Constituent tests for Hindi phrases

A constituent is a sequence of words that grammar treats as a unit. Hindi word order is flexible enough that visual adjacency alone cannot reveal every unit, yet it is constrained enough that words cannot be rearranged independently. Reliable analysis combines several diagnostics: substitution, question-answer pairing, coordination, focus scope, movement, ellipsis, case morphology, and agreement.

No single test is infallible. Pronoun omission, discourse-driven scrambling, prosodic phrasing, and conventional ellipsis can make one diagnostic unclear. A strong analysis looks for convergence.

Establish a neutral baseline

Consider a clause with several nested phrases:

A useful first parse is:

The feminine plural ending on खरीदीं supports नई किताबें as its unmarked controller. Oblique plural बच्चों anticipates के लिए and helps identify the beneficiary boundary. Agreement and case reveal structure that a simple “first phrase, second phrase” description would miss.

Substitution tests whole units

If a sequence can be replaced by a single expression while preserving the rest of the structure, that is evidence for constituency. Replace the beneficiary:

स्कूल के लिए substitutes for उन तीन बच्चों के लिए as a complete postpositional phrase. Replace the object with a pronoun-like demonstrative:

Plural ये replaces the object noun phrase and preserves plural खरीदीं. By contrast, replacing only बच्चों while leaving the compound postposition's के relation malformed would not preserve the syntax.

Substitution is category-sensitive. A location phrase can replace another location phrase, but not necessarily an object NP. Semantic oddness can also spoil an otherwise valid structural test, so choose a replacement that fits the predicate.

Questions and fragments identify answer constituents

A natural short answer normally corresponds to a phrase in the question:

The answer preserves the entire compound postposition, confirming उन तीन बच्चों के लिए as a unit. Object questions identify a different constituent:

Interrogative क्या triggers default masculine singular खरीदा because it does not overtly supply the answer's feminine plural features. The fragment नई किताबें restores those features semantically but has no finite predicate of its own. Question-answer congruence tests information units; it does not require every fragment to display full clause agreement.

Focus particles attach at constituent edges

Particles such as ही and भी reveal scope by following the constituent they associate with:

ही follows the complete beneficiary phrase. Moving it inside the noun phrase changes the focus:

In उन्हीं, focus selects the demonstrative set inside the larger phrase. Both are grammatical, but their scope differs. Particle placement is therefore evidence about nested constituents, not proof that every string before ही is one flat unit.

Negation gives a parallel scope test. बच्चों के लिए किताबें नहीं खरीदीं ordinarily negates buying; contrastive बच्चों के लिए नहीं, शिक्षकों के लिए negates the beneficiary choice. Prosody and parallel alternatives determine which constituent lies under contrast.

Movement must preserve phrase integrity

Hindi scrambling can move a complete postpositional phrase toward the left edge:

The fronted beneficiary remains intact. Splitting off its postposition is impossible:

Movement evidence must be interpreted cautiously. Hindi permits discontinuity in restricted poetic, contrastive, and heavy-phrase constructions, and spoken afterthoughts can appear outside the core clause. These marked patterns have specific prosody; they do not license arbitrary word-by-word permutation. The limits on scrambling page treats those discourse conditions.

Coordination compares like constituents

Natural coordination is evidence that the conjuncts share a category and function:

The two full postpositional phrases coordinate. Hindi can also share a postposition across coordinated nouns—बच्चों और पुस्तकालय के लिए—but that structure coordinates the nominal complements inside one larger postpositional phrase. It does not prove बच्चों alone has the same distribution as बच्चों के लिए.

The coordinated object is mixed-gender plural, conventionally resolved as masculine plural खरीदे. Agreement supplies additional evidence that the coordination as a whole is the controller.

Case morphology marks internal boundaries

Simple and compound postpositions select oblique forms. In उन तीन बच्चों के लिए, demonstrative उन and noun बच्चों are oblique; के लिए closes the phrase. A direct-form sequence cannot simply be followed by an unrelated postposition.

Genitives show a different dependency:

Genitive की and adjective छोटी agree with feminine head बहन, while finite आई confirms that the whole NP रीना की छोटी बहन is the subject. रीना is inside that subject constituent but does not control the predicate.

This nesting matters for semantic intuition. The person most prominent at the beginning of an NP need not be the head, and only the head's features may propagate outward.

Predicate complexes resist internal scrambling

Hindi predicates can contain a lexical stem, light verb, aspectual material, modal, and auxiliary. Their order is constrained:

Feminine plural किताबें controls ली थीं. The sequence खरीद ली थीं forms a structured predicate domain; an object cannot be inserted freely between खरीद and ली, nor can past auxiliary थीं be moved into English auxiliary position.

Constituency here is layered rather than flat: खरीद ली combines before थीं, and negation can take scope at specific layers. Some compound predicates permit limited intervening particles or adverbs, so “adjacent words always form one indivisible constituent” would also be too strong.

Clause substitution and boundaries

A finite कि-clause can occupy a complement position that a nominal expression also fills:

The finite clause and nominal पूरी बात both supply content, but they are not morphologically identical. Feminine NP बात controls बताई; the finite कि-clause does not provide gender-number features, so matrix बताया is default. Substitutability at one semantic slot does not erase category differences.

Ordinary scrambling and agreement remain clause-bound:

Embedded सभी मेहमान controls पहुँच गए only within the कि-clause. Long-distance-looking topics can occur in advanced discourse, but they require independent analysis; the nearest semantic noun is not licensed to control a matrix predicate across a finite boundary.

Why no test works alone

Hindi allows null arguments, shared postpositions, coordination ellipsis, and discourse-driven displacement. A sequence may fail a movement test because it is too heavy or pragmatically odd, not because it is structurally impossible. Conversely, poetic meter can move a noncanonical string without making that order neutral prose.

Use at least two diagnostics:

1. Form: case, agreement, and selection.
2. Distribution: substitution, answerhood, coordination, movement, ellipsis, and focus scope.
3. Interpretation: whether prosody and context preserve the intended reading.

Converging evidence is stronger than an English translation or one acceptable rearrangement.

Common Mistakes

1. Treating a postposition as detachable

2. Letting an embedded noun control the matrix verb

3. Treating a possessor as the noun-phrase head

4. Treating scrambling as free permutation

Key Takeaways

• Constituents are diagnosed by converging evidence, not adjacency or English translation.
• Postpositions, focus particles, case forms, and agreement expose nested phrase boundaries.
• Movement and coordination are useful only when category, case, and discourse remain appropriate.
• Predicate complexes have internal layers and cannot be scrambled like independent noun phrases.
• Finite कि-clauses occupy complement positions but form separate agreement domains.