Fiske 1990
Dedications
To NATASHA for everything. To MATTHEW AND LUCY for keeping quiet (well…fairly) during the cold wet summer of 1980.
Shannon and Weaver: the model of communication
John Fiske introduces Shannon and Weaver’s Mathematical Theory of Communication (1949; Weaver, 1949b) as a foundational seed of the field. The model treats communication as a linear process focused on transmission through channels (phone line, radio waves, etc.). The source selects a message from a set of possible messages, the transmitter converts it into a signal, the signal travels through a channel, the receiver detects it, and the receiver may feed back or respond. Noise is anything added to the signal between transmission and reception that is not intended by the source. The model’s simplicity makes it easy to grasp, but it has attracted critics for oversimplification. The model is used to analyze three kinds of problems, discussed as Levels A, B, and C below, to understand how to improve accuracy and efficiency of communication. The model’s basic elements (source, transmitter, signal, channel, receiver, noise) are instantiated in everyday examples (telephone, conversation), e.g., the mouth as transmitter, air as channel, and ear as receiver. The “signal” becomes the audible sound; the same signal can be transmitted and re-transmitted as it passes through devices (e.g., the handset in a telephone can act both as a receiver and as a transmitter). The idea of “doubling up” of stages in a communicative act (e.g., signal being transmitted and then re-transmitted) is noted as something later models (e.g., Gerbner’s) handle more satisfactorily.
Levels A, B, and C of communication problems
Shannon and Weaver identify three levels of problems in communication:
Level A (technical problems): How accurately can the symbols of the message be transmitted? The focus here is on the channel capacity and the mechanical aspects of transmission.
Level B (semantic problems): How precisely do the transmitted symbols convey the intended meaning? This level concerns the interpretation of the message and the role of culture in meaning.
Level C (effective problems): How effectively does the received meaning influence conduct in the desired way? This level concerns the impact of the message on behavior and actions.
The model originally focused on Level A but is claimed to be applicable to Levels B and C as well; the three levels are not watertight but are interrelated and interdependent. The aim across levels is to understand how to improve accuracy and efficiency of the communication process.
Noise, channel, and signal
Noise in the Shannon–Weaver model is anything that degrades the signal between transmission and reception and is not intended by the source. Examples include crackling on a telephone line, static on a radio, or “snow” on a television screen. Noise can also be non-physical: discomfort in a chair or wandering thoughts during a lecture can act as noise because they interfere with decoding. Shannon and Weaver distinguish between engineering noise (Level A) and semantic noise (Level B). They suggest inserting a “semantic receiver” between the engineering receiver and the destination to address semantic distortions. Noise reduces the amount of information that can be transmitted in a given time, and this limitation applies to the entire process, across A, B, and C.
Information: a technical measure (Level A)
On Level A, information is a measure of the predictability of the signal, not its content. The more predictable a signal, the less information it carries. A signal’s information is measured in bits. If a signal has M equiprobable outcomes, then the information is
For example, a single binary choice (Yes/No) carries 1 bit of information when both options are equally likely. If there are 32 equiprobable possibilities, the information is
Thus information on Level A is about predictability and choice, not content. The unit used is the bit, derived from the binary nature of choices.
A practical illustration: a baby was posited as the outcome of five binary choices in a sequence (old/young; adult/pre-adult; pre-teen/teenager; etc.). If you follow five such binary decisions to identify a category like “baby,” the example yields about 5 bits of information. This is an illustration of how level A can relate to level B as the user moves from signal choices to semantic classification.
The model also discusses the relation between signal and meaning: the content of what is communicated (the meaning) is not necessarily captured by the raw signal, especially when cultural factors shape interpretation. Nevertheless, the information measure remains a tool to analyze how much information can be carried given a channel’s constraints.
Redundancy and entropy: core concepts
Redundancy is what is predictable or conventional in a message. Entropy is the opposite of redundancy and measures the maximum unpredictability or the number of possible signals. The two are inversely related: high redundancy means low entropy; high entropy means low redundancy. An English sentence such as a common greeting (e.g., “Hello”) is highly redundant, whereas a completely entropic, unpredictable message would be unusual in ordinary usage. The English language is estimated to be about $50 ext{%}$ redundant.
