After this chapter you will be able to read the short names (saṃjñās) that every later rule of the Siddhānta-kaumudī uses without explaining them. You will know how a marker letter is spotted and discarded, how a two-letter name such as ac or hal is unpacked into a list of sounds, which sounds count as "the same kind" (savarṇa), what vṛddhi and guṇa mean, what a dhātu, a nipāta, an upasarga and a pada are, and what "heavy" and "light" mean for a vowel. Every one of these names is defined by one of the thirty-three rules below, and each rule is shown at work on a concrete word.
1. a i u ṇ
2. ṛ ḷ k
3. e o ṅ
4. ai au c
5. ha ya va ra ṭ
6. la ṇ
7. ña ma ṅa ṇa na m
8. jha bha ñ
9. gha ḍha dha ṣ
10. ja ba ga ḍa da ś
11. kha pha cha ṭha tha ca ṭa ta v
12. ka pa y
13. śa ṣa sa r
14. ha l
How they are read: each line is a short string of sounds ending in one consonant that is not part of the list but a marker. The final ṇ of line 1, the k of line 2, the l of line 14 and so on are these markers; the grammar calls a marker an it. In line 6 (la ṇ) the a after l is also a marker. The a written after every other consonant (ha, ya, ka …) is there only so that the consonant can be pronounced and is not counted. The whole list exists so that short names like aṇ, ac and hal can be built from it (Sections A and B below).
Before SK 1, Bhaṭṭoji's book opens with passages that are not sūtras: four verses of homage (the last of them, Bhaṭṭoji's own, is explained phrase by phrase by his commentator) and a short passage of four sentences on the Śivasūtras. They say who the grammar comes from, what this book is for, and why the Śivasūtras matter. The commentary on that passage raises the puzzle that SK 1 solves.
A maṅgala is an act of homage set at the start of a work. Bhaṭṭoji's commentators say it is obligatory there. They quote the Bhāṣya (Patañjali's great work on the grammar, see verse 3) on the vṛddhi-sūtra (SK 16): works with a maṅgala at their beginning, middle and end become widely known, and their students flourish.
Verse 1. "Homage to that Pāṇini who, having obtained the inventory of sounds from Maheśvara, proclaimed the entire grammar." The inventory of sounds (akṣara-samāmnāya) is the list of fourteen Śivasūtras. Bhaṭṭoji's commentators cite this verse from the Śikṣā composed by Pāṇini's pupil.
Verse 2. "Homage to that Pāṇini by whom the speech of men was washed with the stainless waters of words, and the darkness born of ignorance split apart."
Verse 3. "I bow to the three munis: Vararuci, author of the vākyas; Patañjali, author of the Bhāṣya; and Pāṇini, author of the sūtras." A muni is a sage. The vākyas are the statements that supplement Pāṇini's sūtras; this chapter calls them vārttikas (see SK 12). The Bhāṣya is Patañjali's work.
Verse 4. "Having bowed to the three munis, and having thoroughly reflected on their utterances, this Vaiyākaraṇa-siddhānta-kaumudī is composed." This verse, munitrayaṃ namaskṛtya …, is Bhaṭṭoji's own maṅgala. With it he announces what he intends to do. The next section explains it word by word.
Watch out. The text itself does not say who composed verses 2 and 3; only verse 1 is traced, by the commentators, to the Śikṣā.
This is verse 4 again, set out on its own. Bhaṭṭoji's commentators explain it phrase by phrase.
"The three munis." munitrayam means "the group of three munis": Pāṇini, Vararuci and Patañjali (verse 3). Their works are the sūtras, the vārttikas and the Bhāṣya.
"Having bowed." namaskṛtya means having pleased them by joining the palms, bowing the head and the like.
"Having thoroughly reflected on their utterances." tad-uktīḥ paribhāvya answers an objection. Older books such as the Prakriyā-kaumudī already collect the grammarians' conclusions, so why write another? Because this book follows the works of the three munis closely, and the older books conflict with them. Bhaṭṭoji argues this at length in his Prauḍhamanoramā. Some read paribhāvya as "having set aside" the older authors' words; the commentators reject that reading as strained.
The title. Vaiyākaraṇa-siddhānta-kaumudī explains itself, part by part:
Moonlight dispels darkness, shows things pleasantly, and relieves the burning heat of the sun. So this book dispels ignorance, lights up the three munis' works without effort, and relieves the strain of poring over vast works such as the Bhāṣya and Kaiyaṭa's work.
