संस्कृतव्याकरणम् · 16 of 24 Home
कृदन्तप्रकरणम्

One letter decides
the whole शब्दः

Every noun and adjective in Sanskrit that is not a pronoun, a number or a plain अव्यय began as a root plus an affix. This page is about the affix half of that — कृत्, the machinery that turns कृ "to do" into a doer, a deed, a duty, an infinitive, a participle. There is no list to memorise. There is one letter on the affix, and it already tells you what is about to happen to the root.

कृ
+
ण्वुल्
→
कारकः
वृद्धिः — the इत् is ण्
धातुः — the root प्रत्ययः — the कृत् affix, इत् letters and all निष्पन्नं पदम् — root and affix joined, इत् gone, marker applied

What कृत् means, and where the boundary is

Every other word class on this site has a positive definition: a pronoun is one of a closed list, a तिङ् ending is one of eighteen. कृत् has none of that. It is defined by what is left over once something else is subtracted.

कृदतिङ्The whole definition, four syllables long

3.1.91 opens a long stretch of the Aṣṭādhyāyī with a one-word scope: everything that follows, until stated otherwise, is prescribed after a root. Inside that stretch, 3.1.93 says the rest of the sentence in three words: an affix that comes after a root and is not तिङ् gets the saṃjñā कृत्. Not "one of these forty affixes" — whatever is left once the तिङ् endings (the eighteen personal verb-endings, त्/सि/मि… — 3.4.78 , walked in full on page 11) are excluded.

The Kāśikā's own test of the boundary. Its gloss on 3.1.93 asks अतिङिति किम्? — "why say not तिङ्?" — and answers with चीयात्, स्तूयात्: two ordinary विधिलिङ् verb forms that also sit after a root, inside this very stretch of rules. Without the exclusion, the grammar would have to call them कृत् too. अतिङ् is not decoration — it is the one word keeping every finite verb form on this site out of this page.

This corpus's own kṛt table, C.krt(), carries affixes across groups — the working slice of a class the Aṣṭādhyāyī itself never bothers to count, because it is not a list, it is everything धातोः अतिङ् from 3.1.91 to the end of the third adhyāya. Page 17 and page 18 already take a double-digit share of that number — the participles, and क्त्वा/ल्यप्/तुमुन् — in real depth. This page does not repeat either. It shows the one mechanism that makes all of them predictable instead of memorised.

Agent nouns — ण्वुल् and तृच्

3.1.133

One sūtra, two affixes, the same job: name the doer. What tells कारकः and कर्ता apart is not the meaning — both mean "one who does" — it is one letter each carries and loses on the way.

ण्वुल्तृचौकृ, one sūtra, two words

3.1.133 — after any root (सर्वे धातवः), कर्तरि, both ण्वुल् and तृच् are licensed at once. The Kāśikā's own examples are this page's own root:

ण्वुल् — इत् ण्, ल् — काशिका: कारकः, हारकः तृच् — इत् च् — काशिका: कर्ता, हर्ता

ण् is the letter that matters. 7.2.115 — before a णित् or ञित् affix, the aṅga takes not merely गुण but वृद्धि. ण्वुल्'s own इत् letter is ण्, so कृ is raised all the way: ऋ's वृद्धि is आर् (1.1.51 keeps the added vowel र्-parked) — कृ → कार्, giving . तृच् has only च् as its इत् — nothing that reaches for वृद्धि, nothing that blocks गुण either — so the default of 7.3.84 applies instead, exactly as it does for a verb stem on page 12: plain गुण, ऋ → अर्, कृ → कर् — giving .

कारक is not only this page's word. Page 08 कारकम् borrows this exact word for its whole topic — "the six roles a noun can play in a sentence." Pāṇini built his own technical vocabulary out of an ordinary agent-noun affix.
कर्तृDeclines exactly like पितृ

तृच् always leaves an ऋ-final stem, and every ऋ-final agent noun declines the same way a kinship term like पितृ "father" does — the same endings, cell for cell:

Action nouns — घञ्, क्तिन्, ल्युट्, अच्, अप्

3.3

Where agent-noun affixes name the doer, this family names the doing itself, or its result — and one of them, क्तिन्, decides a whole grammatical gender before any noun-specific rule gets a chance to.

घञ्भावे — वृद्धि, and a consonant flip besides

3.3.18 — घञ् names the action or state (भावे). Its इत् letters are घ् and ञ् — and ञ् is exactly the letter 7.2.115 watches for, so घञ् raises to वृद्धि precisely as ण्वुल् did. On a consonant-final root the parallel rule is 7.2.116 — वृद्धि of the penultimate अ, not the final vowel:

The इत् does more than pick the vowel grade. 7.3.52 — before an affix marked घ्, ण् or ण्यत्, a root-final च् or ज् becomes its velar counterpart क्/ग्. That is why त्यज् "to renounce" surfaces as त्यागः, not *त्याजः, and पच् "to cook" as पाकः, not *पाचः — the same घ्/ण् marker that forces वृद्धि also reaches into the final consonant.
क्तिन्स्त्रियां — why so many action nouns are feminine

3.3.94 — क्तिन् names the action, in the feminine. The corpus's own kṛt table files क्तिन् under a group it calls स्त्रीलिङ्गवाचिनः, "feminine-denoting," alongside क्यप्, अ, अङ्, युच्, ण्वुल् (in its abstract-noun use), इञ्, ण्वुच् and अनि — ten affixes whose gender the affix decides before any rule about the resulting noun's shape gets a say:

क्तिन्'s own इत् letters, क् and न्, are exactly the letters 1.1.5 watches for — गुण and वृद्धि are both blocked, so the root vowel never moves. That mechanism gets its own section below.

