Collaborative Model of Reference¶

A psycholinguistic theory in which conversational partners jointly establish reference through presentation, acceptance or repair, acknowledgment, and common-ground updating.

Version: v1 · 2026-09-28 · History
Domain-specific #: 8544
Domain group: Humanities
Origin domain: Linguistics & Semiotics
Subdomains: Psycholinguistics, Reference and Grounding, Pragmatics → Linguistics & Semiotics
Aliases: Collaborative model, Conversational model of reference, Collaborative reference

Core Idea¶

The collaborative model treats definite reference as a joint achievement. A speaker presents a description, the addressee displays acceptance, uncertainty, or repair, and the speaker acknowledges or revises the presentation until both have public evidence for the intended referent.

Established descriptions enter common ground and can be shortened or reused, distributing effort across turns. This contrasts with an autonomous literary model in which the speaker supplies a finished description and a passive listener decodes it alone.

How would you explain it like I'm…

Figuring Out Which One

When you tell a friend 'the red one,' they might say 'which one?' Then you add 'the red truck by the door,' and they say 'oh, that one!' You worked it out together. Next time, you can just say 'the truck.'

Pointing Out Things Together

The Collaborative Model of Reference says that when we talk about a particular thing, the speaker and listener figure it out together. The speaker offers a description, the listener shows whether they got it, is unsure, or asks for a fix, and the speaker confirms or tries again. They keep going until both clearly know which thing is meant. Once they agree on a description, they can shorten it or reuse it later, so less work is needed the next time. This is different from thinking the speaker says one perfect sentence and the listener just decodes it alone.

Jointly Grounded Reference

The Collaborative Model of Reference treats referring to a specific thing as a joint achievement between speaker and addressee. The speaker presents a description; the addressee shows acceptance, uncertainty, or a request for repair; and the speaker acknowledges or revises it until both have public evidence that they share the intended referent. Once a description is agreed on, it enters the participants' common ground, so it can be shortened or reused later, spreading the effort across the conversation. This contrasts with an 'autonomous' or literary model, in which the speaker produces a complete description and a passive listener decodes it on their own.

The Collaborative Model of Reference analyzes definite reference as a jointly accomplished action rather than a one-shot transmission. The speaker presents a candidate description, the addressee responds with evidence of acceptance, uncertainty, or a repair request, and the speaker acknowledges or revises the presentation. This cycle continues until both participants have public evidence that the intended referent has been identified. Once a description is established it enters common ground and can later be shortened or reused, so the effort of referring is distributed across turns and becomes cheaper over time. The model contrasts with an autonomous, literary model in which a speaker supplies a finished description and a passive listener decodes it independently. Its key claim is that the addressee's feedback is part of how reference succeeds, not an optional add-on.

Scope of Application¶

Referential communication experiments. Studies repeated description and grounding of unfamiliar objects.
Conversation analysis. Tracks proposals, acknowledgments, and repairs turn by turn.
Dialogue systems. Models confirmations and clarification requests for entity grounding.
Collaborative tasks. Explains how partners establish labels while coordinating action.
Psycholinguistics. Tests audience design and common-ground use.

Clarity¶

Identify the target, each presentation, the addressee's evidence, the speaker's uptake, repairs, and what becomes common ground. Do not infer mutual understanding from successful speaker production alone. Inclusion test: Show a conversational target, a speaker presentation, responsive listener evidence, speaker uptake, and a jointly grounded description or repair sequence. Exclusion test: Exclude monologic writing to an absent audience, one-sided naming, accidental co-reference, and any dialogue where feedback does not affect the reference-establishment process. Nearest boundary: The literary or autonomous model assigns the speaker primary responsibility and treats the listener as independently decoding a completed description. Exit condition: The interaction leaves the model's defining process when reference is fixed without reciprocal evidence or common-ground updating. Common misclassifications: It is not mere conversational cooperation in general. It is not unilateral speaker intention or listener inference. It does not require explicit verbal acceptance on every turn; continued coordinated action can supply evidence. Minimum collaborative effort is not minimum speaker effort alone. Nearest named distinctions: Literary model of reference: Treats reference as speaker-authored and listener-decoded rather than jointly grounded. Common ground: Is the shared-state resource and result, not the whole collaborative procedure. Conversational repair: Corrects trouble broadly; reference grounding is a particular task. Cooperative principle: Provides conversational norms but not the presentation–acceptance model.

Manages Complexity¶

The model reduces a fluid conversation to linked contribution pairs and a changing shared state. It explains why descriptions become shorter, why misunderstandings trigger repair, and why responsibility for reference is distributed across participants.

Abstract Reasoning¶

Specify the candidate referents and partners' initial common ground.
Mark the speaker's presentation.
Classify listener response as acceptance, rejection, or repair request.
Trace speaker acknowledgment or revision.
Record the description provisionally grounded.
Test reuse and effort reduction in later turns.

Knowledge Transfer¶

The model transfers literally across face-to-face, mediated, and human–machine dialogue when reciprocal grounding evidence updates a shared referential state. Teamwork without reference establishment or one-way communication shares only a general collaboration metaphor.

Relationships to Other Abstractions¶

Current abstraction Collaborative Model of Reference Domain-specific

Parents (1) — more general patterns this builds on

Collaborative Model of Reference presupposes Coordination Prime

Collaborative Model of Reference presupposes Coordination because the theory requires conversational partners to align presentations, acceptances, repairs, and common-ground updates.

Hierarchy paths (5) — routes to 4 parentless roots

Collaborative Model of Reference → Coordination → Concurrency

Show alternative paths (4)

Neighborhood in Abstraction Space¶

Collaborative Model of Reference sits in a crowded region of the domain-specific corpus (36^th percentile for distinctiveness): several abstractions share nearly its structure, so a description that fits it tends to fit its neighbors too.

Family — Communication, Learning & Information Practices (15 abstractions)

Nearest neighbors

Appeal to the Stone — 0.89
Direct Method (Language Teaching) — 0.88
Precision marketing — 0.88
Explicature — 0.88
Ambient Authority — 0.87

Computed from structural-signature embeddings · 2026-10-08