Uniform Projection Hypothesis for Psych Verbs #
@cite{kim-2024} @cite{belletti-rizzi-1988} @cite{pesetsky-1995}
@cite{kim-2024}'s Uniform Projection Hypothesis (UPH) applied to the @cite{belletti-rizzi-1988} classification of psych verbs. All Class II (object-experiencer) psych verbs uniformly project Cause + Experiencer; the eventive/stative split comes from the causal source (external percept vs internal representation).
Architecture #
The fragment entries in Verbal.lean set independent fields:
causalSource(external vs internal)subjectEntailments/objectEntailments(proto-role profiles — stimulus-on-subject vs experiencer-on-subject derived viaEntailmentProfile.toRole)opaqueContext(true vs false)
@cite{kim-2024}'s theory predicts these fields must covary:
- All Class II verbs share the same proto-role profile (UPH)
opaqueContextis determined bysubjectIntensionalapplied tocausalSourcecausalSourcedetermines temporal and event-structural behavior
These predictions are captured by the classII_consistent predicate (§ 1),
verified per-verb (§ 2), and then used to DERIVE consequences (§§ 3–7).
Key results #
- Consistency: each Class II entry satisfies
classII_consistent, connecting 4 independently-set fields through @cite{kim-2024}'s theory - UPH derivation: theta-grid uniformity FOLLOWS from consistency
- Opacity derivation:
opaqueContextFOLLOWS fromcausalSource - Temporal prediction: temporal behavior FOLLOWS from
causalSource - T/SM restriction: derived from the Onset Condition on causal chains
- Class I/II theta reversal: derived from the consistency predicates
- Proto-role bridge: theta roles map to canonical Dowty profiles
- Cascade bridge (§ 9): @cite{pesetsky-1995}'s syntactic HMC account and @cite{kim-2024}'s semantic Onset Condition account both predict Cause+SM incompatibility but diverge on Cause+Target
A Class II (object-experiencer) psych verb entry is internally consistent when its independently-set fields agree with @cite{kim-2024}'s predictions:
(a) It has a causal source (external or internal)
(b) Opacity agrees with subjectIntensional applied to the causal source
The existential over CausalSource ties the causal source to the opacity
prediction: changing the causal source field MUST change the opacity field
to maintain consistency.
Equations
- Kim2024_UPH.classII_consistent v = ∃ (cs : Semantics.Causation.Psych.CausalSource), v.causalSource = some cs ∧ v.opaqueContext = Semantics.Causation.Psych.subjectIntensional cs
Instances For
A Class I (experiencer-subject) psych verb entry is consistent with @cite{belletti-rizzi-1988}'s temere pattern: no causal source (the internal/external distinction is Class-II-specific).
Equations
- Kim2024_UPH.classI_consistent v = (v.causalSource = none)
Instances For
Each theorem below connects independently-set fragment fields through
Kim's theory. If causalSource or opaqueContext on a fragment entry
changes, the corresponding theorem breaks — ensuring the fields stay
in theoretical agreement.
Opacity derivation: any consistent Class II verb with internal causal source has an opaque subject position.
This connects two independently-set fields (causalSource, opaqueContext) through Kim's theory: the opacity ISN'T stipulated — it FOLLOWS from the causal source being internal (maintenance relation).
Transparency derivation: any consistent Class II verb with external causal source has a transparent subject position.
UPH within a single verb: worry's eventive and stative readings differ in causal source. This is Kim's strongest test case — same lexical item, two readings.
Temporal derivation (external): any verb with external causal source predicts temporal precedence and a state transition (BECOME). The temporal behavior FOLLOWS from the causal source, not from per-verb stipulation.
Temporal derivation (internal): any verb with internal causal source predicts temporal overlap and no state transition. Cause and effect coexist (maintenance relation).
Per-verb temporal grounding: frighten's fragment datum (external source)
determines specific temporal predictions. Changing the datum to .internal
would change the predictions.
Per-verb temporal grounding: concern's internal source determines temporal overlap and no transition.
UPH at the causal link level: eventive and stative Class II verbs differ in temporal and event-structural predictions despite both being Class II. This is Kim's full claim: the aspectual split is orthogonal to argument structure.
T/SM restriction derived: Cause occupies onset, SM also needs onset, but only one participant can occupy onset → they conflict.
This theorem shows the structural basis: both Cause and SM want the onset position, and there's only one onset slot.
Class II theta roles map to the canonical Dowty proto-role profiles (bridging @cite{kim-2024} UPH to @cite{solstad-bott-2024} proto-role infrastructure). stimulus → causation + independent existence (P-Agent = 2), experiencer → sentience + independent existence (P-Agent = 2).
Class I subject profile matches Class II object profile: both are experiencers (sentience + independent existence).
B&R class → expected subject role mapping is correct for our entries.
Internal causal source implies subject position is intensional (type level).
External causal source implies subject position is extensional (type level).
For Class II verbs, stimulus subtype is DERIVED from causal source
via CausalSource.toStimulusType. No new lexical field needed —
the existing causalSource field determines T vs SM.
These theorems verify that each verb's derived stimulus type
predicts the correct PP frame and Cause-cooccurrence behavior.
Derive a verb's stimulus type from its causal source.
Equations
Instances For
Any verb with external causal source derives a Target stimulus that doesn't conflict with overt Cause.
Any verb with internal causal source derives an SM stimulus that conflicts with overt Cause (Onset Condition).
@cite{pesetsky-1995} and @cite{kim-2024} both predict the T/SM restriction (Cause + SM = ill-formed) but via different mechanisms:
**Pesetsky (syntactic, symmetric)**: CAUS must incorporate into V
via HMC. Both *at* (Target) and *about* (SM) are nonaffixal P heads
that block CAUS movement equally. So BOTH T and SM are predicted to
be incompatible with an overt Cause argument.
**Kim (semantic, asymmetric)**: SM maps to the onset of the causal
chain; Cause also maps to onset → position conflict. But Target maps
to the TERMINUS, so Target does NOT conflict with Cause.
The two accounts agree on Cause+SM being ill-formed but DIVERGE on
Cause+Target: Pesetsky predicts it ill-formed, Kim allows it.
`StimulusType.conflictsWithCause` encodes Kim's asymmetric prediction
(.subjectMatter → true, .target → false) — this is NOT Pesetsky's
symmetric prediction.
Pesetsky's symmetric prediction: both at and about are nonaffixal and therefore block CAUS movement equally.
Kim's asymmetric prediction: SM conflicts with Cause, Target does not.
The accounts agree on Cause+SM: Pesetsky predicts it blocked (nonaffixal about blocks HMC), Kim predicts it blocked (SM and Cause both map to onset), and the empirical data confirms it.
The accounts diverge on Cause+Target: Pesetsky predicts it blocked (nonaffixal at blocks HMC just like about), but Kim predicts it acceptable (Target maps to terminus, not onset).