Skip to content

Triangle of Reference

An Ogden–Richards semiotic model in which a symbol reaches a referent through a user's thought or act of reference rather than an ordinary direct word–thing tie.

Version
v1 · 2026-10-31 · History
Domain-specific #
14592
Domain group
Humanities
Origin domain
Philosophy
Subdomains
Semiotics, Philosophy of Language → Philosophy
Aliases
Semiotic Triangle, Triangle of Meaning, Ogden Richards Triangle

Core Idea

The triangle of reference is Ogden and Richards' model of how a used symbol relates to a referent through a thought or act of reference.[1] The original diagram names those three corners and depicts the ordinary symbol–referent connection as imputed, rather than a direct tie between a word and a thing.[1] The interpreter participates in using a symbol; the interpreter is not a fourth corner or the referent.[1]

The model asks which act or understanding makes a symbol point to what someone is talking about. It does not promise that every sign has one concept and one object, or that every sign relation lacks resemblance: Ogden and Richards explicitly qualify the usual indirect base with cases such as images, gestures, drawings, and imitative sound.[1]

How would you explain it like I'm…

Word, Thought, Thing

When you say the word 'dog,' the word doesn't touch the real dog. Your thinking connects them: you think of the dog, and the word points there through your thought. It's like a triangle with the word, your thought, and the dog at the corners.

The Meaning Triangle

The Triangle of Reference is an idea from Ogden and Richards about how words connect to things. It has three corners: the symbol (like a word), the thought in someone's mind, and the referent (the actual thing being talked about). The word connects to the thing through the thought, not directly, so the bottom side of the triangle is drawn as only an assumed link. The person using the word takes part, but isn't a fourth corner. Some signs, like pictures or sounds that imitate things, do look or sound like what they stand for, and Ogden and Richards noted those cases.

Symbol–Thought–Referent Model

The Triangle of Reference is Ogden and Richards' model of meaning. A symbol (such as a word) relates to a referent (the thing talked about) through a thought or act of reference. In the original diagram, the line between symbol and referent is shown as 'imputed' rather than a direct bond, because the connection runs through the act of referring. The person interpreting takes part in using the symbol, but they aren't a separate fourth corner or the referent. The model asks what understanding makes a symbol point to what someone means. It doesn't claim every sign has exactly one concept and one object, and Ogden and Richards noted exceptions like images, gestures, drawings and imitative sounds, which resemble what they stand for.

 

The Triangle of Reference is Ogden and Richards' model of how a used symbol relates to a referent through a thought or act of reference. The original diagram labels those three corners and depicts the everyday symbol-referent connection as imputed, an indirect relation mediated by the act of reference, rather than a direct tie between word and thing. The interpreter participates in the use of the symbol but is neither a fourth corner nor the referent. The model's question is which act or understanding makes a symbol pick out what someone is talking about. It does not promise that every sign has exactly one concept and one object, nor that sign relations never involve resemblance. Ogden and Richards explicitly qualify the usual indirect base with cases such as images, gestures, drawings and imitative sounds.

Structural Signature

Sig role-phrases: a symbol in use → an interpreter's thought or reference → a referent; the ordinary symbol–referent base is inferred through the mediating reference.[1]

  • Symbol. A word or other sign form is used, rather than merely stored as a mark.[1] In the authors' ordinary example, the word dog supplies this corner.[1]

  • Thought or reference. The middle corner is what the interpreter thinks of or refers to in that use.[1] Remove this mediation and the diagram becomes a direct naming pair, changing its identity.

  • Referent. This is what the reference is directed toward; it is separate from the mediating thought.[1] In the original dog illustration, animals in the street can fill this role.[1]

  • Imputed base. The ordinary word–thing connection is drawn indirectly through use and reference.[1] This is a diagnostic feature of the model's usual case, subject to the authors' resemblance exceptions.[1]

  • Interpretive setting. In an applied terminology system, a symbol can be a term or code while source-specific intensions and represented classes fill adapted middle and referent roles.[2] The mapping must be stated rather than assumed identical to a single person's mental event.[2]

Recognition test: can the analyst name the symbol, the reference made in its use, and what is referred to without collapsing the last two? If only a fixed label–object lookup is known, the triangle has not yet been demonstrated.

What It Is Not

  • Not a speaker–concept–designation diagram. The original printed labels are Symbol, Thought or Reference, and Referent.[1] A speaker uses the symbol but does not replace a corner; a concept is not simply the referent.[1]
  • Not a universal one-to-one code. Different words, interpretations, and objects need not pair uniquely.[2] Campbell and colleagues use the triangle precisely to examine such variation in a biomedical terminology system.[2]
  • Not every kind of sign. The diagram is a particular analysis of interpreted reference.[1] A resemblance-linked image or gesture requires the qualification supplied by the original authors; it should not be forced into an unqualified indirect word–thing rule.[1]
  • Not a historical priority claim. The bounded account rests on the 1923 diagram. The frozen candidate's Bolzano, Aristotle, and problem-of-universals assertions are outside this source-supported identity.

