Psych Verb Syntax-Semantics Interface #
@cite{pesetsky-1995} @cite{kim-2024} @cite{kratzer-1996} @cite{schaefer-2008}## Directions of determination
The psych verb interface involves three layers with distinct directions of determination:
SYNTAX determines SEMANTICS determines
───────────────── ────────────────────
Voice_CAUSE head → ∃ causer argument
↑ ↓
(that's all) DP denotation (percept vs representation)
↓
CausalSource (external vs internal)
↓
┌─────────┴──────────┐
StimulusType opacity, temporality,
(T vs SM) event sort, transition
↓
PP frame (*of* vs *about*)
Cause cooccurrence
Syntax determines one thing: whether there is a causer argument at all. This is the contribution of Voice_CAUSE (= Pesetsky's CAUS head). Agentive/causer Voice projects a specifier; non-thematic/expletive Voice does not. The syntax is "done" after this — it cannot see what kind of entity fills the specifier.
Semantics determines everything else. The DP's referential type (percept vs representation) is a property of its denotation, invisible to the syntax. From this single semantic property, all other properties follow: CausalSource, StimulusType, opacity, temporal structure, event sort, PP frame selection, and Cause cooccurrence.
The zero syntax thesis: the syntactic structure is invariant across the two DP types. No head, feature, or morpheme distinguishes percept from representation DPs. The syntax is blind to the T/SM distinction. All variation is semantic.
@cite{pesetsky-1995}'s CAUS head is identified with Schäfer's Voice_CAUSE (@cite{schaefer-2008}). Both introduce a causer external argument.
This identification follows Occam: the two analyses posit the same structural position (Spec,vP/VoiceP) introducing the same kind of argument (causer).
Equations
Instances For
Syntax determines: Voice_CAUSE introduces a causer argument. This is what the syntax contributes — the existence of a causer in the structure.
Syntax determines: Voice_CAUSE is a phase head, creating a domain boundary.
Syntax does NOT determine: the kind of causer. All θ-assigning Voice heads project the same structural position (Spec,VoiceP with [D] feature). The syntax treats agentive and causer Voice identically in this respect.
Contrast: non-thematic Voice does NOT introduce any argument. This is where syntax makes its one cut: causer argument vs no causer argument.
The referential type of the DP in Spec,Voice_CAUSE.
This is a semantic type, not a syntactic one. The syntax sees only "DP in Spec,VoiceP" — it cannot distinguish percepts from representations. The distinction is a property of the DP's denotation, visible only to the semantic component.
.percept: mind-external entity/event — "the noise" in "the noise frightened John".representation: mind-internal representation — "the problem" in "the problem concerns John" (John's mental representation of the problem, not the problem itself)
- percept : CauserDenotation
- representation : CauserDenotation
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Semantics determines: causal source from DP denotation.
The referential type of the DP determines the causal source. This is a semantics-internal derivation: from what the DP denotes, we know what kind of causation is involved.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Semantics determines: subject opacity from DP denotation.
Composed via CausalSource: denotation → source → opacity.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Semantics determines: causal link profile from DP denotation.
Composed via CausalSource: denotation → source → temporal structure, event sorts, transition.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Semantics determines: stimulus subtype from DP denotation.
Composed via CausalSource: denotation → source → T/SM.
Equations
Instances For
Representation DP → internal source.
Percept DP → transparent subject.
Representation DP → opaque subject.
Percept DP → Target stimulus, of-PP.
Representation DP → Subject Matter stimulus, about-PP.
Percept DP → temporal precedence, BECOME transition.
Representation DP → temporal overlap, no BECOME.
Semantics determines all: one semantic property of the DP (its referential type) determines seven downstream properties. None of these are syntactically encoded.
Zero syntax thesis: the syntactic head projected is the same regardless of DP denotation.
The syntax contributes exactly Voice_CAUSE in both cases. There is no syntactic feature, morpheme, or head that distinguishes the percept configuration from the representation configuration. All seven semantic differences (§ 3) arise from the DP's denotation, not from any syntactic distinction.
The zero syntax thesis stated as a gap: syntax is constant while semantics varies. The two conjuncts make the directions explicit — same syntax, different semantics.
Interface summary: the syntax-semantics division of labor.
Syntax contributes: causer argument exists (θ-role assigned). Syntax does NOT contribute: what kind of causer, opacity, temporality, T/SM, PP frame — none of these are syntactically encoded.
Semantics contributes: all variation between percept and representation configurations.