14. Avyaya-prakaraṇa — Indeclinables: the words that keep one form

After this chapter you will be able to say which Sanskrit words are avyaya, "indeclinable", and why their case-endings vanish. You will know the two ready-made lists that supply most of them (the svarādi words and the particles), the three derivational routes that create more of them (a taddhita form, a kṛt form, an avyayībhāva compound), and the one operation that the name exists for: after an avyaya, the feminine affix āp and the case-ending sup are dropped by luk. You will also be able to explain why atyuccaisau keeps its ending while tatra loses its affix, and to state, in Pāṇini's own terms, what "does not vary" means.

Before you start

• From chapter 1 (Saṃjñā-prakaraṇa) you need the idea of a saṃjñā, a technical name given by a rule, and of anuvṛtti, the carrying of a word forward from an earlier rule. You also need marker letters (it, SK 1–3): an affix is quoted with its markers, which are dropped before it is heard — so ṇamul sounds as am, śadhyai as adhyai, ktvā as tvā, tosun as tos, kasun as as. You also need the pratyāhāra names ec (e, o, ai, au) and sup (the case-endings).
• From chapter 1 you need the name nipāta, particle: an item of the ca-list (cādi) or the pra-list that does not name a substance (SK 19–21). This chapter turns every nipāta into an avyaya.
• From chapter 1 you need the tadanta reading (SK 26): a name given to an affix belongs to the whole word that ends in that affix. Two rules below depend on it.
• From chapter 8 (Ajanta-puṃliṅga-prakaraṇa) you need the twenty-one case-endings su … sup and the name luk (SK 260), the complete disappearance of an affix.
• From chapter 9 (Ajanta-strīliṅga-prakaraṇa) you need the feminine affix āp, the ā that ends stems such as ramā and śālā.
• Three kinds of affix and compound named in this chapter — taddhita, kṛt, avyayībhāva — are taught in full only in later chapters. Here you need only the labels, which are explained where they appear.

A. The off-the-shelf indeclinables — particles and adverbs made avyaya by list — running example: uccaiḥ

Most indeclinables are not derived at all. They are simply listed. One rule hands the name avyaya to two lists at once: the svarādi words (svar "heaven", antar "within", uccais "high", svasti "well-being" and many more) and the nipātas, the particles such as ca "and". So uccaiḥ "loudly, high" is an avyaya straight off the list, with no derivation to perform.

SK 447 · स्वरादिनिपातमव्ययम् · svarādinipātam avyayam (A 1.1.37)

Rule. The words of the list beginning with svar (the svarādi), and the nipātas, receive the name avyaya, indeclinable. A nipāta is a particle as defined in chapter 1: an item of the cādi list, or of the prādi list, that does not name a substance.

The svarādi list begins: svar, antar, prātar, punar, sanutar, uccais, nīcais, śanais, ṛdhak, ṛte, yugapat, ārāt, pṛthak, śvas, divā, rātrau, sāyam, ciram, manāk, īṣat, joṣam, tūṣṇīm, bahis, avas, samayā, nikaṣā, svayam, vṛthā, naktam, nañ, hetau, iddhā, addhā, sāmi. A gaṇasūtra — a rule inside the list itself — adds vat, brāhmaṇavat, kṣatriyavat, sanā, sanat, sanāt, upadhā, tiras, antarā, antareṇa, jyok, kam, śam, sahasā, vinā, nānā, svasti, svadhā, alam, vaṣaṭ, śrauṣaṭ, vauṣaṭ, anyat, asti, upāṃśu, kṣamā, vihāyasā, doṣā, mṛṣā, mithyā, mudhā, purā, mitho, mithas, prāyas, muhus, pravāhukam, pravāhikā, āryahalam, abhīkṣṇam, sākam, sārdham, namas, hiruk, dhik, atha, am, ām, pratām, praśān, pratān, mā, māṅ. The list is an ākṛtigaṇa: an open, illustrative class, so a word of the same kind counts even if it is not printed.

