The word "code" comes from the Latin "codex" or "caudex", meaning "the stock or stem of a tree, a board or tablet of wood smeared over with wax, on which the ancients originally wrote" (Webster's Revised Unabridged Dictionary). Later code came to mean a systematic collection of laws, such as that formed by the Roman Emperor Justinian, or the French Civil Code or Napoleonic Code of 1804 (Oxford English Dictionary). In these cases a code becomes a mechanism by which types of behavior can be restricted and regulated into acceptable norms, a set of injunctions prescribing ways to dress, act, or conduct oneself, as in a dress code, code of honor, or ethical code. Although semiotic systems such as ethics and clothing can constitute sites of positive meaning or expression, it is important to note that in this sense the "code" only refers to its prohibitive function. Codes determine a field in which certain meanings are permissible, but cannot be used in themselves to generate an original message, (disregarding for the moment the metal-linguistic message of transgression, of deliberate violation of a code). A code, so defined, cannot properly be said to be a medium. It acts upon a medium, such as forms of dress or societal behaviors, but does not serve on its own to manifest any sort of meaning.
Similarly, a code can also be the means by which a previously existing language is either obscured or clarified. The function still remains "metalinguistic", acting upon a medium rather than through it. When a communicative act brings attention to the code upon which communication depends, "its function is said to be metalinguistic" ( Glossary of Semiotics , 64). A code can be used, by the military for example, to create arbitrary correspondences between signs, intended solely to render them illegible to any without knowledge of the code. A coded message in this sense can be "decoded", or understood once the elements are sorted or transformed back into their original condition. This still does not constitute a medium, or a language in itself. A verbal language is understood in the Encyclopedic Dictionary of the Sciences of Language as containing three properties: "(a) we can use it to speak of the words that constitute it and, more importantly still, of other systems of signs; (b) we can produce sentences that disallow not only denotation but representation as well...(c) we can use words in a sense previously unknown to the linguistic community and make ourselves understood by means of the context (for example, in using original metaphors" (104). A code, according to Ducrot and Tudorov, only presents the first characteristic, the ability to refer to elements within other signifying systems. It can denote but not represent, nor can it generate new meanings through original use. Such actions can be performed in code, but not by code--a cipher bears only upon the message's legibility, and not its meaning.
Roman Jakobson in his essay "The Speech Event and the Functions of Language" also separates "code" from "message" (73). When language users turn their attention to the code during speech, they are performing a metalingual function, making sure that their messages are comprehensible to one another, to "check up whether they use the same code" (75). Language requires a shared code, but a code in itself does not constitute a language. Codes may obstruct or facilitate the transfer of meaning, but meaning never originates in the code itself.
So far these examples of codes have corresponded to the two "senses of code" "encountered most frequently" "in semiotic texts", according to the Glossary of Semiotics: "In one sense, code means a set of rules prescribing how to act or what to do, and in another, a key (or set of instructions) for translating a message" (64). In both definitions, a code is a way to "read" information (i.e. a legislative code "reads" certain behavior as acceptable or punishable), or in the second instance semiotic information that must be "decoded" to become intelligible. Codes have been capable of recombination and restriction, but as of yet have been restrained from the sphere of semiotic generation--the representation of concepts in themselves, without the mediation of a proper language. Even "dress codes", for example, operate selectively from a pool of already existing representations and connotations--they may endorse or prohibit, but never signify. Moral codes, similarly, can enforce or demarcate within an already existing system of ethics, but they can offer no independent justification for why their precepts should be followed. Information is always lost or disguised in the process of encoding, never generated. To "break" a code is to shatter a layer of obfuscation or distortion, but no meaning is ever lost once the coded message is deciphered, and the code disappears. The significance of a coded transmission never exceeds that of the message it encodes.
