11.A.1 Visarga Sandhi
This is most conveniently presented directly in tabular form:
Next Initial Sound
|-a||-ā||-Vr||-Vr||any vowel other than ă|
|-o||-ā||-Vr-||-Vr-||any other ghoṣa vyañjana|
|-aś||-āś||-Vś||-Vś-||c / ch-|
|-aṣ||-āṣ||-Vṣ||-Vṣ-||ṭ / ṭh|
|-as||-ās||-Vs||-Vs||t / th-|
|-aḥ||-āḥ||-Vḥ||-Vḥ||any other aghoṣa vyañjana|
|-aḥ||-āḥ||-Vḥ||-Vḥ||avasāna (e.g., ॥)|
|Note: The words सः or एषः followed by hrasva अ becomes सोऽ or एषोऽ;
before any other letter the visarga is dropped.
The table is simple enough. Basically a final s or r becomes r before a voiced (ghoṣa) sound (which includes the vowels, of course); the exceptions to this are:
- -as before a ghoṣa consonant becomes -o; the -s is dropped before a vowel, unless that vowel is hrasva अ in which case ॰अस् + अ॰ becomes ॰ओऽ॰.
- where the final is -r and the following word begins with r- (a disallowed combination), the first r is dropped, and the preceding vowel, if a i or u, is lengthened.
And the final s or r becomes a visarga before an unvoiced (aghoṣa) sound (whether a consonant or a pause in sound); the exception to this, is that before c / ch, ṭ / ṭh, or t / th, it is replaced with a sibilant (ś, ṣ, or s) of the same mouth position as that of the following consonant.
One very important point to note about this table, is that the last row: an avasāna is a pause or stop in speech, as for example at the end of a sentence or line of poetry. This also applies when a sentence is split up into its independent words (padāni) by removing the external sandhi, a process called sandhi vigraha. The immediate relevance is that the declension of nouns and conjugation of verbs is given in the form of independent words, which means that sandhi rules applicable to a following avasāna have already been applied. So, when the word is used in a sentence, this sandhi must be removed: where the word is given in the tables with a final visarga, this should be replaced with an s before applying the visarga sandhi.
11.A.2 Consonant Sandhi
As a rule, a word may begin with any vowel or consonant except ḥ ṃ ṅ ñ ṇ ṝ ḷ, and may end (before an avasāna or pause) with one of eight consonants k ṭ t p ṅ n m or ḥ, or with any vowel except ṝ or l. The sandhi of words ending with a visarga (ḥ) were discussed in 11.A.1; this table covers the remaining consonants.
The first four of the final consonants are the alpaprāṇa aghoṣa sparśa (except c) and the remaining three are nasals. As with the visarga sandhi, this table is split according to the following sound being ghoṣa or aghoṣa.
The final aghoṣa (k ṭ t and p) are basically replaced with the ghoṣa alpaprāṇa consonant of the same mouth position when the following sound is ghoṣa, and remain unchanged when followed by an aghoṣa sound; but note that a final -t changes to the mouth position of a following tālavya or mūrdhanya sound (both ghoṣa and aghoṣa), and observe its special changes before l- h- and ś- (in the last case the substitute replaces the following ś- as well). Before an h- (which is ghoṣa) these four are replaced by their ghoṣa equivalents, and the h- is replaced by the mahāprāṇa equivalent of that ghoṣa substitute (e.g., ॰क् ह॰ ⇒ ॰ग्घ॰).
The kaṇṭhya nasal remains unchanged, while the dantya nasal (like the -t) changes to the mouth position of a following tālavya or mūrdhanya ghoṣa sound, and to an anusvāra and sibilant of the following mouth position of a following tālavya, mūrdhanya, or dantya aghoṣa; also note the special changes before l- and ś-. A final -m changes to anusvāra before any consonant (see 8.A.3 for pronunciation of the anusvāra).