The cādi list of nipātas begins: ca, vā, ha, aha, eva, evam, nūnam, śaśvat, yugapat, bhūyas, kūpat, kuvit, sūpatsa, net, cet, caṇ, kaccit, kiṃcit, yatra, naha, hanta, mākiḥ, mākim, nakiḥ, ākim, māṅ, nañ, yāvat, tāvat, tvai, dvai, rai, śrauṣaṭ, vauṣaṭ, svāhā, svadhā, tum, tathāhi, khalu, kila, atho, atha, suṣṭhu, sma, ādaha. Its gaṇasūtra upasarga-vibhakti-svara-pratirūpakāś ca adds words that merely look like (pratirūpaka) a prefix, a case-ending or a particle: avadattam, ahayuḥ, astikṣīrā; the bare sounds a, ā, i, ī, u, ū, e, ai, o, au; paśu, śukam, yathākathāca, pāṭ, pyāṭ, aṅga, hai, he, bhoḥ, aye, dya, viṣu, ekapade, yut, ātaḥ. The cādi too is an ākṛtigaṇa.

Example.

uccaiḥ → it stands in the svarādi list → it is named avyaya; nothing has to be derived.

svasti → it stands in the svarādi list (by the gaṇasūtra) → avyaya.

ca → it is a nipāta of the cādi list → avyaya.

Applies when / blocked by. A cādi word is a nipāta, and so an avyaya, only when it does not name a substance (SK 20). Bhaṭṭoji's commentators, under SK 452, add that this restriction is stricter for the cādi than for the svarādi: a svarādi word is avyaya whether or not it names a substance.

Watch out. Both lists are open. The point of printing them is to show the kind of word meant, not to close the class. Note also that the gaṇasūtra of the cādi makes the bare vowels a, ā, i and the rest nipātas when they are used as particles; do not confuse these with the vowels of the Śivasūtras.

B. A taddhita-derivate declared avyaya — the parigaṇana gate — running example: pacatirūpam

A taddhita is a secondary affix, one added after a finished nominal or verbal form rather than directly after a root. Some taddhita affixes yield words that never show number: they take every case-ending, but only ever in the singular. Such a word is an avyaya. The sūtra states this in general terms, but a supplementary statement then fixes the class by an explicit enumeration, and only the enumerated affixes count. The running example shows the gate at work: pacatirūpam "he cooks admirably" ends in a taddhita and shows no number, yet it is not enumerated, so it is kept out.

SK 448 · तद्धितश्चासर्वविभक्तिः · taddhitaś cāsarvavibhaktiḥ (A 1.1.38)

Rule. A word ending in a taddhita affix, when it is asarva-vibhakti, is also named avyaya. A vibhakti is a case-ending. The word asarva-vibhakti is a bahuvrīhi compound (a compound naming what possesses something) and must be read with care. It does not mean "from which all the case-endings fail to arise"; an avyaya does take all seven case-endings. It means "from which the whole case-ending — the ending in all three numbers — does not arise, but only the singulars arise". The word sarva is taken as "the whole, all the parts", as in sarvaḥ paṭo dagdhaḥ "the whole cloth is burnt".

Which taddhita words these are is fixed by a parigaṇana, an enumeration, stated in a vārttika (a supplementary statement on the sūtra). The enumerated affixes are:

• the group beginning with tasil, up to pāśap — that is, from pañcamyās tasil (A 5.3.7) to dvitryoś ca dhamuñ (A 5.3.45);
• the group beginning with śas, up to the samāsānta (compound-final) affixes — from bahvalpārthāt (A 5.4.42) to avyaktānukaraṇāt (A 5.4.57);
• am (of amu ca chandasi) and ām (of kimettiṅavyayaghād āmu);
• the affixes meaning "times": kṛtvasuc (saṃkhyāyāḥ kriyābhyāvṛttigaṇane kṛtvasuc, A 5.4.17), suc (dvitricaturbhyaḥ suc, A 5.4.18) and dhā (vibhāṣā bahoḥ, A 5.4.20);
• tasi (of tenaikadik … tasiś ca) and vati (of tena tulyam, A 5.1.115);
• nānā and ñau (of vinañbhyāṃ nānāñau na saha, A 5.2.27).

Example.

pacatirūpam → it ends in the taddhita rūpap ("admirably"), and shows no number → but rūpap is not in the enumeration → it is not an avyaya.

pacatikalpam → it ends in the taddhita kalpap ("somewhat"), likewise un-enumerated → not an avyaya.

Any word ending in an enumerated affix → the whole word, by the tadanta reading, is an avyaya.

