Lexical Converse Relation¶
A lexical-semantic relation in which paired predicate senses describe one relational scene from systematically exchanged participant positions, as parent/child, above/below, or buy/sell.
Core Idea¶
A lexical converse relation—also called converseness or a relation between relational opposites—holds between predicate senses that express the same basic relational situation from systematically exchanged participant positions. In the canonical binary case, predicates \(P\) and \(Q\) are converses when
within the relevant senses and contextual conditions. Thus PARENT_OF(x,y) is equivalent to CHILD_OF(y,x), ABOVE(x,y) to BELOW(y,x), and OWN(x,y) to BELONG_TO(y,x). Replacing one predicate by its converse and exchanging the participants yields a semantic paraphrase, not a negation.
Scope of Application¶
Converseness is a standard lexical sense relation in semantics, alongside synonymy, hyponymy, meronymy, complementarity, and other kinds of opposition. It applies across parts of speech: relational nouns (parent/child, employer/employee, advisor/advisee), prepositions and spatial predicates (above/below, before/after, in front of/behind), verbs (own/belong to, lend/borrow, buy/sell), and comparative predicates (older than/younger than, larger than/smaller than).
Clarity¶
The central clarity move is to write the role map instead of relying on the vague phrase “opposite point of view.” For binary predicates:
Both argument exchange and lexical substitution are necessary. Keeping the arguments fixed yields “a is below b,” which contradicts rather than paraphrases “a is above b” in an ordinary vertical order.
Manages Complexity¶
Converseness reduces redundant lexical description by factoring two expressions into one underlying scene plus a role permutation. Instead of treating owns and belongs to as unrelated predicates, a semantic representation can store one possession relation and two lexicalization rules. This reduces duplicated inference rules and makes inconsistencies visible: if a database asserts that Ada owns the violin but denies that the violin belongs to Ada under the same time and legal sense, the inverse mapping exposes a contradiction.
Abstract Reasoning¶
Converse inference. From \(P(a,b)\) and a verified converse link \(P(x,y)\leftrightarrow Q(y,x)\), infer \(Q(b,a)\). The inference is bidirectional.
Double-converse inference. Because role reversal is involutive, the converse of \(Q\) is \(P\). If a lexical resource records only one direction, the inverse link can be generated, subject to the same sense restrictions.
Knowledge Transfer¶
The full abstraction transfers literally to lexicography, formal semantics, computational lexicons, and semantic parsing because those practices all preserve lexical senses, participant roles, and entailment. A dictionary can display inverse definitions; a frame lexicon can map perspectivized roles; a semantic parser can normalize inverse predicates; and a knowledge graph can materialize inverse assertions.
It also transfers to ontology lexicalization with an important layer boundary. An ontology may declare two object properties inverse, while a lexicon links language-specific entries to those properties.
Relationships to Other Abstractions¶
Current abstraction Lexical Converse Relation Domain-specific
Parents (1) — more general patterns this builds on
-
Lexical Converse Relation is a kind of Relation Prime
Relation is the proposed minimal parent.
Hierarchy path (1) — routes to 1 parentless root
- Lexical Converse Relation → Relation
Neighborhood in Abstraction Space¶
Lexical Converse Relation sits in a sparse region of the domain-specific corpus (80th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.
Family — Unclustered & Miscellaneous (1565 abstractions)
Nearest neighbors
- Reciprocal Construction — 0.85
- Dubitative Mood — 0.82
- Lexical Definition — 0.82
- Cross-reference Relation — 0.81
- Social Communication — 0.81
Computed from structural-signature embeddings · 2026-09-08