A syntactic framework that represents sentence structure exclusively as directed binary head–dependent relations between lexical items (words), producing a rooted dependency tree (or forest) in which syntactic functions, word order, and constituency are expressed via links between heads and their dependents rather than by phrase‑structure nodes.

Principle

Syntactic constituency and functional structure are reducible to asymmetrical binary dependency relations: each dependent is syntactically licensed by a single head, and global sentence structure emerges from these local head–dependent links.

Demonstration

Illustrative scenario → Situation: Parse the sentence "The quick fox chased the hare." Recognition: Identify lexical heads (e.g., chased as the main predicate) and link dependents (The, quick → fox; fox → chased; the, hare → chased). Action: Construct a directed tree whose edges encode grammatical relations (subject, object, modifiers). Consequence: The dependency representation captures predicate–argument relations and linear order with a compact, word‑centred graph.

Misapplication

Assuming dependency structures and constituency trees are interchangeable without transformation; the error is to treat phrase nodes as trivial projections of dependency links. That conflates different representational commitments: dependency grammars foreground word‑to‑word relations, while constituency grammars encode hierarchical phrasal nodes.

Consequence

Adopting a dependency view emphasizes head‑dependent relations in syntactic description, often yielding simpler annotation schemes for certain parsing algorithms and clearer alignment with predicate–argument semantics, particularly for languages with relatively free word order.

Reversal

For phenomena that naturally involve nested phrasal coordination, discontinuous constituents, or where certain constituent‑level generalizations are explanatorily central, phrase‑structure representations may be more direct or convenient than strictly word‑to‑word dependency graphs.

Boundary

Clearly within: analyses that model syntactic relations as directed links between words (heads and dependents). Boundary case: multiword expressions or idioms whose internal cohesion can be analyzed either as atomic lexical items or as tightly linked dependency subtrees. Clearly outside: analyses that require explicit phrasal nodes as primitives and do not reduce constituency to head–dependent relations.

Semantic Tension

Tension with constituency (phrase‑structure) grammars over whether hierarchical phrasal nodes or binary head–dependent links are the primitive explanatory objects of syntax; each imposes different analytical priorities.

Synthesis

Dependency Grammar shifts the analytic focus from constituent building blocks to pairwise head–dependent relations, offering a compact, lexical‑centred representation that aligns structural description with predicate–argument relations and surface linear order.