कारकविवेकः — when one participant could be two rôles

Where the choice comes from

1.4.42 · 1.4.49

Page 08 already showed you the two words that make this whole page possible, without dwelling on what they cost. Look again, because everything below is what happens when a sentence gives two candidates equal claim on a superlative.

-तमTwo definitions built on a superlative, not a fact

1.4.49 — the most aimed-at thing becomes कर्म 1.4.42 — the most effective means becomes करण

Neither sūtra says the object or the instrument — both say -तम, most. A superlative presupposes a field of candidates to rank, and ranking is exactly the kind of judgment a rule cannot make for you: it can only say which ranking wins once you have made it. When a sentence has only one plausible candidate for a role, the superlative is invisible — everyone ranks the same way and there is nothing to notice. This page collects the sentences where two candidates are close enough that the ranking becomes a real decision, one the tradition itself records disagreeing over.

This is not the same claim as अनभिहिते. (page 08 · 13) governs whether a case ending appears at all — whether the verb has already spoken for a role. Everything on this page presupposes a role is unexpressed and asks a prior question: unexpressed as what? The two chapters are still separate; this page just shows 1.4 arguing with itself before 2.3 ever gets a turn.

Six places the tradition argues

1.4 · quoted, not composed

Every quotation below is the Kāśikā's own text (or, for अधिकरणम्'s three kinds, the Laghukaumudī, Bālamanorāma and Tattvabodhinī), pulled through C.layer() and trimmed to the definition and the worked examples — nothing composed for this page unless a ·edited· box says so.

अपादानम् · करणम्The fixed point, or the effective means

1.4.24 needs a participant that is ध्रुवम् — fixed, stationary, while the action moves away from it. 1.4.42's करण needs the opposite: the one thing that most actively brings the result about. Most of the time these are obviously different things — nothing about a chariot moves when someone falls from it. But the boundary is not "from vs by"; it is stillness vs effect, and a single word can satisfy either reading depending on what the speaker takes to be doing the work.

गङ्गायां घोषः is the Kāśikā's own trap. "The herdsmen's camps are on the Ganges" sounds instrumental in English — the river is what makes the camps possible. But 1.4.42's own -तम excludes it: the Ganges is not the most effective means of anything happening there, only the ground the camps happen to sit near. gives it सप्तमी instead — this page's next entry, अधिकरणम्, not करणम् at all.

सम्प्रदानम् · कर्मThe recipient, or what the anger is aimed at

1.4.32 defines सम्प्रदानम् as whoever the agent has in mind as the one to benefit — the ordinary "give to". But four sūtras later, 1.4.37 extends the same case to a very different feeling: the person one is angry at. क्रुध्, द्रुह्, ईर्ष्य्, असूय् — anger, malice, envy, fault-finding — all direct their चतुर्थी at "the one the displeasure faces", by the same logic that directs it at a gift's recipient.

Add one उपसर्ग and the recipient becomes an object. देवदत्ताय क्रुध्यति — plain क्रुध्, चतुर्थी, सम्प्रदान. देवदत्तम् अभिक्रुध्यति — the same root with अभि-, and 1.4.38 reassigns the same participant to कर्म, द्वितीया. Nothing about who is angry at whom has changed; only whether the verb is compounded decides which role the case ending will claim.

अधिकरणम्Three kinds of "ground", one kāraka

Every other kāraka in this chapter names one relation. अधिकरणम् — 1.4.45 , "the support the action rests on" — turns out to cover three relations the commentators found worth separating, because a single सप्तमी form cannot show a reader which one is meant.

kind	example	relation
औपश्लेषिकः	कटे आस्ते	direct contact — he sits on the mat
वैषयिकः	मोक्षे इच्छा	a topic, not a place — desire about liberation
अभिव्यापकः	तिलेषु तैलम्	pervading every part — oil throughout the seed

Only the first is "location" in the everyday sense. By the time a reader reaches वैषयिक, सप्तमी has stopped meaning "in" or "on" altogether and started meaning "with respect to" — the same ending, doing unrelated jobs, sorted only by what kind of grip the action has on its ground.