Scope of Application

The primary home is semiotic and language analysis: an ordinary word used to refer, the interpretation that connects it to a referent, and the mistake of treating words as directly attached to things.[1] Campbell and colleagues explicitly adapt the diagram to biomedical terminology, where source terms and codes, differing source intensions, and represented classes need to be kept apart.[2]

That second setting is an application within sign and terminology analysis. It is a formal possible-world account, so its extensional class is not an ordinary physical object encountered by one speaker, and its source intensions are not automatically one person's private thoughts.[2] The triangle may guide analysis in other terminologies only after those roles are established for the actual system.

Clarity

The three corners prevent two common substitutions. First, a person who utters a word is the user of the symbol, not the diagram's “subject” corner.[1] Second, the thought or act of referring is distinct from what is referred to.[1] Calling both “the concept” leaves no way to ask whether two users take the same expression differently while addressing the same thing.

The broken or imputed base gives a practical warning: a word's visible or audible form does not alone reveal its referent.[1] In the original dog illustration, one must consider someone using the word to refer to animals in the street.[1] The warning is about the ordinary word-use relation, with explicit resemblance exceptions.[1]

Manages Complexity

The diagram keeps form, interpretation, and represented target separate when a discussion would otherwise speak loosely of “the meaning.” That compression is useful in terminology integration: Campbell and colleagues show that expressions mapped under one UMLS CUI can carry source-specific conceptual differences.[2] Keeping the corners distinct makes it possible to ask whether a shared code preserves or hides those differences.

The triangle does not itself supply a complete theory of how thoughts form, how every sign is learned, or how a coding system should resolve conflicting sources. It organizes the question and exposes a possible collapse; the source material and use context must answer it.

Abstract Reasoning

For a proposed instance, identify the sign form first. Then ask which act or conceptual understanding connects that use to a referent. Finally, determine what counts as the referent in that setting: an encountered animal in the ordinary illustration or a represented extensional class in a formal terminology model.[1][2]

Run a counterfactual: hold the visible word fixed while changing the user's reference. If the target can change, the word alone did not determine the referent. Conversely, hold a coding identifier fixed while changing source intensions. If those differences still matter, collapsing identifier and thought obscures the interpretive structure Campbell and colleagues analyze.[2]

Knowledge Transfer

The mapping travels from everyday word use to a codified biomedical terminology because the functional question remains: how is a sign connected, through interpretation or source conception, to what it represents?[1][2] The carriers change from an ordinary utterance and encountered animals to terms or codes and a represented class.[1][2] The second account adapts the model; it does not establish that a computer code literally has a private thought.[2]

Transfer is limited by the referent chosen for analysis. A UMLS CUI can represent an extensional class inside the paper's possible-world semantics; it must not silently stand for one tablet, every physical instance, or an individual user's mental state.[2]

Examples

Canonical: the word dog in ordinary use

Ogden and Richards use the ordinary word dog and animals in the street to show why the word is not directly tied to those animals independent of someone using it.[1] This is their conceptual illustration, not a report of measured behavior by a population of speakers.

Mapped back: symbol → dog as used word; thought or reference → someone's act of referring in that use; referent → an animal or animals being referred to; imputed base → the word–animal link depends on that reference.[1] Erase the act of use and only printed marks and animals remain, which cannot explain the intended connection.

Applied: aspirin terms in the UMLS

Campbell and colleagues apply the semiotic triangle to UMLS integration.[2] Their aspirin analysis discusses source words or codes, potentially different source intensions, and CUI C0004057 as a representation of an extensional concept or class of things in the UMLS possible world.[2] A shared CUI therefore does not prove that all source expressions have exactly the same intension, and it is not itself one physical aspirin item.[2]

Mapped back: symbols → aspirin, acetylsalicylic-acid, or source-code expressions; adapted thought/reference → source terminology intensions and interpretive context; adapted referent → the represented class associated extensionally with C0004057 in the model; imputed base → the expressions reach that represented class through source interpretation and integration, without guaranteed one-to-one correspondence.[2] This is the paper's formal adaptation, not a claim that a CUI is a private thought.

Structural Tensions

T1: convenient direct naming versus mediated reference. Treating a word as a stable object label is a convenient shortcut in conversation. Ogden and Richards' ordinary base instead runs through someone's act of reference.[1] Treating every reference as purely private, in turn, would conceal the shared symbol and target. Diagnostic: whose act of reference links this particular sign use to its claimed target?

T2: shared terminology code versus source intension. A common CUI helps integrate expressions from different biomedical sources, while the underlying sources can make distinctions that the shared extension does not preserve.[2] Neither the code nor an individual source term settles both roles by itself. Diagnostic: what difference in source conception is being lost or retained by this mapping?

T3: reusable three-corner form versus domain-specific account. A three-node sketch can be copied into many fields. In this model, the middle corner is Thought or Reference and the symbol–referent base is ordinarily imputed.[1] Erasing those semantic commitments yields a generic mediation diagram rather than the named semiotic identity. Diagnostic: is the purported transfer still about sign use, thought/reference, and what is represented?

