `t`, `r`, `ḷ`, `c`, `g`) ¶21

Abbho abhi (`abbh` from `abhi`) ¶22

Lopañca tatrākāro (elision, then insertion of `a` in place) ¶23

Vagge ghosāghosānaṃ tatiyapaṭhamā. (in `vagga` `ghosa/āghosa` 3rd and 1st) ¶24

kvaci o byañjane (sometimes `o` before consonant) ¶25

o avasse (`o` from `ava`) ¶26

Vaggantaṃ vā vagge. (sometimes, end of `vagga` from ṃ before `vagga`) ¶27

Madā sare. (`m` and `d` before vowels) ¶28

Eheñaṃ. (from `e`, `h` to `ñ`) ¶29

Saye ca. (and with `y`) ¶30

Radānaṃ ḷo (`r` and `d` replaced by `ḷ`) ¶39

Sare, byañjane vā pare binduno kvaci mo. (Sometimes `ṃ` can become `ma` before a vowel or consonant.) ¶40

Binduto parasarāna maññassaratāpi. (after an `ṃ` change to another letter, the following vowel can also change to another vowel) ¶41

Alābūnityādo akāralopo. (eg. `alābūni`, the initial `a` can be dropped) ¶43