But “the exception-ally rich repertoire of definitely coded meaningful units (morphemes and words) is made possible through the diaphanous system of their merely differential components devoid of proper meaning (distinctive features, phonemes, and the rules of their combinability). These components are semiotic entities sui generis. The signatum of these entities is bare otherness, namely, a presumably semantic difference between the mean-ingful units to which it pertains and those which ceteris paribus do not contain the same entity” (1968:15). It would then be more fruitful to call those systems sui generis simply systems, reserving the name code for the correlations between the elements of two different systems. But fre-quently Jakobson speaks of codes in both cases (see, for instance, Jakob-son 1970).
The reason is, I think, rooted in the basic concrete attitude that Jakobson (faithful to his phenomenological inspiration) has always showed. The notion of a purely distinctive and differential system is a rather abstract one and could be considered in isolation only from the standpoint of an ‘algebraic view’ such as that of Hjelmslev. The main object of all of Jakobson’s research is, on the contrary, language in ac-tion. The langue is a theoretical tool useful for explaining why and how langage works. Therefore Jakobson cannot think of a phonological system (or of any semiotic analogue of it) as anything other than something de–, signed for signification. People do not invent phonemes in order to utter them without any intention of signifying (nor in order to contemplate the system without using it); a phonological system takes its form in order to compose words (endowed with meanings and therefore ruled by a code in the full sense of the term).
A l’origine du langage phonique ne se trouvent pas des associations d’élements dépourvus de sens qui présentent par la suite un sens ou sont charges de sens. A l’origine se trouvent bien au contraire des associations de sons qui repoivent leur forme spécifiquement linguistique précisément en vue d’une fonction de signification et qui ne peuvent être définies sans recours à cette fonction de signification. . . . Un phoneme est défini par sa fonction de signe. (Holenstein 1974:96, 202)
Thus, playing on this double sense of code, Jakobson has renounced an emphasis on a sharp methodological distinction in order to preserve the unity of language in action. In many authors who have been inspired by Jakobson, this sense of concreteness has been lost, and there has remained only a sort of imprecise oscillation between two linguistic us-ages of the word.
5.4.3. Semantic s-codes
Structural semantics studies the way in which the content of one or many languages is segmented according to certain criteria of pertinence. In this sense, many of the codes studied by cultural anthropology (kin-ship, culinary, myths) are semantic structures made up with oppositions such as raw vs. cooked, nature vs. culture, male partner vs. female part-ner, and so on. The organization of the content of a given culture and the organization of a portion of the content common to many cultures are s-codes. Let us consider the system of kinship and take into account a triple set of properties: (a) generation hierarchy (Go being the Ego-parameter, G + 1 the individual who generated Ego, and G — 1 the individual generated by Ego); (b) sex (with the opposition male vs. female); (c) lineage. We shall obtain a matrix that can be further ex-panded (Figure 5.1). Such a matrix can analyze all the relationships within the system of kinship even though we have no means for express-ing certain positions, that is, certain content units made up with com-bined features of the system.
It happens that English has a linguistic expression for each of the above positions (from I to 9, Grandfather, Grandmother, Father, Mother, Brother, Sister, Son, Daughter, and Uncle), but there can be a civilization where for two items from the matrix there is a unique term (by the way, English too is able to designate both positions 7 and 9 by the unique term sibling) and a civilization which has either no names, or more names, for one position. Different linguistic codes can correlate different expressions to each of the positions above; however, the sys-tem, the s-code of the kinship positions, remains unchanged through different cultures.
At this point, the difference between a language (as a code) and an s-code is clear: a language correlates the units of an s-code taken as the expression plane to the units of another s-code or more taken as the content plane.
A language is a code because it is, in the first instance, a correlational
123456789 etcetera
(a) Generation
G + 2 + +
G+1 + + + Go + +
G-1 + +
G-2
(b) Sex m
f
(с) Lineage
L1 L2 L3
- class="wp-block-list">
- + + + +
- + + +
- + + + + +
- + +
FIGURE 5.1
device. Is every correlational device a language? Is a language only a correlational device? What is the difference between correlational and institutional codes? Are there correlational codes which are different from a language? What is the difference between an institutional code and an s-code? Such are the questions which should be answered in the following sections.