Subject, purpose, relation, student. The commentators say the verse, through its title, also indicates four things about the book:
After the verses Bhaṭṭoji recites the fourteen Śivasūtras (listed in Before you start). He then adds a short passage of four sentences: iti māheśvarāṇi sūtrāṇy aṇādisaṃjñārthāni. eṣām antyā itaḥ. laṇsūtre akāraś ca. hakārādiṣv akāra uccāraṇārthaḥ. Each sentence answers a question.
Why are they authoritative? They stand outside the works of the three munis. But they are māheśvara, "come from Maheśvara": Pāṇini obtained them from him (verse 1). The commentators also quote Nandikeśvara's commentary on the fourteen sūtras. At the close of his dance, Śiva sounded his ḍhakkā drum fourteen times to uplift Sanaka and the other siddhas (perfected sages); those sounds are the Śivasūtras.
What use are meaningless sounds? aṇādisaṃjñārthāni: they exist for the names aṇ and the rest. Those short names (Section B) are used in the grammar's operations.
Which sounds are markers? eṣām antyā itaḥ: the final sound of each sūtra is it. laṇsūtre akāraś ca: in la ṇ the a after l is it as well. It is named separately because it is not final (SK 3 is the rule for it).
Why is a written after h, y, v, r …? hakārādiṣv akāra uccāraṇārthaḥ: that a is uccāraṇārtha, there only for pronunciation. The a makes h, y, v, r easy to say; otherwise we would stumble over h-y-v-r. Better still, as the Bhāṣya says on uccair udāttaḥ (SK 5), a consonant cannot be uttered at all without a vowel.
The puzzle this leaves. "The finals are it" is only stated here; a sūtra must teach it. That sūtra is SK 1, halantyam, "a final hal is it" (hal, the name for the consonants). But look at what it depends on:
SK 1 speaks of a hal → to know what hal means, SK 2 must build it from the h that opens ha ya va ra ṭ (line 5) up to the marker l that ends ha l (line 14) → but l is a marker only by SK 1 → back to the start.
Each rule waits for the other. This is anyonyāśraya, mutual dependence, and with it nothing can be understood. Bhaṭṭoji answers by reading SK 1 twice; see SK 1.
a i u ṇ and la ṇBefore any name can be read off the Śivasūtras, the grammar must say which sounds in them are markers. This group gives the two rules that do that. In a i u ṇ the final ṇ is stamped as a marker; in la ṇ the nasal a after l is stamped as a marker too. Once a sound is a marker it is written but not counted.
Rule. In an upadeśa — the original statement of any form in the grammar, such as a Śivasūtra, a root, or an affix — a final consonant is it, a marker. A consonant is called a hal; a marker is called an it, and being an it is the it-saṃjñā, "the name it". Bhaṭṭoji reads this rule twice: first to make the l of the Śivasūtra ha l itself an it, so that the name hal ("all consonants") can be formed without circularity (the puzzle set out under Text · mahesvara above); then, with hal now available, to make the final consonant of every upadeśa an it.
Example.
ha l → its final sound l is a consonant → l is it → hal can now name the consonants (Section B).
a i u ṇ → its final sound ṇ is a hal → ṇ is it → the sounds actually listed are a, i, u; ṇ is only a marker.
Likewise the ṇ, k, ṅ, c, ṭ, ṇ, m, ñ, ṣ, ś, v, y, r, l ending the fourteen Śivasūtras are all it.
Applies when / blocked by. Only in upadeśa. Bhaṭṭoji's commentators give one contrast: in the affix thamu, the u is there only for pronunciation; once it has dropped, the m is left final, but it does not become it, because that is no longer the original statement of the affix.
Watch out. The word it carries over into this rule from SK 3 (it is not written in SK 1 itself). Carrying a word forward from a previous rule is called anuvṛtti; you will meet it constantly.
Rule. In an upadeśa, a vowel pronounced through the nose is it. A vowel is called an ac; a nasal sound is anunāsika (defined at SK 9). Which vowels of the grammar are nasal is not marked in writing; it rests on the declared tradition of Pāṇini's school, handed down from teacher to pupil.
Example.
la ṇ → the a after l is pronounced nasal in upadeśa → this a is it.
r (from ha ya va ra ṭ) + this it a → the name ra → it stands for r and l (see SK 2 for how a name is built; scholars discuss whether this name is ever really used).
Applies when / blocked by. SK 1 cannot make this a a marker: it is neither final in its line nor a consonant. That is why this second rule is needed.
