A Reference System for
Communication Design

The rules below codify the editorial practice that the project changelog (README.md) records change by change. They are stated here so that the system can be evaluated, challenged, and reproduced rather than taken on trust.

1 · Selection. A term enters the glossary when it appears in a named source: a workshop corpus, a course vocabulary list, a published glossary or scholarly text, or the Design Incubation keyword corpus. That corpus holds keywords supplied by presenting researchers to describe their own work. Editorial additions made for coverage are permitted, but they always carry the GEN flag. There is no frequency threshold, because the system records the field's working vocabulary as its sources attest it rather than a statistical sample of usage.

2 · Exclusion. Content is excluded when it fails the strip-down guarantee at its own tier. Three kinds fail: claims with no nameable source that are not honest editorial supply; sole-source claims whose source cannot be verified; and passages that argue rather than define. The record names its exclusions. Three passages of the founding proposal were excluded on these grounds when the thesis was integrated (21 Jul 2026), and the log records the reasons.

3 · Synonyms and variants. Synonyms and variant spellings fold under a preferred concept via variant tags, which is the defining operation of a controlled vocabulary. Merges are explicit: where duplicate entries appear across sources, the consolidated definition wins, and the supplement's provenance is appended rather than overwritten. Near-match merges are flagged for review.

4 · Competing definitions. When reputable sources genuinely disagree, the disagreement is preserved and named as a conflict rather than silently resolved; the multiple circulating dates for the Double Diamond are the standing example. Where senses genuinely differ, they are disambiguated into separate concepts (Form (Composition) and Form (Print Production)) rather than blended.

5 · Authority of sources. Three tiers govern every claim. SRC covers publication facts confirmed by live verification at build time. CAN covers widely attested scholarly attribution that this build did not individually re-verify. GEN covers editorial supply, which is always flagged and removable without loss of sourced content. Verified is never conflated with canonical, and the two are labeled separately.

6 · Who decides. One editor (Dan Wong) currently holds decision authority. The record is built so that authority is auditable rather than presumed. Every significant change logs what changed, why, and the verification steps taken. Reversals of the editor's own decisions are logged as named corrections in both directions. See Governance for the challenge pathway and for the system's institutional trajectory.

7 · Revisions and versioning. Changes ship as numbered editions per volume. The changelog is append-only: it is corrected by addition and never rewritten. Named corrections are recorded at the point of use and in the Bibliography, and they include errors found in the system's own sources, such as a misspelled byline in a foundational reference.

8 · Integrity auditing. The glossary's inherited collections were audited for citation integrity on 13 Jul 2026. The method was distinctive-phrase extraction followed by live exact-phrase searches against published glossaries and textbooks. Confirmed wording lineages are now credited on the affected entries, and both the audit and its disposition are part of the record.

9 · Intellectual lineage. Where a definition sits inside a named theoretical tradition, the source line carries a Lineage: note distinct from the wording source. The wording source records the origin of the words; the lineage records the framework to which the concept belongs. Lineage notes are held to the same tiers as any claim: a lineage is SRC or CAN when the attribution is verified or standard, and GEN when it is editorial inference. Four entries carry a Lineage line in the current build as the initial application, and the convention is explained where a reader meets it, in The Lineage line in the glossary guide.

Editorial self-description (GEN) codifies the practice that the project changelog records; the log entries establishing each rule are dated there. Controlled-vocabulary and facet vocabulary follow Hodge (2000), Zeng (2008), and Bruno & Richmond (2003). Rule 9 states a convention now being applied entry by entry, and entries that do not yet carry a lineage note are not thereby claimed to lack one.

The current state, in plain terms. This system has one editor rather than a governance body. The reference works against which it measures itself — the OED, MeSH, the Getty AAT, the Library of Congress Subject Headings — are institutionally governed, and this one is not yet. What it has instead, and what any future governance inherits, is a complete audit trail: every significant change, correction, and reversal is logged with its verification steps in an append-only record. Durable references outlive their creators, so this section states what is policy now, what is invited now, and what is proposed.

