I–V. Saññā and Sandhi

1.¹ Akkharāpādayo ekacattālīsaṁ.

Sūtra can be traced.

The letters beginning with a are forty-one, useful to Suttanta.

They are:
a ā i ī u ū e o,
k, kh, g, gh, ṅ,
c, ch, j, jh, ñ,
ṭ, ṭh, ḍ, ḍh, ṇ,
t, th, d, dh, n,
p, ph, b, bh, m,
y, r, l, v, s, h, ḷ, and ṁ.

3. Tatthodantāsarā aṭṭha.

Of them, the first eight letters ending with o are Vowels.

They are:
a, ā, i, ī, u, ū, e and o.
[These are classified in the following two suttas.]

4. Lahumattā tayo rassā.

The three of soft measure are short (vowels).

Of the vowels, the three of soft measure, viz., a, i, u are short.

5. Aññe dīghā.

The rest are long.

The vowels other than the short ones are long, e.g., ā, ī, ū, e and o. But e and o preceding combined consonants are sometimes regarded as short vowels.
[By ‘combined consonants’ are meant consonants without any vowel intervening them.]
The examples of e and o regarded as short vowels are:
- e in ettha and seyyo;
- o in oṭṭho and sotthi.

6. Sesā byañjanā.

The remaining letters are Consonants.

Setting aside the vowels, the other thirty-three letters, from k to ṁ, are consonants.
The Consonants are:
k, kh, g, gh, ṅ,
c, ch, j, jh, ñ,
ṭ, ṭh, ḍ, ḍh, ṇ,
t, th, d, dh, n,
p, ph, b, bh, m,
y, r, l, v, s, h, ḷ, ṁ.

6. Vaggā pañcapañcaso mantā.

The Vaggas are groups of five of the first twenty-five consonants ending with m.

Of the consonants, the first twenty-five, from k to m, are arranged in five groups, each consisting of five letters.
ka-vagga - k, kh, g, gh, ṅ;
ca-vagga - c, ch, j, jh, ñ;
ṭa-vagga - ṭ, ṭh, ḍ, ḍh, ṇ;
ta-vagga - t, th, d, dh, n;
pa-vagga - p, ph, b, bh, m.
The first and second letters of each Vagga, together with s, are called Āghosa or Surds. The remaining letters ending with ḷ are Ghosa² or Sonants.
The Surds are:
k, kh, c, ch, ṭ, ṭh, t, th, p, ph and s.
The Sonants are:
g, gh, ṅ, j, jh, ñ, ḍ, ḍh, ṇ, d, dh, n, b, bh, y, r, l, v, h, and ḷ.

7. Aṁiti niggahītaṁ.

Ṃ is Niggahita.

The nasal denoted by ṁ, which is sounded after a (in the syllable aṁ), is called Niggahita (arrested letter).³
The vowels i and u may also be placed before it, but not any other vowel.

Sanskrit Grammar, as is indicated in the Sutta “parasamaññā payoge” (vide Kaccāyana, 1. 9). Other terms have been similarly borrowed, e.g. liṅga, sabbanāma, pada, upasagga, nipāta, etc.

The letters of the alphabet can be grouped, according to their seats of pronunciation as follows: —
a, ā, h, and ka-vagga are Gutturals.
i, ī, y, and ca-vagga are Palatals.
u, ū, and pa-vagga are Labials.
r; ḷ and ṭa-vagga are Linguals.
l, s, and ta-vagga are Dentals.
e is Gutturo-Palatal.
o is Gutturo-Labial.
v is Dento-Labial.

II SARA-SANDHI (Vowel-Junction)

Sandhi (Junction) means proximity of two letters without any other letter or pause intervening them.

1. Sarā sare lopaṁ.

There is elision of a vowel before a vowel.

A vowel immediately followed by a vowel is elided, as
- loka + aggo = lok’aggo.
[There is no vowel-sandhi, if there is, intervention of
1. a consonant, as in
  - Maṁ + ahāsi = Maṁ ahāsi
    (here a is not joined with a, because of ṁ intervening),
  or
2. a pause in metre (on account of which the 2nd foot is not joined with the 3rd), as in the following Gāthā:
  
  “Pamādamanuyuñjanti
  bālā dummedhino janā
  Appamādañca medhāvī
  dhanaṁ seṭṭhaṁ va rakkhati.”
  - Janā + appamādaṁ = janā + appamādaṁ,
    where ā of the second foot has not been joined with a of the third foot on account of a pause after the ā of “janā”.
Such is the case in all Sandhis.]

2. Vā paro asarūpo.

There is optional elision of a vowel after a dissimilar vowel.

A vowel after a dissimilar vowel is optionally elided:
- Pana + ime = Pana‘me, Pan’ime.