There are no sandhi changes when a vowel meets a consonant, with two exceptions: when a word ends in a short vowel and the following word begins with ch-, then a c is inserted; secondly, when a word ends in -ṅ or -n preceded by a short vowel, and the following word begins with a vowel, then the nasal is doubled, i.e.:
|-V̆ ch-||⇒||-V̆cch-||-V̆ṅ V̄̆-||⇒||-V̆ṅṅV̄̆-||-V̆n V̄̆-||⇒||-V̆nnV̄̆-|
|Final Consonant (before avasāna|
|-g||-ḍ||-d||-b||-ṅ||-n||-ṃ||g / gh-|
|-g||-ḍ||-j||-b||-ṅ||-ñ||-ṃ||j / jh-|
|-g||-ḍ||-ḍ||-b||-ṅ||-ṇ||-ṃ||ḍ / ḍh|
|-g||-ḍ||-d||-b||-ṅ||-n||-ṃ||d / dh-|
|-g||-ḍ||-d||-b||-ṅ||-n||-ṃ||b / bh|
|-ṅ||-ṇ||-n||-m||-ṅ||-n||-ṃ||n / m-|
|-g||-ḍ||-d||-b||-ṅ||-n||-ṃ||y / r / v-|
|-k||-ṭ||-t||-p||-ṅ||-n||-ṃ||k / kh-|
|-k||-c||-t||-p||-ṅ||-ṃś||-ṃ||c / ch-|
|-k||-ṭ||-t||-p||-ṅ||-ṃṣ||-ṃ||ṭ / ṭh-|
|-k||-ṭ||-t||-p||-ṅ||-ṃs||-ṃ||t / th-|
|-k||-ṭ||-t||-p||-ṅ||-n||-ṃ||p / ph-|
|-k||-ṭ||-t||-p||-ṅ||-n||-ṃ||ṣ / ś-|
11.A.3 Internal Sandhi
The two most common rules of internal sandhi, and which affect the spelling of vibhakti endings in particular, are:
|s||following k r i ī u ū ṛ ṝ e ai o or au|
|is replaced by||even if there is an intervening ṃ or ḥ|
|ṣ||unless it is the final letter or followed by r|
|n||following ṣ r ṛ or ṝ|
|is replaced by||even if ka-varga, pa-varga, y v h or vowel intervene|
|ṇ||when followed by a vowel, m v y or n (which last becomes ṇ)|
The following detailed notes may be used for reference: they need not be studied.
11.B.1 Dvandva Samāsa
The dvandva (lit. “couple”) samāsa is a copulative compound in which the members, if not compounded, would be in the same case (vibhakti) and connected by the conjunction च (and). There are two types of dvandva:
- The members are considered separately; the gender of the compound is the gender of the last member; the number is the sum of the members. For example:
rāmaḥ ca kṛṣṇaḥ ca ⇒ rāmakṛṣṇau (note the dual) = Rāma and Kṛṣṇa
- The members are taken collectively as a unit; it is always neuter singular. Pairs of opposites are often put in this form, for example:
sukham ca duḥkham ca ⇒ sukhaduḥkham (note the singular) = pleasure and pain
11.B.2 Tatpuruṣa Samāsa
The tat-puruṣa (lit. “his man”) samāsa is a determinative compound in which the first member depends on (i.e., has a case relationship to), or modifies, the second. There are several types:
- Also called vyadhikaraṇa-tatpuruṣa, is characterized as having different case endings if the compound is dissolved, i.e., the members are different objects. The compound may be further classified according to the case relationship (dvitīyā through saptamī) of the first member to the second. For example:
vṛkṣamūlam ⇐ vṛkṣasya mūlam (ṣaṣṭhi-tatpuruṣa) = root of a tree, tree-root
- This is a descriptive determinative compound, also called samānādhikaraṇa-tatpuruṣa, and is characterized as having the same case ending if the compound is dissolved, i.e., the members refer to the same object; for example:
pūrṇacandraḥ ⇐ pūrṇaḥ (full) candraḥ (moon) = full-moon
- This samāsa has the same sense as the karmadhāraya, but has a word denoting direction or a numeral as its first member; for example:
एकवचन eka-vacana, singular (lit. one-speaking, from √ vac, to speak) [also dvi- (two), bahu- (many), giving “dual” and “plural”]
- This compound has a dhātu derivative as its second member; for example:
kumbha-kara ⇐ kumbham (pot) + √ kṛ (to do, act, make) = potter (similarly a-kāra etc.)
