THE SEMIOTIC ABSTRACTION
Russell Daylight
Charles Sturt University, Australia
rdaylight@csu.eud.au
Abstract
When we press the “a” key on our computer keyboard, an “a” appears on our screen almost instantaneously. In between those two points there are a number of layers of computer program which communicate with each other: the keyboard controller sends a message to the operating system which is interpreted by a word processor, which then returns a message to the operating system, which communicates with the video controller and the video board sends a message that it needs an “a” and this is mapped as a group of pixels which light up on the screen. Except that none of this actually happens. At the level of physical reality, all that happens is the shifting of magnetic fields and the passing of electrons. The electrons orbiting atoms slide from one atom to the next along wires, among silicon and metal. Entities like computer programs and operating systems are abstractions, layers of abstractions in fact, on top of a brute reality. In this paper I argue that this is the appropriate starting point for understanding the role of semiotics in negotiating reality. Taking examples from computer science, visual perception and language, this inquiry considers what kind of abstraction we want semiotics to be. It is found that “the semiotic abstraction” allows us to understand semiotic systems as machines for creating differences, and semiotics itself as the primordial science of meaning.
Preamble
I wrote a book, called What if Derrida was wrong about Saussure? (2011), in which I tried to discover what was at stake in Derrida’s writings on Saussure, and hence, what was at stake between structuralism and post-structuralism. In an interview with Henri Ronse, Derrida declared that he found it necessary, strategically, to risk saying nothing (1982b: 14). Right to the end of my work on Derrida I found that a puzzling and even disturbing idea. However, if I am somewhat anxious about the proposal contained in this paper, then it's because I also want to risk saying nothing. What I will propose here concerns the role of “abstraction” in semiotics, where abstraction is defined as “considering an object or group of objects from one viewpoint while disregarding all other properties of the object” (Shaumyan 1998: 17). As we shall see, this viewpoint will allow us to understand semiotic systems as machines for creating differences, and semiotics itself as the primordial science of meaning.
Like Derrida also, as he does in “Différance” (1982a: 3), I want to begin with a letter.
The Letter A
I want you to imagine pressing the A key on a computer keyboard and observing the letter “a” which appears on the screen. In between those two points, of course, exist many layers and domains of software and hardware, across which this communication must travel. The kinetic energy of your finger depresses the key, which makes contact with a matrix of electrical receptors underneath the keyboard. This point of contact is registered by the keyboard controller which sends the key number to the operating system. The operating system checks with its window manager, and sends an ASCII keystroke to the open program – let’s say a word processor – which updates both its file structure in memory, as well as its bitmap view of the display. Finally, the operating system then sends a message from its graphics driver to the monitor to change the colour of certain pixels in a certain area of the screen. This chain of communication is simple enough to understand. The only problem is that none of it actually happens.
At the level of brute reality, all that happens is the movement of electrons flowing through metal being controlled by magnetic fields. In other words, that the electrons orbiting, say, copper atoms shift from one atom to the next along wires from the keyboard to the computer to the monitor. Everything can be explained in the physical world as the shift of magnetic fields and the flow of electrons. Entities like files, programs and operating systems are abstractions. They don’t exist in brute reality; they are only a particular conceptualisation of that reality.
Abstraction is a term used in many fields, from philosophy to art history; it is also used by Peirce. However, my semiotic project in general asks us to go back to that moment prior to Peirce and Saussure, when the field of inquiry was still open with possibility. As such, it is helpful to free ourselves of all preceding definitions, even if they share certain themes. The one usage of the term that helps to make this concept distinct is that used by computer science. In computer science, abstraction is explained as the elimination of low-level physical properties in order to isolate only those functions under study. This allows programmers to assume the details of complex physical environments in order to focus on the meaning of those details from a particular perspective. In their (1999) text Concrete Abstractions, Hailperin, Kaiser and Knight ask us to:
Consider, for example, a word processor. When you use a word processor, you probably think that you have really entered a document into the computer and that the computer is a machine which physically manipulates the words in the document. But in actuality, when you “enter” the document, there is nothing new inside the computer – there are just different patterns of activity of electrical charges bouncing back and forth. … Even the program that you call a “word processor” is an abstraction – it’s the way we humans choose to talk about what is, in reality, yet more electrical charges. (1999: ix)
Typically, abstraction in computer science operates in distinct layers. Each layer allows us to assume, and ignore, the physical details in lower levels of abstraction. For example, programs assume the work done by programming languages, programming languages assume the work done by machine languages, machine languages assume binary, and binary assumes electronic circuitry.
