Criteria for Semiosis

Rulon S. Wells

My fundamental message is a message of caution. We must beware of claiming too much for semiotics. And whatever claims we make, we must be prepared to support.

The danger of claiming too much takes various forms; two or three are most frequent.

First, there is the temptation to assimilate. If we are interested in signs (‘signs’ in whatever sense we have already established for ourselves), and are interested in other things, we feel the temptation to construe these other things as signs. (There is the opposite temptation too: the temptation to heighten differences; may it be that I am succumbing to that temptation?) Many of the current attempts to view art semiotically can be explained, though they cannot be justified, by this temptation.

Somewhat more sophisticated is the attempt to generalize.^1. Often this attempt is inspired by the notion that generalizing is one of the things one does in science. Charles Peirce’s semiotics is heavily influenced by this notion and to a considerable extent vitiated by it. There is such a thing as pseudogeneralization, which is empty or sterile generalization, and much of Peirce’s work that he regarded as generalization is really only pseudogeneralization.

In between, there is a process that I may call decoding.^2. It consists in treating what appear to be ordinary, straightforward statements as coded messages. ‘Code’ is used in many senses, but what I have in mind here is the sense in which, say, ‘tanker’ is a code word for stock prices, ‘freighter’ is a code word for bond prices, ‘north’ is a code word for rise, and ‘south’ is a code word for fall. Given this code, the message ‘tanker north’ can be read either as clear or as coded; taken as clear, it says that a tanker is to the north, and taken as coded, it says that stock prices rise. Now there are cases where a statement using the word ‘language’ can be read as if a coded reference to gestures, or to kinship systems, or whatever. When I speak of ‘reading as if coded’, I am using ‘code’ metaphorically, for one who does this doesn’t suppose that the statement was put into code for the sake of secrecy. Rather, he tries the experiment of taking the word ‘language’ as if it were a code word for this or for that, and sometimes the experiment yields interesting results. The result will be interesting only if enough other words be taken as coded in the same code so that entire, extended discourses can be read as if coded.

Codes will have their limits; not every statement containing the word ‘north’ can be read as if coded in the above-specified code; for instance, ‘Nothing is north of the North Pole’ cannot be read this way, nor can ‘North is at right angles with East’. A proposed set of code readings will be the more interesting, the larger the set of discourses that can be read either as clear or as coded in this set of code-readings.

When someone proposes, explicitly or implicitly, to give the word ‘language’ a code-meaning, the ordinary way of describing what he is doing is to say that he is using the word metaphorically. For instance, talk of body language or of the language of flowers, commonly styled metaphorical, may equally well be viewed as a certain kind of coding. A certain kind, not just any kind; the coding system has to be such that, when all the words in a discourse that can be read as coded are so read, the discourse retains the same truthvalue that it has when read as clear. This constraint amounts to requiring an isomorphism between clear and coded readings.

The technique exhibited in a rudimentary way in such double readings was brought to explicitness by nineteenth-century mathematicians. I refer to Ernest Nagel’s fine history of the episode.^3. It led to fruitful results in algebra, geometry, and symbolic logic, and was a step (1) in making mathematics more abstract, (2) in disengaging the intellectual from the intuitive component of mathematical knowledge, (3) in leading up to David Hilbert’s formalism, and (4) in leading up to Hilbert’s use of ‘interpretation’ or modeling to prove consistency. These developments gave geometry a deductive rigor far beyond what it had in Euclid, and they refuted Kant’s thesis that intuition plays an indispensable role in mathematics.

But we should recognize that the technique is different from the technique of generalization. That the same formal system can be interpreted in two (or more) ways does not show that these interpretations have a concept in common; just the opposite. This is because the coding, which makes a double reading possible, has to be systematic. The principle of duality in projective geometry allows us, salva veritate, to take ‘point’ as a code word for line, provided that we simultaneously take ‘line’ as a code word for point. And what point and line have in common is only that they are related to each other by certain symmetries. Point differs from line in the same way as line differs from point. Point and line are like each other in that, in a certain respect, each is different from the other in that same respect.