ल्युट्-अन, the plainest action noun

3.3.115 — ल्युट् names the action too, spelled -अन once its own ट् (इत्) is gone:

अच् and अप् — two more affixes in this same भावार्थक/कर्मार्थक territory, each licensed for a short, specific list of roots rather than सर्वे धातवः (अच् for the पचादि group and a handful of others; अप् for ग्रह्, वृ, दृ, गम् and more). Both are real entries in the reference table below, with every root the corpus's kṛt table names for them — this page does not manufacture an example where the grammar itself reaches for a short, closed list instead.

The इत् letter is the whole story

7.2.115 · 7.3.84 · 1.1.5

Look back at what just happened. Two different root families, five different affixes, five different jobs — agent, action, action-in-the-feminine — and every single vowel-grade outcome traced back to exactly one of three rules, decided by exactly one letter on the affix.

इत् अक्षरम्परिणामःसूत्रम्उदाहरणानि
This is not a special rule for कृत्. 7.3.84 सार्वधातुकार्धधातुकयोः is the exact rule page 12 used for तुद् and भू — because a कृत् affix, like every affix that is not सार्वधातुक, is आर्धधातुक (3.4.114 ), and 7.3.84 covers both. 1.1.5 क्ङिति च is the same blocking rule too. A kṛt-formed word and a finite verb form are built by the same three-way branch on the same letter; this page and page 12 are teaching one mechanism from two directions.

Read this as the page's actual claim, not a summary of it: कृत् affixes are not a list to memorise. They are a small set of इत् letters, applied to a root. Once ण्/ञ् means वृद्धि, क्/ङ् means blocked, and anything else means plain गुण, the 122 affixes in the reference table below stop being 122 things to learn and become one branch, run 122 times.

Where a कृत् word goes next

1.2.45 · 1.2.46

A kṛt affix does not build a finished sentence-word. It builds something one step short of one — and that one step is what lets Sanskrit build its entire vocabulary out of a few thousand roots.

प्रातिपदिकम्The saṃjñā that makes everything after this page possible

1.2.45 — a meaningful form that is neither a root nor an affix is called प्रातिपदिकम्, a nominal stem. A kṛt-formed word does not obviously qualify on its own — it is visibly root-plus-affix, not some independent third thing — so 1.2.46 reaches in and says so directly: कृत्, तद्धित and समास words all count as प्रातिपदिक too. The Kāśikā's own examples on this exact line are, again, this page's own root:

Because कारकः and कर्ता are प्रातिपदिक, three things follow, none of them optional:

  • They take सुप् endings and decline across all seven cases, exactly like any other noun — page 05.
  • The same rule that gives a compound like राजपुरुषः or ब्राह्मणकम्बलः its own single declension — the Kāśikā's own समास examples on 1.2.46 — also covers a compound built with a कृत् word as either member — page 09.
  • They can themselves take a तद्धित affix and become the base of a further word — औपगवः, "descendant of उपगु," is the Kāśikā's own example, a तद्धित word built on a name that was never itself a root.
This is the actual mechanism of Sanskrit vocabulary. A root gives a handful of finite-verb forms directly. Everything else the language has — every abstract noun, every agent noun, every adjective built from an action — passes through 1.2.46 first. कृत् is not a side-topic of verb morphology. It is most of where nouns come from.

One root, eight words

Interactive

कृ, walked through every family this page has taught at once — three agent/action markers side by side, then the two gerundives, and finally the infinitive, the absolutive and the participle that pages 17 and 18 each cover in depth. Step through each: the affix arrives with its इत् letters, the इत् vanishes, the marker decides वृद्धि / गुण / neither, and the word comes out the other side.

Every step above is read off the same कृदन्तानि engine this page's data was built from — derived once over CDP, not composed by hand. For any of the corpus's affixes on any of its 2,259 roots, derive it live and read the engine's own rule-history, not a curated eight-word slice.

All , browsable

Coverage

Every affix the engine implements, grouped exactly as the corpus groups them. Filter by group or search by name; the sūtra column links straight into the reading view.

प्रत्ययःइत्दर्शनम्समूहःसूत्रम्

The twelve groups, by name

समूहःविषयःसंख्या
Where this comes from. Every row is C.krt()['krt'] — the corpus's own kṛt table, as vidyut-prakriya implements it, with इत् letters, group and licensing sūtras exactly as the engine records them. The group table is C.krt()['krtGroups'], unedited. Nothing here is retyped from a printed list.

Check yourself

अभ्यासः

Six questions.