द्विकर्मकाःTwo objects — and which one is the "real" कर्म?

1.4.51 — "and the unspoken-for one, too" — is the sūtra that makes verbs like दुह् (milk) take two accusatives. गो, the cow, would ordinarily be अपादान (what the milk comes from); पयस्, the milk, is कर्म by the ordinary rule. 1.4.51 says: whichever of them the more specific rules (अपादानादि) have not already claimed is कर्म as well. Both become द्वितीया.

The tradition itself does not fully settle which is "primary". Later grammarians call पयस् the प्रधानकर्म — the goal the action is actually about — and गो the गौणकर्म, "unspoken-for" only because 1.4.51 had to reach for it. But the sūtra's own wording treats them identically: both अकथित, both कर्म, both द्वितीया. Whether one of the two is more the "real" object is a question of emphasis the grammar does not adjudicate — which is precisely why दुह्, याच्, रुध्, प्रच्छ्, भिक्ष्, चि, ब्रू, शास् had to be listed as a closed set rather than derived from a general principle.

कर्मवत्कर्ताThe agent, treated exactly like the object it also is

An intransitive root has no कर्म to begin with, so what happens when its agent is also, in every practical sense, what the action is being done to? A piece of firewood that splits "by itself" is स्वतन्त्रः — grammatically the कर्ता, by 1.4.54 — and also the one thing the splitting happens to. 3.1.87 resolves this by fiat: treat that कर्ता exactly as if it were a कर्म for every rule that would otherwise apply to one.

This is why भिद्यते looks passive with no agent expressed. भिद्यते काष्ठं स्वयमेव takes यक् and आत्मनेपद — the कर्मणि shape page 13 derived for हन्यते — even though there is no separate agent standing outside the wood to have been demoted. 3.1.87 is what lets the same morphology serve an event with only one true participant, who is grammatically required to play both parts.

णिजन्ते कर्ताThe causative pushes the old agent down — or lets it choose

Add णिच् — the causative — and the sentence acquires a new agent: the one who makes the old agent act. What happens to the old agent depends on what kind of verb it was. 1.4.52 demotes it to कर्म outright for a named list of root-types (motion, cognition, eating, and a few others). But 1.4.53 — just two roots, हृ and कृ — gives the old agent a choice: कर्म, or तृतीया, अन्यतरस्याम्, "in either way".

णिच् itself is page 14's subject, not this page's. What belongs here is only the कारक consequence: 1.4.52's list is not optional (गमयति माणवकम् — always द्वितीया), while 1.4.53's two roots genuinely let a speaker choose how demoted the old agent sounds — हारयति भारं माणवकम् (as fully demoted as any other कर्म) or माणवकेन (as if it were still, in some diminished sense, doing the work). See page 14 for how णिच् itself is built.

Three names for the same twenty-two words

1.4.59 · 1.4.60 · 1.4.83

Page 15 already covers उपसर्ग vs कर्मप्रवचनीय in depth and names गति in passing without pursuing it — "a second name for the same twenty-two, not covered further here". What that page leaves open, and what belongs on a कारक page rather than a word-formation page, is this: of the three names one of them, alone, carries a 2.3 case rule. The other two are morphology, not कारक — and it is worth being able to tell which is which before assuming any prefix-shaped word is "just" a preposition.

उपसर्गः · गतिःWord-formation labels — no case rule of their own

1.4.59 names उपसर्गसंज्ञा; 1.4.60 names गतिसंज्ञा for an overlapping set, under the same condition, क्रियायोगे — joined to a verbal action. A word can carry both names at once, or neither, depending only on whether it is doing that job in this sentence.

