Words and Models - a systems approach to linguistics

What is the unit of semantic analysis? A word by itself does not mean a great deal. It means more when associated with other words, i.e. when used in a specific context. Together they can be regarded as constituting a model. When building a model one must choose the appropriate variables for them. Words suitable for one context or model are not necessarily appropriate for another. Thus the words used in ordinary language are not always appropriate for the more objective models built by scientists; they were developed for a more subjective model that has come down to us as part of our cultural heritage.

The fact that contexts are often implicit rather than explicit makes it possible to use words in such a way as to convey to them their meaning when used in a totally different context. This is the basis of many ambiguities and verbal tricks. It is facilitated by our tendency towards 'nominal realism', i..e. towards regarding words as real things as opposed to units of a model designed to represent real situations for specific purposes. It is only by regarding them in the latter way that such ambiguities can be avoided.

The only possible reason for building models is to make predictions. The value of a prediction will depend on the model on which it is based, which in turn will depend to a large extent on the choice of its variables. It is possible to formulate the qualities that are required of those variables that will be used in an effective model.

I shall content myself, however, with taking one of them: their most basic feature, they must be functional. They must have a particular function to play within the model. For this reason when building a model the tendency is for scientists to develop those variables that they require rather than make use of those that happen to be available.

Such concepts as the atom and the pi-meson did not come about because they happened to be there or corresponded to any observable phenomena, but because they were required as variables for a model. In the same way, an engineer building a motor car does not use odd pieces of machinery that he might find in his backyard, which were designed for quite a different purpose, e.g. for building a lawn mower or a washing machine.

Unfortunately, most of the classifications, in terms of which we are accustomed to think in less satisfactory sciences than physics, such as the various social sciences, are not designed in this manner, but simply correspond to concepts that are part of our cultural heritage and that were in fact designed for totally different models of a simpler and more subjective nature.

In order to grasp why such terms are so unsuitable, it is important to understand the relationship between words and things. It can then be shown why we have a tendency to attach importance to existing words and how this blinds us to the real issues at stake, which cannot, in fact, be examined in terms of them.

The relationship of words to things is merely one of representation. There is clearly no necessary connection between a word and a thing. Any number of different words could be devised and in fact, are, to represent any given thing.

A word is in fact nothing more than a symbol that has been chosen to represent something. This symbol may have been chosen for various reasons, etymological, even onomatopoeic. Whatever they may be, it still remains true that any word can theoretically be abandoned and replaced by another. For instance, as it has been pointed out, if Lord Cardigan had invented the sandwich, and Lord Sandwich the cardigan, we would be wearing sandwiches and eating cardigans!

Ogden and Richards [1] illustrate this principle in the following way:

A thought is connected to the symbol or word that represents it. The thought is also connected to the referent or thing. However, there is no connection between the symbol and the referent save indirectly through the thought. If the thought is placed at the apex of a triangle with the symbol and the referent at the other two angles, the triangle will have no base:

Nominal realism

To attribute a necessary connection between the word and the thing is tantamount to attributing some sort of reality to the former. This fallacy is normally referred to as 'nominal realism' and is committed to a greater or lesser degree by all save the most scientific writers. It is particularly pronounced in children and also in primitive man.

This is because, with the appearance of language, the environment to which man learns to react is no longer composed exclusively of things, but of things and words. It is thus only to be expected that behaviour towards words should, in the first instance, be determined by the same mechanism as behaviour towards things.

Even when the representational role of words is fully accepted, as must undoubtedly be the case with modern philosophers, unconsciously they will still continue to regard words as things and as a result this fallacy persists, as is evident in the case of Oxford linguistic philosophy. I think it is worth looking a little more closely at the more extreme forms of this fallacy.

The fact that children's thinking is coloured by nominal realism is pointed out by Piaget. [2]

"Concerning words, the theories of Sully, Campayre and many others are well known, according to which it is maintained with much justice that to a child's eye every object seems to possess a necessary and absolute name, that is to say, one which is part of the object's very nature. M. Luque has shown that many children's drawings bear a title simply because of this peculiarity: 'The addition of a title has we consider no other meaning than that of expressing the name of the object, which is regarded by the designer as a property as inherent in its essence and as worthy of being produced as its visual characteristics......'

Asks Bert (7):

"Are you called Albert?"

"Yes."

"Could you have been called Henry? Would it have been the same?"

"No."

"Why not?"

"Because they are not the same thing."

"And could the moon have been called the sun and the sun moon?"

"No."

"Why not?"

"Because the sun makes it warm and the moon gives light."

Roc (6½): admits that God might have changed the names:

"Would they have been right then or wrong?"