Watch out. Inside a name built from the Śivasūtras, the markers are not counted as members. Bhaṭṭoji's proof: the word anunāsika is used in this very rule. If the k ending ṛ ḷ k counted as one of the vowels named by ac, then in anunāsika the i before k would be treated as a vowel before a vowel, would be changed to y by the rule iko yaṇ aci (see SK 13) and then dropped, and we would hear anunāska. We hear anunāsika, so markers are not members.
aṇ (a, i, u) and ac (all nine vowels)Now that the markers are known, the grammar can build its short names. A name made by taking a first sound and a marker and meaning everything from the one to the other is a pratyāhāra, literally "a drawing-together". The two names built here, aṇ and ac, are used in almost every later rule.
Rule. An initial sound, pronounced together with a final it, is the name of itself and of every sound in between. The name so formed is a pratyāhāra.
Example.
a (initial, Śivasūtra 1) + the it ṇ (end of Śivasūtra 1) → aṇ → names a, i, u.
a (initial, Śivasūtra 1) + the it c (end of Śivasūtra 4, ai au c) → ac → names a, i, u, ṛ, ḷ, e, o, ai, au, the nine vowels.
ha (initial, Śivasūtra 5) + the it l (end of Śivasūtra 14) → hal → names every consonant.
Applies when / blocked by. "Initial" and "final" are counted over the stretch of sounds you have in mind, not over one Śivasūtra line. So i of line 1 can be the initial of ik (i, u, ṛ, ḷ) although it is not first in its line. Only an it can close a name; a consonant that is not a marker cannot. Forty-four such names are actually used in the grammar.
Watch out. The markers lying inside the stretch (the ṇ of line 1 inside ac, for instance) are not among the sounds named; see SK 3. Also, aṇ is ambiguous: two Śivasūtras end in ṇ (lines 1 and 6). Unless a rule says otherwise, aṇ is read with the earlier ṇ and means a, i, u; only SK 14 reads it with the later ṇ.
a, sorted into 3 durations × 3 pitches = 9 kindsA vowel is not one thing but a family. The five rules of this group name its three durations and three pitches, so that "a" comes out in nine varieties. Rules about vowels later on need these names.
Rule. A vowel whose duration equals that of u, ū or u3 is named, in that order, hrasva (short), dīrgha (long) or pluta (protracted). The unit of duration is the mātrā: u lasts one mātrā, ū two, u3 three. The sūtra's ū is the three sounds u, ū, u3 run together.
Example.
a of one mātrā → same duration as u → hrasva.
ā of two mātrās → same duration as ū → dīrgha.
a3 of three mātrās → same duration as u3 → pluta.
Watch out. Each of the three is itself threefold by pitch (SK 5–7), which is how the vowel becomes ninefold.
Rule. A vowel produced in the upper part of its place of articulation (the palate, the throat and so on, each of which has parts) is named udātta, "raised". The word ac is carried forward here from SK 4 by anuvṛtti.
Example.
ā ye (the opening of the Vedic verse ā ye mitrāvaruṇā) → the ā and the e are produced high in their places → both are udātta.
Rule. A vowel produced in the lower part of its place of articulation is named anudātta, "not raised". Bhaṭṭoji says only "clear", since the rule mirrors SK 5.
Example.
arvāṅ (opening a Vedic verse) → its first a is produced low → that a is anudātta.
Rule. A vowel in which the two properties udātta-ness and anudātta-ness are gathered together is named svarita.
Example.
the ī of agnim īḷe → one vowel that is partly raised and partly not → svarita (the parts are fixed by SK 8).
Rule. Of a svarita vowel, the first half is udātta; the remainder, by elimination, is anudātta. The sūtra says "half a short", but Bhaṭṭoji says this mention of "short" is not essential: the first half of any svarita, long or short, is udātta.
Example.
the svarita ī of agnim īḷe → its first half is udātta → its second half is anudātta.
In kva vo 'śvāḥ, rathānāṃ na ye 'rāḥ, śatacakraṃ yo hyaḥ → the svarita is followed by an udātta or another svarita → its second half is heard clearly as anudātta.
In agnim īḷe → no udātta or svarita follows the ī → its second half sounds raised (the udātta-śruti described in the Prātiśākhya, an old treatise on Vedic recitation), though it is anudātta by rule.
Watch out. Count the varieties so far: 3 durations × 3 pitches gives 9 kinds of every vowel. Bhaṭṭoji adds that each of these 9 is again twofold, nasal or non-nasal; that gives the 18 of SK 9.
daṇḍa + āḍhakam → daṇḍāḍhakam (lengthening fires) against dadhi + harati → dadhi harati (change blocked)Rules of sandhi speak of a sound "of the same kind" as another. This group defines that sameness (savarṇa), fixes which sounds a bare letter in a rule stands for, and then draws the boundary: a vowel and a consonant are never of the same kind. The two running examples show the boundary at work. In daṇḍa + āḍhakam the a and ā are of one kind, so they merge into ā. In dadhi + harati the i and h are not, so the i stays.
