64 3. The Code Model Decoded
Despite distinctions between the information theoretic notion of information and the common notion, which includes meaning, many theoreticians have attempted to apply the
theory to semantics Chater 1994
:1687. Even Shannon’s colleague Warren Weaver seems to have struggled with the necessary exclusion of semantics. The fact that these
efforts have met little success should offer no surprise. As Langacker notes, “One has no guarantee that a seemingly apt metaphor will actually prove appropriate and helpful when
pushed beyond the limited observations that initially inspired it”
1991 :507. Shannon
himself commented upon such misapplications during a 1989 interview. As the inter- viewer John Horgan records: “Shannon himself doubted whether certain applications of
his theory would come to much. ‘Somehow people think it can tell you things about meaning,’ he once said to me, ‘but it can’t and wasn’t intended to’”
Horgan 1996 :
207– 208. Everett Rogers similarly comments: “Shannon saw information theory as limited
to engineering communication and warned the scientific world against applying it more
broadly to all types of human communication. Nevertheless, communication scholars have not paid much attention to Shannon’s warning”
Rogers 1994 :428.
3.2.3.1.2. Message
As may be anticipated, Shannon’s restricted definition of ‘information’ is closely related to his restricted definition of ‘message’. He does not elaborate on the technical
definition, only stating that “The significant aspect is that the actual message is one selected from a set
of possible messages” Shannon 1948
:379, 1949
:3. Shannon’s ability to statistically evaluate the success of a transmission required that the actual message
represent a possible message. Indeed, without this condition, it would be practically impossible for the transmitter to send a signal which corresponded to that message and
statistically impossible for the receiver to “narrow down” the possibilities involved in reconstructing a message from that signal. For example, if a woman wished to send, via a
Morse Code-based telegraph, a note to a friend in a distant city, the system could handle either of the first two messages, but not the third:
1 The characters “
XXXOOO
”
37
2 The characters and spaces “
I LOVE YOU
” 3 The shape “
” This is because Morse Code is not designed to recognize the heart shape and relate
that shape to a string of “dots and dashes.” But Morse Code does support signals that correspond to the orthographic conventions of the Roman alphabet. Note that these
particular constraints on the set of possible messages might not apply if the woman were to employ a different system of transmission. For example, if she had wished to employ a
facsimile machine, rather than the telegraph, the system could have easily handled any or all of the three messages she suggested. The point here is that each system of communi-
cation defines its own set of “possible messages.”
37
“XXXOOO” is sometimes used to represent “Hugs and Kisses.”
3. The Code Model Decoded 65
Note, as well, how the meaning the woman may intend through use of these symbols or arrangements of characters is totally irrelevant from the engineering perspective. One
or another of the messages may seem more fitting to her, but the system treats the message as raw information; it is only concerned that the message be selected from a set
of possible messages. This does not mean that the woman is limited to the respective subsets or sequential order of characters reflected in “
XXXOOO
” and “
I LOVE YOU
.” Rather, it means that the message must be assembled from a particular set of components,
which in this case would generally be the Roman alphabet and punctuation. A system may
limit the subset or sequential order of characters in its definition of “possible messages,” but such a requirement would be system specific. In the case of Morse Code,
the system will support a potentially infinite arrangement of characters, provided the message is assembled from a restricted set of possible characters. A system might,
however, be composed of only one switch, so that a closed circuit represents one message and an open circuit another. From the perspective of information theory, the sort of
meaning a reader may associate with such an on-off signal is irrelevant. The system is concerned only with the success or failure of the signal transmission and reception.
It should be evident at this point that Shannon applied a further restriction of the definition of ‘message’, and again his definition conflicts with common usage. Within
information theory, the message is not the material that is transmitted. Rather, the message is pre-transmission material. Strictly speaking, within information theory, it is
completely impossible to “send a message”; one can only send a signal. The following illustrations may help to explain the significance of this definition.
A woman has a message she wants to send to her boyfriend. At the telegraph office she scribbles the message on a piece of paper. The telegraph operator looks at it, then sends it. At
station in another city, a second operator receives the message then writes it out. The boy- friend stops by the office later to pick up his message and he is elated. Unfortunately, his dog
eats it that night, but no matter; he still savors it for weeks.
In the description provided, the term ‘message’ is used in its common usage. As was discussed in relation to the conduit metaphor, the common usage of ‘message’ involves a
significant use metonymy. The following version eliminates that metonymy and high- lights the definitions of ‘message’ within information theory as “primary message” and
“constructed message.”
A woman has a [mental abstraction] she wants to [share with] her boyfriend. At the telegraph office she scribbles [a primary message, i.e., an arrangement of ink spots] on a piece
of paper. The telegraph operator looks at [the ink spots], then sends [a corresponding set of electrical signals]. At a station in another city, a second operator receives the [somewhat
distorted, but mostly intact electrical signals] then writes [a constructed message, i.e., a corresponding arrangement of ink spots]. The boyfriend stops by the office later to pick up his
[paper bearing the telegraph operator’s ink spots, which correspond to a signal received, which to some extent reflects the signal sent, …] and he is elated. Unfortunately, his dog eats [the
paper and ink] that night, but no matter; he still savors [the mental abstraction he generated after observing the telegraph operator’s ink spots] for weeks.
66 3. The Code Model Decoded
From an information theoretic perspective, the semantic intentions of the woman are irrelevant.
The transmitter and receiver are only concerned with how the constructed message may be related to the set of possible primary messages in this case the set of
characters composing the English alphabet and the signal alphabet in this case the set dots and dashes defined by Morse Code.
For readers not directly concerned with the information theoretic perspective on communication, this distinction between message and signal may seem superfluous—a
“quibble over semantics.” Indeed, most linguists have handled it so, as the code model quotations presented in section
2.4 illustrate. This practice has reflected a grave
misunderstanding, for as will be seen, it is a very significant distinction. Indeed, this misunderstanding andor disregard for this distinction has been one of the major factors
contributing to the integration of the conduit metaphor with Shannon’s information theory.
3.2.3.1.3. Transmitter and receiver
