वाक्यरचना — agreement, order, and the usual errors

What page 08 and 13 were for

recap

कारकम् (08) named the six roles a word can play and the sūtra that gives each its case when the verb has not already claimed it. प्रयोगः (13) named the one choice — कर्तरि, कर्मणि, भावे — that decides which role the verb speaks for. Both are machinery for building one sentence at a time. This page is where you start turning the crank yourself.

A live drill exists for exactly this. Everything on this page is prose and a fixed set of worked examples — enough to see the rules once, clearly. अभ्यासः — write Sanskrit, sentence by sentence is the companion app: it generates a fresh sentence from this same course lexicon, has you type the answer, and grades every word against the sūtra that licenses it. This page teaches; that page drills. Use both.

Three agreements, and all three are total

1.4 · 2.3 · 1.1.26 ff.

"Agreement" in Sanskrit is not a percentage. A word either matches every feature the rule asks for, or the sentence is wrong — not awkward, not informal, wrong. There are exactly three places this bites.

विशेषणम्An adjective or participle agrees in liṅga, vacana AND vibhaktiall three, always

गत — गम् + क्त, "gone" — is a participle: it declines exactly like a noun, because a कृदन्त is a प्रातिपदिक once the affix is on (see 16–17). Whatever noun it describes, it must match in all three features at once, not the ones that happen to be convenient:

liṅga	प्रथमा एक.	प्रथमा द्वि.	प्रथमा बहु.	द्वितीया एक.

Paired with real nouns from the course lexicon, in the cases those nouns actually need:

The boy went. (m., प्रथमा एक.)

The two boys went. (m., प्रथमा द्वि. — see below, the dual is not optional.)

The girl went. (f. — a different vibhakti-set entirely: -आ,-ए,-आः, not -अः,-औ,-आः.)

The teacher sees the boy who went. (accusative now, because गुरुः दृश्यति needs a direct object — and गत follows बालम् into द्वितीया, not just gender and number.)

Case agreement is the one learners forget. गत tracks whatever case its own noun is in — nominative when the noun is the subject, accusative when the noun is an object, exactly as शुद्ध adjectives do in every other language. "गतः" describing an accusative noun is not a minor slip; it is a different, ungrammatical sentence.

क्रियापदम्A verb agrees with its kartā in puruṣa and vacana — never liṅga1.4.21 · 1.4.22

Rāma goes.

Sītā goes — the identical verb form.

The verb does not see gender at all. Compare that against गत above: the participle changed shape for सीता (गता, not गतः); the finite verb गच्छति did not, because a तिङन्त ending only ever encodes पुरुष and वचन. Confusing the two — expecting गच्छति to have a "feminine form" — sends a learner looking for something that does not exist in this grammar.

वचनम्	बालः / बालौ / बालाः	गच्छति-group

सर्वनामA pronoun agrees with its antecedent in liṅga and vacanapage 07 · तद्

तद् — page 07's own suppletive pronoun — tracks the noun it stands for exactly the way an adjective does, in gender and number (not, notably, in case: the pronoun takes whatever case its own role in the new clause needs, same as any noun):

Rāma laughs. He goes.

Sītā laughs. She goes. — , not .

A third form exists for a neuter antecedent — — the same suppletive stem, a third gender's endings. सः / सा / तत् is one paradigm, not three different words.

Free is not the same as random

क्रमः

Page 08 already showed that the four orders of a sentence say the same thing, because the endings — not the positions — carry the grammar. What that page did not say is that a Sanskrit author still chooses an order, and the choice is not arbitrary: it is how emphasis is marked without changing a single ending.

विध्यतिOne event, four foregroundingsव्यध्, इण्-frame

The unmarked order puts the कर्ता first and the verb last, with whatever falls between them in whichever sequence the sentence's other business wants — कर्ता–करण–कर्म–क्रिया here:

Unmarked. Rāma pierces the deer with an arrow — a plain report, nothing foregrounded.

Object first. Foregrounds the deer — "it is the deer that Rāma hits with the arrow," as if answering "what did he hit?"

Instrument first. Foregrounds the means — "it is with an arrow that Rāma strikes the deer," as if the manner is what is in question.

Verb first. Foregrounds the act itself, ahead of who or what — common at the start of a clause or a verse line.

This is convention, not a sūtra. Pāṇini's grammar fixes no word order at all — every one of these four is equally grammatical, and the emphasis reading is the tradition's own reading of prose rhetoric and metre, not a rule this site can cite a sūtra for. What is a grammatical fact is the one page 08 already stated: the case endings are unchanged in all four, so the identity of who struck whom never becomes ambiguous, however the words are shuffled.

प्रथमपदम्The first word is not "the subject"the single commonest beginner error

Take order 2 above out of context and a reader trained on English word order reaches for the wrong parse on reflex:

"The deer pierces..." — wrong: मृगम् is accusative, and an accusative noun cannot be a कर्ता no matter where it stands. रामः is nominative wherever it appears in the sentence, and that is what makes it the agent. Read the ending on the first word before deciding what job it does.

The errors people actually make

छह् सामान्यभ्रान्तयः

Every one of these has a rule behind it — none is a matter of taste. The wrong form is struck through beside the right one.

कृत्यकर्ताA gerundive's agent is instrumental — always

तव्यत् — "should be done" — is a कृत्य affix, and 3.4.70 puts it in the very same basket as क्त: कर्मणि/भावे meaning. A learner who has just met क्त्वा and सामान्य verbs reaches for the nominative out of habit — but a कृत्य no more takes a nominative agent than a passive क्त participle does:

Category error: अहम् is प्रथमा, and प्रथमा marks nothing here — the sentence has no verb that expresses कर्तृत्व for it to be bare-named by.