Rule. A sound uttered by the mouth together with the nose is named anunāsika, nasal.
Example.
a nasal a → uttered by mouth and nose → anunāsika.
a, i, u, ṛ → each has 9 kinds (SK 4–8) × nasal or not → 18 varieties each.
ḷ → has no long form → 12 varieties.
e, o, ai, au (the group ec) → have no short form → 12 varieties each.
Rule. Sounds that share the same place of articulation in the mouth (sthāna) and the same internal effort (ābhyantara-prayatna) are named savarṇa, "of the same kind", with respect to each other.
The places, as Bhaṭṭoji lists them:
Effort is of two kinds, internal and external. The internal effort is fourfold: spṛṣṭa (full contact) for the stops, īṣatspṛṣṭa (slight contact) for the semivowels, vivṛta (open) for the spirants and the vowels, saṃvṛta (closed) for the short a — but only in finished words; while a derivation is running, the short a is treated as open, like every other vowel. Bhaṭṭoji says Pāṇini himself signals this, and shows it at SK 11–12.
Example.
daṇḍa + āḍhakam → a and ā both belong to the throat and both are open → they are savarṇa → the sandhi of two savarṇa vowels gives one long vowel → daṇḍāḍhakam.
Watch out. Sameness means sameness of the whole set of places in the mouth; the nose does not count. So i (palate) and e (throat and palate) are not savarṇa, and in bhavaty eva the i and e do not merge; but ṅ (throat, plus nose) is savarṇa with k (throat). Scholars discuss the details.
Rule. The last sūtra of the whole grammar. It restates the short a, which the Śivasūtras taught as open, and prescribes the closed a in its place. In other words, finished words have the closed a; the grammar's own sound-list has the open a.
Example.
daṇḍa as a finished word → its a is closed (saṃvṛta), as in ordinary speech.
daṇḍa inside the derivation of daṇḍāḍhakam → SK 11 has not yet "happened" (see SK 12) → its a is still open → still savarṇa with the open ā → daṇḍāḍhakam is obtained.
Applies when / blocked by. This rule is asiddha — treated as not having taken effect — with respect to the entire Aṣṭādhyāyī before it, by SK 12. That is why, from the grammar's point of view, the a stays open all through a derivation.
Rule. This is an adhikāra, a heading that is read into every rule that follows it rather than a rule on its own. Its effect: the last three quarters of the Aṣṭādhyāyī (A 8.2 to 8.4, the Tripādī) are asiddha — treated as not having taken effect — with respect to the seven and a quarter books before them (the Sapādasaptādhyāyī); and within the Tripādī, a later rule is asiddha with respect to an earlier one. What is suspended is the rule itself, not merely its result.
Example.
SK 11 (a a, A 8.4.68) → lies in the Tripādī → asiddha for every earlier rule → while SK 10 is at work, a is still open → daṇḍāḍhakam.
nis + ay- (see SK 23) → the rule turning final s into ru is A 8.2.66, later in the Tripādī than the l-rule A 8.2.19 → the ru is asiddha for that l-rule → no l, hence nirayate.
Watch out. Under this sūtra Bhaṭṭoji finishes his account of the sounds. External effort (bāhya-prayatna) is elevenfold: vivāra, saṃvāra, śvāsa, nāda, ghoṣa, aghoṣa, alpaprāṇa, mahāprāṇa, udātta, anudātta, svarita. The first and second sounds of each stop-class (the group khay), their yamas, the jihvāmūlīya, the upadhmānīya, visarga and ś, ṣ, s have vivāra, śvāsa and aghoṣa (unvoiced); all other sounds have saṃvāra, nāda and ghoṣa (voiced). The first, third and fifth of each class, the yamas of the first and third, and y, r, l, v are alpaprāṇa (light breath); the rest are mahāprāṇa. A yama is the echo of a stop heard between it and a following nasal, as in palikknīḥ, cakhkhnatuḥ, aggniḥ, ghghnanti. The sounds from k to m are the stops (sparśa); y, r, l, v the semivowels (antastha); ś, ṣ, s, h the spirants (ūṣman); the vowels are svara. The two half-visarga signs before k and p are the jihvāmūlīya and the upadhmānīya; in aṃ and aḥ the elements after the vowel are anusvāra and visarga. External effort plays no part in the savarṇa name; it matters later, when a substitute must be chosen as the closest sound.
