Split ergativity and perfective aspect

Hindi has split ergativity: its alignment changes according to aspect and transitivity. In ordinary nonperfective clauses, an unmarked subject controls agreement. In a canonical perfective transitive clause, the agent takes ने, becomes unavailable for agreement, and an eligible unmarked object controls instead. Perfective intransitives normally keep their subject unmarked.

The system is “split” because Hindi does not treat every subject alike in every tense-aspect form. English has no comparable case alternation, so translating “the doer” directly into an unmarked Hindi subject gives the wrong result precisely where perfective aspect activates the ergative pattern.

Three alignments in one table

Let S mean the sole argument of an intransitive verb, A the agent-like subject of a transitive verb, and O its object.

Compare one intransitive and one transitive habitual:

Unmarked feminine लीला controls लौटती as S and पढ़ती as A. Nonperfective alignment groups those two subjects together.

Perfectivity changes the transitive clause

Now make both events perfective:

The intransitive S remains unmarked and controls feminine लौटी. The transitive A becomes लीला ने. Feminine पढ़ी now agrees with unmarked feminine O ख़बर, not with Leela. This is the alignment split: perfective transitive A behaves differently from both intransitive S and nonperfective A.

The split follows aspect, not calendar time

Past habituals and progressives describe past situations but remain nonperfective, so the subject stays unmarked:

A present perfect includes a perfective participle and therefore takes ने despite present auxiliary है:

The contrast proves that ने is not a past-tense ending. Habitual बनाता था is past without ने; present-perfect बनाया है has ने without being a simple past.

Object marking determines the second half of the pattern

After ने blocks A, the predicate searches for an eligible unmarked O. If O bears को, it cannot control and masculine singular default appears:

Unmarked छात्रा controls feminine देखी in the first clause. Specific छात्रा को is ineligible in the second, so देखा is default; it does not agree with a hidden masculine noun. This agreement contrast diagnoses the structural effect of differential object marking.

Case must be chosen for independent grammatical reasons. Do not omit required को merely to obtain visible object agreement. The differential object marking page explains when को is selected.

Negation and focus do not switch the alignment

Perfective morphology continues to license ने under negation:

Feminine unmarked रिपोर्ट still controls पढ़ी. Negation takes scope over the completed-event claim; it does not convert the clause into a nonperfective alignment.

Focus also preserves case licensing:

Particle ही associates with रिपोर्ट, while the object remains unmarked and controls agreement. If a focused object independently requires को, focus cannot make it eligible again.

Compound predicates inherit a case frame

A compound verb's final light verb contributes event structure and often bears visible perfective agreement. Canonical transitive combinations such as पढ़ लेना retain the ergative pattern:

Masculine plural अध्याय controls लिए. The light verb लेना is not freely interchangeable with every other vector; changing it can alter aspectual nuance and sometimes argument structure. Learn attested combinations as predicates, then calculate case and agreement from that construction.

Perfectivity alone does not force ने on every apparent “doer.” Passives, dative-experiencer predicates, and lexicalized complex predicates have their own case frames. Split ergativity is a productive system, not a license to attach ने to any completed action.

Each finite clause chooses independently

An embedded clause does not inherit the matrix clause's alignment:

Matrix transitive perfective मीना ने बताया has an ergative A and default बताया before its finite content clause. Embedded intransitive perfective आरव लौट गया है keeps unmarked S आरव, which controls masculine गया.

The embedded predicate is now transitive perfective. Embedded A bears ने, and masculine plural टिकट controls ख़रीदे हैं. Complementizer कि marks a new domain in which aspect, transitivity, case, and agreement are calculated again.

The productive rule and its lexical edge

The clean rule covers ordinary transitive verbs such as पढ़ना, लिखना, खोलना, बनाना, and countless compound predicates. Some predicates sit at the transitivity boundary or permit an agentive construal in particular varieties. खेलना and बोलना, for example, can occur with different case frames depending on whether an object is expressed, whether the event is construed as a controlled activity, and regional or colloquial usage.

Those are not counterexamples to aspectual split ergativity. They are cases where speakers differ over the predicate's argument structure. The next page treats optional and variable ने directly.

Common Mistakes

1. Treating ने as a general past marker

2. Omitting ने from a canonical perfective transitive

3. Keeping subject agreement after ने

4. Letting matrix case spread into an embedded clause

Key Takeaways

• Nonperfective clauses normally show unmarked-subject agreement, whether the verb is transitive or intransitive.
• Perfective intransitives normally retain an unmarked agreeing subject.
• Canonical perfective transitives mark A with ने and shift agreement to an eligible unmarked O.
• A को-marked object cannot control, producing default masculine singular when no other controller is available.
• Negation, focus, tense auxiliaries, and matrix predicates do not replace the local aspect-transitivity test.