Pāṇini's grammar is a machine for producing words. You feed it a root or a stem plus what you want to say, and its
3,983 rules build the finished word step by step. This page shows what goes in, what comes out, how
the eight chapters split the work, and how a derivation hands off from one chapter to another.
What is the Aṣṭādhyāyī, fundamentally?
Yes, it is a generative grammar: a finite set of rules that derives every correct Sanskrit word from a small set of
inputs. In Wikipedia's words, it "takes material from lexical lists (Dhātupāṭha, Gaṇapāṭha) as input and describes
algorithms to be applied to them for generation of well-formed words." Kiparsky describes the rules as mapping what
the speaker means onto sound in stages: meaning → kāraka relations → abstract morphemes → pronounced form.
Inputs
Verbal roots (dhātu)
A list outside the sūtras, the Dhātupāṭha (2,229 entries in this site's copy). 1.3.1भूवादयो धातवः names its members dhātu.
Nominal stems (prātipadika)
Any meaningful word that is not a root or affix (1.2.45), and stems made by kṛt, taddhita and compounding (1.2.46). Gaṇa lists such as sarvādi (1.1.27) and uṇādi words (3.3.1) supply more.
When two rules compete, the later one wins (1.4.2) and a specific rule beats a general one. The last three pādas are a one-way final pass: 8.2.1पूर्वत्रासिद्धम् makes them invisible to every earlier rule.
Sentences are strings of such padas. Compounds and derived nouns are made by running the engine more than once and feeding one output back in as a new stem.
Is it true that chapters 3–5 add the suffixes and 6–7 do the heavy lifting?
Mostly yes, with two refinements. Here is the conventional division, with every range taken from the heading (adhikāra) sūtras in this site's data:
1
Definitions and metarules; kāraka (1.4.23), ātmanepada/parasmaipada (1.3), person/number (1.4). The heading 1.4.1 runs to 2.2.38.
2
Compounds (2.1.3 to 2.2.38), case endings for kārakas (2.3.1 to 2.3.72), number/gender of compounds, root substitutes and luk deletions (2.4).
3
The affix section opens: 3.1.1प्रत्ययः runs to 5.4.160 (1,821 sūtras). Within it 3.1.91धातोः (to 3.4.117) covers affixes after roots: sanādi, vikaraṇas, kṛt, and the lakāras with their tiṅ endings.
4–5
Affixes after nominal stems: 4.1.1ङ्याप्प्रातिपदिकात् runs to 5.4.160: case endings sup (4.1.2), feminine affixes (4.1.3 to 4.1.81), and taddhitas (4.1.76 to 5.4.160).
6–7
Operations on the string: reduplication and sandhi (6.1, 6.1.72 to 6.1.157), compound accent and first members (6.2–6.3), and the stem section 6.4.1अङ्गस्य which runs to 7.4.97 (613 sūtras): augments, guṇa/vṛddhi, affix replacements.
8
Word repetition and sentence accent (8.1; 8.1.16 runs to 8.3.54), then the Tripādī: 8.2.1 to 8.4.68 (295 sūtras) of final phonology.
Refinement 1. Adhyāya 3 is "verbal" in the sense of what the affix attaches to. Many of its affixes (kṛt) make nouns from verbs: पाचक "cook" and गत "gone" get their affixes in Adhyāya 3 (3.1.133, 3.2.102) and their case endings in Adhyāya 4.
Refinement 2. "Heavy lifting" depends on what you count. In the 833 sample derivations this site's build runs through vidyut, Adhyāya 1 fires the most steps (58% of all rule applications), because its it-marker deletion and labelling rules repeat at every stage. Adhyāyas 6–7 fire only 10% of the steps, but they supply 53% of the different sūtras those derivations use: they are where the word gets its particular shape.