The example is significant as proving that likeness (resemblance) cannot always be reduced to sharing a property. That kind of likeness which is displayed by the clear meaning and the coded meaning of a code word is not generic likeness; and therefore the possibility of setting up a code in which ‘language’, ‘vocabulary’, ‘syntax’, ‘phrase’, etc., have coded meanings alongside of their clear meanings does not suffice to establish that there is a semiotic genus of which speech is one species and the phenomena signified by the coded meanings are another species. The formalistic approach is revealing, but it does not reveal generic properties.

There is a particular version of formalism that is proving attractive to semioticians. Every sign has two ‘faces’ (Saussure): its form and its content. Here I must digress briefly on a point of terminology. In general it is convenient to follow Hjelmslev in drawing two distinctions within the sign; in his terminology the two faces are called expression (not form) and content, and then each of these two is subject to a second distinction between its form and its substance. The only inconvenience of this terminology is that the term formalism becomes inappropriate, and ‘expressionalism’’ wouldn’t do because it is preempted. So I will accept the inconvenience and retain the now inappropriate term ‘formalism’ to signify concern with the expression-face in abstraction from the content-face.

As one of our important concerns in this Colloquium will be zoosemiotics, I want to call attention to the possibility that when we ascribe signs to brute animals, we are speaking formalistically. I do not lay this down as a thesis; my role here is to ask questions. May it not be that the so-called animal signs resemble human signs in their expression-face and yet lack a content-face, in other words, have no meaning?

The hypothesis that wherever there are phenomena that appear to be the expression-faces of signs, there will also be found phenomena that appear to be the content-faces of signs, is a respectable hypothesis; but if it is to be a scientific hypothesis, it must be susceptible of verification, i.e., of confirmation or refutation; and, if it is confirmed by the consequences that it predicts, then it must pass the further test of being compared for simplicity with alternative hypotheses.

What if no alternative hypotheses can be thought of? In the case that concerns us, when the expression-structure is elaborate, it is tempting to say that the only alternative to the hypothesis of an accompanying content-structure is the hypothesis that what appears to be an expression-structure is due to chance, and that the more elaborate the structure, the more improbable is the explanation by chance. Since we are speaking of living organisms, the alternatives can be stated more pointedly in terms of function. Most organic structures have functions, and the more elaborate a structure is, the more likely it is to have some discoverable function.

But in addition to the fact that cases are known of elaborate biological structures that have no justifying function as far as we can see, it is appropriate at this juncture to bring in an entirely new consideration, namely, diachrony. The injection of the diachronic viewpoint affects semiotics in two main ways. First, there is the inertia often displayed in evolution, whereby something—the dorsal plates of dinosaurs, the sabre-teeth of tigers or of boars, the horns of various reptiles and mammals—continues, blindly following along a certain line, to get bigger and bigger, to the point where it ceases to be an asset and even becomes a liability; the phenomenon, in short, that moved Theodor Einer to posit alongside of natural selection a cause which he called ‘orthogenesis’. Besides this extreme inertia which is, viewed animistically, blind and intransigent, there is also a lag which would only be called sluggish, but it need not here be distinguished as a separate case; any inertia whether long-or short-lasting spoils a synchronic functional analysis à la Malinowski. The second main way in which the injection of the diachronic viewpoint affects semiotics (and everything else) is that things may change function. An organism may perform (accomplish, fulfill) function A by means of structure S, then S may come to serve also, incidentally, function B, then the need for function A may wane, or the need for function B may wax, and function B amay come to be the main function served by structured. Speaking with specific reference to human purposes, Wilhelm Wundt called this phenomenon the heterogony of ends; Hans Vaihinger called it the preponderance of means over end; Gordon Allport, Robert Merton, and others have discussed it.^4. In the first half-century or so of response to Darwin, this phenomenon was put forward as an alternative to finalistic (teleological) explanations; at the present time it deserves consideration as an alternative to semiotical explanations. The hypothesis that plants and brute animals act purposefully was challenged as anthropomorphic; a responsible defense of the hypothesis that they act semiotically will address itself to the same challenge. A teleological explanation of biological phenomena would conceive Nature as a purposive agent, guiding the course of organic evolution; an alternative explanation along the lines I have sketched would see each change as caused a tergo.

