# Communication Models: Shannon, Jakobson, Example, and Limits

Communication models: compare Shannon and Jakobson with an example of an instruction misunderstood despite successful transmission.

- Canonical URL: https://www.harmate.com/en/blog/communication-models-shannon-jakobson-example-and-limits
- Author: Harmate Team
- Published: 2026-10-04
- Updated: 2026-10-04T00:20:44.025407+00:00
- Language: en

## Content

# Communication Models: Shannon, Jakobson, Example, and Limits

A communication model represents selected parts of an exchange to help frame a question. Shannon's model describes, among other things, how a signal travels from a source through a channel; Jakobson's diagram relates six factors of verbal communication—addresser, addressee, message, context, code, and contact—to different language functions. Neither model alone proves that two people assigned the same meaning to an instruction.

Imagine a trainer who clearly says at the end of a workshop: “Send the form after review.” Everyone hears the words. Yet some people understand that a peer should review it; others wait for a manager's approval. No obvious channel failure is described; the meaning of “review” and the reviewer’s role still need clarification.

## Shannon and Jakobson: what is the difference?

The models address different questions. Shannon formalizes problems of information transmission in engineering: how can a source, message, signal, channel, and noise affecting the received signal be represented? His 1948 paper is a contribution to the mathematical theory of communication ([C. E. Shannon, *A Mathematical Theory of Communication*](https://onlinelibrary.wiley.com/doi/10.1002/j.1538-7305.1948.tb00917.x)). The model is useful for discussing transmission, but it does not aim to describe every part of a human relationship.

In his work on linguistics and poetics, Roman Jakobson sets out factors of verbal communication and functions of language. He relates an addresser, addressee, and message to context, code, and contact ([Jakobson, “Closing Statement: Linguistics and Poetics”](https://commons.princeton.edu/wp-content/uploads/sites/41/2017/09/Jakobson-Linguistics-and-Poetics.pdf)). Context and code help frame questions that a transmission diagram does not foreground: what is the message about, and do the people share enough reference points to interpret it?

```text
Source → Transmitter → Channel → Receiver → Destination
                          ↑
                        Noise
```

Here, noise is a possible disturbance to the signal in the channel, not a general explanation for every misunderstanding. Jakobson relates six factors to language functions:

| Factor | Associated function | Illustrative example |
|---|---|---|
| Addresser | Emotive | “I am worried about this deadline” — the addresser’s stance |
| Addressee | Conative | “Send the form tomorrow” — an appeal to the addressee |
| Message | Poetic | “Review to write better” — attention to form |
| Context | Referential | “The form has three sections” — information about context |
| Code | Metalingual | “By review, I mean a peer check” — clarifying the code |
| Contact | Phatic | “Can you hear me?” — checking contact |

These functions are lenses, not exclusive boxes: one instruction can involve several at once.

| Lens | Shannon | Jakobson |
|---|---|---|
| Useful question | Was a signal transmitted or affected by the channel? | Which factors and functions are involved in verbal communication? |
| Main elements | Source, message, transmitter, signal, channel, noise, receiver, destination | Addresser, addressee, message, context, code, contact |
| Example of difficulty | Audio drops out; the instruction cannot be heard | A familiar word has different references or context |
| What the model cannot establish | That the recipient assigned the intended meaning | That people share an interpretation or have agreed |

Educational accounts sometimes vary in their terminology or historical attribution. For practical use, do not reduce Jakobson to a simple “more complete Shannon model.” He offers another way to describe verbal communication. The [ENSFEA teaching resource introduces Jakobson's factors while discussing the limits of a linear model for interpersonal communication](https://escales.ensfea.fr/ressources-pour-lenseignant-e-animateur-trice/les-cours/communication-humaine/premieres-approches/).

## Example: one instruction, two readings

**Fictional situation:** During training on writing an intervention form, the trainer says: “Send the form after review.” The audio is clear, no one interrupts, and the message appears on the projected slide. A week later, two participants describe different next steps.

> **Trainer:** “Send the form after review.”  
> **Participant A:** “I reviewed it myself and sent it.”  
> **Participant B:** “I did not send it because I am waiting for my manager to approve it.”  
> **Trainer:** “I meant: ask a colleague to check the information, then send the form.”

In this case, the details describe no obvious channel failure: the instruction was audible and visible. The divergence was not about volume or audio quality; “review” could mean a self-check, a peer review, or formal approval. The instruction also did not say who should do it.

Shannon helps frame the question of signal transmission, but does not by itself prove that the signal was faithful. Jakobson's diagram invites questions about the work context and what shared code was needed to understand it. But the diagrams cannot tell us which reading each participant made. Their explanations—and other contextual evidence—are needed to check.

A trainer might ask: “For this form, what would you do after drafting it? Who would review or approve it?” Then they could clarify: “Ask a colleague to check the information, then send the form; no manager approval is needed for this exercise.” A paraphrase or task example can check whether the clarification resolved the ambiguity, without proving that future misunderstandings are impossible.

## Transmission, interpretation, and agreement are different

A transmission model helps describe how a signal travels. Understanding also requires connecting words to an object, situation, shared knowledge, and purpose. Agreement adds another step: people may understand a proposal and still reject it or prefer a different action. “I heard it,” “I understood it,” and “I agree” are not interchangeable.

Confusing these stages leads to poor diagnosis. If an instruction is unclear, repeating it louder may not clarify its referent. If participants understand but lack the time or authority to act, the primary issue is not channel noise. Before blaming a channel or a person, specify the question: access to the message, understanding a term, agreement about the action, or ability to carry it out.

## Choose a check that fits your question

To check transmission, ask whether the message was accessible—for example, whether an attachment opened or audio was audible. To check meaning, request a paraphrase or example: “What does ‘review’ mean in your situation?” To measure comparable choices, a closed question may help, such as selecting who should review the form. But that measure assumes the important options are already known. It will not surface an omitted role or a different interpretation on its own. Agreement and feasibility call for another question: “What might stop you from following this instruction?” can reveal a refusal, a lack of time, or an approval step required elsewhere in the organization.

Open questions can help discover what people understood and the context of their answers. They do not automatically make a sample representative, and they do not replace measurement when the question concerns a defined frequency. Start by naming the decision, as in the [actionable questionnaire guide](/en/blog/actionable-questionnaires-start-with-the-decision-not-the-questions), then decide whether to explore meanings, measure an option, or do both at distinct stages. The [5W1H guide](/en/blog/5w1h-method-a-practical-framework-for-framing-qualitative-research-questions) can help frame an inquiry; it cannot decide which interpretation is true.

## What the models do not explain on their own

A diagram simplifies. It can isolate a channel, code, or context to make a question visible, but it cannot replace examination of the situation. The same words can refer to different practices by occupation, team, language, timing, or stakes. A written sentence may also be ironic, elliptical, or a response to earlier turns that the diagram does not show.

Do not use a model to diagnose people: a misunderstanding does not establish incompetence, resistance, or bad faith. Nor do several people giving the same answer prove an entire group shares a belief; ask who responded, who is missing, and under what conditions the answers were collected. A closed measure and an open account provide different information.

In training or consulting, a tool may support response collection or import, surface themes with their supporting verbatims, and prepare a report. The practitioner checks meaning and context, keeps disagreements visible, and decides what to do next. A surfaced theme does not establish shared understanding or a cause, and it is not a pedagogical recommendation.

Communication models remain useful when treated as partial maps, not as the whole reality. Shannon helps frame questions about transmission; Jakobson draws attention to the factors of language. To know what an instruction meant in a specific situation, you still need to listen to people, compare possible readings, and check which reference points were missing.