Structural–Framed Character

Evaluative weight: the diagram diagnoses a relation among signs, reference, and referents; it does not judge whether a representation is truthful or desirable.

Human-practice dependence: the original account requires a user's reference, while the UMLS paper supplies a qualified formal adaptation through terminology sources and their interpretations.[1][2]

Institutional origin: the original work is a semiotic account, and the UMLS application exists within a designed biomedical terminology system.[1][2] Vocabulary travel: “symbol,” “thought,” and “referent” have roles in this account; relabeling arbitrary three-part systems with these words is not an instance. Import versus recognition: the model is recognized where an actual sign use and represented target are connected through an identifiable reference relation, not where the name is merely pasted onto a diagram.

Skeletal core: a sign is connected to what it represents through an intermediate interpretive role. Domain accent: the intermediate role is thought or reference, and the ordinary base is an imputed symbol–referent relation. The resulting character is framed and domain-specific: it is a disciplined analytic view of semiosis, with carefully stated exceptions.

Structural Core vs. Domain Accent

The general skeletal pattern “A reaches C through B” is widespread; it is too thin to carry this entry's identity. The distinctive account concerns a symbol in use, a thought or act of reference, and a referent, with a qualified ordinary indirect base.[1] A different three-part chain that lacks signification and interpreted reference may resemble the triangle graphically while being a different abstraction.

Future Prime question: could an independently evidenced cross-domain relation preserve these exact necessary functions without borrowing semiotic words metaphorically? The two supported settings here remain sign and terminology settings. They establish a reusable model inside that domain, not a substrate-independent Prime. The entry therefore remains domain-specific while a broader representation or interpretation concept may supply a separate foundation if a typed graph test warrants it.

This entry presupposes Sign.

  • Sign. Strict prerequisite: the diagram analyzes a sign in use and its relation to what it refers to. This is a composition/presupposition edge, since the model is not itself a species of sign.
  • Representation. The reviewed Triangle→Sign edge passed the official graph check, so the triangle inherits Sign's existing Representation parent. This connection does not establish an additional direct subsumption edge from the triangle to Representation.
  • Interpretation. Related because the symbol–referent relation depends on a use or interpretive act. The precise typed relation requires a separate graph test.
  • Symbolic Representation. A nearby general account of convention-bound symbol use. It cannot simply absorb this model: the mediating thought/reference and original work's resemblance qualifications are part of the present identity.
  • Signifier–Signified Duality. A different two-corner semiotic model that brackets the external referent; it is a contrast, not an alias or asserted parent.

Relationships to Other Abstractions

Local relationship map for Triangle of ReferenceParents appear above the current abstraction, mutual partners to the right, and children below. Node labels state whether each abstraction is prime or domain-specific; colors identify relation types.Triangle of ReferenceDOMAINDomain-specific abstraction: Sign — presupposesSignDOMAIN

Current abstraction Triangle of Reference Domain-specific

Parents (1) — outgoing typed relations

  • Triangle of Reference presupposes Sign Domain-specific

    The triangle analyzes a sign's reference relation; without a sign standing for or indicating something to an interpreter, the model has no subject.

Hierarchy path (1) — routes to 1 parentless root

Neighborhood in Abstraction Space

Triangle of Reference sits in a sparse region of the domain-specific corpus (67th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Language, Mind & Meaning-Making (73 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-31

Not to Be Confused With

  • Signifier–Signified Duality. Tell: is an external or represented referent a necessary corner, or is analysis confined to the signifier–signified pair?
  • Sense and Reference. Tell: is “sense” a Fregean objective mode of presentation, or is the middle role Ogden and Richards' thought or act of reference?
  • Denotation. Tell: is a semantic value assigned under a formal interpretation, or is the interpreter-mediated word–referent connection itself under study?
  • A terminology concept diagram. Tell: does it establish reference through sign use, or simply list designation, concept, and object as fixed fields?

References

[1] C. K. Ogden and I. A. Richards, The Meaning of Meaning (1923), printed pp. 10–12, especially the diagram on p. 11 and the dog discussion and resemblance qualification on p. 12. Original scan hosted by MIT: https://courses.media.mit.edu/2004spring/mas966/Ogden%20Richards%201923.pdf . The cited printed pages were checked through indexed primary-page text; local facsimile extraction was unavailable. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o ↩p ↩q ↩r ↩s ↩t ↩u ↩v ↩w ↩x ↩y ↩z ↩27 ↩28 ↩29 ↩30 ↩31 ↩32

[2] Keith E. Campbell, Diane E. Oliver, Kent A. Spackman, and Edward H. Shortliffe, “Representing Thoughts, Words, and Things in the UMLS,” Journal of the American Medical Informatics Association 5, no. 5 (1998): 421–431, https://doi.org/10.1136/jamia.1998.0050421 . Full text and figures: https://pmc.ncbi.nlm.nih.gov/articles/PMC61323/ . Relevant locators: Figure 1; “The UMLS as a Possible World”; “Words,” Table 1; “Thoughts,” Figures 3–4; and “Things,” Figure 5. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o ↩p ↩q ↩r ↩s ↩t