We may consider a series of species together and find that, when considered together, it is as if they made up different and successive stages of one individual. When the old ‘teleologische Naturbetrachtung’ did this, the counterexplanation was put forward that each stage must be considered separately in its own terms, which means in terms of its needs and its means. The insects considered in one of Professor Sebeok’s papers^5. furnish an example. The third species in the series seems, considered as a later stage of the first and the second, to be fooling his wife; he presents her, it seems, with a package that is all wrapping and no gift, but she discovers this only too late, after he has made his getaway. Now the sceptic would be at once suspicious of the fact that this episode occurs at each copulation. Invariably the wife is fooled. What has happened to the philosopher’s insight that deception and lie can only occur against a background of truth? Deception depends on ambiguity; and only he who can tell the truth can deceive (Plato). And the sceptic will shrewdly observe that the semiotic plausibility of the example depends on construing Insect number 3 as a later stage, a more sophisticated and worldly stage, as it were, of numbers 1 and 2. Number 2 was exploratory; he stumbled on a ruse that rescued him, and (for whatever reason) then moved into stage 3, in which, as a steady routine, and without giving the matter any thought, he employed this ploy.

I have cautioned against three dangers: the dangers of assimilating, of generalizing, and of coding. The main question to which I am addressing myself is this: What is the proper business of semioticians; or rather, how do we recognize our business? How do we recognize a sign? Here I have to say that Peirce and Morris gave the inquiry a wrong turn. Besides trying to generalize the concept of sign, they tried to define it. Peirce tells Lady Welby that “I despair of making my own broader conception understood.”^6. What reason to think he has a conception? More specifically, what reason to think that there is a conception of something common to the icon, the index, and the symbol? It is conspicuous that in his studies of semiotics, Peirce uses as his chief heuristic guide not any inductive study and not any feeling for the distinctively semiotic, but rather the combinatorial possibilities offered him by his three categories.

But it is another aspect of Peirce’s philosophy that I concentrate on here. Peirce’s philosophy has many aspects and many themes; and it is not a mere heap of ideas from which one can deftly pick out some while leaving the others quite behind. His concept of sign involves interprétant; his concept of interprétant involves mind; his concept of mind is generalized (to include ‘quasi-mind’) beyond recognition along panpsychist lines dictated by his synechistic idealism. His pragmatism and his anti-Cartesianism must be mentioned in the same breath.

Where a strict Cartesian would limit mind to consciousness, a Peircean would construe it far more widely. Instead of admitting it only where he has to, he would admit it everywhere that he can't be prevented from admitting it. (Descartes seeks to restrict; Peirce, to extend. Descartes seeks, as it were, discontinuities; Peirce, continuities.) Their respective views of transspecific communication nicely exhibit their differences. Peirce socializes with his horse and his canary. Descartes grants to brute animals feelings but not thoughts.^7. No wonder, then, that Peirce refuses to separate thought in the strict sense—logos, ratio—from feeling and from will. His synechism directs him to say that in brutes, the rational element is not lacking altogether but is merely present in imperceptibly low degree.

Without attempting an exhaustive or a conclusive proof that semiosis cannot be defined, I will mention three representative attempts.

First, surrogate. A Sign is a surrogate, or substitute, for that which it signifies. At best this gives the genus of a definition, but not the specific difference, since even if all signs are surrogates, not all surrogates are signs. So next we must seek the specific difference. But it is manifestly idle to define a sign as a significant or semiotical surrogate.