Neither संज्ञा appears anywhere in chapter 2.3. What they unlock instead is elsewhere in the grammar entirely — sound changes at the join (नुम्, इत्त्व, णत्व/षत्व after specific उपसर्गs), and compound formation (गति-समास) for गति. Whatever role the verb's own participant plays is settled exactly as if the prefix were not there.

कर्मप्रवचनीयःThe one of the three with its own कारक-shaped rule

1.4.83 opens a wholly separate अधिकार. The same words — प्र, अनु, अभि and the rest — get this third name precisely when they are not joined to a verb at all, but govern a noun directly, the way a preposition governs its object in English. And unlike उपसर्ग and गति, कर्मप्रवचनीय comes with its own case-assignment sūtra, sitting right there in chapter 2.3 alongside the six kārakas' own:

कर्मप्रवचनीय is not a seventh kāraka. 1.4.83's अधिकार begins well outside 1.4.23 कारके's span — the corpus's own inheritance graph shows it with no आधिकारिक ancestor back to कारके at all. It gets a 2.3 rule because Pāṇini chose to give it one, the same way he gave षष्ठी one at 2.3.50 for relations that are not kārakas either. A word being "prefix-shaped" tells you nothing about which of these three names, if any, applies — only क्रियायोगे or its absence does.

अनभिहिते once more — expressed by a कृत्, not a तिङ्

2.3.65 · 2.3.71

Every अनभिहिते example so far — page 08's passive, page 13's तिङन्त triple — turned on whether a finite verb ending had already spoken for a role. A कृत्-affix word can speak for a role too, and when it does, अनभिहिते applies to it in exactly the same way — except that here the affix does not just remove a case, it can offer two.

कृत्याःThe gerundive's agent: तृतीया, or षष्ठी by special dispensation

कृत्य affixes — 3.1.95–96 , "should be done" — carry कर्मणि/भावे force exactly as क्त does (page 13, 3.4.70): a कृत्य word expresses its कर्म in agreement, the way हतः expresses its. That leaves the agent अनभिहित, and the general rule for a कृत् affix's participants, 2.3.65, would give it षष्ठी — the usual case a कृदन्त takes its "subject" in (रामस्य गमनम्, "Rāma's going"). But 2.3.66 reserves that षष्ठी for the कर्म instead when both would otherwise want it — which should leave the agent with the ordinary unexpressed-participant case, तृतीया, by 2.3.18.

2.3.71 — "of कृत्यs, optionally in the कर्ता" — reopens षष्ठी as a second option for exactly this agent, specifically for कृत्य affixes and no other कृत्. The result is one participant, one role, and a live choice between two case endings for it:

Two sūtras, two different kinds of "choice". हृ/कृ's अन्यतरस्याम् (1.4.53, above) is a choice about which role the old agent plays after a causative — कर्म or करण-like तृतीया. 2.3.71's वा is narrower: the role is fixed (still कर्ता), and only the case that role gets is optional. This page's title promises rôles, plural; this last entry is the reminder that not every contested choice in 1.4–3.1 is a rôle contest at all — some are case contests layered on an uncontested rôle, and telling the two apart matters for what a sentence is actually claiming.

One event, one word, two licensed readings

Interactive · composed for this page

रामः leaves ग्रामः. The vakya lexicon's own two frame-entries for त्यज् — TRANS and ABL — license two different sentences for the identical event, and each is a real reading a speaker could intend, not a mistake either dial position corrects. Switch the reading; watch the case change on ग्राम alone.

ग्रामः is…

रोल

विभक्तिः

licensed by

Every contested pair, at a glance

Reference

One row per pair on this page, plus the two-vs-none case-rule split among गति / कर्मप्रवचनीय / उपसर्ग, and the hard-side अनभिहिते entry. Every sūtra links into the reading view.

contest	what is at stake	sūtras

Where this page's data comes from

honesty

Check yourself

अभ्यासः

Six questions. Every one is answered by a sūtra, not by feel.