The Computer and Traditional Semiotics

[Presented at the Modern Language Association convention in Chicago, 1985]

For literature is like schizophrenia: a process and not a goal, a production and not an expression. --Deleuze and Guattari, Anti-Oedipus (133)

Suppose that one day an omniscient Mephistopheles appeared at your terminal and asked, "What do you want to know about the text of Proust, or Joyce, or Edgar A. Guest, or the Sears Roebuck catalog?" Your first question must absolutely be: "Where is the text?" and your second question, "What is it possible to know about the text?" For any other question about interpretation, elucidation, or explication will be based on a theory of literature and a theory of the text which will reveal that you have already made up your mind about what the text is and, if not what it means, then at least the philosophical and methodological parameters within which the text and literary studies in general are to be understood.

For reasons I will try to suggest, the computer analyst has pretty well made up his mind about the text and has been forced to make up his mind because of the demands of turning natural language into a computer language in order to be made available to the binary circuitry--a binary circuitry which very nearly becomes a modern definition of a universal characteristic of which Leibniz dreamed--but a characteristic which handles everything on an atomistic level with incredible speed and efficiency. Hundreds of articles and abstracts written over the last several years indicate that researchers in the field of computers and the humanities have learned more about the computer than the text.

In the course of this essay, we ought not to be surprised at or indignant about the fact that one of the finest products of the technological society, a machine which has been created within the epoch of the metaphysics of presence, a tool which seems to imitate the functioning of the brain, should bring with it the values of the technological society and preassumptions about the nature of language, representation, signifying, and even power--on which our society is built.

Computer analyses of a text depend to a great extent on identification and lemmatization of linguistic entities, to a degree that the computer in text analysis is an adjunct of linguistics and operates according to categories identified by linguists. Efforts to go even farther in identifying entities may result in such things as a "conceptual index" or "conceptual dictionary."

Conceptual indices can be compared to the "yellow pages" in the phone book versus the alphabetically organized "white pages" of the traditional indices and concordances. If it may be assumed that every literary analysis should begin with a thorough and complete inventory of the entire conceptual material of a work, we need a tool that can get us to that point without the usual drudgery of reading over and over again, taking notes and progressively collecting more data depending on shifts of focus and alertness for detail, each time necessarily favoring certain aspects over others./1/

I have quoted this paragraph in order to demonstrate what the computer in the epoch of the logos may require of the researcher: the creation of a reading which does not allow for "shifts of focus...necessarily favoring certain aspects over others." That is, requires the identification of an Ur-text with an unvarying meaning, which the computer may then approach from the point of view of a super-exegete. Such a literary analysis which gives a "thorough and complete inventory of the entire conceptual material of a work" necessarily presupposes not only an Ur-text but also a fixed and transcendental self as author, placing such a computer analysis firmly within the epoch of the logos, the transcendental self, and the metaphysics of presence, with its obvious theological implications. It effectively creates external lexemes, doing away with the Barthesian texte riche in favor of a privileged reading.

Computer analysis does not escape the usual problems of representation. Such analyses cannot be excluded from the general atmosphere of semiotics. Any attempt at such exclusion would be to make claims for a privileged point of view outside the traditions of Western civilization, even outside language.

It does not make very much difference whether, in looking at computer analysis, we accept the Saussurian duality of signifier-signified with its arbitrary but surreptitiously privileged relationship between signifier and signified, or whether we tend to admire the Derridean rejection of a transcendental signified and see the origin of the possibility of meaning in the scandalous supplement and the play of differance. Putting the computer within such a semiotic atmosphere, we may see that what goes in one side comes out the other side, frequently expressed by the trite phase "garbage in, garbage out." In other words, whether we see the input as signifier and the output as signified, or the input as signifier and the output as another signifier in the signifier chain, does not change things much.

One way or another the computer cannot escape semiotic considerations, but it is important to note the necessary relation between input and output. Such lack of arbitrariness in the input-output and in any program which deals with the stored data is consistent with the traditional conceptions of the Western logos. The fact that garbage-in-garbage-out introduces necessity between the input and output, the fact that binary machine language becomes a kind of perfect encodement and universal logic, means that a unique event has taken place in the signifier chain: there are two signifiers whose relation is no longer arbitrary (for natural language, computer language, and machine language are not identical entities). The computer becomes a mirror of the logos and the literary researcher who uses it takes his place in the Western tradition: Garbage-in-garbage-out, Logos-in-logos-out. The researcher cannot reject his culture, in any case, and ought not to be too upset about it, but he ought to wonder if the nature of his intuitions is not to some extent controlled by the nature of his research and his research tools. It is well-known in the physical sciences that pure science does not precede technology, that scientific knowledge does not express itself in technology, but rather that the development of technology determines the direction science will take. We know the things our tools allow us to know and our knowledge conforms to the requirements of the tools. The computer may thus be seen as a conservative, even a reactionary, force in the acquisition of knowledge and in dealing with the text. Weizenbaum pointed out in Computer Power and Human Reason that the computer's ability to keep records on public welfare tends to perpetuate the system of public welfare which might otherwise fall to pieces and demand some other solution to the problem of poverty. It thus is a conservative social force as well, tending to preserve the status quo of power.