AdhyāyaShare of the textShare of rule applicationsShare of different sūtras used
19%58%28 · 11%
27%1%10 · 4%
316%17%45 · 18%
416%3%2 · 1%
514%0%0 · 0%
618%5%56 · 22%
711%5%76 · 30%
89%11%33 · 13%
Sample: 833 derivations (40 common roots in all ten lakāras plus a passive form, and 16 noun stems
in all cases and numbers). Adhyāyas 2, 4 and 5 are under-represented because the sample has no compounds and few
derived nouns. Across the 14 curated examples below, the split of Aṣṭādhyāyī steps by adhyāya is
1: 185 · 2: 0 · 3: 45 · 4: 8 · 5: 0 · 6: 26 · 7: 15 · 8: 31 (of 310).
Run the engine
Pick a word. The derivation is computed by vidyut,
a software implementation of the Aṣṭādhyāyī, when this site is built. Watch the token hop between stations in the
factory: it does not roll neatly from 1 to 8, because labels and marker deletion (Adhyāya 1) are needed again after
every new affix, and the finishing rules of Adhyāya 8 come last.
How the factory maps onto the grammar
Inputs: the crates are the root and stem lists (Dhātupāṭha, gaṇas); the order slip is the speaker's intent.
Station 1, the control room: definitions and metarules. It labels parts (dhātu, it, aṅga, pada) and strips marker letters.
Station 2: compounds, case assignment, zero-deletions.
Stations 6–7: the big presses: reduplication, sandhi, guṇa/vṛddhi, augments, stem changes.
Station 8, behind one-way glass: the Tripādī. It sees everything the earlier stations did; they cannot see its work (8.2.1).
The token is the word in progress; its label shows the current form and the rule that just fired.
Verbs (tiṅanta)
Nouns from verbs (kṛt)
Nouns from nouns (taddhita)
Inflected nouns (subanta)
Input
root भू (Dhātupāṭha 01.0001, "to be")(a dhātu by 1.3.1) intent: present tense · 3rd person · singular · active
Outputभवति
The classic verb pipeline: the present-tense lakāra laṭ is added (3.2.123), its l is replaced by the ending tip (3.4.78), the vikaraṇa śap is inserted (3.1.68), the root takes guṇa (7.3.84: bhū → bho), and sandhi turns o + a into av + a (6.1.78). Marker letters are stripped by 1.3.9 again and again along the way.
step 1 / 20
1.3.1Definitions & metarulesभूवादयो धातवः1.3.1 · dhātu, it-markers, voiceThe words भू and others, when used to denote an action, are called 'धातु'.
Every square is one sūtra, in order: eight columns for the adhyāyas, four blocks each for the pādas. Zoom in for pāda
summaries and heading brackets (grey lines beside the cells mark the reach of each adhikāra), further for sūtra
numbers, and further still for the text. The sūtras used by the example above are outlined.
Loading the map…
Chapters and pādas
Eight adhyāyas of four pādas each. The colour is each part's main role in a derivation: definitions, compounds & case, verbal affixes, nominal affixes, stem operations, word level, tripādī.
1
Definitions & metarules
351 sūtras · Definitions
Sets up the vocabulary the rest of the grammar runs on (vṛddhi, guṇa, it, dhātu, prātipadika, pada, kāraka, person and number) and the metarules that say how rules are to be read and applied.
vṛddhi and guṇa (1.1.1–3), savarṇa, pragṛhya, sarvanāma, avyaya; metarules for substitution: the genitive means "in place of" (1.1.49), a substitute behaves like the original (1.1.56), locative/ablative mark the right/left context (1.1.66–67).
kit/ṅit status, accent, prātipadika· 73
Which affixes behave as kit or ṅit (1.2.1–26, an atideśa block; kit/ṅit affixes block guṇa by 1.1.5), the accent terms udātta/anudātta/svarita, the definition of a nominal stem (prātipadika, 1.2.45–46), upasarjana, and ekaśeṣa (one word standing for several, 1.2.64).
dhātu, it-markers, voice· 93
Defines dhātu (1.3.1); marker letters (it, 1.3.2–8) and their deletion (1.3.9); then which roots take ātmanepada endings (1.3.12 onward) and which take parasmaipada (1.3.78).