Applies when / blocked by. Only the enumerated affixes give the name. Without the enumeration the rule would over-apply (ati-vyāpti): to pacatirūpam and pacatikalpam, and also to the word ubhaya "both", which ends in a taddhita and shows no number but is not an avyaya. Bhaṭṭoji's commentators note that the tasi of pratiyoge pañcamyās tasiḥ is already covered by the śas group, and that, since the enumeration alone settles the class, the sūtra itself could strictly have been dispensed with.

Watch out. The wrong reading — "an avyaya takes no case-endings" — is the classic mistake. An avyaya takes all seven, each in the singular only. That is exactly why SK 452 below names the whole pratyāhāra sup: the endings are all there to be dropped. Bhaṭṭoji's commentators ground the "singular only" on the Mahābhāṣya's reading of the number rules: the singular is the general case, and the dual and plural override it only when duality or plurality is meant; where no number is meant at all, the singular of every case is what arises.

C. A kṛt- or compound-derivate declared avyaya — running example: smāraṃ smāram

A kṛt is a primary affix, one added directly after a root. Three rules now declare certain kṛt-formations, and one kind of compound, to be avyaya. The first picks out kṛt affixes by their final sound (m or an ec vowel); the second names three kṛt affixes outright (ktvā, tosun, kasun); the third adds the avyayībhāva compound. The running example smāraṃ smāram "remembering again and again" is a kṛt-formation ending in m, doubled to express repetition.

SK 449 · कृन्मेजन्तः · kṛnmejantaḥ (A 1.1.39)

Rule. A kṛt affix that ends in m, or that ends in an ec vowel (e, o, ai, au), is named avyaya. The sūtra is divided kṛt — m-ec-antaḥ, "a kṛt whose end is m or ec". Since a bare affix is never used on its own, the name passes, by the tadanta reading, to the word that ends in such a kṛt.

Example.

√smṛ "remember" + ṇamul (heard as am; added in the sense of repetition by ābhīkṣṇye ṇamul ca, A 3.4.22) → vṛddhi of the root vowel by aco ñṇiti (A 7.2.115): ṛ becomes ā, and the substitute for ṛ is followed by r (raparatva, SK 70), giving ār → smāram → doubled by nityavīpsayoḥ (A 8.1.4) → smāraṃ smāram; the kṛt ends in m → avyaya.

√jīv "live" + ase (by tumarthe sesenase…, A 3.4.9) → jīvase; the kṛt ends in e → avyaya.

√pā "drink" + śadhyai (heard as adhyai; by the same rule) → because the affix is śit it is a sārvadhātuka → the root takes the substitute piba by pāghrādhmā… (A 7.3.78) → pibadhyai; the kṛt ends in ai → avyaya.

Watch out. pibadhyai takes no śap, the affix that normally comes before an agentive sārvadhātuka. The reason, on the settled view, is that a kṛt-avyaya of this kind expresses the bare action (bhāve), not an agent; scholars discuss this.

SK 450 · क्त्वातोसुन्कसुनः · ktvā-tosun-kasunaḥ (A 1.1.40)

Rule. A word ending in any of the three kṛt affixes ktvā, tosun or kasun is named avyaya. The sūtra means "that which ends in these" (etad-anta).

Example.

kṛtvā "having done" → it ends in ktvā → avyaya.

udetoḥ "until sunrise" → it ends in tosun → avyaya.

visṛpaḥ → it ends in kasun → avyaya.

SK 451 · अव्ययीभावश्च · avyayībhāvaś ca (A 1.1.41)

Rule. And an avyayībhāva compound is named avyaya. The word avyaya is carried into this sūtra by anuvṛtti from the rules above, and ca "and" adds the compound to what it names; the predicate "receives the name avyaya" is supplied. The avyayībhāva is a class of compound taught in the next prakaraṇa, which follows this chapter.

Example.

adhihari "in Hari" → it is an avyayībhāva compound, formed in the sense of a case-relation and meaning harau → by this rule it is an avyaya.

D. The payoff — luk of the āp and sup enjoined after an avyaya — running example: tatra

Now the name is cashed out. Five rules have said which words are avyaya; one rule says what follows: the feminine affix āp and the case-ending sup that are enjoined after an avyaya disappear by luk. This is why an avyaya has one form only. The running example is tatra "there": in the sense "in that hall" (śālāyām) the word would be feminine and would call for āp and a case-ending, yet we hear only tatra.