Two supplementary statements (vārttikas) close the account. First: ṛ and ḷ are to be treated as savarṇa with each other, although their places differ. Second: by SK 10 as it stands, a and h, i and ś, ṛ and ṣ, ḷ and s would each come out savarṇa (same place, both open). The next rule removes that result.
Rule. A vowel (ac) and a consonant (hal) are never savarṇa of each other, whatever their place and effort. Bhaṭṭoji reads the sūtra's ac as āc, "ac together with ā", so that the prohibition covers ā as well.
Example.
dadhi + harati → i (palate, open) and h (throat, open; but a consonant) → not savarṇa → i does not become y → dadhi harati.
dadhi + śītalam → i and ś both palate and open, but vowel and consonant → not savarṇa → no merging into a long vowel → dadhi śītalam.
dadhi + ṣaṣṭham, dadhi + sāndram → likewise unchanged.
viśvapābhiḥ → the ā after p is not savarṇa with h → the rule replacing h by ḍh (ho ḍhaḥ) does not touch it.
Applies when / blocked by. Without this rule, h, ś, ṣ, s would count as vowels whenever a rule says "a", "i", "ṛ", "ḷ", by force of SK 14 — just as the long vowels do — and the i of dadhi would change before them.
Rule. A sound of the group aṇ, or a sound marked with u as its it (an udit), when it is not itself the thing being prescribed, names its savarṇas as well as its own form. Here, and only here, aṇ is read with the later ṇ (Śivasūtra 6), so it covers the vowels and h, y, v, r, l. The udits are ku, cu, ṭu, tu, pu, each naming its whole class of five stops. This is the grahaṇaka rule, the rule by which a letter "grasps" its kind.
Example.
the i in a rule such as iko yaṇ aci → is an aṇ, and is not what the rule prescribes → names all 18 kinds of i (short, long, pluta, three pitches, nasal or not).
a, i, u → each names 18. ṛ → names 30 (its own 18 plus the 12 of ḷ, the two being savarṇa by the vārttika under SK 12). ḷ → likewise 30. each of e, o, ai, au → names 12.
ku → names the whole k-class, the five stops beginning with k.
non-nasal y → names the nasal and the non-nasal y; likewise v and l.
Applies when / blocked by. Not when the sound is the thing prescribed: an a being taught as a substitute is just a. And never across the vowel–consonant line (SK 13): i names ī, never ś. e and ai, o and au are not savarṇa with each other; if they were, Pāṇini would not have begun the separate Śivasūtra ai au c (scholars discuss this reason). So no ec vowel names 24.
Watch out. The rule says aṇ rather than ac precisely to bring in y, v, l, which have nasal and non-nasal varieties.
Rule. A sound that has t after it, or stands after t (either is tapara), names only savarṇas of its own duration. This restricts SK 14: the marker t keeps out every variety of a different duration; for at, it, ut that means the long and protracted ones, and for a tapara long vowel it means the short ones. A rule that limits another in this way is a niyama.
Example.
at (in adeṅ guṇaḥ, SK 17) → tapara → names only the six one-mātrā a-sounds (three pitches × nasal or not) → not ā, not a3.
it, ut → likewise six each.
ṛt → six of ṛ plus six of ḷ → twelve.
Watch out. The aic of SK 16 is also tapara, standing after the t of āt. With SK 15 the toolkit is complete: names are built (SK 2), each letter grasps its kind (SK 14), and t trims the grasp to one length (SK 15).
Two names for sets of vowels, used whenever a later rule says "the vowel takes guṇa" or "takes vṛddhi". Pāṇini opens his grammar with them; Bhaṭṭoji delays them until the names they are written in (āt, aic, at, eṅ) can be read.
Rule. āt and aic — that is, ā, ai, au — are named vṛddhi.
Example.
āt → the t after ā removes any doubt → the long ā only (scholars discuss whether SK 15 is needed here).
aic → ai (initial, Śivasūtra 4) + the it c → ai, au.
together → ā, ai, au → each is a vṛddhi vowel.
Watch out. Though this is the first sūtra of the Aṣṭādhyāyī, it needs SK 2, SK 14 and SK 15 to be read, which is why it comes sixteenth here.
Rule. at and eṅ — that is, a, e, o — are named guṇa.