One label each; kāraka; person & number· 110
Under 1.4.1, labels such as laghu/guru, aṅga (1.4.13), pada (1.4.14) and bha (1.4.18); the conflict rule 1.4.2; the six kārakas (1.4.23 kārake: apādāna 1.4.24 … kartṛ 1.4.54); nipāta, upasarga and gati (1.4.56ff); persons and numbers of verb endings (1.4.99–108).
2
Compounds, case & deletions
268 sūtras · Compounds & case
Compounds (samāsa) and their types; which case ending expresses which kāraka relation; then the number and gender of compounds, root substitutes before ārdhadhātuka affixes, and zero-deletions (luk) of affixes.
Only connected words combine (2.1.1). Under 2.1.3 samāsa: avyayībhāva compounds (2.1.5–21), then tatpuruṣa (from 2.1.22), including dvigu (2.1.52) and karmadhāraya types.
Compounds: bahuvrīhi, dvandva, order· 38
The rest of tatpuruṣa, then bahuvrīhi (2.2.23) and dvandva (2.2.29), and which member is placed first (2.2.30–38).
Case for each kāraka· 73
Under 2.3.1 anabhihite ("if not already expressed"): accusative for the object (2.3.2), instrumental for agent and instrument (2.3.18), dative, ablative, locative; nominative for the bare stem meaning (2.3.46); genitive for the rest (2.3.50).
Number/gender of compounds; root substitutes; luk· 85
Singular and gender of certain compounds (2.4.1–31); substitutes for roots before ārdhadhātuka affixes (under 2.4.35, e.g. as → bhū 2.4.52); luk deletion of affixes, e.g. of a case ending inside a stem (2.4.71) and of śap after ad etc. (2.4.72).
3
Affixes after roots
631 sūtras · Verbal affixes
Opens the long affix section (3.1.1 pratyayaḥ, 3.1.2 paraś ca, running to the end of 5.4). Adhyāya 3 covers affixes added after a dhātu: sanādi affixes that make new roots, the vikaraṇas, kṛt affixes (which make nouns and participles from verbs), and the lakāras that express tense and mood, replaced by tiṅ endings.
Opens the affix headings (3.1.1–3). Affixes that make new roots: desiderative san (3.1.7), intensive yaṅ (3.1.22), causative ṇic (3.1.26); 3.1.32 calls the results dhātus. Then vikaraṇas (sya/tās 3.1.33, cli 3.1.43, śap 3.1.68, u 3.1.79 …). From 3.1.91 dhātoḥ: kṛtya affixes (3.1.95) and agent affixes ṇvul/tṛc (3.1.133).
kṛt with upapadas; past and present· 188
kṛt affixes after a root with a dependent word (3.2.1 karmaṇy aṇ …); past-time affixes (3.2.84 bhūte), including niṣṭhā kta/ktavatu (3.2.102) and laṅ/liṭ (3.2.111, 3.2.115); laṭ for the present (3.2.123) and its participles; habitual agents (3.2.134ff).
uṇādi, future, action nouns, moods· 176
uṇādi affixes (3.3.1); future lṛṭ and luṭ (3.3.13, 3.3.15); action and instrument nouns: ghañ (3.3.18 bhāve), feminine ktin (3.3.94), lyuṭ (3.3.115); then the moods: liṅ (3.3.161) and loṭ (3.3.162).
Gerunds; lakāra → tiṅ endings· 117
Infinitive and gerund affixes (ktvā 3.4.21, ṇamul); voice of the lakāra (3.4.69); then under 3.4.77 lasya the l is replaced by the 18 tiṅ endings (3.4.78) with their adjustments; sārvadhātuka vs ārdhadhātuka (3.4.113–114).
4
Affixes after stems (I)
635 sūtras · Nominal affixes
Affixes added after a nominal stem (4.1.1, in force to 5.4.160): the 21 case endings (sup), feminine affixes (ṭāp, ṅīp …), and the start of the taddhitas: derived nouns for "descendant of", "dyed with", "deity of", "born in", "made of" and many more.
