Evolution of Symbol Conventionality

Vasily Ogryzko

1. Axiological approach to conventionality

One of the most intriguing properties of semiotic phenomena is that one and the same information can be carried by signs of completely different material nature. So, at least, in some cases the material nature of a sign has absolutely nothing to do with what it signifies. This "conventionality of sign" is particularly striking in cases of human language and the genetic code.

In both of these semiotic systems, we have strings of discrete elements, combining independently with each other, which have no resemblance or physical (material) relation with whatever they are supposed to mean.

It is intriguing that twice in the process of its evolution, nature developed this characteristic independence. It is tempting to think that this reveals some deep principles of evolution or of the functioning of semiotic systems.

On other hand, in both "human" and "biological" cases we see the multitude of material forms that the same information can take: DNA, RNA, protein in an organism; speech, written text, silicon chips in the case of human culture. I would like to demonstrate that some forms of material carrier are more "conventional" than others.

Since it is not clear what exactly we should understand by the meaning of a genetic text (is it a protein sequence, or phenotypic trait, etc.?), I propose an approach to sign conventionality which is independent of a theory of meaning. It relies upon the concept of the value of the sign (Sharov 1995). In other words, it is an axiological (pragmatical), and not a semantical approach. The value of a sign can be more objectively characterized than its meaning, and this concept can be applied to both human and biological cases.

We consider a semiotic system as a part of a natural system that uses symbolic information for its survival. The fitness (value) of a text, therefore, can be objectively estimated, comparing the frequency of observing it to all other similar texts. We observe, for example, that "Don Quixote" by Cervantes can be found in any library in the USA, whereas most other "texts", which by all statistical parameters are undistinguishable from "Don Quixote", cannot. Then we can ask whether the value of a text, expressed in particular material form, depends on its material differences with other texts expressed in the same material form? If it does not, then this material form is conventional.

2. Levels of conventionality

In the cases of both genetic code and human language, different material carriers of information vary in the above mentioned aspect. For simplicity of argument, I compare only two different forms for each case, using them as convenient reference points to emphasize a tendency which actually covers the whole spectrum of various material carriers. I call one form language-memory (DNA, written text) and the other one language-action (protein, speech).

Both of these languages code something. Even in the case of proteins, one can say that the function of an enzyme (transformation of substance A into substance B) is its meaning. In other words, the presence of a protein "means" that A = B. However, there is a coding asymmetry between language-memory and language-action. The former is used (was designed, or evolved) to code the latter, and the opposite is untrue (a hierarchical treatment of this relationship is more appropriate, see Ogryzko 1995).

Second, language-action is more active, more involved in practice. For example, the spoken word is forced (imposed) upon us; we do not need to look for it in a book. It is also more essential in direct interaction between people and in the process of thought. Protein's main function is an active enzymatic transformation of substances. Language-memory, for its part, is more passive: a text needs to be read actively to make a difference; it serves as an information repository. For this reason it is more stable, less transient than language- action.

Finally, language-memory is more conventional. In the case of DNA all the apparatus of genetic enzymology (replication, recombination, repair, etc.) seems to be designed to treat all genetic texts regardless of their particular sequence, and therefore of any material (in this case, physico-chemical) differences between them. The very existence of viruses is evidence for this formal treatment of any nucleotide sequence, since it allows selfish DNA (RNA) to use the host apparatus for its own needs.

To the contrary, the value of a particular protein sequence undoubtedly depends on its physical properties, since it directly defines the function of the protein molecule.

One can see a similar pattern in the case of human language. Material restrictions on the spoken word can never be eliminated. First, not all combinations of sounds can be pronounced. Therefore, the values (probabilities to find) of particular sound combinations have partly material origin (e.g., in the organization of the speech apparatus). Second, sounds themself possess some elements of meaning (e.g., in the opposition "bing-boong", which word corresponds to a heavy-slow-large object and which one to a light-fast-tiny one?). On other hand, written text is designed to be free from all these restrictions. Formal and computer languages can be viewed as an ultimate result of this trend, which started with the appearance of written text. Again, computer viruses are evidence for this formality.

3. Why conventionality?

It is easy to speculate that the evolutionary trend towards conventionality is just a byproduct of the search for more stable and convenient information carriers, which by definition will have a new material nature. I think, however, that the very emergence of the phenomena of life and consciousness were at stake here. Symbol conventionality is a necessary prerequisite for digitality.

Two main breakthroughs in the global evolution of nature: the emergence of life and of human consciousness, were associated with a new semiotic system (genetic code, human language) which both have a developed digital component. The digital nature of information processing and storage implies: 1. Independent combination of symbols with each other; 2. A mechanism for selection of the essential ("meaning" of a symbol) from nonessential (noise). Recognition phenomena in molecular biology is the basis for this mechanism in the case of the genetic code. In the case of human language, the ability of the brain to distinguish between sounds of speech, depending on what is the native language of an individual, is well established.

The digital nature of language and the genetic code was of crucial importance in the two "big bangs" which created life and human consciousness. The independence of recombination between symbols allowed the exponential compression of information and created a new space of possibilities, necessary for the evolution of complex systems. Stability of this space and therefore the maintenance of the complex patterns necessary for the existence of life and consciousness would be impossible without the filtering mechanisms of information storage and processing.

Thus, the role of digitality is in providing freedom of combinations between stable units, thereby creating a new space for evolution. To ensure this freedom entails that the value of a particular sequence should not depend upon its physical (material) characteristics, since all combinations should be represented without bias. This, however, is synonymous with our definition of conventionality, confirming the idea mentioned in the beginning of this section.


Ogryzko, Vasily 1995 Digital and nondigital information in genetic langauge, in: Rauch, I. and G.F.Carr (eds), Semiotics Around the world. Synthesis in diversity, Berlin: Mouton de Gruyter.
Sharov, Alexei 1995 Value and meaning in biosemiotics, in: Rauch, I. and G.F.Carr (eds), Semiotics Around the world. Synthesis in diversity, Berlin: Mouton de Gruyter.