Jakobson's conception of a linguistic "code", however, already offers a hint of how codes can be construed as a medium in themselves. A "code", taken to be a way of making sense of utterances within a linguistic community (Umberto Eco also conceives of language as a multiplicity of codes (48-150), draws from Ferdinand de Saussure's analysis of the linguistic sign as an arbitrary relationship between a signifier (sound pattern) and a signified (concept). Saussure never uses the term "code", but in Jakobson it seems to mean the system of signs allowing language users to consistently reference signifieds by means of their signifiers. As Saussure points out, however, the elements of a given linguistic entity are meaningless apart from their relation to one another (144). It is not a coding of previously existing signifieds in terms of predetermined signifiers, but rather a determination of meaning by setting a given linguistic sign apart from those that surround it. Just as sounds have meaning only when placed in a field of relation with other sounds, "no ideas are established in advance, and nothing is distinct, before the introduction of a linguistic structure." (155). A language "code", then, cannot be seen merely as metalinguistic means of distinguishing meaning, but rather a system that creates meaning at the same moment it allows it to be expressed. A linguistic "code" is a medium through which signs are created.
Computing and genetics also offer two cases where the definition of "code" cannot contain the operations they enable, without admitting to it the functions of a medium. Judging from the Oxford English Dictionary, the first uses of the word "code" in the context of Cybernetics in the 1940's suggest an encryption of instructions to a computer, a code in the traditional sense: "Nature 26 Oct. 568/1 The brains of the machine lie in the control tape, which is code-punched in three sections...1948 Electronics Sept. 111/2 Orders to various parts of the machine..can be expressed conveniently as numbers in some arbitrary code" (emphasis mine). The computers are imagined to make use of coded instructions in the same way that German U-boats might make use of coded commands sent through the enigma machine. The difference that this analogy lets pass, however, is that the computer never "decodes" its instructions into its original, human, language. Rather than existing as a screen of distortion to be added and removed by sender and receiver, the "code" in cybernetics becomes a "programming language" used to "distinguish instructions from data" (The Free On-Line Dictionary of Computing). The message does not emerge on the other side of decryption in its original form, but is used as instructions by the computer to mediate between the user and a deeper level of encryption, performing operations another step removed from the original system of representations. It becomes easy to imagine the point at which the code used ceases to become "arbitrary", indeed, it becomes dictated by the demands of the second level of mediation, that between the code and the data itself. The one-to-one correspondence between a line of code and a command in a "natural" language breaks down, and the signifier of code and signified (on the level of manipulation of data) form representations and signs independent of the natural language--machine language level of encryption. Interestingly, the OED lists the definition for the verb "code" in Cybernetics (1.c.) as "To put into the form required by a code. " This phrasing suggests the code has ceased to be "arbitrary", and has taken on the inflexibilities of an independent semiotic system. [see information]
This problem is even more clearly illustrated with the "code" of genetics. Instead of a consciously generated code becoming independent through a process of evolution and self-mediation (computer language changing according to its own internal logic), the sequences of RNA used to generate sequences of amino acids can only be called a "code" by force of analogy--an assumption that their arrangement signifies a kind of one-to-one correspondence that can be deciphered and reduced to a natural language (i.e. such and such a gene sequence "codes" for Parkinsons), or to use the an OED example, "Sci. Amer. Sept. 52/2 There is a locus on one pair of homologous chromosomes that codes for eye color." Even more strikingly than with cybernetic code, genetic code eludes systematic representation on any but its own terms. It is, to state the matter simply, not a code to be broken, but a language to be learned. Again, to turn to the OED: "1964 G. H. HAGGIS et al. Introd. Molecular Biol. Viii. 212 The question of how a sequence made up from four different bases might specify unambiguously a sequence made up from twenty different amino acids is known as 'the coding problem'". If three long, three short, and three long signals in Morse code did not represent "S.O.S.", but rather the first paragraph of the Constitution, the use of the term "code" to describe the phenomena would undoubtedly be called into question.
The examples of genetics and cybernetics are instructive in that they show how the term "code" has come to apply to systems that exceed a merely prohibitive or metalinguistic function. While legislative or ethical codes can do nothing other than police the appearance of pre-established significations, computer and genetic codes can signify operations and interactions on a level irreducible to any language other than their own. As far as the study of media in general are concerned, these examples suggest that other "codes", such as those operating in linguistic or aesthetic systems, can be thought of not as referential, but generative, actively constituting meanings on a level beyond the strict correspondence of terms. "Codes can be artistic--not merely encryptions to be broken and swept away, but complex systems to be developed and explored.