Shannon’s classical theory of communication

In his paper from 1948, The Mathematical Theory of Communication, Claude Shannon presents the foundation of classical communication theory and differentiates between three conceptual levels: the syntactic, semantic and pragmatic. As we will see, these levels can be adapted to all main concepts within information theory, such as noise, redundancy, etc.

The syntactic level, the data level, deals solely with the internal relation and linking between signs used. This includes rules for building up word and sentences in a formal manner. It is the study of the logic and grammar of sign systems. Syntactics concerns the physical form rather than the content of signs. The semantic level, the information level, is the level of application and of general understanding of the significance of signs used to relate to things, actions and the outside world. It concerns the meaning of the signs and the association between signs and behaviour. It gives the meaning of the signs for the system’s goal. The pragmatic level, the knowledge level, is the level of the user world, of the personal and psychological impact of communication. It concerns the role of the signs in controlling and regulating the function of the system in decisionmaking and makes practical applications from knowledge, hereby forming wisdom. Here questions of meaning, results and value for both sender and receiver are actualized. The attribute of meaning to a message can only be considered if the receiver is taken into account. Pragmatics thus is the study of mutual understanding. Empirics, finally, tells us about the physical characteristics of the communication media. This can be different communication cannels e.g. sound, light and electronic transmission.

A brief summary thus tells us that syntactics concerns the formalism used to represent the message, semantics the meaning of the message, pragmatics the issue of intentions and empirics the signals used to code and transmit the message. Figure 4.1 shows the interrelation between the three levels and the pertinent information processing.

Figure 4.1 Information processing in human communication.

Meaning and probability are related concepts. Probability is always probability to somebody. Thus, an interprétant is implied who somehow constructs the meaning. Every meaning is therefore unique for the human being interpreting it. The meaning cannot be forecasted by someone else. In the definition of meaning, it is assumed that both the source and receiver have previously coded (and stored) signals of the same or similar referents, such that the messages may have meaning and relate to behaviour. That is, the used symbols must have the same signification for both sender and receiver. If not, the receiver will create a different mental picture than intended by the transmitter. Meaning is generated by individuals in a process of social interaction with a more or less common environment. It is a relation subsisting within a field of experience and appears as an emergent property of a symbolic representation when used in culturally accepted interaction. The relation between the symbolic representation and its meaning is random. Of this, however, the mathematical theory has nothing to say. If human links in the chain of communication are missing, of course no questions of meaning will arise.

According to Shannon, all meaning is interpreted outside the transmission of signals and the transformation of signals into a “message” is left outside Shannon’s definition. His concept of information relates not so much to what is said as to what could be said (or is not said). Therefore, information must not be confused with meaning and has no value in itself. The value of information comes out mainly in connection with human action or as an indirect relation. The real value of a message is the amount of work which has been done by its creator and which the receiver is spared the effort to repeat.

The three conceptual levels of communication may be compared with the three stages in handling a spoken language, implying two intermediate translations between the outside world and the subjective reception of information (see Figure 4.2).

The first level consists of the acoustic pattern taken physically as vibrations in the air. The second level consists of the various phenomena in the inner ear and pertinent parts of the nervous system. The third represents the conversion of the information pattern by the brain into an experience of individual meaning.

The three conceptual levels of communication by Shannon are transformed into three related problems. The first is the technical problem: how accurately can the symbols used in communication be transmitted? The second is the representation problem: how accurately do the transmitted signs represent the intended message? The third is the efficiency problem: how efficiently does the received message influence the behaviour of the receiver.

Figure 4.2 The three stages in handling a spoken language.

In his work, Shannon explicitly states that the presented theory relates only to the syntactic level and the technical problem. The theory thus concerns the probability of the reception of certain signs under various conditions in the transmission system. Information is an entity regarded as neither true nor false, significant nor insignificant, reliable nor unreliable, accepted nor rejected. The coding of experience into a set of communication symbols and its recall after transmission by decoding, the very content, is irrelevant and outside the scope of the theory.

Although the representation and efficiency problems are irrelevant for the technical problem, the technical problem is highly relevant for the representation and efficiency problem. All calculations of representation and efficiency are dependent upon the precision in the technical. Whatever its form, the message has first to be received properly before the content can be perceived.

In Figure 4.3, the main concepts of Shannon’s theory are presented. The theory is completely general and the communication process is seen as a transaction between terminals with the sole aim of generation and reproduction of symbols. It may be applied to a person communicating to himself (writing a memo) or to an unintentional interceptor of a message. If humans form parts of the communication system, the mathematical theory is relevant only in the technical part of the system.

Figure 4.3 Concepts of Shannon’s communication theory.

Every message intended to be communicated to someone has a source. The message which consists of a sequence of symbols from a certain repertoire is forwarded to the transmitter and then sent through a channel to the receiver. The receiver is connected to the destination. In this process the information is first coded and then decoded. It is thereby assumed that both the source and the destination have previously agreed on a code with the same or similar referents, used in such a way that the messages have meaning and relate to behaviour. In reality, the restrictions of the used codes creates the reciprocal, intelligible message. The code may be generally known or secret to prevent interpretation of an outsider. Note that every message in this communication system has a sender and a recipient. A message without recipient is inconceivable per definition. Both transmission and coding take advantage of the spatial, temporal or other classificatory ordering of the elements carrying the information.