- A compound with a negative particle (na-, an-, or a-) as its first member, giving a negative or privative sense; for example:
a-jñānam ⇐ a- (negation or absence) + jñānam (knowledge) = ignorance
11.B.3 Avyayībhāva Samāsa
The avyayībhāva (lit. “an unchanging nature”) samāsa is indeclinable (avyaya) and functions as an adverb. The first member is an indeclinable (preposition or adverbial prefix), and the last a noun (nāman), and the whole takes the form of the neuter singular; for example:
|sakrodham||⇐||sa- (the sense of accompaniment)||+||krodha (anger)||=||with anger, angrily|
|pyathāśraddham||⇐||yathā- (the sense is proportion)||+||śraddha (faith)||=||according to (one’s) faith|
11.B.4 Bahuvrīhi Samāsa
The bahuvrīhi (lit. “(having) much rice”) samāsa is a descriptive compound forming an adjective (viśeṣaṇa) agreeing with a noun (expressed or understood); for example:
|padmākṣa||⇐||padma (lotus)||+||akṣa (eye)||=||whose eyes are (like) lotuses, lotus-eyed|
The difference between the tatpuruṣa and the bahuvrīhi is that the former remains a noun, while the latter becomes an adjective or epithet. In the Vedic Sanskrit the determinative and descriptive compounds were distinguished by accents (see 9.A.1):
|rāja-putrá||⇐||rājan (king)||+||putra (son)||=||the son of the king, the king’s son (tatpuruṣa)|
|rājá-putra||=||whose son is king (bahuvrīhi)|
- Practice sounding the alphabetical order while following the consonants on the alphabet chart in 9.A.4.
- Write out the alphabet chart in 9.A.4 once per day from memory.
- Write the following sentences in devanāgarī, applying sandhi rules as necessary — and it will be necessary quite often! — and then translate them into English. For example:
naraḥ aśvaḥ ca vṛkṣān labhete नरो’श्वश्चाल्पान्वृक्षाल्ँलभेते The man and horse take the small trees.
- narau alpam vṛkṣam agnim aśvāt vahataḥ
- bālā aśvam naram ca vṛkṣāt labhate
- phalāni aśvam vahati iti guruḥ bālāḥ vadati
- gurū alpam naram vṛkṣāyāt śīghram gacchataḥ
- naraḥ vṛkṣam agnim bālāyai aśvena vahati
- bālā aśvam alpām nadīm vṛkṣāt nayate
- naraḥ vṛkṣān phalebhyaḥ aśvena gacchati
- guruḥ agnim narāt gacchati iti alpā bālā vadati
- bālā alphaḥ aśvaḥ ca agnim narāt gacchataḥ
- alpebhyaḥ phalebhyaḥ sundareṣu vṛkṣeṣu gacchāvaḥ
The answers to #3 can be found in Appendix 2.
-Vs = any vowel except a or ā before the final s. ↩
-Vr = any vowel before the final r. ↩
A र् followed by another र् is elided, and a preceding अ, इ, or उ lengthened. ↩
The nasal doubles to -ṅṅ or -nn if the preceding vowel is short. ↩
This is a nasalized l, i.e., ॰न् ल॰ becomes ॰ल्ँल॰. ↩
-ñś- may also become -ñch-. ↩