"Wrong."

"Why?"

"Because the moon must be the moon and not the sun, and the sun must be the sun."

Asks Fran (9):

"Could the sun have been given another name:"

"No."

"Why not?"

"Because it is nothing else but the sun, it couldn't have another name."

The importance attached to words by primitive man is revealed by the fact that in many cultures a man's name is considered to be an integral part of himself. We are taught, in western society, a dualism that divides us into two parts; mind and body. Among Eskimos it is said that man consists of three parts - body, soul and name. In ancient Egypt, man had his Ka, or double, as well as his name, and of these three elements according to Cassirer, the name was the most important expression of a man's self. Even in Roman law we find the same importance attached to a name. Thus Cassirer writes:

"In Roman law the concept of the 'legal person' was formerly articulated, and this status was denied to certain physical subjects; those subjects were also denied official possession of a proper name. Under Roman law a slave had no legal name, because he could not function as a legal person." [4]

Even more striking is the case of the Australian aborigine, on which subject Stanner writes:

"We tend to consider body and mind as separate. The aborigine does not seem to think this way. The distinctiveness we give to 'mind', 'spirit' and 'body' and our contrast of 'body' versus 'spirit' are not there, and the whole notion of 'the person' is enlarged. To an aborigine, a man's name, spirit and shadow are 'him', in a sense which to us may seem passing strange. One should not ask an aborigine 'What is your name?' To do so embarrasses him and shames him. The name is like an intimate part of the body, with which another person does not take liberties. They do not mind talking about a dead person in an oblique way, but for a long time they are reluctant even to breathe his name. In the same way, to threaten a man's shadow is to threaten him. Nor may one treat lightly the physical place from which his spirit came; by extension, his totem, which is also associated with that place, and with his spirit, should not be lightly treated."

It is in this way that we can explain the fact that Egyptian pharaohs were known by sort of nick-names, their real names being kept a secret. Herodotus did not dare mention the name of Osiris. The true name of Allah is apparently kept secret. Jews are not supposed to pronounce the word 'Jahweh', who is referred to as 'Adonai', which means Lord or Master, and undoubtedly the Christian admonition against using the name of the Lord in vain can be traced to the same origin.

The attribution of reality to words has coloured the works of philosophers from the earliest time. Heraclitus considered words as embodying the nature of things just as Pythagoreans did numbers. Plato developed his theory of universal or abstract ideas in accordance with which such terms as 'duty', 'goodness', 'virtue', constituted the only reality; a preposterous theory that, unfortunately, has had considerable influence on subsequent thought. Aristotle was also guilty of the same error, according to Mauthner: [6]

"He was perhaps more than any other notable writer in the whole history of philosophy, superstitiously devoted to words. Even in his logic he was absolutely dependent on the accidents of language, on the accidents of his mother tongue. His superstitious reverence for words was never out of season."

No individual thinker has exerted such influence on subsequent thought as did Aristotle. As a result, Mauthner writes:

"For full 2,000 years human thought has remained under the influence of this man's catchwords, an influence which has been wholly pernicious in its results. There is no parallel instance of the enduring potency of the influence of words."

Modern thought is affected by the tendency towards nominal realism in the following way:

• First, if words have a reality of their own, it is implied that they must have a meaning in vacuo and thus independently from the context in which they are used.
• Second, if words have a reality of their own, the purpose for which in fact they were devised is totally forgotten, and words designed as units for a subjective model of the system are used as units of a model aspiring to greater objectivity.

The unit of analysis

The next question is whether or not a word is the suitable unit for representing things in the world, or whether a smaller or larger unit should be used. This takes us back to the question of representation. If a model is used to represent a situation and if it succeeds in doing so sufficiently well for whatever purpose representation is required, this does not mean that the model can be cut up into the same number of sub-models; each one of which will constitute a satisfactory representation of the situation.

Let us take a model of the world contained in a digital computer. Since such a computer uses a binary code, a particular value will be represented by a corresponding sequence of zeros and ones. It is clear that the minimum unit of analysis must be at least a complete sequence, and to extract from such a sequence a sub-sequence of zeroes and ones would not give any information about the sequence as a whole.

The same is true in the case of a model using the medium of an analogue computer such as a slide-rule. Once again, particular measures represent analogous ones in the environment; nevertheless, it is meaningless to suggest that a subdivision of this measure has any capacity to represent anything in the real world.

This problem must be quite clear to professional interpreters. In order to render a particular idea from, say, English into French, the technique involved does not consist of translating each word separately, but rather in the reformulation in French of the idea contained in the English text. In many cases a text formulated in one language will require a much larger, or, conversely, a much smaller number of words, if it is to be properly rendered in the other language.

