A Smuggling Approach to the Passive in English #
@cite{collins-2005} @cite{chomsky-2001} @cite{legate-2003} @cite{rizzi-1990}
Connects the empirical passive data in Phenomena.ArgumentStructure.Passive
to the smuggling theory in Theories.Syntax.Minimalist.Movement.Smuggling
and the Voice/phase infrastructure in Theories.Syntax.Minimalist.Voice.
Collins's central claims #
@cite{collins-2005} argues for a smuggling derivation of the passive that combines aspects of the principles-and-parameters approach (no specific rules, no downward movement) with @cite{chomsky-1957}'s Syntactic Structures analysis (the arguments are generated in the same positions in active and passive, satisfying UTAH).
- §2 (UTAH): The external argument is merged into Spec,vP in the passive just as in the active — no special downward movement or theta-role absorption. Same θ-positions in both voices.
- §3 (PartP movement, not head movement): The verb does not raise through Part to v to Voice; rather, PartP (= [Part V DP]) undergoes XP-movement to Spec,VoiceP. Evidence: particle stranding (15–16) and pseudo-passives (18–19) are incompatible with head-movement.
- §4 (by heads VoiceP): The preposition by is the head of a functional projection VoiceP, not the head of an ordinary PP. By is composed entirely of uninterpretable features and checks accusative Case on the external argument in Spec,vP. The Case feature of v dissociates to Voice (eq. 32).
- §5 (Smuggling): PartP smuggles the internal argument past the external argument in Spec,vP, freeing it for A-movement to Spec,IP without violating Relativized Minimality.
- §6 (Short passives): The external argument is structurally present even when not phonologically realized — an empty pronominal in Spec,vP.
- §8 (by-DP is a constituent): Coordination data (63–64) shows by John behaves as a unit; this is derived as VoiceP coordination with deletion, not as a stipulated [PP by DP].
§1. Voice properties Collins defends #
The four core properties of voicePassive: it does not assign an external
θ-role (which v does, per §2 UTAH), it does check accusative Case (per §4
Case dissociation), it is not a phase head (which is what permits smuggling
per §5), and it has a D-feature (it is a Voice head, not an expletive).
@cite{collins-2005} §2: passive Voice does not assign external θ. The external θ-role stays on v in both active and passive.
@cite{collins-2005} §4: passive Voice checks accusative Case. The Case feature dissociates from v and projects on the Voice head (whose lexical realization is by).
@cite{collins-2005} §5: passive Voice is not a phase head, which is what permits PartP to smuggle past the external argument in Spec,vP.
@cite{collins-2005} §5: passive Voice permits smuggling.
§2. Active vs. passive: feature dissociation (Collins §4 eq. 32) #
Collins's central theoretical move: the active and passive v are identical in θ-assignment (both assign external θ in Spec,vP). They differ in where the accusative-Case feature is checked. In the active, v checks Case directly. In the passive, the Case feature dissociates from v and is added to the numeration as part of the Voice head. The two heads are then in complementary distribution on (θ-assigning, Case-checking).
@cite{collins-2005} §4 eq. 31: active and passive have the same θ-assignment but distribute the Case-checking feature differently.
The phase status correlates with where Case checking lives. Active v* is a strong phase (it checks Case); passive v is not (Case has dissociated to Voice). The non-phase status is what permits smuggling.
§3. PartP movement is XP-movement, not head movement #
Collins argues against the head-movement analysis of the passive on the
basis of (a) particle stranding and (b) pseudo-passives. Both diagnostics
fall out from the assumption that the constituent moved is PartP, not
just the verb. The data live in Passive.particleData and are derived
here from the smuggling configuration.
@cite{collins-2005} §3 (15–16): in passive with a particle verb,
the only grammatical order is V Prt EA (summed up by the coach),
not V EA Prt (*summed by the coach up). This follows from the
smuggling configuration: PartP = [Part V DP] moves as a constituent
to Spec,VoiceP, taking the particle along with the verb.
@cite{collins-2005} §3 (18–19): pseudo-passives (John was spoken to by Mary) require the preposition to be inside the smuggled PartP
(P stays adjacent to V, not stranded after EA). Same XP-movement
diagnostic as particles.
§4. Without PartP there is nothing to smuggle #
The smuggling derivation requires PartP to be present (something to move). Without it, no derivation is licensed.
§5. Short passives (§6 of paper) #
@cite{collins-2005} §6 (45–47): even when the external argument is not
phonologically realized (The book was written), it is structurally
present as an empty pronominal in Spec,vP. Evidence: implicit arguments
bind reflexives (42a Such privileges should be kept to oneself), license
depictives (44b Breakfast is eaten by the campers nude), and pattern
exactly like overt external arguments (44).
In the formal terms of this file, a short passive has the same Voice configuration as a long passive — the only difference is whether Spec,vP is filled by an overt DP or an empty pronominal.
§6. By-DP as a VoiceP constituent (§8 of paper) #
@cite{collins-2005} §8 (63–64) defends by-DP as a constituent on the
basis of coordination: The book was written by John and by Bill is
grammatical — coordination of two by-DP strings. Under Collins's
analysis there is no [PP by DP] constituent (since by heads VoiceP,
not PP), so the apparent coordination is analyzed as VoiceP coordination
with deletion of the second PartP.
The substantive claim relevant to this formalization: the entity that
behaves as a coordination target is VoiceP (with PartP-deletion),
not [PP by DP].
The coordination test bears on Voice-head identity: if by heads
VoiceP (Collins §4), then [by John] is the surface form of a
[VoiceP PartP [Voice' by vP]] and the coordination in (63b) is
VoiceP-coordination with PartP-deletion (Collins §8 eq. 64). The
substantive constraint is that both conjuncts use the same Voice
head, which voicePassive represents.
§7. Inverse-voice family membership #
The smuggling derivation Collins develops for the passive is the same
mechanism @cite{storment-2026} extends to QI and LI. The shared
structural invariant — non-phase Voice permits VP/PartP smuggling —
is captured by Minimalist.InverseVoiceConstruction in the Theories
layer. Collins's passive instance is Minimalist.passiveCanonical.
The passive is one canonical instance of the inverse-voice family.