The Lasswell and Shannon–Weaver Models

Story

1948: two texts, one word, opposite directions

In the same year, in two different worlds, two texts appeared, and both were about “communication”.

The first came out in New York, in a book collected from lectures given at a religion-and-society institute over the winter of 1946–47: The Communication of Ideas, edited by Lyman Bryson. In his chapter, the political scientist Harold Lasswell proposed describing communication with five questions: who says what, in which channel, to whom, with what effect? Lasswell’s formula At the heart of these questions are the content and the effect of what is said.

The second appeared in July and October of the same year, in two parts, in a telephone company’s technical journal, the Bell System Technical Journal. Its author, Claude Shannon, wrote in the introduction that messages often have meaning, but that “These semantic aspects of communication are irrelevant to the engineering problem”.

So one asks “what was said, and what did it do?”, while the other asks “leave meaning aside: how accurately does the message get carried?” You can read the first years of mass communication theory through the tension between these two sentences. This lesson introduces both, and also shows how both were later misread.

Story

Why did a telephone company leave meaning out?

Imagine a telegraph line with a telegrapher at each end. One day the line carries a company’s most heartfelt letter; the next day it carries signals produced by a ball rolling at random into one of three pockets. For the line there is no difference between the two; it has to carry both with the same care.

This thought comes from twenty years before Shannon, from R. V. L. Hartley, an engineer at Bell Laboratories. In 1928, in the Bell System Technical Journal, he looked for a measure of “information” based on physical considerations, and wrote that psychological factors had to be eliminated for this. In his example, a sequence of signals sent by an operator working the key of a submarine telegraph cable could just as well be produced by a mechanism driven by a ball rolling into one of three pockets; for the cable the two are the same.

Shannon inherited this legacy. His 1948 paper says its basis lies in the work of Nyquist and Hartley, and calls Hartley’s logarithmic measure the “most natural choice”. The engineer’s job is not to guess which message will be sent but to prepare the system for every possible message: which one will be chosen is unknown at design time.

So leaving meaning out was not indifference but a job description: whoever runs the line has to carry what is said without damaging it, whatever it says.

Card

Three names, and who is who

The three main names of this lesson are not much alike.

• Harold Lasswell (1902–1978): a political scientist. His Propaganda Technique in the World War, a study of wartime propaganda, appeared in 1927; in later years he became a professor of law and political science.
• Claude Shannon (1916–2001): a mathematician and electrical engineer; he joined Bell Telephone Laboratories in 1941.
• Warren Weaver (1894–1978): a mathematician, but above all a science administrator; from 1932 to 1955 he headed the natural sciences division of the Rockefeller Foundation.

Weaver was not Shannon’s co-worker but the interpreter who opened his paper to the general reader. In the book published in 1949 by a publisher in Illinois, two texts stand side by side: Weaver’s introductory essay and Shannon’s paper. The name “Shannon–Weaver model” comes from this partnership; but it is Shannon who drew the system diagram, and Weaver’s essay interprets and extends it.

A short timeline:

• 1927: Lasswell’s propaganda book
• 1928: Hartley’s measure of information
• 1941: Shannon at Bell
• 1948: Lasswell’s chapter (in a book) and Shannon’s paper (July–October)
• 1949: the Shannon–Weaver book
• 1956: Shannon’s “Bandwagon” editorial
• 1960: Wright’s functional analysis article

Card

Five questions, five fields of research

What Lasswell proposed is not a theory but a list of questions: a convenient way to describe an act of communication is to answer these five. Who? Says what? In which channel? To whom? With what effect? Lasswell’s formula

Imagine this morning’s newspaper headline. Who wrote it, what did it say, in which paper did it appear, who read it, and what did they do after reading it?

Each question is a door into a field of research:

• Who? → control analysis (the institutions and people who start and steer communication)
• Says what? → content analysis
• In which channel? → media analysis
• To whom? → audience analysis
• With what effect? → effect analysis

In textbooks this list often appears as a “model” drawn with boxes and arrows. But the purpose of the chapter is to organise the scientific study of communication; some theorists therefore prefer “formula” or “construct” to “model”. Lasswell’s own emphasis, too, is less on splitting the act of communication into parts than on seeing it within the whole of the social process; the next card shows this.

Note the last question as well: “with what effect?” is a question. Whether the effect is large or small, direct or indirect is not written into the formula.

Card

Three functions: how does a society inform itself?