As a representative of the technological society, computer analyses are likewise consistent with the ideas of Jaques Ellul concerning the self-augmenting, universalizing activity of technique. Since the program must forego connotation, psychology, sociology, and the varied readings of the rich text, it may be suggested that computerization demands a return to a rigid text which yields its meaning to exegesis, a state of affairs ironically preceding the concept of representation and style, a situation which Foucault places before the 17th century. A computer program is the equivalent of an exegetical technique which always works within its own givens, producing a result electrically and necessarily consistent with the program. It is thus consistent with the assertion of James Ziegler, in a recent conversation, that the United States is still intellectually in the Enlightenment, suggesting why some American intellectuals find modern French thinkers a little annoying. As computerists we are still classicists and we suspect that order exists somehow, that we can identify it, that we can find it in the text, and attach meaning to it.

Since we must situate the computer within the problem of representation, we must take into consideration the ideas of Deleuze and Guattari on representation. Briefly, the unconscious as a desiring machine does not need representation for production, even social production. Deleuze has criticized Oedipus because it is an artificial belief forced on the autoproductive unconscious by a bourgeois-capitalist movement aimed at control and at identifying the schizophrenic as ill...a situation in which the psychiatrist becomes a cop.

Oedipus has been forced onto the ignorant and autoproductive unconscious, so that the self exists as a function of the family triad of mommy-daddy-me, the family being the very structure of control. This oedipalization of the autoproductive unconscious is of the nature of representation. We may even describe the Oedipus in Derridean terms as an entity which can exist only within the play of difference in the family triad, the Oedipus existing as the structure of difference.

The schizophrenic does not succumb to oedipalization, he "scrambles the codes," involves himself in a "concerted destruction of the signifier." He "knows" that the signifier is a bourgeois invention to keep him under control. Literature is likewise oedipalized by its reduction to representation.

Let us pursue the problem along Deluzian lines: if the unconscious is autoproductive and the level of representation is a bourgeois development, a capitalist oedipalization to deny the unconscious its freedom, then the computer and its analysis, by injecting another level of representation between reader and his appreciation of the text is a bourgeois tool, reflecting a bourgeois semiotic, which does not really add extra dimensions of interpretation and understanding to the text, but actually restricts the understanding and the imagination to that controlled by representation, the demands of representation, of machine language, of input and output. The computer analysis is yet another text slipped between the reader and meaning, between reader and production, between reader and meaning-product. The computer reading, as another text, takes its place in the grand play of intertextuality, and in an intertextuality complicated by every value of the Western logos, the metaphysics of presence, the technological society, and maybe even of capitalism.

A lemmatized text is no longer dangerous--it no longer roves about clamping onto things, it does not float as a signifier, it is fixed and controllable, and it is the end of connotation. It does not attend to Verlaine's admonition:

Logos is the semiotic which lies behind power as the basic phenomenon and the prime motive of life. Logos-in-logos-out is the reading of power, a theological reading, and the computer reading is the very image of the Western text reading itself and entrenching itself as ideology. We ought not be surprised that the computer is the prime instrument in missile guidance systems, in general military applications, and in record keeping by the utility companies. We must recognize, however, that the instrument we develop by our technology reflects the values and the philosophy of that technology and we must likewise go on using it. For even within the preassumptions of computerized research we may have only scratched the surface of the knowledge. But we cannot expect that it will be a revolutionary tool which will allow us startling insights into the text, or into the human condition, because the computer has already "made up its mind" about those things.

"Within a certain historical epoch," wrote Derrida, "there is a profound unity among infinitist theology, logocentrism, and a certain technicism. The originary and pre- or meta-phonetic writing that I am attempting to conceive of here leads to nothing less than an 'overtaking' of speech by the machine."/5/ Phonetic writing lends itself to programming. In the computer analysis of a text, we contemplate ourselves and our culture as oedipalized, capitalized, technicized, theologized, logocentric, disciplined, bourgeois...and proud of it.

The Computer and Traditional Semiotics: Logos In, Logos Out