If a word by itself is not a unit of analysis, what is? The answer is a context. Korzybski [8] points to the existence of what he calls multiordinal terms:

"The main characteristic of these terms consists of the fact that on different levels of orders of abstraction they may have different meanings, with the result that they have no general meaning; for their meanings are determined solely by the given context, which establishes the different orders of abstraction."

I maintain that all terms are 'multiordinal'. A word by itself may contain a general set of instructions for its use - but no more.. To determine the exact meaning of a word, it is necessary to examine it, in its context.

Let us take the term 'individual'. What constitutes an individual? Clearly the word is only meaningful if the context be specified. To give an absurd example, is a two-headed woman one individual or two? To a brain-specialist, a hairdresser or a hatter, she is clearly two; to a gynaecologist, a Chinese pedicure, or a shoe-maker, she is one.

To a theologian, the problem can only be resolved by determining the exact location of the soul. If it is situated above the neck, she is two; if below the neck, one, while the criterion applied by a hostess will undoubtedly be whether or not she is capable of carrying on two different conversations at the same time.

The extent of the context required will vary for different behaviour purposes. Thus, if we are told to expect the visit of King Leo XV, this proposition will convey sufficient information for some of the functions involved in receiving this monarch, e.g. ladies will know that they must curtsy to him. For other purposes, however, a larger context is required. For instance, it must be determined whether King Leo XV is a reigning monarch or a monarch in exile.

In the former case, it is assumed that some official reception will be arranged for him and that he will be greeted by some member of the royal family. For other functions involved in his reception, more information will be required. Thus, the size and importance of the country of which he is king is relevant in determining whether he will be received by the Queen herself or by some less important member of the royal family.

For other purposes, for instance, in order to determine whether the Prime Minister should be present at the reception, it might be necessary to know whether the king was in fact running his country, as does the King of Jordan or that of Saudi Arabia; or whether his functions were of an honorary or constitutional nature, as is the case with the kings of Norway, Sweden and Denmark.

From this it must be apparent that the term 'context' is in fact nothing else than a convenient word for 'linguistic model'. What I have said about a context can be said about a model, i.e. a unit of a model is meaningless outside the model.

Implicit context

One of the reasons why the notion of 'context' has been badly understood is that it has always been equated with explicit context. However, the statement "King Leo of Ruritania arrives tomorrow at Southampton" will 'mean' something different to different people, depending both on the nature and the amount of relevant information they possess.

Thus, the information value of the statement will be much greater for the king's nephew currently reading PPE at Oxford, and the Chief of Protocol who will be responsible for his reception, while it will be much lower for Joe Bloggs who is a greengrocer at Nuneaton. In each case, a very different context will be implied.

Ogden and Richards seem to consider [9] that words are only valid within a specific and restricted context, which they refer to as a "universe of discourse". They write:

"All definitions are essentially 'ad hoc'. They are relevant to some purpose or situation, and consequently are applicable only over a restricted field or 'universe of discourse'.

"Taken out of the universe of discourse, it becomes a 'metaphor' and its use is no longer possible in a scientific context: [10]

"When a word is taken from one universe of discourse to another one, we obviously have the case of a very apparent metaphor, which is unlikely to lead to any verbal ambiguity save that which might be exploited in the grossest of puns. The resemblance between the use of the same word in two universes of discourse, being so flimsy, only the vaguest analogy obtains."

They therefore accept the notion of a 'context' - but by stating that only a specific type of context is acceptable for a particular word, they attach too great an importance to the meaning of the word and not enough to that of the context.

Ogden and Richards' view is, of course, consistent with the unfortunately still prevalent view that the various disciplines constitute independent fields of enquiry that are unrelated to each other.

To avoid 'verbal ambiguity', what is important is not to restrict the use of words to a limited number of different contexts but to specify the context in which they are used and in terms of which, their precise meaning will be apparent.

Indeed, the use of words connoting or implying a particular context in a different, explicit, one is the basis of most linguistic ambiguities. To understand this point, one must comply with Korzybski's dictum of concentrating on the actual events in the world, that a particular proposition represents, rather than on the words constituting this proposition. We will take an example. Let us consider the following propositions:

1. England has declared war on China.
2. England has beaten the West Indies at Lords.
3. England has 56 million inhabitants.
4. England is the shaded area on the map.

The first proposition refers to a certain number of middle-aged men sitting round a table who, together, constitute the British Government. The second refers to a team of 11 youngish men dressed in impeccable white flannels playing cricket at Lords; the third to an island occupying about 90,000 square miles off the Northern coast of the continent of Europe and the fourth to the small area on the page of an atlas.