What is policy now. (1) Permanent version history. Every volume carries a numbered edition, and every significant change is logged with its verification steps. (2) Deprecated content is preserved. Superseded definitions and reversed decisions remain in the record as named corrections. This includes reversals of the editor's own prior calls, which are logged in both directions. (3) Conflicts coexist. Genuine source disagreements are published as conflicts rather than resolved by authority. (4) Provenance is tiered and visible. Live-verified, canonical, and editorial content are labeled distinctly on every entry.

What is invited now — challenges and proposed revisions. Any entry, placement, or source attribution may be challenged. A challenge should name the contested claim and bring source evidence meeting the system's own standard: a verifiable publication (SRC), or a widely attested scholarly attribution (CAN). Every challenge receives a logged disposition. It is either accepted as a named correction that credits the challenger, or declined with reasons, and the record carries it either way. The contact route will be specified alongside the governance-council pathway below.

What is proposed — under review. The governance model that would carry this system past its single-editor phase is under active consideration, pending two conditions. First, the system's editorial rules and taxonomy must undergo further peer review, including the review process this response report addresses. Second, a governance council or committee must be established to steward the reference on behalf of the community it serves. The proposal under development names Design Incubation as the institutional host. Design Incubation's colloquium keyword corpus supplied a founding vocabulary for this system's glossary, its scholarly-workshop practice supplied the Field Atlas, and its editor community is the natural first constituency for a governing council. Under the proposal, a Design Incubation-hosted council would ratify editorial policy, admit new editors, adjudicate contested revisions that the single-editor pathway cannot resolve, and hold the transfer of stewardship when it happens. Until that constitution is settled, the honest description of this project's governance remains single-editor, fully documented, and open to challenge.

Why the transfer is designable in advance. The system is built for it. It rebuilds from a portable data spine (data.json and the concept/relationship corpus), its editorial rules are recorded above and in the changelog, and its full decision history is on the record. An incoming board would therefore inherit an audit trail rather than a black box. That is what makes single-editor governance defensible now: the trust the system asks for is not authorial confidence but a verifiable record. Governance can be added here without a rebuilding of trust, because every decision to date is already on that record.

On the standard spectrum of knowledge organization systems — term lists, controlled vocabularies, classification schemes, and relationship models (Hodge 2000; Zeng 2008) — this is a faceted classification system that carries a controlled vocabulary at its base and an encyclopedic apparatus above it. Three features place it there. The glossary's variant tags fold synonyms under preferred concepts, which is the defining operation of a controlled vocabulary. The six taxonomy roots, the breadcrumb hierarchy, and the four facets place each concept as specifically as possible. This is faceted classification, and it follows Bruno & Richmond's facets-as-metadata recommendation together with the thesis's placement rule. The Theories, atlas, and map volumes treat concepts discursively — key texts, connections, disputes, sources — which is encyclopedic rather than lexicographic work.

In orientation the system is descriptively sourced but normatively structured, and it is research-oriented in purpose. The data layer records actual usage: workshop language kept verbatim, coursework vocabularies, published glossaries, and conflicts preserved. The structure imposed on that record (domains, roots, Kinds, preferred terms) is a set of editorial commitments. Those commitments serve the thesis's retrieval aim: that design scholarship can be positioned, found, and extended by later work. Its scope is pedagogical and professional. Its view is synchronic: it maps the field's present conceptual state and holds historical entries without periodizing how meanings have changed.

Two labels it deliberately does not claim. It is ontology-informed but not a formal ontology, because it has no axioms and no machine-enforced subsumption. The open question of whether the six upper roots should align with the eleven Kind values sits precisely on that line, and Why this taxonomy in the glossary guide carries the structural defense. It is also graph-shaped but not a knowledge graph, because its concept relationships are rendered for human reading rather than exposed as a queryable semantic store with external identifiers.

Classification vocabulary follows Vandamme (2015) and the KOS-spectrum literature: Hodge (2000), live-verified 21 Jul 2026, and Zeng (2008), canonical. The facet apparatus follows Bruno & Richmond (2003). The placement rule and the concept-over-term principle come from the project thesis, above. Full citations: Bibliography → Thesis sources. The application of these categories to this system is editorial self-description (GEN).

The thesis claims that communication design lacks a classification system. That claim is only meaningful against the systems that partially occupy the territory.

