Skip to content

2.1: `sandhi` (How words join)

· 7 min read

Euphonic joins of multple words, eg. cāyaṁ, natthi, kiñci, evametassa, …dhamman”ti, and how to undo each: a vowel dropped or lengthened {12}–{15}, a ṁ changed {31}–{40}, a consonant put in {35} or doubled {28}–{29}.

Pāḷi was based on a spoken rather than written language, and when speakers speak the words they can seem joined together and sometimes the pronunciation of words may be altered to make the words blend into each other more seamlessly. These changes are preserved when Pāḷi texts were written down.

When a word ending in a vowel or in ṁ meets the next word, the two run together and the seam changes shape. The joined form is a sandhi (a join), and the texts of chapters 3, 5, 7, 9 and the bonus chapter have one in almost every line: yo cāyaṁ is yo ca ayaṁ, natthi is na atthi, kiñci is kiṁ ci, and the ”ti that closes every quotation is iti with its vowel gone.

So the first question to ask when looking up a word in the dictionary and not finding it: is it two words? The table lists the most common joins, what each was before it was joined, and the rule that makes it; the rule number links to Kaccāyana’s statement with its examples, and Appendix E tables all the rules {12}–{51}.

pattern example before join joined examples rule
the first word’s final vowel missing pana idaṁ panidaṁ, natthi (na atthi), vivicceva {12}
a long vowel for two short ones: the first dropped, the second lengthened ca ayaṁ cāyaṁ, tāvāhaṁ (tāva ahaṁ), yannūnāhaṁ {12}, {15}
o’ti, ova: iti, iva after o, the i dropped abhisambuddho iti abhisambuddho’ti, sūriyova {13}
ī”ti: iti after i, the first i dropped, the second lengthened hoti iti hotī”ti, kiñcī’ti {12}, {15}
n”ti, mpi: iti, api after ṁ, the first vowel dropped, the ṁ turned into n or m (the word’s own ending is -aṁ, never -an) nirodhadhammaṁ iti, maraṇaṁ api nirodhadhamman”ti, maraṇampi, yampicchaṁ (yaṁ api icchaṁ) {40}, {31}
ṅ, ñ, ṇ, n, m before k, c, ṭ, t, p: a ṁ turned into the nasal of that group kiṁ ci kiñci, imasmiñca, saṅkhittena (5: 2209 keeps the ṁ: saṁkhittena {30}) {31}
m or d between a word in ṁ and a vowel evaṁ etassa evametassa, ayamantimā, yadidaṁ (yaṁ idaṁ) {34}
a y (or v, m, d, n, t, r, ḷ) between two vowels, neither dropped tassā eva tassāyeva, nayidaṁ (na idaṁ) {35}
a doubled consonant after a short prefix or first word vi payogo vippayogo, appaṭivattiyaṁ {28}
kkh, cch, ṭṭh, tth, pph: an aspirate doubled with the first letter of its group (jjh, ddh, bbh with the third) dukkha khandha dukkhakkhandhassa, pañcupādānakkhandhā {29}
pacc-: the prefix pati before a vowel pati aññāsiṁ paccaññāsiṁ {19}
  1. Find the seam: a long vowel before a short tail (cāyaṁ), an apostrophe or quotation mark, an m, d, n, ñ or ṅ no dictionary form has (evametassa, kiñci), a doubled consonant after a one-syllable prefix (appaṭivattiyaṁ), a y between two vowels (tassāyeva). Then undo what the rule did: put the dropped vowel back, shorten the lengthened one, turn the nasal back into ṁ, take the y out, make the double consonant single.
  2. Try the small words first: ca, eva, api, iti, pana, na, ayaṁ, idaṁ, ahaṁ or a pronoun in ṁ (yaṁ, taṁ, evaṁ) is on one side of nearly every join.
  3. Look both halves up: the Digital Pāḷi Dictionary lists the common joins as headwords (cāyaṁ: ca + ayaṁ) and splits the rest. If no split gives two dictionary words, it is one word, a compound (chapter 6) or a prefixed word (2.6) with the join inside it: dukkhakkhandha, vippayoga.

Yo cāyaṁ kāmesu kāmasukhallikānuyogo

*️⃣🤟🚹⨀①(ya) ⏏️(ca) *️⃣🤟🚹⨀①(ima) 🚹⨂⑦(kāma) 🚹⨀①🆎(kāma+sukhallikā+anuyoga)
which / and / this / in sensual pleasures / pursuit of sensual happiness
cāyaṁ = ca ayaṁ: the a of ca dropped {12}, the next a lengthened {15}; 5: 2207.

Taṁ kho panidaṁ dukkhaṁ ariyasaccaṁ pariññeyyan’ti

*️⃣🤟🚻⨀①(ta) ⏏️(kho) ⏏️(pana) *️⃣🤟🚻⨀①(ima) 🚻⨀①(dukkha) 🚻⨀①🆎(ariya+sacca) 🚻⨀①🔽(ṇya)(pariññeyya) ⏏️(iti)
that / indeed / now / this / suffering / realisation / to be fully understood / thus
panidaṁ = pana idaṁ: the a of pana dropped {12}; pariññeyyan’ti = pariññeyyaṁ iti: the i of iti dropped {40}, ṁ becomes n {31}; 5: 2210.