Worse than idle, though, it is also not true. Some signs function in some ways as surrogates, but in no case are we warranted in saying that to be a sign is to be a surrogate. I will show what I mean by making reference to Morris’s example (Signs, Language, and Behavior, p. 6) of a road sign. How does the message ‘The road ahead is blocked by a landslide’ function as a sign to the automobile driver? Morris answers that ‘‘in some sense. . . the words control . . . behavior... in a way similar to (though not identical with) the control which would be exercised by the . . . obstacle. . . . The . . . words are in some sense ‘substitutes’. ...” I find myself obliged to point out that not only does Morris fail to show, but no one before or since him has succeeded in showing, what that sense is.

One sees that Morris hopes to define sign; that he hopes to do it in terms of the concept ‘substitute’; and that he hopes to work out a criterion for substitution along these lines, ‘ x is a substitute for y if x is similar to y in a certain way’. Now why is it that no one has been able to achieve that aim by specifying (in behavioral terms) what the ‘certain way’ is? The answer is simple. The requisite certain way in which x is similar to y is that it is taken to be substitutable fory. But ‘taking’ is a mental, not a behavioral notion. Far from defining sign in terms of substitute, we are now thinking of defining substitute in terms of taking; we might as well bypass substitution altogether, and, if we still aspire to define the sign, define signifying in terms of taking. This would be a genuine genus-and-species definition, in that taking-as-a-sign is a species of taking, and ‘x is a sign of y to z’can be paraphrased as ‘z takes x as a sign of y’. As my last word upon quitting the subject, I mention that the inspiration for my definition comes from Peirce. Peirce considers the connection between similarity and the judgment of similarity, and controverts the usual view. The usual view is that our minds judge A and B to be similar because they are similar; Peirce holds that, in the last analysis, if A and B are similar it is because our minds judge them so.^8.

My first objection to the surrogate-conception has been that its assumption about similarity is untenable because it assumes a hysteron proteron. My second objection is that the sense or the way in which a sign is a surrogate is far removed from the ordinary sense or way. Consider Morris's road sign. If I come to a landslide, I back up, turn around, and take some detour to my destination. If, a mile before I would have come to the landslide, I come to a sign, ‘Landslide one mile ahead’, and I believe the sign, I stop traveling on that road, and take some detour to my destination. The landslide makes me stop and take a detour; the believed sign makes me stop and take a detour; here is the similarity between the believed sign and that which it is a sign of. But why should we call this substitution, or surrogation? The sign is not in the same place as the landslide; if it were, it would be superfluous, and if it were put where the landslide was, and the landslide were taken away, then the sign would become false.

Let us not confound signs with replicas. A glass imitation of a diamond is like a diamond in glitter and transparency, though not in the detailed quality of its glitter and not in other physical and chemical properties. An opera singer’s understudy passably resembles her in voice, if not in appearance; a movie actor’s double resembles him in appearance, if not in the range of his acting abilities. But none of these replicas is a sign; none of them signifies that which it replicates.

As a second representative attempt to define, reduce, or analyze the concept of sign, I choose Dewey’s proposal to reduce signifying to inferring. Not that the proposal is original with Dewey; it is the Epicurean concept of sign, set forth by Philodemus in his treatise on signs and copiously reported in Sextus Empiricus. My point of criticism, made with all possible brevity, is that the proposal of Epicurus and Dewey forces on signs a unity that they do not have.

My third choice is thirdness. Thirdness is for Peirce the category of law and of mediation. And it might seem that mediation is what the surrogateconception was trying to achieve. After all, one of the oldest definitions of the sign is that it ‘stands for’ its object. But there would be an important difference between Morris and Peirce, even though both talk of substituting and both talk of mediating. It is that whereas the surrogation-theory, as in Morris, points up the identity between sign and object, the thirdness- or mediation-theory points up the distinctness. In between the interpreter and the object stands the sign.