Lasswell’s chapter has another idea as important as the list of questions: what communication is good for in society. In his account communication does three jobs:

1. Surveillance of the environment: reporting the threats and opportunities facing the community and its parts.
2. Correlation: coordinating the parts of society so that they respond to the environment as one.
3. Transmission: passing the social heritage from one generation to the next.

Imagine a storm approaching a coastal town. The weather warning and the wave reading at the pier are surveillance. The municipality, the fishermen and the school agreeing on the same precautions, and the “what should we do?” debate in the local paper, are correlation. The story elders tell children about “what happened last winter” is transmission.

The language of functions asks a different question from the five questions: not what individual messages do, but what communication does for a society. This is the language of the functionalist tradition: society is thought of as an interdependent, self-regulating system, and each part is asked what it contributes to the whole. The personal question “why am I following this news?” belongs to a later approach, uses and gratifications (Uses and Gratifications). The functions are treated in more detail in Functions of the Media.

A warning: Lasswell’s list has three items. “Entertainment”, often added as a fourth, is not his; we will see this in the deep dive the three social functions of communication.

Card

Shannon’s diagram: reproducing at one point a message chosen at another

Shannon opens his paper by stating the “fundamental problem” of communication: reproducing at one point, either exactly or approximately, a message selected at another point.

For this he draws a system Shannon–Weaver model:

• Information source: produces the message or sequence of messages (a sequence of letters in telegraphy, a time-varying sound wave in telephony).
• Transmitter: turns the message into a signal suited to the channel (sound pressure into electric current in telephony, letters into dots and dashes in telegraphy).
• Channel: the medium that carries the signal (a pair of wires, a coaxial cable, a band of radio frequencies, a beam of light).
• Receiver: does the inverse of the transmitter and rebuilds the message from the signal.
• Destination: the person or thing for whom the message is intended.

The drawing also contains a noise source, which mixes into the signal in the channel noise.

Weaver explains the same arrangement with human speech: when I talk to you, my brain is the information source, my voice mechanism is the transmitter, the air is the channel, the apparatus formed by your ear and the nerve attached to it is the receiver, and your brain is the destination. So although the boxes are borrowed from engineering language, they seem adaptable to human communication too. Shannon himself will look at this extension with caution; we will see this in the critique section.

Lab

Lab: measuring a bit with twenty questions

Let us play a game: think of a number between 1 and 16 and I will find it with yes/no questions.

First question: “Is it greater than 8?” Whatever the answer, the options are halved. Since every question halves the options, 16 options come down to one number in 4 questions (16, 8, 4, 2, 1).

Shannon’s bit is exactly this: the uncertainty of one yes/no answer. The word was suggested by J. W. Tukey; it is short for binary digit. Weaver likewise describes a situation with 16 equally likely options as 4 bits.

Note: here “information” is not the meaning of the answer but the number of options and their probabilities. Weaver’s striking example: you are choosing between two messages; one is the whole text of the King James Bible, the other is just “Yes.” Since information depends on the choice situation and not on the content of the message, this situation, if the two are equally likely, counts as 1 bit.

If the probabilities are not equal, the measure falls: if one message is chosen almost for certain, both the uncertainty and the information are small. In a desert where it never rains, the news “no rain today” carries almost no information; “rain today” carries a lot. (This last example is ours; the principle is Weaver’s.) For formulas, see the deep dive entropy.

Card

Weaver’s three levels: technical, semantic, effectiveness

In opening Shannon’s mathematics to the general reader, Weaver first draws a map: problems can arise at three separate levels of communication Weaver’s three levels.

• Level A, the technical problem: how accurately can the symbols be transmitted?
• Level B, the semantic problem: how precisely do the transmitted symbols convey the desired meaning?
• Level C, the effectiveness problem: how effectively does the received meaning affect conduct in the desired way?

Imagine a flood-warning SMS sent in a district. On some phones the message arrives with garbled characters: level A. The message arrives intact but the word “evacuate” sounds to some like “get ready”: level B. Everyone understands correctly, but nobody leaves home: level C.

Weaver shows how deep the semantic problem can go with his own example: the meaning a Russian viewer takes from a US newsreel picture. The effectiveness level he ties explicitly to propaganda theory; in his view communication either affects conduct or has no discernible effect. This view stands close to Lasswell’s “with what effect?” question (this is our interpretation).

Shannon’s theory belongs directly only to level A. Weaver’s claim is that the limits found at A also apply to B and C, and that the level-A theory can to a significant degree be counted as a theory of B and C too. This claim was much debated later (see the deep dives).