5.5. Cryptography and natural languages
5.5.1. Codes, ciphers, cloaks
The most elementary example of a correlational code is a cryptographic one. In cryptography a code is a set of rules which transcribe a plaintext (in theory a conceptual content, in practice a sequence expressed in some semiotic system, be it linguistic or else) into an encoded message so that the receiver knowing the transcriptional rule can map backward from the encoded message to the plaintext.
Transcriptions can be realized either by transposition or substitution. Transposition does not require specific rules, except a sort of meta-instruction warning that the encoded sequence has to be worked out in order to find again the original order of the expression; typical examples of transposition are anagrams and palindromes. Substitution is allowed either by ciphers or by cloaks.
A cipher substitutes every minimal element of the plaintext with the element of another set of expressions; for instance, every letter of the Latin alphabet with a number, or with a letter of the Greek alphabet, and so on. Ciphers are clearly working upon the expression-planes of two different semiotic systems. The Morse code is a cipher.
A cloak makes entire strings of a given content correspond to the strings or to the units of another semiotic system. A cloak can work from content to content; for example, a cloak establishing that the encoded message /the sun also rises/ means «the D-day will be tomorrow» could work very well even though the expressions were written or spelled out in Chinese or in French. There are, on the other hand, cloaks working from expressions to content: an English dictionary is a cloak of this type (/bachelor/ «unmarried male adult», where the definition could also be expressed in French without changing the correlational rule); a bilingual dictionary makes the expression of a given language correspond to a con-tent expressed in a second language (definition) or to another expression of the same second language taken as absolutely synonymous (if any).
The boundaries between ciphers and cloaks are frequently imprecise. To which category belongs, for instance, the following code invented in 1499 by Thritemius?
A- In the Heavens
В Forever and Ever С- World without End
D In an Infinity . . . (and so on)
With such a code /bad/ can be translated as «Forever and Ever, In the Heavens, In an Infinity». It represents an expression-to-expression cipher, but it works also if the sentences are translated into another lan-guage (as a matter of fact, the original code was in Latin); therefore it seems to permit also an expression-to-content mechanism.
In any case, so far we can say that there is only a category of codes which are blatant instances of pure correlation, that is, ciphers matching an expression to another expression, as the Morse code. We can call these ciphers substitutional tables. Being uniquely correlational, they do not imply any interpretation, and they instantiate a case of minimal semiotic level. We shall see, however, in the following section, that, except for the Morse code, certain codes used by secret agents, and alphabets (where a given graphic signs correspond to a given sound), substitutional tables have a very restricted and nearly theoretical domain of usage. Every true code always correlates an expression to a series of
contextual instructions and triggers inferential processes (interpretation).
5.5.2.Prom correlation to inference
One can say that even substitutional tables involve some inference. In a Animal cipher, p is equivalent to q, but only if p is considered the token of a type belonging to the expressive plane of a given code a. If by chance the code were β, then p would be the token of another express-ive type and would represent the expression of a different sign-function (see Eco 1976, 2.1.)· Such an introductory choice represents a case of overcoded abduction. In the same way, to recognize the written letter /e/ as the equivalent of one sound (or more!) in English implies a certain abductive labor; in Italian the same letter would correspond to a differ-ent sound.
Beyond this unavoidable inferential character of any communicational approach, there are cases in which a seemingly correlational cipher is in fact intermingled with inferential instructions. Consider, for instance, the case of a computer receiving instructions in a binary language accord-
ing to the following cipher a:
Character 0 00 0000
1 00 0001 2
3
4 00 0100 5 00 0101 6
7 00 0111 8 00 1000
9 00 1001
Zone Numeric
000010 000011
00 0110
The operator programs instructions in a ‘numeric operation code system’
that we shall call code β:
00
01
02
03
Unconditional Jump
