Non-eventive Namespaces

Purpose and orientation

This document specifies the Layer-2 anchor for the non-eventive namespaces. It reuses the apparatus of the event-structure document — the two-layer model, the qualia-role vocabulary (AGENTIVE / TELIC / CONSTITUTIVE / FORMAL, §3 there), and the semantic types Sentient / State_of_affairs (§8 there) — but makes one central claim that changes the construction:

For @attribute, @entity, and @relational, FrameNet-Brasil already has schematic top frames that serve as the meta-frame. The model designates the existing frames rather than inventing parallel ones, and the specific frames relate to them through ordinary Inheritance, not the new Meta relation.

Frame names below are given in English; the #id is the authoritative, language-neutral reference (the webtool URL is the same regardless of the reader's display-language choice).

Scope. This revision covers @attribute, @entity, and @relational. It defers @situation and @pragmatic (§4).

Why no invented meta-frames here

The event-structure document had to coin meta-frames (Causative_meta, Inchoative_meta, …) and a dedicated Meta relation because FrameNet has no pre-existing schematic frame for an event view such as "the causative perspective." Those abstract structures had to be built from scratch.

The non-eventive namespaces are different. FrameNet-Brasil already contains the schematic frames — Attributes, Entity, Relation — each created to anchor inheritance, and each already functioning as an inheritance parent for the specific frames of its namespace. They are the abstract reference structures the event document had to invent. Reusing them is more FrameNet-native and avoids replication.

Two consequences:

• The linking relation is Inheritance, not Meta. Specific attribute / entity / relational frames already inherit from these anchors; no new relation is introduced for these namespaces.
• The qualia-role reading is descriptive overlay, not new machinery. The tables below map each anchor's actual FEs onto the qualia vocabulary so the connection to the event-structure model stays visible — but nothing is created; the frames already exist.

A small typing detail worth noting: in all three anchors the FORMAL element (the classifying/joining FE — Attribute/Value, Relation_type) is core but unexpressed, exactly as the master-table structure notes prescribe ("FE core-unexpressed for attribute"; "FE core joining the concepts").

The anchor frames

For each namespace: the existing frame (name, id, type), its definition, its FEs with status, and the qualia-role / DUL reading.

2.1 @attribute — Attributes (frame #485, non-lexical)

An Entity has a particular Attribute with some Value. This frame was created primarily to structure inheritance relations.

This anchor is where the DUL Quality / Region distinction lives: Attribute = Quality (the dimension), Value = Region (the value). This is the concrete reason the ontological type stays the coarse .attribute rather than splitting (event doc §4) — the dimension/value refinement is carried here, by the anchor's FEs. The bearer's semantic type is the only thing separating the entity-attribute sub-kind (Entity) from the modal-attribute sub-kind (State_of_affairs: Possibility, Probability; event doc §6.8). Already inherited by Gradable_attributes, Measurable_attributes, Importance, Shapes, and many others.

@entity — Entity (frame #226, lexical)

This frame is for words that denote highly schematic entities.

The thinnest anchor: a single core FE. An entity frame names a kind — it neither predicates nor relates. The three classifying axes from the Entity classification guide are not FEs; they surface as the inheriting subframes:

Already inherited by Artifact, Biological_entity, Location, Containers, Languages, Representation, …, and used by Existence_cycle_scenario. Layer 1 (.entity) already does most of the work for this namespace; the anchor's contribution is mainly to root the Reality × Nature × Essence inheritance tree.

@relational — Relation (frame #224, lexical)

A relation holds between Entity_1 and Entity_2.

This anchor is the Layer-2 home of the .relation ontological type (event doc §4.2). Relation_type is the FORMAL "joining" element (the master-table "FE core joining the concepts"); the Excludes/Requires constraints between Entity_1, Entity_2, and the collective Entities encode the dyadic ↔ collective alternation natively. Already inherited by Cognitive_connection, Media, Inherent_purpose.

How specific frames relate to the anchors

• Through ordinary Inheritance. Each anchor is already an inheritance parent; a specific frame inherits from it and binds its own core FEs to the anchor's FEs through the FE-to-FE bindings that Inheritance already carries. No Meta relation is introduced for the non-eventive namespaces.
• The qualia-role columns are analysis, not a mechanism. They let these three namespaces connect to the same high-level vocabulary as the event meta-frames, but add no new relation or constraint.
• Layer 1 ↔ Layer 2. The anchors sit at Layer 2 for their namespaces; the coarse ontological types .attribute, .entity, .relation are their Layer-1 counterparts (event doc §4).

Deferred: @situation and @pragmatic

These two namespaces are deliberately out of scope for now, and this document declines to speculate about them:

• @situation — scenarios and image-schemas are comparatively late additions to the FrameNet apparatus and need more analysis before their schematic structure can be fixed.
• @pragmatic — pragmatic frames belong to a separate, newer project that must be understood better on its own terms first.

When they are taken up, the question to ask first is the one this document answers for the other three: does FrameNet-Brasil already have a schematic frame that serves as the anchor? — before any meta-frame is invented.

Open points

1. Empirical check: confirm that the specific frames of each namespace in fact inherit (directly or transitively) from the designated anchor, and flag those that do not.
2. Whether @entity's anchor, being a single-FE frame, earns an explicit Layer-2 treatment at all, or whether Layer 1 (.entity) plus the inheritance tree already suffices.
3. The eventual treatment of @situation and @pragmatic (§4).