In each of these four cases, the use of the word 'England' is quite legitimate but in each case it refers to something very different. All things denoted by it, though structurally very different, are, in other respects sufficiently similar to justify their being referred to for specific purposes as "England". If, however, the term is used for other purposes, ambiguity must result.

In each of these four cases, the term 'England' stands for an implicit model. We know that when we say "England beat the West Indies", we are referring to a cricket team and not to the Government, nor part of a page on a map, nor to the entire British people. The information value of this statement is quite high to any inhabitant of England but particularly high to those who know something about cricket.

If, however, a Martian were to hear the latter phrase used out of context, he would 'legitimately' conclude that if England can beat the West Indies, their chances of beating China, if war were declared, would be gauged by examining China's cricket record against the West Indies.

This may be considered an absurd example but it is, nevertheless, illustrative of the point I am trying to make. Let us take some less ridiculous ones. This same error characterises, without exception, all the highly admired dialogues of Plato. When Socrates asserts his intellectual superiority over his opponents, he is merely showing himself up as being a greater master of the art of using words out of context. Look at this passage from the Symposium: [11]

"Consider, then, whether when you way 'I desire what I possess' you do not really mean, 'I wish that I may continue to possess in future things which I possess now.' If it were put to him like this he would agree, I think."

"Yes", said Agathon.

"But this is to be in love with a thing which is not yet in one's power of possession, namely the continuance and preservation of one's present blessings in the future."

"Certainly."

"Such a man, then, and everyone else who feels desire, desires what is not in his present power or possession and desire and love for that object, things of qualities which a man does not at present possess but which he lacks."

"Yes."

"Come then," said Socrates, "let us sum up-the points on which we have reached agreement. Are they not first that Love exists only in relation to some object and second that that object must be something of which he is at present in want?"

"Yes."

"Now recall also what it was that you declared in your speech to be the object of Love. I'll do it for you if you like. You said, I think, that troubles among the gods were composed by love of beauty for there could not be such a thing as love of ugliness. Wasn't that it?"

"Yes."

"Quite right, my dear friend and, if that is so, love will be love of beauty will he not, and not love of ugliness?"

Agathon agreed.

"Now we have agreed that Love is in love with what he lacks and does not possess."

"Yes."

"So after all Love lacks and does not possess beauty?"

"Inevitably."

"Well, then, would you call what lacks and in no way possesses beauty, beautiful?"

"Certainly not."

"Do you still think then that Love is beautiful, if this is so?"

"It looks, Socrates, as if I didn't know what I was talking about when I said that."

"Well, it was a beautiful speech, Agathon. But there is just one more small point. Do you think that what is good is the same as what is beautiful?"

"I do."

"Then if Love lacks beauty and what Is good coincides with what is beautiful, he also lacks goodness."

"I can't find any way of withstanding you, Socrates. Let it be as you say."

"Not at all, my dear Agathon. It is truth that you find it impossible to withstand: there is never the slightest difficulty in withstanding Socrates."

In this nonsensical argument, Socrates is transferring the connotations of the phrase "lacks and does not possess" from one context to another, as if it had a definite meaning on its own, rather than as part of a given context.

Logic has, on the whole, been a fairly fruitless pursuit. Korzybski goes so far as to say:

"Academic philosophy and formal logic have hampered rather than advanced knowledge and should be abandoned." [12]

The main thing wrong with logic is that it considers words by themselves, as if they have a reality of their own, and can thus be transferred with total abandon from one context to another.

Take a stock example of a mediaeval syllogism:

• Major premise: No cat has eight tails.
• Minor premise: Every cat has one more tail than no cat.
• Conclusion: Therefore every cat has nine tails.

It needs no lengthy explanation to point to the fact that the word "no" is merely used in a different way in the major premise than in the minor premise, providing us, thereby, with a totally absurd conclusion. If one again abides by Korzybski's dictum of finding the referent, i.e. looking beyond the words at what in fact is going on, no ambiguity is possible. In such a case, one would be using the term 'no' with its full information value in each case, i.e. by considering the model of the world implied by the use of the term 'no', and not just regarding the explicit symbol 'no', which by itself communicates no more information than does any isolated unit of a model.

Let us take another example: a paradox formulated by the mathematician Frege:

"In a village there was only one barber, who shaved only those who didn't shave themselves. The question arises if the barber shaves himself; if we say Yes, then he didn't shave himself; if we say No, then he shaved himself." [13]

The solution to this problem is quite simple - the term 'shaved' is a high order absubstraction; if we think of the processes or events that it describes rather than the word itself, we can imagine totally different processes, which fall within this linguistic classification. Among these processes are two in particular which are relevant to this case: one is shaving a customer seated in front of him for a fee; the other is shaving himself in front of a mirror. In most contexts, the use of a single term "shave" for these processes is quite sufficient.

