Original at

Ancient Images and New Technologies: THE SEMIOTICS OF THE WEB by Philippe Codognet

This paper tries to develop a semiotic analysis of computer-based communication, and especially human-to-human communication through an electronic medium. Semiotics is a very interesting and powerful tool in order to rephrase information theory and computer science and shed a new light on this global phenomenon. Along with semiotics, we will nevertheless also base our study on a more classical historical analysis, sometimes verging on historicism, to point out the deep roots of contemporary computer communication. We will therefore retrace the history of the ‘universal language of computer’, that is, binary notation, and link it to that of the ‘universal language of images’, that is, a long tradition in the history of ideas going back to Cicero’s Art of Memory and various Renaissance curiosities.

Computers are artifacts aimed at storing and manipulating information -- information being basically anything that could be algorithmically generated -- encoded in various ways. Information theory can be thought of as some sort of simplified or idealized semiotics : a cyphering/deciphering algorithm represents the interpretation process used to decode some signifier (encoded information) into some computable signified (meaningful information) to be fed to a subsequent processing step. As could be this process, semiosis is, of course, unlimited.

Communication between computers follows the same scheme. As data have to be transmitted through some external (usually analog) medium, a further encryption scheme (semiotic system) has to be devised and applied: the communication protocol. The current success of the World Wide Web protocol on the internet (http) is mainly due to its ability to manipulate images and sound in addition to simple alphanumeric text.

As humans communicate through this medium and exchange cultural signs, some problematic issues should be raised. The self is mutilated and disintegrated into conventional signs, in a deeper and much more dramatic way than oral communication.

The success of the Web goes with its semantic poverty, and is heading toward some "zero degree" of communication. The Entropy Principle warns us that we are converging towards a stable state where everyone will be connected and fully informed of a homogeneous and therefore null content.

Computers can compute and using binary notation for representing numbers is certainly of great interest, but there is nevertheless another key issue for making them able to process higher-level information. The first step was to code alphabetical symbols, therefore moving from the realms of numbers to the realms of words.

Nowadays, computers employ the ASCII encoding of letters and numbers that represent each character with 7 bits (or 8 for extended ASCII, which includes accentuated letters). Being able to handle numbers and letters, the computer soon became the perfect data-processing machine, the flawless artifact of the information technology age.

Another landmark is digitalization and binarization of pictures, which marked the opening of the realms of images to the computer. Computers might handle images composed of millions of pixels (‘’small dots’’) with millions of colors, requiring about 100 times more bits of binary codes.

Indeed, this ability to manipulate pictorial information proved to be the main reason for the current explosion of cyberspace and the internet. Without images, with human-to-computer and human-to-human (through computer) interactions limited to the alphanumeric set, electronic communication was circumscribed to computer professionals and a few crucial business/military applications.

It is therefore not surprizing that when computers came to the realm of images, a new dimension was added to Cyberspace (literaly indeed, from 1D to 2D) and then the term ‘’Virtual Reality’’ started to be more than a daydream. 

Pierce’s classical classification of signs as Icons, Indexes and Symbols, which is very useful in understanding the different ways in which signs operate and semiosis is performed. Let us take Arthur Burk’s presentation of this trichotomy :

" We can best do this in term of the following examples : (1) the word ‘red’, as used in the English sentence, ‘the book is red’ ; (2) an act of pointing, used to call attention to some particular object, e.g. a tree ; (3) a scale drawing, used to communicate to a machinist the structure of a piece of machinery . A sign represents its object to its interpretant symbolically, indexically, or iconically.

Web pages are so-called hypertexts, possibly linked to other (hyper)texts, and so on and so forth. The reader can navigate through the whole text in a non-linear manner, by activating so-called hot links or anchor points that are linking some piece of text to some other.

These links are an obvious example of indexes, with a word pointing to (referring to) its definition or to some related piece of information. However in order to act as an index, a sign has to be recognized as such, i.e. the index has to exhibit itself as a reference. This is done in hypertext by marking the hot links in blue ink, in order to make the reader aware that he can jump to another piece of hypertext or image, therefore using a conventional symbol in order to ‘’show’’ the index as such.