3. Kvacāsavaṇṇaṁ lutte.

In case of elision, the succeeding vowel sometimes becomes dissimilar.

If the preceding vowel is elided, the succeeding vowel is sometimes changed into a dissimilar vowel, so that i and u are respectively changed into e and o owing to the affinity of their pronunciation:\
- Bandhussa + iva = Bandhuss’eva. (So also ī is changed into o.)
- Na + upeti = N’opeti. (So also ā is changed into o.)

4. Dīghaṁ.

There is lengthening of the succeeding vowel.

If the preceding vowel is elided, the succeeding vowel is sometimes lengthened:
- tatra + ayaṁ = tatr’āyaṁ
- yāni + idha = yān’īdha
- bahu + upakāraṁ = bah’ūpakāraṁ
- saddhā + idha = saddh’īdha
- tathā + upamaṁ = tath’ūpamaṁ

5. Pūbbo⁴ ca.

There is lengthening of the preceding vowel.

If the succeeding vowel is elided, the preceding vowel is sometimes lengthened:
- kiṁsu + idha = kiṁsūdha

p. 14). I have left the spelling just as it occurs in manuscripts.

6. Yamedantassādeso.

The final e is changed into y.

The final e (of the preceding word) followed. by a dissimilar vowel is sometimes changed into y:
- te + ajja = tyajja
But,
- te + ahaṁ = tyāhaṁ.
[Here a is lengthened, because “a vowel followed by a consonant is sometimes lengthened” (vide Kaccāyana, 1. 3. 3).]

7. Vamodudantānaṁ.

The final o and u are changed into v.

The final o and u (of the preceding word) followed by a (dissimilar) vowel are sometimes changed into v:
- so + assa = svassa
- anu + eti = anveti
But,
- tayo + assu = tay’assu
- sametu + āyasmā = samet’āyasmā

8. Do dhassa ca.

Dh is sometimes changed into d.

If a vowel follows, dh is sometimes changed into d:
- idha + ahaṁ = idāhaṁ,
  
  [Here a of “ahaṁ” is lengthened, because the preceding a is elided (vide Sutta 4 above).]
But,
- idha + eva = idh’eva.
Even when followed by a consonant, dh is sometimes changed into d:
- idha + bhikkhave = ida bhikkhave.

9. Ivaṇṇo yannavā.

I and ī may or may not be changed into y.

If a (dissimilar) vowel follows, the preceding letters i and ī may or may not be changed into y:
- vutti + assa = vutyassa.
[If there is ty, it is sometimes changed into cc; as,
- pati + antuṁ = patyantaṁ = paccantaṁ.]
But,
- pati + aggi = paṭaggi.
  
  [Here t becomes ṭ by the sutta ‘‘ Pati patisseti’” (vide Kaccāyana, 1. 5. 7).]
The word “vaṇṇo” is used to include in all cases both short and long vowels; for instance, the “vaṇṇa a” implies both a and ā; the “vaṇṇa i” implies both i and ī; and the “vaṇṇa u” implies both u and ū.

10. Evādissa ri pubbo ca rasso.

E of eva is optionally changed into ri, and the preceding vowel is shortened.

If eva follows a vowel, the e of eva may optionally be changed into ri, and the preceding vowel into its corresponding short form:
- Yathā + eva = yathariva, yath’ eva:

11. Yavamadanataraḷā cāgamā.

Y, v, m, d, n, t, r, and ḷ are the optional insertions.

When a vowel follows, the letters y, v, m, d, n, t, r, and ḷ may be optionally inserted. G also may come as an insertion:
- na + imassa = na-y-imassa
- ti + aṅgikaṁ = ti-v-aṅgikaṁ
- lahu + essati = lahu-m-essati
- atta + atthaṁ = atta-d-atthaṁ
- ito + āyati = ito-n-āyati
- tasmā + iha = tasmā-t-iha
- sabbhi + eva = sabbhi-r-eva
- cha + abhiññā = cha-ḷ-abhiññā (Sometimes there is no sandhi.)
And, putha + eva = putha-g-eva (Sometimes there is no sandhi.)
Also, pā + eva = pa-g-eva (Sometimes
But, abhi + uggato = abbhuggato.

III. BYAÑJANA-SANDHI (Consonant-Junction)

In the following suttas the words byañjane (‘when a consonant follows’) and kvacit (‘sometimes’) are understood.

1. Lopañca tatrākāro

There is elision of the vowel, with an insertion of a in its place.

A vowel followed by a consonant is sometimes elided and a is inserted in its place:
- So + bhikkhu = sa bhikkhu.
U and o may also come as insertions:
- Jānema + taṁ = jānemu taṁ.
- Kaccinu + tvaṁ = kaccino tvaṁ.
But, So + muni = so muni (without Sandhi).