General-purpose vocabularies reach the field's objects, not its research concepts. The Getty Art & Architecture Thesaurus is a faceted vocabulary of eight facets: Associated Concepts, Physical Attributes, Styles and Periods, Agents, Activities/Processes/Techniques, Materials, Object Types, and Brand Names. It was built to catalog works, materials, and techniques. Its own editorial guidelines state that terminology outside the visual arts is excluded, except as needed to catalog or discover information about visual works. The Library of Congress Subject Headings have been maintained since 1898, and the 47th edition carries data current to April 2026. They position published works within a universal bibliographic scheme. Both systems classify things about design — artifacts and books — rather than the concepts of which design research is made. Neither is maintained by, or answerable to, the communication-design field.

Discipline-owned schemes show what the field is missing. Computing classifies its own research through the ACM Computing Classification System, now in its 2012 revision. Roughly 120 computing specialists built that revision, a third of them ACM Fellows. Its structure is poly-hierarchical, and an open community-review mechanism keeps it current. Philosophy maintains PhilPapers' taxonomy of research topics, which holds 6,153 categories across five hierarchical levels. Nearly 1,000 volunteer academic editors curate it, and it is open both to public browsing and to editor applications. These are the structural models for what this system attempts: concept-based, research-oriented, expert-maintained, and open. Communication design has no equivalent at this scale or with this governance, though the note on the Typography & Graphic Design Thesaurus below describes the source that comes closest and still falls short.

Adjacent thesauri stop at adjacent borders. ERIC, the education-research thesaurus, has no "graphic design" or "communication design" descriptor at all; the literature it indexes is filed under "Graphic Arts," "Art Education," and "Design" instead. That absence confirms that even a thesaurus adjacent to design pedagogy stops at its border rather than entering the field's own conceptual vocabulary. Professional-body resources — competency frameworks and practice taxonomies such as the AIGA Types of Design Practice, cited on the Domain Intersection Map — describe the profession's shape rather than its research concepts. Published design glossaries and dictionaries sit at the term-list end of the KOS spectrum (Hodge 2000): definitions without conceptual structure, hierarchy, or governed revision.

The closest precedent, and why it still falls short. A search for prior art turned up one genuine near-miss: the Typography & Graphic Design Thesaurus (TGDT). It holds roughly 1,900 preferred terms plus 600 lead-in terms for typography, graphic design, web design, and the book arts. Those terms are organized in nine facets, and the thesaurus was built to conform to ANSI/NISO Z39.19-2005. It is the only source found that attempts a controlled vocabulary specific to this territory, and it is serious work. On inspection it is also exactly the kind of artifact this system's governance section is built to avoid becoming. It was a single-author project completed as graduate coursework in 2006. It remains at version 1.1, with no evidence of revision in the twenty years since, despite its own stated plans to continue development. It has neither an editorial board nor a succession plan. Its scope is also deliberately narrower than this system's. It indexes vocabulary for cataloging works about typography and graphic design, rather than the theories, methods, and disciplinary debates a research classification must carry. Its structure is monohierarchical, and its own introduction names polyhierarchy as a future revision that never came. Honest engagement with the TGDT sharpens the claim rather than undermining it. The nearest predecessor to this project stalled at precisely the point that this system's governance section exists to answer: single editor, undocumented process, no succession.

Therefore. The gap is specific. What is missing is a classification of communication design that is concept-based, research-oriented, expert-maintained, and openly governed; that takes as its units the field's own research concepts; that is structured for retrieval; and that publishes its editorial rules. No standing classification meeting this description was found, whether maintained by AIGA, the International Council of Design (ico-D), the Design Research Society, or a comparable body. That is what this system proposes, and the differences named above — including the one precedent that came closest and still fell short — are the measure by which it asks to be judged.

All six external systems were live-verified 22 July 2026, and the TGDT was additionally examined in full from its published PDF on 24 July 2026. Full citations for AAT, LCSH, ACM CCS, PhilPapers, ERIC, and TGDT are in the Bibliography. The comparison itself — which systems count as adjacent, and what "filling the gap" would require — is editorial (GEN).