Phase theory proposes that syntax builds in cycles—phases are syntactic domains (vP, CP) at which derivations are transferred to the phonological and semantic interfaces. Phases explain why certain movements are impossible (phases are opaque to operations outside them) and provide a principled account of locality constraints in generative grammar.
Examine phase edges (the highest element in a phase) and how they remain accessible for further movement while phase internals become inaccessible. Compare predictions of phase theory against traditional locality constraints like the subjacency condition.
You've worked through the Merge operation — the basic structure-building mechanism in the Minimalist Program — and the core minimalist commitments: that syntax is an optimal computation mapping sound to meaning, using only what is necessary. Phase theory is the Minimalist Program's answer to one of syntax's oldest puzzles: why can't elements move freely to any position in a sentence? Why are some movement dependencies impossible across certain structural boundaries, no matter how far away the relevant positions are?
The classic puzzle is island sensitivity. You can form a question from a simple embedded clause: "She believes [that John ate what]" → "What does she believe that John ate?" But you cannot form a question from a complex noun phrase: "She believes [the claim that John ate what]" → *"What does she believe the claim that John ate?"* (ungrammatical). Something about the noun phrase "the claim that John ate what" blocks movement out of it. Traditional accounts posited stipulated constraints like subjacency to describe these patterns, but they gave little insight into *why* these particular boundaries should be opaque. Phase theory offers a principled explanation.
The key claim is that syntactic derivations build in phases. After each phase head — the canonical phase heads are v* (the light verb that introduces agents in transitive clauses) and C (the complementizer that heads clauses) — completes its derivation, the interior of that phase is transferred to the phonological and semantic interfaces and becomes opaque to further syntactic operations. Only the phase edge (the specifier and head of the phase) remains accessible to higher operations. This is the Phase Impenetrability Condition (PIC). If an element needs to move from inside a phase to a higher position, it must pass through the phase edge — it cannot jump directly from the interior of a completed phase to an external position. Movement out of a phase must proceed in steps, landing at the edge of each phase along the way.
The elegance of Phase Theory is that it unifies a range of apparently different locality effects under a single principle. Island violations occur when an element fails to reach a phase edge before the phase interior is transferred — at that point, the element is inaccessible and movement is impossible. The clause boundary (CP phase) creates one type of island; the vP domain creates another. Phase theory also connects locality to efficiency: transferring the phase interior to PF and LF as derivation proceeds means the computational system does not need to hold an entire sentence in memory simultaneously. Syntax can proceed incrementally — building and dispatching one phase before beginning the next — which has implications both for the theoretical elegance of the minimalist program and for how syntactic processing is understood in real-time language comprehension.