"[It] should be done by me" — मया is अनभिहित-कर्ता, and gives it तृतीया, the identical rule that gave the passive agent its case on page 08 and 13. गम् → गन्तव्यम् and नम् → नन्तव्यम् work exactly the same way (page 18's own table).

भिन्नकर्तृकं क्त्वाक्त्वा with two different agents needs a locative absolute

Page 18 covers this rule in full; here is the trap in one line. 3.4.21 requires one agent for both the क्त्वा clause and the main verb:

"Having eaten, [he] goes" — one person does both. Fine.

* भुक्त्वा ब्राह्मणे व्रजति देवदत्तः

Attempting "the Brāhmaṇa having eaten, Devadatta goes" with क्त्वा — but क्त्वा always attaches to whoever does the main verb, so this can only mean Devadatta ate, not the Brāhmaṇa. Two agents, one क्त्वा: broken.

The fix is a locative absolute — a सप्तमी participle agreeing with its own subject, unconnected to the main verb's कर्ता — taught in full on page 18.

अकर्मकक्तक्त on an intransitive root is active, not passive

Page 13 derived this from the engine directly; the trap is assuming क्त always means "passive". 3.4.72 pulls गत्यर्थ (motion) and every अकर्मक root out of निष्ठा's कर्मणि default:

not "[a place] gone to" — गम् has no object here to be "gone to"; गतः is कर्तरि, "the one who went". Contrast कृतः — कृ is transitive, so its क्त stays कर्मणि by the ordinary rule, "the thing done", not "the one who did". Same शब्द-आकृति, opposite role — full derivation on page 13.

द्विवचनम्The dual is obligatory, not optional

1.4.21 and 1.4.22 are not two styles a speaker picks between — exactly two things get द्विवचन, and "several" is not exactly two:

Two boys, marked बहुवचन — wrong, the way "the two boys goes" is wrong in English but for the opposite reason: English lacks the category, Sanskrit has it and enforces it.

Exactly two agents get exactly the दुवचन verb ending, every time — not "close enough to singular", not "close enough to plural".

लिङ्गम् आङ्ग्लभाषातःGender agrees with the Sanskrit noun, not the English glosspage 06 · लिङ्गम्

English "fruit" carries no grammatical gender at all, so a learner translating word-by-word has nothing to anchor a participle's gender on and defaults to masculine out of habit. But फल is a Sanskrit word with a fixed gender — neuter — and every word agreeing with it must follow that, not English:

फलः does not exist — फल has no masculine paradigm in this corpus at all; the gender is not a free choice.

पतित follows फल into नपुंसकलिङ्ग. Compare the same participle on the course's other genders, same root, same rule:

Page 06 said this already, in general. A noun's gender is a fact about the Sanskrit word, established once when the stem is learned — never re-derived from what the referent's English name happens to imply.

प्रथमपदम्The first word is not the subjectsee §3 above

Covered above under word order — repeated here only so the six traps are in one place. Read the ending, not the position.

इति

subordination without a conjunction

Page 20 gave इति's four jobs in full — quoted speech, quoted thought, a title, a stated reason — and showed the mechanism: इति closes off whatever precedes it and hands the whole closed stretch to the surrounding verb as a single unit. That is how Sanskrit subordinates a clause with no word meaning "that". One fresh example, from this page's own lexicon:

वाक्ये वाक्यम्A whole sentence as another sentence's object1.4.57 चादयोऽसत्त्वे

"Rāma said, 'I go to the village.'" — गच्छामि is a complete finite clause, first person, with its own implicit agent; इति closes it, and अवदत् — third person, agreeing with रामः exactly as any ordinary verb agrees with its कर्ता — governs the whole quoted clause as its content. No word here corresponds to English "that"; इति's position, immediately after the closed material, does the entire job.

The outer sentence's agreement does not change. रामः is still nominative, अवदत् still agrees with it in पुरुष and वचन, exactly as in every plain sentence on this page. Embedding a clause with इति does not touch the outer sentence's own grammar at all — which is the point: page 20 has the other three jobs and the full argument.

Building one, in order

meaning first, endings last

The order that works is the reverse of how this page has been reading finished sentences. Composing one runs forward:

१ · घटनाpick the event — who does what to what २ · प्रयोगःकर्तरि, कर्मणि, or भावे — page 13 ३ · कारकाणिassign each participant its role — page 08's 1.4 ४ · विभक्तयःonly now do endings appear — 2.3, mechanically, from step 3

Endings are the last step, never the first. A learner who starts by asking "what case is this?" before deciding what role the word plays is solving the problem backwards — 2.3 cannot run until 1.4 has already decided, for every word, whether the verb has claimed its role or left it for a case ending to mark.

Build one yourself

Interactive

Pick a frame. Every ending you see is explained by its sūtra below the sentence; the two side controls flip word order and प्रयोग without touching the roles or the event — watch which endings move and which do not.

frame

क्रमः

प्रयोगः

This is the fixed, worked half of the exercise — seven frames, one verb each, always the same participants. For a sentence that is different every time, with your own answer graded word by word, go to अभ्यासः.

The seven frames

Reference · from

Every frame this course uses, its participant slots, and the template behind its English gloss. A verb's own lexicon entry restricts which frames it can take and which semantic types can fill each slot — see the builder above for one verb per frame in practice.

frame	slots	gloss template

Check yourself

अभ्यासः

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