Sandhi and Word Formation
What the exam tests
This domain tests external sandhi at word boundaries and the internal sandhi that appears when a root meets a suffix. Questions usually ask you to join two padas, split a given string, or name the sandhi that applies.
Core rules
- Ac sandhi (vowels): savarṇa-dīrgha (a/ā + a/ā → ā; i/ī + i/ī → ī); guṇa (a + i/ī → e; a + u/ū → o; a + ṛ → ar); vṛddhi (a + e → ai; a + o → au); yaṇ (i/ī → y, u/ū → v, ṛ → r before a vowel); pūrvarūpa (e/o + a → e/o + avagraha).
- Visarga sandhi: ḥ before a vowel or voiced consonant → o (rāmaḥ api → rāmo'pi); ḥ + t/th → s; ḥ + c/ch → ś; ḥ + ṭ/ṭh → ṣ.
- Anusvāra and parasavarṇa: ṃ before a consonant becomes the nasal of that class (taṃ ca → tañ ca); before a vowel it becomes m.
- Cerebralisation: ṣatva and ṇatva (rāmeṇa, rāmāṇām) are triggered by specific preceding sounds and are tested as recognition, not free choice.
Worked example
Join viṣṇu + ālayaḥ. The final u meets initial ā: yaṇ gives v + ā, so viṣṇvālayaḥ (‘Viṣṇu’s temple’). Join rāmaḥ + gacchati: ḥ stands before the voiced g, so ḥ → o and the result is rāmo gacchati. Join tat + ca: final t before the palatal c becomes c, giving tac ca; this is not a vowel sandhi at all.
Traps
Do not apply guṇa where savarṇa-dīrgha is required: devānām, not devaonām. Do not drop the avagraha in te'pi. Do not treat a compound as a sandhi problem; rājaputraḥ is a tatpuruṣa, not two words joined by visarga sandhi.