Card

What both silently take for granted

The two models ask different questions but live in the same house. Their silent assumptions are these:

• The message exists beforehand. The source chooses the message, then sends it. There is no picture in which the message is built jointly on the way.
• The flow is one-way. The boxes are lined up in sequence, the arrows go forward. There is no feedback in the main diagrams.
• Communication is an act. An event with a beginning and an end: sent, arrived, affected.
• Effect is the aim. Weaver says communication either affects conduct or has no discernible effect; Lasswell’s formula also reserves its last question for effect.

But the differences matter too:

• Lasswell thinks at the level of society: communication is the set of functions by which a society adjusts to its environment. This approach is associated with the functionalist tradition and brings forward the function of keeping the system stable.
• Shannon thinks statistically: information is defined not through the content of a single message but through the set of possible messages. In Weaver’s phrasing, information has less to do with what you do say than with what you could say.

Both describe well what the communicator does; the receiver is often left in a box as just the ‘destination’ or ‘audience’. Later interaction models (Schramm’s, for example) replaced these linear models for many purposes (Process Models: Schramm, Berlo, Westley–MacLean, Newcomb) linear transmission model.

Step by step

A telegram’s journey, step by step

Let us put a telegram message into the system and follow it step by step:

1. The source chooses. The source picks one message from the set of possible messages: “ARRIVING”.
2. The transmitter encodes. The letters are turned into a sequence of dots, dashes and spaces. This operation is called encoding.
3. The channel carries. The sequence of signals travels along the wire.
4. Noise mixes in. The line adds something the source did not want: static, a wrong dash.
5. The receiver decodes. The receiver does the inverse of the transmitter and turns the signals back into letters.
6. The destination reads. The message reaches the destination identical or close to what was sent.

The truly surprising result is hidden in what could be called step seven: Shannon showed that a channel has a capacity; that as long as the information rate produced by the source does not exceed this capacity, suitable codes can make the error as small as we like; and that if the capacity is exceeded, no code can bring the error below a certain lower bound. So noise does not make communication impossible; it limits how much can be carried.

One more note: in the discussion section of his paper Shannon himself writes that this result is a kind of existence proof, and that building good codes in a concrete way is a separate job.

Lab

Lab: measuring a telegram with two rulers

Now let us apply the two models to the same documented event. Try it yourself first, then compare.

The event. In early 1917 the German Foreign Minister Arthur Zimmermann sent a coded telegram to the German representative in Mexico: if the US entered the war, an alliance was proposed to Mexico. British intelligence deciphered the message and presented it to the US government on 24 February; the American press published it on 1 March, and the US declared war on 6 April 1917.

Task 1: Lasswell. Who: Zimmermann. What: a coded proposal of alliance. Channel: telegraph lines; Britain had cut Germany’s cables in 1914, Germany used lines borrowed from neutral countries, and the British listened in on these lines. To whom: the intended audience was the Mexican government; in practice it reached American public opinion. Effect: the aim was to keep the US busy on its southern border; the result was growing American anger.

Task 2: Shannon. Source: Zimmermann choosing the message. Transmitter: the process that encoded it and put it on the line. Channel: the line. Receiver: the decoding at the far end. Destination: the Mexican government. Noise? In the sources I read there is no detail about line corruption; saying you do not know is also a finding.

Task 3: The gap. Which box will you put the British, who listened to the line and decoded the cipher, in? Neither noise nor receiver: neither diagram has a box for them, yet they changed the course of the event.

Card

Five questions, five filing drawers

The ‘proof’ of a formula is not that it turns out true but that it works. Lasswell’s questions worked for sorting research: you can put many studies into one of five drawers.

• Control analysis: Who? Who owns the paper, under what pressures does the editor work?
• Content analysis: What is said? Which theme appears how often?
• Media analysis: What does the channel itself change?
• Audience analysis: Whom does it reach, who is watching?
• Effect analysis: What does the audience think and do?

Look at Lasswell’s own 1927 book with this eye and you see an interesting order: one chapter is given to the organisation of propaganda, a run of chapters to the themes propaganda worked with (war guilt, demonising the enemy, the illusion of victory…), and the last chapter to “The Results of Propaganda”. This is our reading; Lasswell does not present the chapters as answers to the five questions.

You also see the formula’s traces in later work: in 1958 Braddock added two questions (under what circumstances, for what purpose); in 1960 Wright turned the three functions into functional analysis. The Shannon–Weaver diagram, for its part, is described in a Turkish textbook as the foundation of the mainstream understanding of mass communication and as the model that brought the idea of noise into the process; the sequel is in Process Models: Schramm, Berlo, Westley–MacLean, Newcomb.