SK 452 · अव्ययादाप्सुपः · avyayād āp-supaḥ (A 2.4.82)

Rule. After an avyaya, the āp and the sup that are enjoined after it undergo luk, complete elision. āp-sup is a collective compound (samāhāra-dvandva) of the two affixes, put in the genitive. The word luk is not written in the sūtra; it is carried by anuvṛtti from ṇyakṣatriyārṣañito yūni luk (A 2.4.58). Bhaṭṭoji reads the sūtra as "of the affix enjoined after an avyaya" (avyayād vihitasya), not "of the affix that merely follows an avyaya".

Example.

tatra "there", used of a feminine, śālāyām "in the hall" → the feminine affix āp is due → tatra is an avyaya → āp undergoes luk → tatra. (śālāyām is written only to show the feminine gender; nothing of it enters the form.)

atha → a case-ending sup is due after it → atha is an avyaya → the sup undergoes luk → atha. (Bhaṭṭoji's commentators supply this example for the sup.)

Applies when / blocked by. The qualifier "enjoined after" (vihita) blocks the rule in atyuccaisau "two who have gone beyond uccaiḥ". The analysis is uccair atikrāntaḥ; the compound is formed by atyādayaḥ (A 2.2.18, the ati-group in the sense "surpassing"). Here the case-ending does stand after the avyaya uccaiḥ, but it is enjoined after the compound aty-uccais, not after the avyaya. So there is no luk, and the dual ending au stays.

Nor is the whole aty-uccais itself an avyaya, though it ends in one. Only the bare word uccaiḥ is in the svarādi list. The tadanta reading does apply, in principle, to the avyaya name; but it does not reach a subordinate member (gauṇa, upasarjana) of a compound. Bhaṭṭoji's commentators trace this to the Mahābhāṣya: the bar "not for a subordinate member" that limits the name sarvanāman carries into the avyaya name as well.

Scholars discuss how an avyaya such as uccaiḥ can enter a compound in an accusative sense at all; the answer Bhaṭṭoji's commentators give need not detain a first reading.

The vārttika. āb-grahaṇaṃ vyartham aliṅgatvāt: the mention of āp in this sūtra is idle, because an avyaya has no gender (aliṅgatva). Without gender no feminine affix is ever due, so there is nothing for the rule to drop. Bhaṭṭoji's commentators quote the Mahābhāṣya's description: words that do not swerve from feminine and neuter, from the qualities of a substance, and from singular, dual and plural — that is an avyaya.

The old definition and its reading. A Vedic verse cited in the Mahābhāṣya defines the avyaya: sadṛśaṃ triṣu liṅgeṣu sarvāsu ca vibhaktiṣu / vacaneṣu ca sarveṣu yan na vyeti tad avyayam — "alike in the three genders, in all the case-relations and in all the numbers, that which does not vary is an avyaya". Read at face value this contradicts "no gender". Bhaṭṭoji's answer is that the verse's purport is the absence of gender, case-role (kāraka) and number. His commentators explain: its locatives (liṅgeṣu …) are the locative of disregard, taught by ṣaṣṭhī cānādare (A 2.3.38): "disregarding gender, case-role and number, that which does not vary but keeps one form is an avyaya". Here vibhakti and vacana mean the case-role and the number, not the affixes; otherwise one of the two words would be redundant.

Bhaṭṭoji's commentators add a limit. The absence of gender, case-role and number holds strictly of the nipātas; some svarādi words do enter a case-relation, as in svasti vācayati, svasti vācyam, kṣīṇe puṇye svaḥ patati, prātar yajate. This matches the Mahābhāṣya on SK 447: of the cādi, only those not naming a substance are avyaya; of the svarādi, both those that name a substance and those that do not.

A verse of Bhāguri, added in passing. vaṣṭi bhāgurir allopam avāpyor upasargayoḥ / āpaṃ caiva halantānāṃ yathā vācā niśā diśā: the teacher Bhāguri wants the a of the prefixes ava and api dropped, and the āp of consonant-final feminine stems dropped as well. So vagāhaḥ beside avagāhaḥ, pidhānam beside apidhānam; and vācā, niśā, diśā as feminines without āp. In ava only the first a is meant, by its pairing with api. Bhaṭṭoji's commentators doubt that this verse really stands in the Mahābhāṣya, and hold that vācā as a form without āp is simply incorrect (asādhu); scholars discuss this. With this the indeclinables close, and the Avyayībhāva-prakaraṇa follows.