Not all information emitted by the transmitter in a communication system is obtained by the receiver. Here, significant degrading phenomena are damping and distortion. Also, the receiver attains some information that was never sent by the transmitter. The very existence of a channel with its noise transforms the message. Noise, which is always present in the channel, interferes with the transmission, degrading its quality to a greater or lesser extent. The total influence of noise can be measured as the resulting difference between the input and the output message.

In contrast to Shannon’s classical theory, modern information theory makes a distinction between channels of communication and channels of observation. This is based on the fact that communication requires at least two persons, while observations and measurements require only one (nature does not communicate to us with language or signs).

In the communication system presented in Figure 4.3, some critical conditions must be fulfilled for optimum performance.

  • According to its intentions, the information source must provide adequate and distinct information.
  • The message must be correct and completely coded into a trans­missible signal.
  • Taking into account different kinds of noise and the needs and aims of the destination, the signal has to be transmitted in a sufficiently rapid and correct form.
  • Received signals must, in spite of disturbances, be translated into a message in a way that corresponds to the coding.
  • The destination must be able to convert the message into the desired response.

Figure 4.4 Hierarchic levels of a written message in an electronic channel.

An examination of the communication itself during the transmission of a written message will show a general hierarchical structure of the process. Imagine the short electronic message, ‘remember today’s meeting’ mailed from one computer to another. To analyze this message, four levels will be used according to Figure 4.4.

Pressing the letter R on the keyboard activates a certain binary pattern generated by the computer, here defined as the zero level. This pattern is assigned to a particular letter or keyboard function, now interpreted on the screen as the letter R on the next level one. When additional letters are formed they build the first word on level two. Level three consists of complete sentences, eventually building a paragraph at level four (see Figure 4.4).

Being occupied only with the first two levels of the hierarchy, Shannon’s theory concerns the binary pattern representing the letters (coding). Effects of intervening noise occurring in the channel and the application of redundancy to neutralize the subsequent existing errors are also dealt with. The hardware is totally indifferent to the choice of letters used. The combining of these letters to create words is directed not by the hardware but by  the rules of the language and the linguistic customs of the human user who wishes to be understood. Words are then arranged into sentences according to the syntax of the language and these in turn may be further arranged in paragraphs to emphasizing and distinguishing different topics.

Yet one has to bear in mind that the words are poorer than our thoughts and our understanding is poorer than the words we use. Exchanges between different levels of knowledge are necessarily irreversible and they have different meaning for partners of different history. Several prelinguistic processes have, however, been completed before the specific message is chosen and typed out. The origin of human communication begins on levels far higher than those existing in Figure 4.4. The move from the highest descriptive to lower levels clearly shows how certain systematic changes occur.

Given his background as an electrical engineer, the restriction of Shannon’s work to the technical problem is natural. He must, however, have been aware of the extreme difficulties present in the development of a theory for the other two levels. A complete theory of communication must deal with the structure of the message and its encoding, communication, decoding, and understanding, of the inherent meaning. Finally, the theory must calculate how efficiently the message will contribute towards desired behaviour.

As we now understand it, classical information theory tells us as little as does the hardware used about the incorrect choice of words or use of a vague structure in a sentence. A vague structure may even be intended by the source and it should be possible to transfer every kind of message, even sheer nonsense (genuine or as used, for example, in cryptography). Shannon’s concept of information therefore excludes meaning per se; what is significant is only that the actual message is selected from a set of possible messages. How the source chooses its messages is irrelevant to the theory.

Although the concept of meaning lies outside formal information theory, the pragmatic aspects influencing information must not be forgotten. The concept of information is thus inseparable from that of meaning. Meaning is always the meaning for someone, defined in terms of the person or system receiving the message. Meaning is a relationship between the message and the receiver and no inherent property of the message itself. It is thus that meaning can differ from one receiver to another. We can only attribute a meaning to a message if the response of the receiver is taken into account.

Of the many attempts to define meaning in a more formal way, D. MacKay’s (1969) is one of the most useful. MacKay distinguishes between the following aspects of the term:

  • the meaning intended by the sender
  • the meaning understood by the receiver
  • the common meaning

He then defines the meaning of a message as its selective function on a specified set of responses, something valid for all the three aspects.

Every message is accompanied by certain implicit instructions for the receiver as to how to interpret it and relate to the sender. This kind of metamessage is always superior to the pertinent content of the message and is generally conveyed analogically, for instance by intonation, facial expression or bearing.

C. Cherry (1966) points out that the pragmatic qualities of a message are dependent on:

  • Earlier experiences by the sender and receiver
  • Present circumstances
  • Individual qualities

While Shannon’s theory mainly belongs to the area of communication, he presents a pragmatic view of the highly abstract concept of information. The transitory nature of information has traditionally made it an integrated part of the media used. Separating information from its material carrier is nevertheless a prerequisite for the understanding of its nature. Information is not dependent upon any specific technology for production, distribution and use. Only when separated from its technology will information become an adequate measurable entity.

Source: Skyttner Lars (2006), General Systems Theory: Problems, Perspectives, Practice, Wspc, 2nd Edition.

Leave a Reply

Your email address will not be published. Required fields are marked *