Of course, thirdness is many things for Peirce. Perhaps I should invert the statement and say that for Peirce many things are thirds. His favorite example of thirdness is law, but his theory of categories is largely given over to working out the complications, the proliferations, the ramifications of the deceptive utter simplicity of his initial display of one, two, three. As regards the theory of signs, I make only passing mention of his sixty-six varieties of signs and his conception (modeled after mathematics) of degenerate cases, in order to concentrate on the application of the categories to one another (the thirdness of secondness, etc.). After all the complications entailed by this are noted and reckoned with, there will remain the simple fact that the realm of thought is distinct from the realm of actuality. The expression in his terms of that simple fact will be that actuality (as a whole) is a second to thought (as a whole). Now the fact that semiosis is unanalyzable—that it may be treatable as a variety of thirdness, but that we are unable to specify in a non-redundant way what variety of thirdness it is—will manifest itself in Peirce’s system this way, that there will be some formula containing the word ‘thirdness’, such that the proposition consisting of ‘semiosis is identical with’ followed by that formula will be added to Peirce’s system (not deducible within it), and will have the status of a synthetic a priori proposition, not of an analytic proposition. Peirce is able to disguise from himself the need for synthetic a priori propositions connecting his thought-world with actuality by assigning such propositions the status of hypotheses needing further investigation, with the hope that sufficient investigation will present them no longer as synthetic a priori, but as analytic. For the purposes of the present colloquium, it is enough to say that so far the attempt to define semiosis as a species of thirdness has not succeeded and shows no prospect of succeeding.

One last point about attempts to define the sign: there may be circles of interdefinability. Cases are known where there is a set of concepts such that any member of that set may be defined in terms of one or more other members of the set, but no member can be defined otherwise. (Example: In logic, ‘or’ can be defined in terms of ‘not’ and ‘and’; also, ‘and’ can be defined in terms of ‘not’ and ‘or’.) It is a familiar doctrine in analytic philosophy that ethical notions form such a set; and R. M. Chisholm^9. has argued that behavioral definitions of purpose and the like appeal to a notion of intentionality, in which case intending, purposing, seeking, trying, etc., form an inter-definability-set. Peirce expressly (4.530) rejects the definition of a sign as a substitute, but in the definition (in a letter to Lady Welby, already quoted) he himself offers, he defines it in terms of determination. But determination, if it is to yield a result that is neither too wide nor too narrow, must be understood to mean something like intention. And Peirce himself was aware of the interdefinability of ‘sign’ and ‘mind’. Here, then, is an interdefinabilityset: intention, sign, mind.

It might be thought that all definition is circular in this way, but it is not so. For example, when Russell defined the concept of number it was urged against him by Joseph^10. that his definition of number was circular because he defined the number one as a class and a class is one class, so that he was defining one in terms of one. Russell successfully refuted this objection, and therefore I am not making a point that can be made of all definition when I say that the concept of sign can only be defined circularly as a member of some interdefinability-set. And interdefinability is tantamount to indefinability, since we think of definability as being an asymmetrical relation.

More modest than attempts to define signs are attempts to classify them. Ernst Cassirer, for example, in An Essay on Man, classifies signs into signals and symbols. Cassirer says that symbols have a new dimension; this dimension is, however loath he is to admit it, the dimension of intentionality explored by Brentano, Meinong, and Husserl, and applied to present-day philosophy by Chisholm, as remarked above. I mention, but do not develop, the objection that Cassirer neither justifies nor illuminates his metaphor of dimension; instead, I propose to accept Cassirer’s distinction and to give it a turn that he never thinks of giving it.

To explain myself I must sketch a philosophy of classification. When there is a genus with two species, it may be that species A is characterized by two positive properties, the generic property that it shares with B and its specific difference that differentiates it from B; and correspondingly for B. This according to Aristotle’s biological writings is the structure of a good classification. But what if the structure is that A has just one positive property, the generic property, and B has two positive properties, the generic property and also a differentia specifica? We can, if we like, redescribe this by saying that each of the two species, A and B, has two properties, one generic and one specific, provided that we go on to say that they differ in that the specific property of A is a negative, or privative, property, and the specific property of B is a positive property. This is the way in which Aristotle treats plant, brute animal, and man. (In other words, Aristotle’s treatment of the three kingdoms does not conform to his own strictures on classification.) Plant is living; brute animal is plant plus something (mobile and sensitive); man is brute animal plus something (rational).