Story

A promise made, kept forty-five years later

In 1948 Shannon made a promise: even a noisy channel can be used with as few errors as you like, as long as the information rate stays below the capacity. The promise had a flaw: it did not show how. In the discussion section of his paper he himself writes that, apart from some simple cases, no explicit description of codes approaching the ideal had been found.

Practical codes approaching the limit were developed over time. In 1993 Berrou, Glavieux and Thitimajshima presented a new class of codes called “turbo-codes”; in terms of bit error rate their performance was close to the Shannon limit. Between the theoretical promise and the practical construction there were about forty-five years.

This is the theory’s real evidence: it works in engineering. Shannon’s obituary says that all communication lines are today measured in bits per second and that it was his notion of channel capacity that made this precise. But this does not mean “the theory also explains human communication”. Shannon warned explicitly in 1956: that a human being acts in some situations like an ideal decoder is an experimental, not a mathematical, fact, and must be tested in experiments.

In short: the evidence is very strong in one place (lines and codes), and in another place it is a question not yet asked (people and meaning).

Card

The linear-model objection: who said what, and on what grounds?

In textbooks, Lasswell and Shannon–Weaver are often put in the same bag: “linear, one-way models”. In one example from Turkish literature, the Lasswell model is described as seeing the target audience as passive and as well suited to propaganda analysis, while the Shannon–Weaver model is said to ignore people’s emotional state completely. The objections are usually these linear transmission model:

• No feedback: the receiver only receives, does not respond.
• A passive receiver: the audience waits at the end of the box.
• Meaning and context are left outside: especially for Shannon–Weaver, though Lasswell’s list of questions does not mention context either; the “under what circumstances?” question Braddock added to the formula in 1958 puts context forward.
• Communication = transmission: James Carey called this view the “transmission view” and contrasted it with a “ritual view” that sees communication as sharing and sustaining a common world (The Ritual View of Communication: Carey).

What do the defenders say? According to some authors, most of these objections are aimed at later presentations of the formula drawn with boxes and arrows; as McQuail and Windahl noted in 1993, until boxes and arrows are drawn, what you have is a formula. Sapienza and colleagues also conclude that, despite all the misreadings, the model is still relevant and useful today. On the Shannon–Weaver side, Weaver himself had proposed adding meaning boxes to the diagram (deep dive).

Story

Shannon’s own brake: the 1956 editorial

When a theory becomes popular, the first person to sound the alarm is sometimes its founder. That is exactly what Shannon did in 1956, in a one-page editorial in an engineering journal. Information theory, he wrote, had become “something of a scientific bandwagon”; from biology to psychology, from linguistics to economics, everyone was carrying words like “information”, “entropy” and “redundancy” into their own field.

His warning comes down to a few points:

• The basic results of the subject are aimed in a very specific direction; they need not be relevant to psychology, economics and the other social sciences. Its hard core is mathematics.
• Applying it to other fields is “not a trivial matter of translating words to a new domain”: it is a slow, tedious process of hypothesis and experiment.
• Keep your own house in order: rigour in publications, research before show.

Seven years earlier, Weaver’s essay had been more hopeful: it had argued that the level-A theory could also be meaningful for levels B and C. In one profile, Shannon is reported to have stressed that his theory could not address the question of meaning and was not designed for it.

So even inside the “Shannon–Weaver” approach that textbooks put in one box, there are two tones: the engineer’s caution and the interpreter’s hope.

Myth check

Was Lasswell a “hypodermic needle” theorist?

What people assume. Lasswell was the first theorist to see the media as a hypodermic needle (or magic bullet) acting directly and instantly on the audience. One Turkish textbook also writes that theories such as the “silver bullet” and the hypodermic needle were developed on the basis of his propaganda studies.

What the evidence says. There are two sides to it.

On one side, Lasswell does not underestimate the power of propaganda: in his 1927 book he calls it one of the most powerful instruments in the modern world.

On the other, the “results” chapter of the same book treats the effect not as a settled fact but as a contested matter: it says the precise effect of propaganda is largely a matter of surmise, that defeated countries exaggerate the influence of propaganda, that it is hard to separate the influence of propaganda from the other means of control closely allied to it, and that propaganda was present on both sides of the contested fronts. The 1948 formula, too, puts effect not as an answer but as a question.