2. Vagge ghosāghosānaṁ tatiya-paṭhamā.

The sonant and surd consonants of a vagga get doubled by taking before them the third and the first letter respectively of the vagga.

The fourth and the second letters of a vagga after a vowel are in some instances doubled through conjunction with the third and the first letter respectively of the vagga:
- U + ghoso = ugghoso.
- Ā + khātaṁ = akkhātaṁ. [Here ā is shortened according to the Kaccāyana sutta “Rassaṁ” (1. 3. 4), which lays down: “A vowel followed by a consonant is sometimes shortened.“]
Exceptions: —
1. O is sometimes inserted when a consonant follows:
  - Para + sahassaṁ = parosahassaṁ.
  - Atippa + kho = atippagokho. [There is insertion of o here after g has been added according to the sutta “Yavamadanataraḷā cāgamā” (pp. 14, 15).]
2. Ava is sometimes changed into o when a consonant follows:
  - Ava + naddhā = onaddhā.
  - But, Ava + sussatu = avasussatu.

IV. NIGGAHĪTA⁵-SANDHI (Nasal-Junction)

1. Vaggantaṁ vā vagge.

The Niggahīta followed by a consonant of a vagga is optionally transformed into the last consonant of that vagga.

Ṁ followed by a consonant of a vagga is optionally changed into the last consonant of the same vagga:
- Kiṁ + kato = kiṅkato.
- Saṁ + jāto = sañjāto.
- Saṁ + ṭhito = saṇṭhito.
- Taṁ + dhanaṁ = tandhanaṁ.
- Taṁ + mittaṁ = tammittaṁ.
But, Na + taṁ + kammaṁ = nataṁkammaṁ (without Sandhi).
Ṁ may be changed into l when followed by a l:
- Puṁ + liṅgaṁ = pulliṅgaṁ.

2. Madā sare.

[The Niggahīta is optionally changed into] m or d if a vowel follows.

Ṁ followed by a vowel is sometimes optionally changed into m or d:
- Evaṁ + assa = eva + (ṁ→m + a) + ssa = evamassa
- Etaṁ + avoca = eta + (ṁ→d + a) + voca = etadavoca
But, Maṁ + ajini = maṁ ajini.

3. Eheñaṁ.

When e or h follows, [the Niggahīta is optionally changed into] ñ.

Ṁ followed by e or h is optionally changed into ñ:
- Taṁ + eva = taññeva [the doubling of ñ takes place according to the Kaccāyana sutta “Paradvebhāvo ṭhāne” (1. 3. 6)]; also, — tameva.
- taṁ + hi = tañhi; also, — taṁhi (without Sandhi).

4. Saye ca.

The Niggahīta together with y, which follows, is also optionally changed into ñ.

Even when followed by y, the ṁ together with the y is optionally changed into ñ:
- saṁyogo = saññogo [ the doubling of ñ takes place according to the Kaccāyana sutta (1. 3. 6), quoted above]; also,— saṁyogo.

5. Niggahītañca.

The Niggahīta is also [sometimes optionally] inserted.

Ṁ is also sometimes optionally inserted when a vowel or a consonant follows:
- Cakkhu + aniccaṁ = cakkhuṁaniccaṁ.
- Ava + siro = avaṁsiro.
[The opposite is also noticeable (vide Kaccāyana, 1.4.9 “Kvaci lopaṁ”):
1. Ṁ is sometimes elided when a vowel follows:
  - Vidūnaṁ + aggaṁ = vidūna + aggaṁ = vidūnaggaṁ
  - Tāsaṁ + ahaṁ = tāsa + ahaṁ = tāsāhaṁ (Here, the second vowel of tāsāhaṁ is lengthened according to the sutta “Dīghaṁ” see ante, p. 10, sutta 4.)
2. There is occasional elision even when a consonant follows:
  - Buddhānaṁ + sāsanaṁ = buddhāna + sāsanaṁ = buddhānasāsanaṁ
  - Saṁ + rāgo = sa + rāgo = sārāgo (For the lengthening of the first vowel of sārāgo, vide Kaccāyana, 1.3.3, — “Dīghaṁ.”)]

6. Paro vā saro.

The succeeding vowel is sometimes optionally elided.

Sometimes the vowel after ṁ is optionally elided:
- Bījaṁ + iva = bījaṁ va.

7. Byañjano ca visaññogo.

The consonant is also simplified.

When the vowel is elided after ṁ, the succeeding conjunct is simplified:
- Evaṁ + assa = evaṁ + ’ssa = evaṁ ’sa.

V. VOMISSAKA-SANDHI (Miscellaneous Junction)

1. Anupadiṭṭhānaṁ vuttayogato.

The foregoing suttas are to be suitably applied to instances not yet discussed.

