The claim
A75 §7: “In formal concept analysis, concepts arise from the relationship between objects and attributes. A formal concept is a pair of an extent (objects sharing attributes) and an intent (attributes shared by those objects), connected by a Galois correspondence … Rather than positing primitive entities such as teasing, admonition, or complicity, we can ask which conjunctions of observations support stable communicative categories. These categories can be partially ordered by inclusion of their defining constraints. The resulting structure is closer to a concept lattice than a strict hierarchy. For example, affectionate teasing and sarcastic condemnation may share linguistic and rhetorical attributes but differ in a small set of social-relational constraints. Neither must be located above or below the other in a simple abstraction hierarchy.”
Where it went
The chunk-5 reader finds it absent from A81’s outline, C6 and the manuscript, and Ganter and Wille appear again only in the U33 bibliography. The record holds Formal Concept Analysis: Mathematical Foundations (LIT-344), deferred and unread. The concept lattice is the intersection picture of CLAIM-042 made into a structure. It would give the manuscript’s “affectionate teasing” (§3) a place among its neighbours.
A93 §2, the bibliography (U33), recommended it again: formal concept analysis “directly addresses your account of objects emerging from intersections of observable constraints, through concept extents, intents, and lattice structure”. It had already fallen out of the proposals, and proposal v6 (A100) did not restore it.
What its own source says
Ganter and Wille’s preface (LIT-344, read from the publisher’s free front matter only; the book is not read) says formal concept analysis “does not strive to explain conceptual thinking”. Applying it to human concepts is “a matter for the experts in the respective science”. The lattice mathematics is available (Fong and Spivak’s Galois connections and closures, LIT-803), but the claim that communicative categories are a concept lattice is not the book’s, and it would need psychology’s evidence.
What the embedding literature does and does not supply
One route from text to such a lattice is now partly worked out. THEORY-185 (Korchinski et al., LIT-863) shows that binary attributes acting independently on co-occurrence become linear directions in a word embedding, from which an object–attribute incidence, and so a concept lattice, can be read off; the Lattice Representation Hypothesis (LIT-267) assumes as much of language models. But the independence that makes the derivation work makes every combination of attributes possible, and the lattice Boolean. This claim needs the opposite: categories that share most attributes and are told apart by a few constraints that rule combinations in or out. Whether distributional statistics with that structure still give linear attribute directions is open (QUESTION-025). Until it is answered, the embedding results neither support nor tell against this claim.
Note of 2026-10-30: what holds it back
No judged communicative categories have been tested for closure under formal concept analysis’s attribute–object correspondence, or for lattice-predicted confusions against a flat feature list, which is what the defeat condition asks. The claim’s own source disclaims applying the lattice to human concepts, and QUESTION-025, whether constraint-structured attributes still give linear directions, is open.