After 4.1.1 ṅyāp-prātipadikāt: the sup endings (4.1.2); feminine affixes under 4.1.3 striyām (ṭāp 4.1.4, ṅīp 4.1.5 …); 4.1.76 taddhitāḥ opens the taddhitas, 4.1.83 makes aṇ the default, and 4.1.92 starts "his descendant" (apatya).
Taddhitas: dyed, deity, group, place· 145
Taddhitas meaning "dyed with" (4.2.1), "whose deity is" (4.2.24), "a group of" (4.2.37), place names (4.2.67), and from 4.2.92 śeṣe the "remaining" senses such as "belonging to / coming from".
Taddhitas: born in, his, made of· 168
Continuing the śeṣa senses: "born there" (4.3.25), "this is his" (4.3.120), "a product of" (4.3.134), with many stem-specific affixes.
ṭhak and yat senses· 144
Two default-affix blocks: ṭhak for senses like "plays / digs / wins with" (4.4.2) and "lives by" (4.4.12), from 4.4.1 to 4.4.74; and yat for senses like "what bears it" (4.4.76), from 4.4.75 to the end of the pāda.
5
Affixes after stems (II)
555 sūtras · Nominal affixes
More taddhitas: "good for", "bought with", "similar to", abstract nouns (-tva, -tā), possessives (-mat/-vat), pronominal adverbs (tatra, tataḥ), comparatives and superlatives (-tara, -tama), and affixes that close compounds (samāsānta).
Default-affix blocks: cha (5.1.1), e.g. for "good for" (5.1.5); ṭhañ (5.1.18), e.g. for "bought with" (5.1.37); vati "like" (5.1.115); abstract nouns in -tva/-tā (5.1.119).
Fields, measures, possessives· 140
Affixes for fields of a crop, measures and quantities (5.2.41), and the possessive senses "who has / in which is" with matup (5.2.94) and its rivals (-in/-ika 5.2.115, -vin 5.2.121 …).
Pronominal adverbs, comparison· 119
Under 5.3.1, case-like affixes on pronouns (tasil 5.3.7, tral 5.3.10: tataḥ, tatra); comparatives and superlatives (5.3.55, 5.3.57); ka for "unknown" and diminutive (5.3.70 heading; 5.3.73, 5.3.86).
Meaning-preserving affixes; compound finals· 160
Affixes that do not change the sense (svārthika) and the samāsānta affixes added at the end of compounds (5.4.68 to the end of 5.4), which closes the affix section opened at 3.1.1.
6
Reduplication, sandhi, accent, stem changes
736 sūtras · Stem operations
Operations on the string once affixes are in place: reduplication and saṃprasāraṇa, vowel sandhi at junctions (6.1.72 saṃhitāyām), accent, changes to the first member of compounds, and from 6.4.1 aṅgasya the long section of changes to the stem (aṅga) before an affix, which continues to the end of Adhyāya 7.
Reduplication (6.1.1, in the perfect 6.1.8); saṃprasāraṇa (y/v → i/u, e.g. 6.1.15); sandhi under 6.1.72 saṃhitāyām: yaṇ (6.1.77), ay/āv (6.1.78), single replacements under 6.1.84 (guṇa 6.1.87, long vowel 6.1.101); deletion of su/ti/si after consonants and ī/ā (6.1.68); accent from 6.1.158.
Accent of compounds· 199
Which syllable of a compound carries the udātta accent: first member keeps its accent (6.2.1 …), then headings for initial and final accent of the second member (6.2.64, 6.2.92, 6.2.111, 6.2.143).