The foregoing suttas of vowel-junction and consonant-junction are to be applied in a suitable manner to the prefix, indeclinable and other kinds of junction, not yet discussed:
For instance, yadi + evaṁ = yady + evaṁ = yajjevaṁ — (first, by the sutta “Ivaṇṇo yannavā” (see ante, p. 13), di of yadi is changed into dy; secondly, by the principle (implied in the same sutta) that “ty is sometimes changed into cc,” dy is changed into jj). Similarly, Bodhi + aṅgā = bodhy + aṅgā = bojjhaṅgā.]

The following minor rules are applicable in Miscellaneous Junction:—

Asadisa-saṁyoge ekasarūpatā ca.
- In junction, the preceding consonant is assimilated to the succeeding dissimilar consonant:
  - Pari + esanā = pary + esanā (vide sutta 9, p. 18) = payyesanā (r being assimilated to y).
Vaṇṇānaṁ bahuttaṁ, viparītatā ca.
- In some cases there is multiplication of a letter, and in certain cases there is inversion too. For instance,
- In sa + rati = sumarati, there are insertion of ma (II. 11) and transformation of a into u (III. 1).
- In iti + evaṁ = itvevaṁ, there is transformation of i into v by extension of sutta II. 7 (p. 11).
- In sā + itthi = sotthi, the succeeding vowel i having been elided, the preceding ā is changed into o by extension of sutta II. 5 (p. 10).
- In busā + eva = busamiva, m is inserted (11. 11), the preceding vowel ā is shortened (vide Kaccāyana, 1. 3. 4. - “Rassaṁ”) and the succeeding e is changed into i by extension of sutta II. 10 (p. 14).
- In bahu + ābādho = bavhābādho, u being followed by the dissimilar vowel ā is first changed into v (II. 7), and then there is inversion between and v.
- In adhi + abhavi = addhābhavi, there is application of the principle according to which “adhi is sometimes changed into addha,” and of sutta II. 4 (p. 10), which lays down, “If the preceding vowel is elided, the succeeding vowel is sometimes lengthened.”
Radānaṁ ḷo.
- R and d are changed into ḷ:—
  - Pari + bodho = paḷibodho.
  - Pari + dāho = pariḷāho.
Sare, byañjane vā pare binduno kvaci mo.
- If a vowel or a consonant follows, ṁ is sometimes changed into m:
  - Maṁ + ahāsi = mamahāsi (vide sutta IV. 2);
  - Buddhaṁ + saraṇaṁ = buddham saraṇaṁ.
Binduto parasarāna maññassaratāpi.
- The vowels after ṁ may be changed into dissimilar vowels:
  - Taṁ + iminā = tadaminā — (i is changed into a, after ṁ has been replaced by d according to sutta IV. 2).
  - Evaṁ + imaṁ = evumaṁ — (ṁ is elided, according to the Kaccāyana sutta “Kvaci lopaṁ” 1.4.9, and then i is changed into u.
  - Kiṁ + ahaṁ = kehaṁ - (ṁ is elided, and i is changed into e).
Vākyasukhuccāraṇatthaṁ, chandahānitthañca vaṇṇalopopi.
- Letters are sometimes omitted for the sake of euphony and metre:
- Paṭisaṁkhāya + yoniso = paṭisaṁkhā’yoniso — (the preceding ya is elided for the sake of euphony).
- Alāpūni + sīdanti = ’lāpūni sīdanti — (the initial a of alābūni is elided on account of the exigency of metre).
Vutyabhedāya vikāropi.
- There is exception even to the above sub-rule for the sake of metre:⁶
- Akaramhase te = akaramhasa te - (the long vowel e of se is changed into the short vowel a, metri causa).

Dīgho saṁyogādi-pubbo rasso ca garū, lahu tu rasso.
Metre is the harmonious order of letters. Measure is the regulation of heavy and light vowels.
A heavy vowel is a long vowel, and so is also the short vowel which precedes a double consonant or the Niggahīta; e.g., the italics in the following:— ā, assa, aṁ.
A light vowel is a short vowel; e.g., the vowels a, i and u.

These figures refer to the Book, the Chapter and the Sūtra respectively of Kaccāyana’s Pāli Grammar, to which the Bālāvatāra ↩
The technical terms Ghosa and Āghosa have been taken from ↩
The following verse gives a full definition of Niggahita: Bindu cūḷā-maṇ’-ākāro Niggahitan ti vuccate | Kevalass’ appayogattā a-kāro sannidhīyate || i.e., The point resembling a small gem is called Niggahita, As it is not employed alone, a is placed before it. ↩
The word is always spelt with a short u in Pāli (vide Sutta 10, ↩
See p. 5, sutta 1.7, where the word is spelt with a short i. ↩
Akkharaniyamo Chandaṁ, garū-lahu-niyamo bhave Vutti,\ ↩