Fig. 1: Stacked abstractions in computing.
At the core of a computer, at the simplest or lowest level of abstraction, are electronic circuits. In physical reality, a circuit is a microscopic arrangement of metal within a silicon platform holding two different voltage potentials in place with magnetic fields; but conceptually, it is a binary decision maker: what's called a logic gate. And from a logic gate, you can build anything, from calculators and word processors to movies and climate models.
Computer abstraction is a fascinating topic in itself – well, fascinating to me as an English Lecturer – but what is important here is thinking about how abstraction can help us understand the properties of semiotic systems. But first, and in less detail, I want to briefly suggest how the neurological processes of perception are analogous to these computer processes.
Abstraction in General
One of the simplest and most common examples of semiosis occurs when we see a tree. Indeed, it was the first example Ferdinand de Saussure could think of (1983: 65). We, the viewer, being of sound mind and body, and of unitary consciousness, observe the shape and colour of a singular, discrete, object. That information is sent for visual processing, looked up in memory, cross-checked with language, and we emerge with the concept and name of a tree. But once again, of course, at the level of brute reality, none of that happens. In the same way as a computer, a physical process – in this case, electromagnetic waves striking our retinas – is converted into electrical signals by the photoreceptor cells, and sent as information along neurons. Just as with the word processor, nothing new is entered into the brain: the neurological data is transmitted in the form of changes in electrical potential, and is absolutely indistinguishable from other sensory, cognitive or physiological instructions.
It is impossible, however, to solve any problems in perception, cognition, signs and language if we continually have to refer to electrical charges leaping across synapses, and so we abstract out lower levels of information to concentrate on the issue at hand. According to Bear, Connors and Paradiso, in their (2007) text Neuroscience, the layers of abstraction on top of this basic reality move from the molecular to the cellular to the behavioural to the cognitive (2007: 13–14).
Fig. 2: Abstraction in Neuroscience (Bear, Connors and Paradiso).
Interestingly enough, they also talk of “voltage potential” (2007: 60) and “gates” (2007: 59) as the basic material of brain functioning. And in the same way as the abstractions of computer science, cognitive neuroscience (which studies consciousness and language) assumes the work done by behavioural neuroscience, behaviour assumes systems, systems assume cells, and cellular models assume and ignore the details of molecular functioning.
We need to ask an essential question at this point: if each of these layers defines a useful level of abstraction to make discoveries about the brain in total, then which is the appropriate level of abstraction for thinking about communication, information, and signs? Information is most obviously an abstraction which ignores the physical details of a chemical-electrical flow in order to focus on the meaning of that flow as data or code. So that kind of abstraction might sit between the systems and behavioural layers, or perhaps between the behavioural and cognitive layers. On the contrary, the answer is that semiotic work is being done at every level of abstraction, whenever we eliminate physical properties in order to focus on the meaning of those properties within a system. With visual perception, for example, we could focus on information at a low level – such as the work of rods and cones in colour perception – or at a very high level – like the aesthetic pleasure of autumn leaf colours. It's important to remember, however, thatevery layer of abstraction will be built on top of the same neurological reality.
Clearly there is a great difference between thinking about the semiotics of seeing a tree, the semiotics of colour perception, and the semiotics of voltage potential leaping across synapses. Most often, though, in this field of study called semiotics, we begin at the very top layer of abstraction, with unitary consciousness and discrete objects. Let’s say a tick smells the warm blood of a mammal and falls onto its prey. That’s already at a very high level of abstraction, perhaps already at the highest possible level of abstraction, making it impossible to understand what language and signs actually do. Such a starting point neglects, and takes for granted, the great majority of abstractions – indeed a stack of abstractions – that could potentially be semiotic systems. My view, then, would be that abstraction is currently insufficiently explored in the study of meaningful systems. The question is: what kind of abstraction should semiotics be? In other words, what are the essential characteristics of a semiotic abstraction?
To answer that, let's return to logic gates and to language.
Semiotic Abstraction
The only linguist I have discovered to be interested in this idea of abstraction is Sebastian Shaumyan – the Armenian-American structuralist (1916–2007) – who defined abstraction as: “considering an object or group of objects from one viewpoint while disregarding all other properties of the object” (1998: 17). This much is consistent with the view of abstraction I have presented so far, and indeed, Shaumyan understands abstraction in the same layered sense as with computer science and neuroscience.
Fig. 3: Linguistic abstraction according to Shaumyan.