Now by Cassirer’s account the relation between signal and symbol is that symbols are signals plus something. They fit the second pattern, rather than the first, which is to say that sign, if it deserves to be called a genus at all, is a genus of the second sort, not of the first. The new turn that I propose to give to Cassirer’s account is to align it with Saussure’s. Let us consider identifying Cassirer’s signal with Saussure’s signifiant, and his symbol with Saussure’s signe. Then the property which, added to a signal, gives a symbol would be the property of having a signifié. The dimension which, according to my suggestion, symbols add to signals is the dimension of content, or meaning.

It is not my claim that this is what Cassirer really had in mind; neither is it my claim that we ought as I suggest to identify Cassirer’s new ‘dimension’ with Saussure’s second ‘face’. My self-chosen role in this paper being the role of questioner and not of answerer, I content myself with having made the suggestion. I do, of course, make this much of a claim, that the questions I pose are serious questions, worthy of investigation, not idle or wild questions.

If we do match up Cassirer’s contrast with Saussure’s, then certain adjustments in Saussure’s conceptual framework and in his terminology are called for. To speak of a signifiant that doesn’t signify, a signifié that isn’t signified, an expression (Hjelmslev) that doesn’t express, a content that isn’t the content of anything—every one of these locutions seems to be a contradictio in adjecto. The reason is the same for all: all are relative terms; to be related is built into them. But we can perform on built-in properties an operation of subtraction, such as Hume does when he writes: ‘‘Every effect necessarily presupposes a cause; effect being a relative term, of which cause is the correlative. But this does not prove that every being must be preceded by a cause; no more than it follows, because every husband must have a wife, that therefore every man must be married.’’^11. There being no terms in ordinary language for the derelativized results of these four semiotic terms, as there is ‘man’ for ‘husband’ derelativized, we might use some prefix and speak of quasi-expressions or semi-expressions or the like.

But it may seem that in Saussure the real difficulty remains untouched, and demands something more than ‘adjustment’. According to him (Part II, Ch. 4), signifiant and signifié presuppose each other so intimately that each would be shapeless without the other. But Saussure himself resolves the difficulty.‘‘Linguistics then works in the borderland where the elements of sound and thought combine; their combination produces a form, not a substance.’’ (This passage is the textual basis for Hjelmslev’s distinguishing, as noted above, between the contrast of form and substance and the contrast of expression and content.) It is not the expression-substance and the content-substance that presuppose each other, but their forms; when an expression-substance is used to signify a content, then expression-form and content-form emerge. Form and substance are not correlative, i.e., do not presuppose each other; the same substance may (functioning in different systems) have different forms, or it may function in no system and so have no form; and the same form may be the form of different substances.

Having introduced the Saussure-Hjelmslev distinction between form and substance, we may now say that it furnishes us the terms that we hitherto lacked. Instead of a quasi-expression or a semi-expression, the same purpose will be served if we speak of an expression-substance, and similarly of a content-substance.

The import for semiotics is momentous. Semiotics is primarily the study of signs, and (as a built-in property) a sign has a signifiant and a signifié, an expression and a content. But there may be expression-substance without expression, and the semiotician may include them in his study, even though they are not among his primary objects, because they are relevant to his primary objects. The semiotician will want to know which facts of semiosis are necessary, which contingent; and a part of this question is the question, which expression-substances could be expressions and which ones could not be. The evolutionary version of this question would be, which systems of expression-substances that are not expressions might evolve into systems of expression-substances that are expressions, and, in particular, from what did genuine human expressions evolve? And the study of what Cassirer calls signals might be relevant to semiotics even if signals are not signs. I will return to the subject of evolution at the end of this paper.