Example.
at → the tapara a → the short a only.
eṅ → e (initial, Śivasūtra 3) + the it ṅ → e, o.
together → a, e, o → each is a guṇa vowel.
pra + √nī → praṇayati; and the nis/nir splitSix rules name classes of words: roots, particles, and the prefixes that attach to verbs. The running example shows why the names matter. pra before √nī is an upasarga, and only an upasarga turns the n of nī into ṇ: praṇayati. The list of prefixes also keeps nis and nir apart, and the last rule shows the different forms that result.
Rule. The items of the list beginning with bhū (the Dhātupāṭha), when they denote an action, are named dhātu, root. Bhaṭṭoji reads the sūtra's vā as "and items like vā", that is, action-denoting ones.
Example.
bhū → first item of the list, denotes "being" → dhātu.
nī (in praṇayati) → an item of the list, denotes "leading" → dhātu.
Applies when / blocked by. "Denoting an action" keeps out the particle vā "or", which is spelled like the root vā but denotes no action. "Beginning with bhū" keeps out hiruk, which denotes exclusion but is not in the list.
Rule. An adhikāra: everything taught from here up to, but not including, the sūtra adhir īśvare (A 1.4.97) is named nipāta, particle. The word nipāta is carried by anuvṛtti through all the rules in between (SK 20–23 among them).
Example.
ca (SK 20), pra (SK 21) → taught inside this stretch → each is a nipāta.
Watch out. The sūtra says rīśvara, quoting adhirīśvare with its r, so that an earlier sūtra containing īśvare (A 3.4.13) is not taken as the end-point.
Rule. The words of the list beginning with ca, when their meaning is not a substance (sattva, dravya), are named nipāta. A substance is something that carries gender, number and a case-role; "and", "only", "not" carry none.
Example.
ca "and", eva "only" → meaning is not a substance → nipāta.
paśu in chāgaḥ paśuḥ "the goat is an animal" → in the ca-list, but here names a substance → not a nipāta.
paśu in puṣṭaṃ paśu manyate "he rightly deems it well-fed" → means "rightly" → nipāta.
Rule. pra and the rest (the prādi list, given in full under SK 23), when their meaning is not a substance, are likewise named nipāta.
Example.
pra (in praṇayati) → a prādi, not naming a substance → nipāta.
Rule. The prādis, when construed with a verbal action, are named upasarga, prefix.
Example.
pra + √nī → pra is construed with the action "lead" → pra is an upasarga → the ṇ-change that holds "after an upasarga" applies → praṇayati, praṇeyam.
Applies when / blocked by. Only in connection with an action; a prādi standing alone is a nipāta but not an upasarga.
Rule. And the prādis, in the same connection with an action, are also named gati. The ca "and" makes the two names, upasarga and gati, hold together — an exception to the general principle that one item gets only one name — and the prādis keep the name nipāta as well. The prādis are: pra, parā, apa, sam, anu, ava, nis, nir, dus, dur, vi, āṅ, ni, adhi, api, ati, su, ud, abhi, prati, pari, upa.
Example.
pra + √nī → pra is gati and upasarga at once → both the operations stated for an upasarga (the ṇ of praṇeyam) and those stated for a gati apply.
nir + ay- "go" → the r of an upasarga becomes l before √ay (A 8.2.19) → nilayate; likewise dulayate.
nis + ay- → nis ends in s, which becomes ru only by A 8.2.66 → that rule is later in the Tripādī, so asiddha for A 8.2.19 (SK 12) → no l → nirayate; likewise durayate.
Watch out. This is why nis and nir, dus and dur are listed separately: the s-forms and the r-forms behave differently.
er ac and agner ḍhakA paribhāṣā is a rule about how to read other rules. Three of them are placed here. They settle what "optionally" means, whether a word in a rule means itself or its synonyms, and whether "after i" means "after a stem ending in i".
Rule. The name vibhāṣā belongs to negation and option taken together: "not; or". In ordinary usage vibhāṣā means bare option; in this grammar it means "first no, then optionally yes". By extension, every word for option in the grammar (vā, anyatarasyām and the rest) is read the same way.
Example.
√śvi in the perfect (liṭ), with vibhāṣā śveḥ → in the dual and plural endings a vowel-change (saṃprasāraṇa) is already compulsory by another rule; in the singular it is not available at all → the "not" first cancels it where it was compulsory, so the whole perfect is uniform → then one uniform option runs → śuśāva / śiśvāya, śuśuvatuḥ / śiśviyatuḥ.
Applies when / blocked by. The word iti in the sūtra matters: without it, the name would belong to the words na and vā themselves (by SK 25), and they would be substituted as if they were forms. Scholars discuss whether the sūtra is needed at all.