Historiography also questions the story. According to Sproule, the standard story that describes interwar propaganda critics as “magic bullet” theorists is a myth that emerged after 1940. For the hypodermic needle model itself, see the lesson The Hypodermic Needle and Magic Bullet Model.

In short: Lasswell took the possibility of strong effects seriously; but the certainty of a “hypodermic needle theory” later attached to him is not in his texts.

Card

Today: Shannon in your pocket, Lasswell in your notifications

Shannon is in your pocket. A corner of a QR code is smudged and your phone still reads it: the reason is deliberately added redundancy. A QR code corrects errors by adding a Reed–Solomon code to the data; Denso Wave’s page offers four levels: roughly 7, 15, 25 and 30 per cent of the code’s codewords can be repaired. Shannon’s 1948 idea (redundancy left in a code protects against noise) has turned into a square on paper. A thesis also lists a great many technologies, from CDs and mobile phones to modems and MP3, as carrying Shannon’s ideas.

Lasswell’s five questions have not disappeared in the digital world, only got tangled. What follows is not a finding documented in sources but our own reasoning: in a viral post the ‘who?’ question multiplies into an account, a bot, a brand; the ‘in which channel?’ question is entangled with an algorithm’s ranking and therefore also determines ‘to whom?’; and ‘with what effect?’ remains the hardest question to measure. The formula is still useful as a checklist, but the answers do not fit into one box.

A small experiment: ask the five questions of the last three notifications on your phone. How long did it take to find the answer to ‘who?’ If a notification’s channel also decides which notification reaches you, is it enough to call the channel ‘just a carrier’?

Card

This is how exams ask it

Frequent patterns

• “Who does the formula ‘who says what in which channel to whom with what effect’ belong to?” → Harold Lasswell (1948).
• “Whose model has the elements source, transmitter, channel, receiver, destination and noise?” → Shannon and Weaver.
• Matching: Lasswell → five questions and three functions; Shannon–Weaver → bit, entropy, noise; Weaver → three levels.
• “List Lasswell’s three functions.” → surveillance, correlation, transmission.

Tempting wrong answers

• “Lasswell is the theorist of the hypodermic needle model” → a label attached later; the sources are more complicated.
• “Entertainment is Lasswell’s fourth function” → no; it was added later, generally attributed to Wright.
• “Weaver worked at Bell Laboratories” → no; the Rockefeller Foundation.
• “The Shannon–Weaver model measures meaning” → no; it keeps meaning outside the engineering problem, and Weaver puts meaning at level B.
• Years: Lasswell 1948; Shannon’s paper 1948 (July–October); the Shannon–Weaver book 1949. Sources may date the model 1948 or 1949; both rest on a fact.

Alternative names in Turkish course texts (useful if you read Turkish sources)

• Lasswell model, Lasswell formula, Lasswell’s general communication model; the spellings “Laswell” and “Lasswel” also occur.
• Shannon–Weaver: Information Theory (Enformasyon Kuramı), Mathematical Communication Model/Theory, bilgi kuramı.
• Parts: information (enformasyon) source, transmitter (verici), channel (kanal), receiver (alıcı), destination (hedef), noise (gürültü) source; redundancy (fazlalık), entropy (called “enformasyonda eksiklik” in some books).
• The five analyses: control (source), content, media (channel), audience and effect analysis.

Summary

Summary card

• 1948: Lasswell’s five questions (in a book) and Shannon’s paper (July–October) come from separate worlds, with opposite divisions of labour: one asks about content and effect, the other leaves meaning outside the engineering problem.
• Lasswell: who says what in which channel to whom with what effect; five fields of research (control, content, media, audience, effect analysis). Three functions for society: surveillance, correlation, transmission. Entertainment is added later.
• Shannon–Weaver: source, transmitter, channel, receiver, destination and noise; bit, entropy, redundancy; below channel capacity the error can be made as small as one likes.
• Weaver: technical (A), semantic (B) and effectiveness (C) levels; the semantic receiver and semantic noise proposal. Weaver is Shannon’s interpreter, not his co-worker.
• The hypodermic misreading: Lasswell counts propaganda as powerful but also writes that its precise effect is guesswork; the label was attached later.
• Critique: linear, one-way, no feedback, no meaning; Carey’s transmission/ritual distinction. In 1956 Shannon warned against overextending his own theory.
• Today: error correction from QR codes to modems; Lasswell’s questions still work as a checklist but do not fit into boxes.
• Exam: three functions (not four), the years 1948/1949, alternative Turkish names.