Documentation

Linglib.Theories.Semantics.Events.ProtoRoles

@cite{dowty-1991} Proto-Roles: Theory Bridges and Canonical Profiles #

@cite{levin-2022}

Pure theory file connecting the entailment profile theory (EntailmentProfile.lean) to other semantic modules: @cite{cruse-1973} agentivity, @cite{krifka-1998} SINC, change-of-state, causation, and @cite{levin-1993} meaning components.

Verb-specific entailment profiles (canonical instances of EntailmentProfile) are defined here as theory-level constants. Bridge theorems connecting these profiles to English Fragment entries live in study files under Phenomena/.

die: no volition, no causation, 3 P-Pat → unaccusative.

arrive: no volition, no causation, 1 P-Pat → unaccusative. This FIXES the anomaly in @cite{dowty-1991}'s flat counting, which predicted unergative because pAgentScore (2) > pPatientScore (1).

Kick subject passes the do-test (has volition + causation + movement).

Die subject does NOT pass the do-test (no volition, causation, or movement).

Variable agentivity = the broom sense has strictly more P-Agent entailments than the basic sense.

Wind as effector: movement + independent existence.

Kick object is a force recipient (causally affected + stationary).

@cite{levin-1993}'s meaning components are a coarse-grained projection of Dowty's entailment profiles. Three of the four components correspond to specific Dowty entailments:

| Levin component | Dowty entailment | Argument |
|---|---|---|
| `causation` | P-Agent (c) `causation` | Subject |
| `changeOfState` | P-Patient (a) `changeOfState` | Object / unacc. subject |
| `motion` | P-Agent (d) `movement` | Subject |
| `contact` | — (no direct Dowty counterpart) | — |

The key divergence: Levin's `causation` is *narrow* — it diagnoses
specifically causation of a change of state (via the causative/inchoative
alternation). Dowty's `causation` is *broad* — any event where one
participant causes something in another. 

All canonical verb profiles satisfy the internal well-formedness constraint (volition → sentience).

Kick satisfies the causation↔CoS paired constraint but NOT the IE↔DE constraint (the ball exists independently of the kicking).

@cite{grimm-2011}'s agentivity lattice reformulates proto-role entailments as privative features with lattice structure. This section connects the entailment profiles to Grimm's lattice, combining results from multiple theory bridges (@cite{cruse-1973} agentivity, @cite{krifka-1998} SINC).

Cross-theory: kick subject passes the do-test (@cite{cruse-1973}), has maximal agentivity on @cite{grimm-2011}'s lattice (⊤), and maps to NOM in an accusative system. Three theories converge.

Cross-theory: build object is an incremental theme (@cite{krifka-1998}), maps to exPersEnd on @cite{grimm-2011}'s persistence scale (creation), and falls OUTSIDE the canonical ACC region — creation verb objects are non-prototypically transitive.

Cross-theory: arrive subject has movement but not instigation → Grimm's lattice places it outside the NOM/ERG region, consistent with the priority-based ASP predicting unaccusative.