We have hit upon an example where, to avoid ambiguity, two different terms describing these two different processes are necessary. The Eskimos have four words for snow; the Navaho differentiate between three different types of cause, as did Aristotle, and if the Germans distinguished between the two different uses of the word "shave", then Frege's paradox could no longer have been formulated.

Implied contextual transfer

I shall refer to this as the principle of 'implied contextual transfer'. It is of paramount importance and, as I have already said, the basis of most verbal ambiguities. Let us take the word 'exist'. The word is usually used to refer to things like tables, chairs, telephones, etc., as having in common the fact that they occupy space/time and whose behaviour is explicable in terms of the natural sciences. The term is also used to apply to things which are very different indeed, such as the particles of an atom.

If we say that the 'positron' exists we cannot help but transfer to it, unconsciously at least, some of the characteristics of tables, chairs and telephones. This is clearly a great disadvantage because of the fact that atomic particles have very little in common with such objects. Scientific progress, in fact, as we have shown, has been made in proportion to our ability to understand things in terms other than those of our immediate experience.

In fact, the term 'exists' is incorrect. Rather, we should say that it is possible to build a model of our environment by dividing it into objects and processes, such as chairs, tables and telephones, and sitting, eating and telephoning, which is satisfactory for our day-to-day requirements. However, for certain scientific purposes, a more complex model is required, in which the world is divided into atoms and their various four-dimensional constituents. Neither chairs, tables, nor telephones, nor atomic particles, exist. They are in fact just words or models, which are used to represent the world for specific behavioural functions.

The term 'exist', however, is not only restricted to chairs and tables and atomic particles but also to 'universals' or 'abstract ideas', such as 'beauty', 'truth', 'God', etc. Again, we are using the term 'exist' in a very different way. These terms have very little in common with chairs, tables and atomic particles. They are terms of very considerable subjectivity, designed for building a simple model of the world around us, for the purposes of day-to-day behaviour.

Nevertheless, it is obvious that when we say "God exists" we are transferring the connotation of concreteness and solidity that we associate with chairs and tables to this concept of God. We are automatically supposing that God, chairs, tables, and atomic particles, have enough in common to justify the common use of the term 'exist'.

Theologians might reply that God has nothing in common with these other two categories. In fact, they accentuate the abstract nature of God and resent any tendency towards anthropomorphism, which is usually considered as highly blasphemous. Thus Antonio Perez, a minister of King Philip II of Spain, was accused by the Inquisition of threatening to cut off God's nose. It was not so much the threat that upset them as the implication that God, like we poor ordinary mortals, had a nose. If theologians disapprove of anthropomorphism, why then use the term 'exist' to apply to God?

Thus we use the word 'tree' because there are many contexts in our everyday conversation where differences between oak-trees, palm-trees, beech-trees, apple-trees, are of no consequence. It is the fact that it is a tree that is important to us. Certain Indian tribes, we are told, have never recognised the usefulness of these classifications, for which they apparently have no need in the context of day-to-day conversation and, consequently, though they have separate words for each different type of tree, they have none for "tree".

Thus we have words with very little in common, such as 'typewriter', 'obstinacy' and 'influenza' and it would be quite conceivable to invent a new word representing a category into which all three of them could be fitted. Let us call this category 'typobin'. The trouble with a 'typobin' is that there are few statements that we can make about it that are denotatively very significant, since the concepts falling within this category have practically nothing in common. Since we can find no use for 'typobin', let us swallow our pride, admit that it was a mistake ever to coin the term, and eradicate it from our already over-encumbered vocabulary.

The word 'exists' is in this respect very much like the word 'typobin'. The things that are said to fall within the class of 'existing' things, such as chairs and tables, atomic particles, goodness, truth and God, have no more in common than typewriters, obstinacy and influenza. It is thus denotatively as useless as 'typobin' yet we continue to use it - why?

The answer is that, although it has not denotative meaning, it has acquired a powerful connotative one and in this respect the inclusion of such diverse concepts in the category of existing things is no hazard. Indeed, when we say "God exists", we are unconsciously transferring the connotation of concreteness which the term 'exists' imparts to material things such as tables, chairs and telephones to the concept of God. How do we avoid this?

The answer is by refraining from looking at words in vacuo. Instead they must be regarded as the functional units or variables of a model whose role is to represent some aspect of the world for specific behavioural purposes.

References