First member of compounds· 139
Changes to the first member before a second member: non-deletion of its case ending (6.3.1 aluk), mahat → mahā (6.3.46), lengthening and other substitutions, and (under 6.3.114 saṃhitāyām) lengthening of a final vowel before certain words.
aṅga operations begin· 175
6.4.1 aṅgasya opens the stem section (to 7.4.97). Lengthening, nasal deletion (6.4.37), the past-tense augment a- (6.4.71), and from 6.4.22 a block treated as asiddha among itself; bha-stem operations (6.4.129ff, e.g. 6.4.148).
7
Stem & affix operations (aṅga)
438 sūtras · Stem operations
Still under 6.4.1 aṅgasya: replacing affixes after the stem (yu/vu → ana/aka, ṭā → ina), augments such as iṭ and num, guṇa and vṛddhi of the stem, changes of noun stems before case endings, and the rules for the reduplicated syllable.
Replacements of affixes after particular stems: yu/vu → ana/aka (7.1.1), jh → ant (7.1.3), bhis → ais (7.1.9), ṭā → ina (7.1.12); the augment num (7.1.58 …).
iṭ augment; vṛddhi; pronouns· 118
vṛddhi in the s-aorist (7.2.1); the iṭ augment and when it is blocked (7.2.10, 7.2.35); pronoun stem substitutions (7.2.102); vṛddhi before ñit/ṇit affixes (7.2.115–117).
guṇa; noun stems before endings· 120
Taddhita vṛddhi variants (7.3.1–31); k/g for c/j (7.3.52); gam → gacch (7.3.77); guṇa before affixes (7.3.84); stem changes before case endings (7.3.102 lengthening, 7.3.103 -e-, 7.3.119).
Causative aorist; the reduplicated syllable· 97
Shortening in the causative aorist (7.4.1), various stem substitutions, then under 7.4.58 the shape of the reduplicated syllable: short vowel (7.4.59), only the first consonant kept (7.4.60), k → c (7.4.62).
8
Words in the sentence, then the Tripādī
369 sūtras · Tripādī
Pāda 1 repeats words and fixes the accent of words in a sentence. From 8.2.1 pūrvatrāsiddham the last three pādas form the Tripādī: each of its rules is treated as not applied (asiddha) by all earlier rules, and by earlier rules within it. This is where final-consonant changes, rutva/visarga, retroflexion (ṣ, ṇ) and assimilation happen.
Doubling of words (8.1.1, e.g. for "every" 8.1.4); 8.1.16 padasya heads the word-level rules to 8.3.54; accent of words in a sentence, e.g. a finite verb loses its accent after a non-verb (8.1.28).
Tripādī begins: final consonants· 108
8.2.1 pūrvatrāsiddham. Deletion of final n of a stem (8.2.7) and of the last of a final cluster (8.2.23); c/j → k/g (8.2.30); voiced finals (8.2.39); niṣṭhā t → n (8.2.42); s → ru (8.2.66); pluta vowels (8.2.82ff).
Visarga, anusvāra, s → ṣ· 119
ru/r → visarga (8.3.15), final m → anusvāra (8.3.23), and the retroflex ṣ (8.3.55 heading; 8.3.59 after i/u etc.).
n → ṇ; assimilation; last rule· 68
ṇatva (8.4.1, 8.4.2); assimilation of s/dentals to palatals (8.4.40) and of stops in voicing (8.4.53, 8.4.55); svarita accent (8.4.66); and the last sūtra 8.4.68 a a, restoring the closed short a.
George Cardona, Pāṇini: A Survey of Research, The Hague: Mouton, 1976; and Pāṇini: His Work and its Traditions, vol. 1, Delhi: Motilal Banarsidass, 1988 (2nd ed. 1997).
Rama Nath Sharma, The Aṣṭādhyāyī of Pāṇini, vol. 1: Introduction to the Aṣṭādhyāyī as a Grammatical Device, New Delhi: Munshiram Manoharlal, 1987.
Sūtra text, English glosses and heading scopes: ashtadhyayi.com open data. Derivations: vidyut-prakriya (ambuda-org), run at build time.
Summaries of chapters and pādas are this site's, written from the heading sūtras and the sūtras of each pāda (see 3.1.1 and the other headings linked above).