Redundancy serves practical and social functions:
Technical role: redundancy helps accuracy of decoding and serves as a check against errors. For example, spelling errors can be caught because of redundancy (e.g., context and expectations). If a language were non-redundant, a single letter change could create an entirely different word, making errors harder to detect. The context contributes to redundancy since language is not equiprobable and relies on convention.
Social role: redundancy helps manage communication with diverse audiences and noisy channels. More entropic messages may be used when addressing a specialist audience, while more redundant messages aid broad, less homogeneous audiences (e.g., popular art and advertising).
These concepts are interwoven with convention: convention tends to reduce entropy and increase redundancy by imposing standardized forms, patterns, and expectations. Rhythmic patterns in poetry, or syntactic constraints in a sonnet, reduce the number of possible next words, increasing redundancy and facilitating decoding. Conversely, breaking conventions can be entropic in the short term but may later establish new conventions and thus paradoxically increase redundancy for future audiences.
Phatic communication is a further extension of redundancy into social function: greetings like “Hello” carry little new information but maintain social bonds and relationships. Phatic communication is highly redundant and audience-centered: it helps hold a community together even though it doesn’t convey new information. The idea connects to Jakobson’s notion of the social function of language and to subcultures that are defined partly by conventional art forms and contained redundancies (e.g., folk songs).
In art, the balance between entropy and redundancy relates to audience accessibility. Popular art tends to be more redundant; highbrow art can be entropic but often becomes conventional as it is learned and accepted. For example, the Impressionists initially faced resistance, but as conventions formed around their style, redundancy increased within that new framework.
An analysis prompt often used is to examine an image (Plate 1a) and determine if it is entropic or redundant. The photography may appear conventional (redundant) at first glance but can carry entropic readings when interpreted against social conventions (e.g., police as aggressors vs. victims). Editors may increase redundancy by providing captions or context (Plate 1b) to align reader interpretations with conventional views, thus achieving both technical decoding ease and social bonding among readers.
A related social function is that redundancy supports audience stability and alignment with prevailing norms, often reinforcing the status quo. Entropy challenges audiences and can provoke change, but it is harder to communicate effectively.
Channel, medium, and code: how messages travel
Two additional key concepts that complement the model are channel and code, and they relate to what we call the medium. The channel is the physical means by which the signal is transmitted (wire, air, light, sound, neural pathways). The medium refers to the technical or physical means of converting a message into a signal that can move through the channel. The medium can be categorized into three main groups:
1) Presentational media: the voice, the face, the body. They use natural language and require the presence of the communicator; they produce acts of communication in the here-and-now.
2) Representational media: books, paintings, photographs, writing, architecture, interior design, etc. They create a text that can be read or interpreted independently of the communicator.
3) Mechanical media: telephones, radio, television, telexes. They transmit both presentational and representational content, but are subject to greater channel constraints and level-A noise.
Categorization is a heuristic; media can leak into each other and be merged for analysis. The relationship between medium and code is reciprocal: the physical properties of channels constrain which codes can be transmitted, while the codes determine how messages are encoded for a given medium.
A code is a system of meaning shared by a culture or subculture. It includes signs (physical signals) and rules that govern how those signs are used. The same code can be transmitted through different channels by various secondary codes (e.g., Morse code, semaphore, Braille, handwriting). The channel shapes which codes are admissible, while the medium constrains the range of codes that can be used.
A notable study by Katz, Gurevitch, and Hass (1973) used a circular model (Figure 3) to map the interrelationships among the five main mass media, based on audience needs. They found that audiences perceive neighboring media as most similar and use media to satisfy different needs (e.g., newspapers, radio, and TV for social connection; books and films for escaping reality). Education level correlates with media choice, and the researchers produced a table (Table 1) linking audience needs to media preferences (B=books, C=cinema, N=newspapers, R=radio, T=television).
The relationship between code and channel is crucial: primary codes (verbal language) map onto channels with constraints and may be transformed into secondary codes to fit the channel (e.g., semaphore or Braille). Medium and code boundaries overlap; dress, for example, functions as both a practical and communicative artifact. Clothing codes reveal how material culture serves communicative aims, and the same principle applies to houses, cars, and furniture, where design communicates social meaning in addition to fulfilling a functional purpose.