For Shaumyan, the infinite complexity of human sounds are abstracted to form phonemes, phonemes are abstracted to form words, and so on (1998: 15). Each layer can make its own unique discoveries about language by taking for granted the mechanisms at lower levels of abstraction.
Shaumyan's task is to discover the semiotic properties of language. In order to do that, he states that: “Our problem is to consider human language under the one viewpoint of its semiotic properties, while disregarding all its other properties. … Singling out the semiotic properties of language I call semiotic abstraction” (1998: 17). So this would be the guiding principle of my argument in this paper: that isolating the semiotic properties of some system or phenomenon is what we should call “semiotic abstraction”.
Now, in between the molecular layer – where ions are transferred between atoms – and the cognitive layer – in which “I” “see” “a tree” – or between the infinite complexity of human sounds and their abstraction as arguments, are many possible layers and types and purposes of abstraction. And if we are dealing with multiple layers then it gives us a chance to think about which layers – if any or all – are semiotic in character, or conversely, which properties would need to be isolated in a semiotic abstraction.
To answer that, I’m going to return to computer science for a moment. Because I have an intuition, let’s call it, that studying perception, experience and language is such muddy ground for us all – it has been so trampled underfoot by philosophical inquiry – that it becomes far more fruitful in the first instance to deal with something at some remove from those familiar questions. And so returning to the example of computer science, we may as well start from the most elementary level of abstraction, which is the logic gate, and which, reassuringly, does actually feel like a semiotic system par excellence. Saussure, and indeed Shaumyan, describe language as the most representative of all semiotic systems; but I think that the logical, or binary, gate is a much more likely candidate. It’s a system of only two possible states of energy: the minimum number to create a distinction, which we can abstract to being yes/no, black or white, zero or one. In other words, a computer is a distinction machine: a machine for creating differences. And this is what I believe we should think of a semiotic system as: a machine for creating differences. With the ability to create differences, we can create anything.
Can we say that the “zero state” or the “one state” of voltage potential has no meaning at all outside of its binary opposition, that is, its semiotic system? Well, in most senses, no: it is a flow of electrons and it has an impact on the world, it exists, it can’t be nothing. But my point would be that it is not semiotic; the two states, taken together, is semiotic. For a state to carry that special property we call meaning – even if it is perceived by a non-intelligent being like a transistor – it must be able to be different from something else.
The semiotic abstraction, then, is to filter out all details except for meaningful differences within a system, i.e., for value. It's not much, but it's something: a single criterion. Which to me is reassuring, as I’m uncomfortable with that particular discourse of semiotics that wants to include every field of inquiry within a semiotic purview: to gather developmental psychology, evolutionary biology, quantum physics, under one overarching theory. In a sense this might be true, that everything in the universe – both living and non-living – is available for semiotic analysis. But if semiotics is to be anything at all, it has to be the universe viewed from a particular point of view, and that is the view produced by the semiotic abstraction.
Implications
I want to end not with a conclusion but with a series of implications, and I apologise if some of these rely on ideas and arguments from other works of mine, particularly my (2012) essay “The Difference Between Semiotics and Semiology.”
The first implication is that abstraction in general allows us to see that all scientific inquiry is built upon the same basic reality.
Fig. 4: Abstraction of the sciences
Physics, chemistry and biology, for example, do not study different objects; they are simply specialised, and stacked, views of the same object. Biological systems come into view only when we abstract chemical information, chemical systems come into view only when we abstract physical information, and physics appears only by making abstractions of physical reality.
This understanding is important in determining the boundary of semiotic inquiry. Because, from this point of view, there is no possible distinction between living and non-living systems. Indeed, they are the same objects, simply observed from different points of view. It would be controversial to say that biology creates cells, or that chemistry creates molecules, or that physics creates atomic forces. But it is necessary to abstract physical information to allow molecules to appear. In the same way, it might be controversial to say that semiotics creates signs. But the semiotic abstraction is necessary for us to see signs. More prosaically, and at the risk of saying nothing, it's only in looking for signs that we find them.
A number of implications arise from the fact that the semiotic abstraction produces a set of meaningful differences within a system. The first is that it is a mechanism which is incapable of comparing the sameness of entities, their essences or qualities. It is incapable of addressing habits or patterns which form over time. These differences, orvalues in the Saussurean sense, are taken at a single moment in time, and hence are systematic, not temporal. Furthermore, the relationship between, say, visual perception and emotional response, is one of multiple ways of conceptualising the same basic phenomenal or neuroscientific reality, and hence is best understood simultaneous layers of abstraction, rather than as a temporal chain of semiosis. Indeed, if physical properties are abstracted for meaning within a system, then the essential partnership of semiotics is not the stand-for relation but the abstracted-as relation. The full set of potential electromagnetic waves do not stand for short, medium and long wavelengths, but are abstracted as short, medium and long wavelengths.