Rule. In a rule, a word names its own form — not its meaning, and not its synonyms — unless the word is a technical name of the grammar itself (a śabda-saṃjñā).
Example.
agner ḍhak → agni names the word agni only → the affix ḍhak is added after agni, not after any synonym for fire.
vṛddhir ādaic → vṛddhi is a grammar-name → it names ā, ai, au (SK 16), not the sound-string vṛddhi.
Watch out. The exception is still needed even though SK 16–17 define the names; scholars discuss whether the sūtra is needed.
Rule. A word used as a qualifier in a rule names both what ends in it and itself. This is the tadanta reading, "that which ends in it".
Example.
er ac (this ac is an affix, a different thing from the vowel-name ac) "after i, the affix ac" → i names roots ending in i and the root i itself → cayaḥ (from √ci), ayaḥ (from √i).
Applies when / blocked by. Two vārttikas: in a rule prescribing a compound (samāsa) or an affix (pratyaya), the tadanta reading is barred — except where the qualifier is an ugit (an item named after its marker; the class is not defined in this chapter) or a single sound (varṇa). So ṅīp by ugitaś ca still follows the ugit-ending stem in atipacantī, and iñ by ata iñ still follows the a-ending stem in dākṣiḥ.
śikṣā (its cluster kṣ, and the weight of the preceding i); contrast sudṛṣatprāsādaThe last seven names describe where sounds stand and how heavy a vowel is. In śikṣā the k and ṣ form a cluster, and the i before the cluster counts as heavy; that heaviness is what licenses the affix that makes the word. The contrast shows that a name for the cluster belongs to the whole cluster, not to one consonant.
Rule. The cessation of utterance — the absence of any further sound — is named avasāna, pause.
Example.
a word spoken last, with nothing after it → after its final sound no sound follows → that position is avasāna → rules stated "in avasāna" apply there (for instance kharavasānayoḥ, which puts visarga in place of r).
Watch out. Bhaṭṭoji's commentators give two readings of what exactly bears the name (the final sound itself, or the absence after it); the result for the rules is the same.
Rule. The utmost closeness of sounds — no gap beyond the half-mātrā that is unavoidable — is named saṃhitā. This is the condition under which sandhi operates.
Example.
daṇḍa āḍhakam spoken with no pause → the a and ā stand in saṃhitā → the sandhi of two savarṇa vowels (a later rule) is available, because SK 10 has made a and ā savarṇa → daṇḍāḍhakam.
Rule. What ends in a sup or a tiṅ is named pada, a finished word. sup is the pratyāhāra from su, the first case-ending, to the p of the last (the list sv-au-jas-am-auṭ …); tiṅ is the set of verb-endings.
Example.
rāmaḥ → ends in a case-ending → pada.
bhavati → ends in a verb-ending → pada.
Watch out. sup here is the whole set of case-endings, not just the one ending spelled sup (the locative plural).
Rule. Consonants with no vowel between them are named saṃyoga, a cluster. Two are enough; where more stand together, each pair is also a saṃyoga.
Example.
śikṣ → k and ṣ stand together with no vowel between → kṣ is a saṃyoga.
Applies when / blocked by. The name belongs to the group as a whole, not to each consonant singly. So in sudṛṣatprāsāda "palace of fine stone", the final t of sudṛṣat, though next to the p of prāsāda, is not itself "a saṃyoga", and the rule that deletes a word-final saṃyoga does not delete it.
Rule. A short vowel is named laghu, light. Bhaṭṭoji leaves it unglossed, the sense being plain.
Example.
the i of śikṣ → a short vowel → laghu in itself.
Rule. When a saṃyoga follows, a short vowel is named guru, heavy. The word hrasvam "short" is carried forward from SK 31.
Example.
the i of śikṣ → followed by the saṃyoga kṣ → guru → the affix a prescribed "after a consonant preceded by a guru" (guroś ca halaḥ) is available → śikṣā.
Rule. A long vowel too is named guru, and needs no following saṃyoga: the word saṃyoge is not carried forward from SK 32.
Example.
the ā of śikṣā → a long vowel → guru, cluster or no cluster.