Feedback: a cybernetic view of communication
Feedback is the return of the receiver’s reaction to the sender, enabling adjustments to the message. Although Shannon and Weaver did not emphasize feedback, later theories use feedback to describe a cybernetic loop that improves communication efficiency. In practice, feedback is easier to achieve in interactive channels (two-way radios, telephones, face-to-face), while mass media impose constraints that limit feedback (the BBC cannot be constantly fed back to by the entire audience). In face-to-face communication, simultaneous transmission and reception (speaking and listening) is possible, whereas some channels limit access to feedback. Good speakers are sensitive to feedback; overly domineering speakers may filter it out. Feedback makes recipients feel involved and can reduce frustration and noise by signaling reception and agreement or disagreement.
Suggestions for further work
1) Apply Shannon and Weaver’s levels A, B, and C to diverse examples (e.g., a job interview, a news photograph, a pop song) and evaluate how widely they apply and how useful the analysis is.
2) Examine the problems of transferring level A information concepts to level B (meaning); discuss whether meaning can be numerically measured. See references in the text (Smith 1966; Cherry 1957).
3) Explore the claim that the English language is about $50 ext{%}$ redundant; analyze how redundancy affects comprehension and speed of communication. See Cherry (1957) and Smith (1966).
4) Outline the main communicative functions of redundancy and consider how convention facilitates understanding.
5) Discuss how writers/artists break or extend conventions and how this affects communication and audience reach.
6) Consider various media (book, photograph, record, live play, film) and categorize them as media; consult Guiraud (1975) for foundational ideas.
7) Analyze multiple media and channels; discuss whether there is a significant relationship between medium and channel or if they operate independently.
Publication history and series context
First published in 1982 by Methuen & Co. Ltd. Second edition published 1990 by Routledge. This edition also exists in e-formats (e-Library, 2002) and print (pbk).
This book is part of the Studies in Culture and Communication series edited by John Fiske, which includes titles on popular culture, television, language and society, and related fields. The series in the same catalogue includes: POPULAR CULTURE: THE METROPOLITAN EXPERIENCE; UNDERSTANDING RADIO, SECOND EDITION; ADVERTISING AS COMMUNICATION; UNDERSTANDING TELEVISION; UNDERSTANDING NEWS; TEXTUAL POACHERS: TELEVISION FANS AND PARTICIPATORY CULTURE; AN INTRODUCTION TO LANGUAGE AND SOCIETY, SECOND EDITION; and others listed in the front matter.
Foundational principles and real-world relevance
The model emphasizes that communication is shaped by technical constraints (channels, noise), semiotic conventions (codes and signs), and social/cultural factors (audience expectations, phatic functions, subcultural practices).
It shows how form (redundancy and convention) affects decoding and audience reception, and how content interacts with social relations and power dynamics (e.g., dominant media frames can suppress entropy and reinforce the status quo).
The interplay of channel, medium, and code helps explain why different media convey similar messages differently and why audiences respond to media in diverse ways depending on their needs and contexts.
The discussion of phatic communication, audience-centered redundancy, and subculture membership connects communication theory with sociology and cultural studies, illustrating that media analysis is as much about social relations and power as it is about messages and signals.
Key symbols and formulas (recap)
Information on Level A (signal predictability): where M is the number of equiprobable signal outcomes. Example: ;
Redundancy and entropy (qualitative): Redundancy increases decodeability and reduces entropy; Entropy increases with the unpredictability of signals. A language is said to be about $50\%$ redundant: In formal terms, redundancy can be related to the fraction of information that is predictable given context: where and for M equiprobable outcomes.
Channel, medium, and code relationships are described qualitatively, with secondary codes (e.g., Morse, Braille) used to adapt messages to various channels and media.
These notes summarize the major and most of the minor points in the transcript, with emphasis on how Shannon and Weaver’s framework has shaped subsequent thinking in communication studies, how redundancy and entropy shape message design and reception, and how channels, mediums, and codes interact with social context to produce different meanings and effects in real-world communication.