A further implication is that the best and most characteristic semiotic systems will be finite and complete systems. In a logic gate, we know the entire set of possible outcomes, and we know all of the individual elements. You can’t know the system until you know all the elements and you can’t know the value of the individual elements until you know the entire system. This is something that could never be perfectly achieved with natural languages, which are open and indistinct. In other words, the precision and reliability of meanings are dependent upon the precision of both the frame and the distinctions within it. So we can add that the second ideal characteristic of a semiotic system would be the precision of the distinctions, or commutations. With a logic gate, the distinction between the zero and one values is perfectly clear and consistent: either the connection between the two circuit points is made, or it isn’t. With languages and other communicative systems, the set of values varies between individuals and the commutations are variable and approximate.
However, the most important implication of defining semiotics as a machine for creating differences is that it establishes semiotics as an originary or primordial science. Difference is the source, the prerequisite, of logic, of mathematics, of the natural sciences. It is a primary science, the foundation of all others. For Peirce, semiotics is logic. But in the same way that Derrida showed that the difference involved in signs is the source and constitution of Husserlian phenomenology, we would say that difference is the source and constitution of Peircean logic.
My final point is that I believe that semiotic abstraction has the potential to unite theoretical and applied semiotics. Applied semiotics might well be the Frankenstein's monster of semiotics – throwing together whatever pieces of Peirce, Saussure, Eco and Sebeok lie to hand, but it has the distinct advantage of being useful: of providing undergraduates with a better sense of how we make meaning in the world. Theoretical semiotics, on the other hand, retains its conceptual purity only by studying imaginary objects, and hence, at the great cost of abandoning the real world. Semiotic abstraction is capable, perhaps, of bridging the two domains. It is true that some systems are simple, like a logic gate with only two possible values. Many systems in the natural world, like photoreceptor cells, produce results with a manageable level of complexity. Language is very much more complicated, and a traffic intersection more complicated still. But the relationship between theory and practice remains sound; it is non-contradictory. Like the relationship between Newtonian Physics and the real-world movement of weights down a slope: simple laws can produce physical complexity beyond the scope of the greatest supercomputers. The same relationship would pertain to the laws of semiotics and the situation at a busy intersection, when you decide if it safe to cross the street. It’s a massively complex semiotic system, but one that is based on simple and self-contained laws.
In summary, semiotics is increasingly interested in bridging the gap between the word and the world, or between brute reality and some representation of it. But this is the oldest task of philosophy, and is caught up in long standing questions of mind and representation. Caught up, also, in the scientific inquiry into living and non-living matter. As a self-contained field of inquiry, semiotics should not depend on a priori notions of logic or scientific method, but create its own rules from the ground up. I would encourage humility in the face of this task, but suggest that abstraction offers a unique perspective, and a powerful analytical tool, in developing semiotics as the science of meaning.
References
BEAR, Mark, Barry W. CONNORS, and Michael A. PARADISO. 2007. Neuroscience: Exploring the Basics. 3rd edition. Baltimore, MD: Lippincott Williams and Wilkins.
DAYLIGHT, Russell. 2011. What if Derrida was wrong about Saussure? Edinburgh: Edinburgh University Press.
DAYLIGHT, Russell. 2012. The Difference between Semiotics and Semiology. Gramma/Γράμμα: Journal of Theory and Criticism. (20): 37–50.
DERRIDA, Jacques. 1982a. Margins of Philosophy. Trans. Alan Bass. Chicago: University of Chicago Press.
DERRIDA, Jacques. 1982b. Positions. Trans. Alan Bass. Chicago: University of Chicago Press.
HAILPERIN, Max, Barbara KAISER & Karl KNIGHT. 1999. Concrete Abstractions. Pacific Grove, CA: Brooks/Cole Publishing Company.
SAUSSURE, Ferdinand de. 1983. Course in General Linguistics. Trans. Roy Harris. London: Duckworth.
SHAUMYAN, Sebastian. 1998. Two Paradigms of Linguistics: The Semiotic Versus Non-Semiotic Paradigm. Web Journal of Formal, Computational and Cognitive Linguistics. (2): 1–72.