With this the saṃjñā-prakaraṇa ends: these are the names from the first book of the Aṣṭādhyāyī that the sandhi rules will use.
| SK | Sūtra | What it does | Example |
|---|---|---|---|
| Text · mangala | yenākṣarasamāmnāyam … munitrayaṃ namaskṛtya … | Four verses of homage: two to Pāṇini, one to the three munis; verse 4 is Bhaṭṭoji's own maṅgala. | Pāṇini, Vararuci, Patañjali |
| Text · intro | munitrayaṃ namaskṛtya taduktīḥ paribhāvya ca | Bhaṭṭoji's maṅgala: the book follows the three munis; its title gives subject, purpose, relation and student. | kaumudī = moonlight |
| Text · mahesvara | iti māheśvarāṇi sūtrāṇy aṇādisaṃjñārthāni … | The Śivasūtras come from Maheśvara and exist for names like aṇ; their finals and the a of la ṇ are it; the other *a*s are for pronunciation. | the commentary draws from it the SK 1 ↔ SK 2 circularity |
| 1 | halantyam | A final consonant in upadeśa is a marker (it). | ṇ of a i u ṇ |
| 3 | upadeśe 'j anunāsika it | A nasal vowel in upadeśa is a marker. | a of la ṇ |
| 2 | ādir antyena sahetā | Initial sound + final marker names the stretch between. | aṇ = a, i, u; ac = the nine vowels |
| 4 | ūkālo 'j jhrasvadīrghaplutaḥ | Vowel of 1 / 2 / 3 mātrās is hrasva / dīrgha / pluta. | a, ā, a3 |
| 5 | uccair udāttaḥ | Vowel produced high is udātta. | ā ye |
| 6 | nīcair anudāttaḥ | Vowel produced low is anudātta. | first a of arvāṅ |
| 7 | samāhāraḥ svaritaḥ | Vowel combining both is svarita. | ī of agnim īḷe |
| 8 | tasyādita udāttam ardhahrasvam | First half of a svarita is udātta, the rest anudātta. | agnim īḷe |
| 9 | mukhanāsikāvacano 'nunāsikaḥ | Sound uttered by mouth and nose is anunāsika; vowels are 18-fold. | nasal a |
| 10 | tulyāsyaprayatnaṃ savarṇam | Same place and internal effort = savarṇa. | daṇḍa + āḍhakam → daṇḍāḍhakam |
| 11 | a a | Finished words have the closed a; the sound-list has the open a. | daṇḍa |
| 12 | pūrvatrāsiddham | The Tripādī is asiddha for earlier rules; later for earlier within it. | a a inert during derivation |
| 13 | nājjhalau | A vowel and a consonant are never savarṇa. | dadhi harati (no change) |
| 14 | aṇudit savarṇasya cāpratyayaḥ | An aṇ or udit letter names its savarṇas too. | i = 18 kinds of i; ku = k-class |
| 15 | taparas tatkālasya | A tapara letter names only its own duration. | at = six short a |
| 16 | vṛddhir ādaic | ā, ai, au are vṛddhi. | ā, ai, au |
| 17 | adeṅ guṇaḥ | a, e, o are guṇa. | a, e, o |
| 18 | bhūvādayo dhātavaḥ | Action-denoting items of the bhū-list are dhātu. | bhū, nī |
| 19 | prāg rīśvarān nipātāḥ | Heading: up to adhir īśvare, items are nipāta. | ca, pra |
| 20 | cādayo 'sattve | ca etc., when not naming a substance, are nipāta. | ca, eva |
| 21 | prādayaḥ | pra etc., likewise, are nipāta. | pra |
| 22 | upasargāḥ kriyāyoge | Prādis with a verbal action are upasarga. | pra + √nī → praṇayati |
| 23 | gatiś ca | …and gati as well; the prādi list. | nilayate vs nirayate |
| 24 | na veti vibhāṣā | vibhāṣā = negation and option together. | śuśāva / śiśvāya |
| 25 | svaṃ rūpaṃ śabdasyāśabdasaṃjñā | A word in a rule means its own form, unless it is a grammar-name. | agner ḍhak |
| 26 | yena vidhis tadantasya | A qualifier names what ends in it and itself. | er ac → cayaḥ, ayaḥ |
| 27 | virāmo 'vasānam | Cessation of utterance is avasāna. | end of an utterance |
| 28 | paraḥ sannikarṣaḥ saṃhitā | Closest joining of sounds is saṃhitā. | daṇḍāḍhakam |
| 29 | suptiṅantaṃ padam | What ends in sup or tiṅ is pada. | rāmaḥ, bhavati |
| 30 | halo 'nantarāḥ saṃyogaḥ | Consonants with no vowel between are saṃyoga. | kṣ of śikṣā |
| 31 | hrasvaṃ laghu | A short vowel is laghu. | i of śikṣ |
| 32 | saṃyoge guru | A short vowel before a saṃyoga is guru. | i of śikṣ → śikṣā |
| 33 | dīrghaṃ ca | A long vowel is guru. | ā of śikṣā |