Terms used in this chapter

• avyaya — indeclinable: the name given by SK 447–451 to the words whose āp and sup are dropped by SK 452.
• saṃjñā — a technical name given by a rule (chapter 1).
• anuvṛtti — carrying a word forward from an earlier sūtra into a later one (chapter 1).
• pratyāhāra — a short name for a stretch of sounds or affixes; here ec (e, o, ai, au) and sup (chapter 1).
• sup — the pratyāhāra naming the twenty-one case-endings su … sup (chapters 1 and 8).
• nipāta — a particle: a cādi or prādi word not naming a substance (chapter 1, SK 19–21).
• tadanta — "that which ends in it": the reading by which a name given to an affix belongs to the whole word ending in it (chapter 1, SK 26).
• luk — the complete disappearance of an affix (chapter 8, SK 260).
• āp — the feminine affix ā that ends stems such as ramā, śālā (chapter 9).
• svarādi — the list of indeclinables beginning with svar (SK 447).
• cādi — the list of particles beginning with ca (SK 447).
• prādi — the list of prefixes beginning with pra (chapter 1, SK 23).
• gaṇasūtra — a rule placed inside a word-list, extending the list.
• ākṛtigaṇa — an open, illustrative list: words of the same kind belong to it even if not printed.
• pratirūpaka — look-alike: a nipāta that has the form of a prefix, a case-ending or a particle.
• sattva / dravya — a substance: a meaning that carries gender, number and a case-role (chapter 1).
• taddhita — a secondary affix, added after a finished nominal or verbal form (taught in later chapters).
• vibhakti — a case-ending; also, in the old verse, the case-role it expresses.
• asarva-vibhakti — "from which the whole three-number ending does not arise": taking every case only in the singular (SK 448).
• bahuvrīhi — a compound naming the thing that possesses what its members describe.
• parigaṇana — an explicit enumeration fixing the members of a class.
• vārttika — a supplementary statement on a sūtra, correcting or completing it (chapter 1).
• ati-vyāpti — over-application: a rule reaching cases it should not.
• samāsānta — an affix added at the end of a compound.
• kṛt — a primary affix, added directly after a root (taught in later chapters).
• ec — the pratyāhāra e, o, ai, au (chapter 1).
• ṇamul, ase, śadhyai, ktvā, tosun, kasun — kṛt affixes whose formations are avyaya (SK 449–450).
• vṛddhi — the name of ā, ai, au (chapter 1, SK 16).
• raparatva — the rule that the substitute for ṛ (or ḷ) is followed by r (or l), so vṛddhi of ṛ gives ār (SK 70).
• it — a marker letter written with an affix and dropped before it is heard (chapter 1, SK 1–3).
• śit — an affix whose marker is ś; such an affix is a sārvadhātuka.
• sārvadhātuka — a class of verbal affix, taught later; before it the root pā is replaced by piba.
• śap — the affix that normally precedes an agentive sārvadhātuka; absent in pibadhyai.
• bhāve — "in the sense of the bare action", as opposed to the agent.
• avyayībhāva — a class of compound, taught in the next prakaraṇa; adhihari is one (SK 451).
• samāsa — a compound (chapter 1).
• samāhāra-dvandva — a collective compound, "āp and sup taken together".
• vihita — enjoined after: an affix taught with the avyaya as its base, not merely standing after it (SK 452).
• gauṇa / upasarjana — a subordinate member of a compound; the avyaya name does not reach it.
• sarvanāman — the name of the pronoun class; its bar on subordinate members is shared with the avyaya name.
• aliṅga / aliṅgatva — without gender; the reason the mention of āp is idle.
• kāraka — the case-role a word plays toward an action (agent, object, location …).
• śruti — a Vedic statement; here the old verse defining the avyaya.
• ṣaṣṭhī cānādare / anādara — the rule (A 2.3.38) giving the genitive and locative in the sense of disregard.
• al-lopa — loss of the sound a; Bhāguri's vagāhaḥ for avagāhaḥ.
• upasarga — a prefix construed with a verbal action (chapter 1, SK 22).
• asādhu — incorrect: not a form the grammar licenses.
• Mahābhāṣya — Patañjali's great commentary on Pāṇini, which Bhaṭṭoji's commentators cite for several points above.

Summary