Evametassa kevalassa dukkhakkhandhassa nirodho hotī”ti

⏏️(evaṁ) *️⃣🤟🚹⨀⑥(eta) 🚹⨀⑥(kevala) 🚹⨀⑥🆎(dukkha+khandha) 🚹⨀①(nirodha) 🟢⨀🤟▶️(hoti) ⏏️(iti)
thus / of this / of the entire / of the mass of suffering / cessation / is / (end quote)
evametassa = evaṁ etassa: ṁ becomes m before a vowel {34}; dukkhakkhandhassa = dukkha + khandha: kh doubled as kkh {29}; 3: 4.

yo tassāyeva taṇhāya asesavirāganirodho

*️⃣🤟🚹⨀①(ya) *️⃣🤟🚺⨀⑥(ta) ⏏️(eva) 🚺⨀⑥(taṇhā) 🚹⨀①🆎(asesa+virāga+nirodha)
which / of that / very / craving / remainderless fading away and cessation
tassāyeva = tassā eva: neither vowel dropped; a y put between them {35}; 5: 2209.

Each phrase, from chapters 5, 9 and the bonus chapter, has a join in bold: split it into its parts, say what the join dropped, lengthened, changed or doubled, and name the rule.

  1. natthi dāni punabbhavo
  2. Imasmiñca pana veyyākaraṇasmiṁ bhaññamāne
  3. Yampidaṁ cakkhusamphassapaccayā uppajjati vedayitaṁ
  4. yaṁ kiñci samudayadhammaṁ sabbaṁ taṁ nirodhadhamman”ti
  5. ‘anuttaraṁ sammāsambodhiṁ abhisambuddho’ti
  6. So vivicceva kāmehi
  7. piyehi vippayogo dukkho

Note: please do the exercises first before checking the answers.

  1. natthi dāni punabbhavo

⏏️(na) 🟢⨀🤟▶️(atthi) ⏏️(dāni) 🚹⨀①🆎(puna+bhava)
not / there is / now / appearing again
natthi = na atthi: the a of na dropped {12}, nothing lengthened; 5: 2215.

  1. Imasmiñca pana veyyākaraṇasmiṁ bhaññamāne

*️⃣🤟🚻⨀⑦(ima) ⏏️(ca) ⏏️(pana) 🚻⨀⑦(veyyākaraṇa) 🚻⨀⑦🔽(māna)(bhaññamāna)
in this / and / moreover / detailed exposition / being spoken
imasmiñca = imasmiṁ ca: the literal final ṁ in the ⑦ inflection form becomes ñ before c {31}; 5: 2216.

  1. Yampidaṁ cakkhusamphassapaccayā uppajjati vedayitaṁ

*️⃣🤟🚻⨀①(ya) ⏏️(api) *️⃣🤟🚻⨀①(ima) 🚹⨀⑤🆎(cakkhu+samphassa+paccaya) 🟢⨀🤟▶️(uppajjati) 🚻⨀①🔽(ta)(vedayita)
what / too / this / with eye-contact as cause / arises / feeling
yampidaṁ = yaṁ api idaṁ: the a of api dropped {40}, ṁ becomes m before p {31}, the i of pi dropped {12}; 9: 75.

  1. yaṁ kiñci samudayadhammaṁ sabbaṁ taṁ nirodhadhamman”ti

*️⃣🤟🚻⨀①(ya) *️⃣🤟🚻⨀①(kaci) 🚻⨀①🆎(samudaya+dhamma) *️⃣🤟🚻⨀①(sabba) *️⃣🤟🚻⨀①(ta) 🚻⨀①🆎(nirodha+dhamma) ⏏️(iti)
whatever / anything / subject to origination / all / that / subject to cessation / (end quote)
kiñci = kiṁ ci: ṁ becomes ñ before c {31}. nirodhadhamman”ti = nirodhadhammaṁ iti: the i of iti dropped {40}, ṁ becomes n {31}; 5: 2216.

  1. ‘anuttaraṁ sammāsambodhiṁ abhisambuddho’ti

🚺⨀②🆎(na+uttara) 🚺⨀②🆎(sammā+sambodhi) 🚹⨀①🔽(ta)(abhisambuddha) ⏏️(iti)
unsurpassed / perfect awakening / fully awakened / thus
abhisambuddho’ti = abhisambuddho iti: after o the i of iti is dropped {13}; 5: 2214.

  1. So vivicceva kāmehi

*️⃣🤟🚹⨀①(ta) 🔽(ya)(viviccati) ⏏️(eva) 🚹⨂⑤(kāma)
that / secluding oneself from / just / sensual pleasures
vivicceva = vivicca eva: the a of vivicca dropped {12}; Bonus: 38, 697.

  1. piyehi vippayogo dukkho

🚹⨂⑤(piya) 🚹⨀①(vippayoga) 🚹⨀①(dukkha)
from the loved / separation / suffering
vippayogo = vi + payogo: the p doubled after the prefix vi {28}; 5: 2209.