Web pages are usually full of small images that act as user-friendly and aesthetically appealing ways of navigating through the network. These are symbolic signs, in the sense that their object must be conventionally established in order to help the reader to orient himself in a homogeneous and unlimited cyberspace.

Shape of (waiting clock or time thing) somewhat represent we need to wait, the program is loading. Or a pointer arrow on the computer screen shows us where the mouse located in order to select an icon or a task. The arrow turns in to (I) cursor while you are on the word processing page. When we click on the page it becomes a blinking (I). On a web page, it becomes a pointed index finger. 

"…. the shape of the mouse pointer (the device used to move on the computer screen through the displayed text) changes when passing upon such a link, and it becomes ... a small hand with a pointed index. Thus, an index is used to identify an icon as such. But note also that an icon (pointed finger) is then used to identify the index as such. "

In general, all pages at one Web site (physical/logical place hosted by some institution) are homogenized in order to use the same symbols to designate basic moves in the hypertext documentation (usually at the top or bottom of the pages), in such a way that the reader can quickly learn their conventional meaning.

Symbols for hypertext links tend to become icons, as if it was their only means to get rid of the textual tautology.

A closely related shift from symbolic signs to iconic signs can be observed in electronic mail communication with the use of ‘’smiley’’ signs. These signs can be considered as conventional symbols from the cyber-jargon that are used within an alphanumeric text to express, in a very rudimentary manner, some personal emotions or personal facts. But they indeed become icons if the text is turned 90 degree. The range of feelings that could be expressed is rather limited on the web and this points out in a rather crude manner the poverty and standardization of the virtual communication towards which we are concretely going.

Another example of iconic signs currently used on the Web is given by the Sony’s ‘Cyberpassage’ software, which makes it possible for several users to wander in a common ‘virtual social space’ represented in real-time 3D. Users can communicate in this virtual world by two means : either written language in a small window shared by the people meeting virtually, or by using predefined expressions that can decorate their so-called ‘avatars’

The final example of E-Play’s home page, an Italian fashion company, is a good instance of a fully iconic Web page. No text duplicates the iconic links, which indeed ‘’talk’’ by themselves, that is, ‘’exhibit their objects’’. After careful reading and a bit of thinking, one could indeed identify links to a male fashion gallery, a female fashion gallery, a list of shops stocking the products, etc.

However pure iconography is not possible, as icons have to present themselves as such, to display their own icon-ness. A sign is not iconic until the interpreter recognizes it as such. To that purpose the shape of the mouse pointer (the device used to move on the computer screen through the displayed text) changes when passing upon such a link, and it becomes ... a small hand with a pointed index. Thus, an index is used to identify an icon as such. But note also that an icon (pointed finger) is then used to identify the index as such. As is well-known, pure iconicity is not possible.

As in all semiotic systems, we have seen that the web is a mesh of icons, indexes and symbols, with each type of the trichotomy indeed depending on the others, even for its own definition. Although the web in its essence and its success relies heavily on images, the dream of a 'perfect language of images' cannot be reified with this medium either. Pure iconicity is always fading away in the distance...

W. V. Quine’s famous counter-example showing the impossibility of learning a language from scratch. Imagine yourself as an adventurous linguist marooned in some unknown territory hosted by a primitive tribe (to give an exotic thrill). You are trying to learn their language, and at some point your host points his index towards a rabbit running in the distance and says ‘’gavagai’’. What could you deduce from that ? That ‘’gavagai’’ means rabbit ? Or ‘’in the distance’’ ? Or ‘’brown-grey’’ ? Or ‘’this is our meal for tonight’’ ? Or that this rabbit is called ‘’gavagai’’ ? Or that it means ‘’Go home !’’ ? If the indexical sign is not clearly identified, (and how could it be identified ?), this iconic way of learning a language (word meanings by analogy with existing things) cannot work. Observe nevertheless that we have yet a fairly strange semiotic system ‘’in reverse’’ where the linguistic signs are the signified and the ‘’real’’ things are the signifiers, in so far as things could be real, are the signifiers ...