I have considered two ways of delimiting signs. The first way, definition, went to its doomed failure. The second way, classification, encountered a major question: might it not turn out that an alleged classification of signs into species A and B could be regarded as a distinction between signs proper, A, and certain non-signs, B, that were potentially the substance for signs? The possibility was illustrated by Cassirer’s distinction between symbols and signals. Now, in the last part of my paper, I will consider a third way of delimiting signs, the way that I regard as most promising. Whether or not we agree that symbols (in Cassirer’s sense) are the only signs, they are certainly signs par excellence. Let us then examine the properties of symbols, other than the built-in property of being intentional, and see whether, on the basis of these, we can find any warrant for calling other things signs, though not signs par excellence. The method is proposed by Plato in the Phaedo (100a), where Socrates says,‘‘This was the method which I adopted: I first assumed some principle which I judged to be the strongest, and then I affirmed as true whatever seemed to agree with this ...; and that which disagreed I regarded as untrue.’’ But I modify this method in a Peircean way; I replace the dichotomy, agree or disagree, by a series: wholly agree, largely agree, somewhat agree, no more agree than disagree, etc. The resulting modified method is called by Aristotle the method of more and less.

A being that has intentions also has volitions, has freedom, is capable of arbitrary decisions, has open-ended or infinite power, understands mere possibility, and is capable of signifying negation. At least a human being has these several powers. But must every being that has one of them have all?

In general, when we specify more than property as a criterion, or as a necessary condition, for some concept, we should anticipate the possibility that there are things having n of these properties but not the remaining m-n properties. A. Ingraham, as reported by Ogden and Richards, entertainingly described the possibility using the example ‘wall’. C. D. Broad and Hilary Putnam have developed the idea further, the former using the metaphor of ‘degrees of freedom’ and the latter coining the label ‘cluster-concepts’.^12. The import of these insights for my touchstone is that we must not expect a single touch of the touchstone to answer the question whether this or that is a sign, nor must we expect that the answer will be either yes or no. A phenomenon examined by the touchstone may turn out to have some but not all of the properties that speech-signs, the signs par excellence, have; there will be degrees of closeness to the paradigm case.

The first two properties of a signifier, intending and willing, are both expressed in ordinary language by ‘mean’. When communication using language-signs occurs, the speaker wills to emit something that will be taken as a sign by the hearer. In other and plainer words, the speaker means something that is taken as meaningful by the hearer. According to the scholastic language reintroduced by Brentano into modern philosophy, when the hearer takes something, x, as a sign of something, y, he intends y, and y is called the intentional object of x. This technical sense of ‘intend’ is easily distinguished from the more familiar sense in which for example someone intends to keep a promise and in which ‘intention’ means the same as ‘intent’ and ‘intentional’ means the same as ‘deliberate, on purpose’. The familiar and the technically philosophical sense of the verb ‘intend’ and derivatives are easily distinguished, but in semiosis par excellence there is a certain necessary connection between them: the speaker intends in the familiar sense, and the hearer intends in the philosophical sense: the speaker intends _f that the hearer shall intend _psuch and such.

When the communication is nonlinguistic, there are analogues to speaker and hearer: in plastic art, artist and spectator (observer, esthete); in music, musician (split into composer and performer) and audience.

It is common parlance to speak of signs that are not intended in the familiar sense; a cloudy sky as a sign of rain is an easy example. Human beings may exhibit signs of this sort: a red face is a sign of being hot (though ambiguously, because it is also a sign of being angry, of being embarrassed, and of being strangled), a bluish-white face is a sign of being cold. These will not be signs par excellence, because there is no signifier; there is not, for example, Mother Nature.

Taking the touchstone seriously will lead to surprising and seemingly objectionable results; such as, that not all speech is signification par excellence. Things that are said ‘spontaneously’, that are ‘blurted out’, might not count as willed. But why should this perplex us? There is nothing odd about saying that only speech, but not all speech, is semiosis par excellence. But no doubt every utterance is an utterance that could have been uttered intentionally (deliberately, on purpose), and this is what people have in mind when they say that speech is essentially intentional.

The next-mentioned property after intention (deliberateness) and intentionally is freedom. For my part, it is not my intention, nor my pretension, to solve the questions of freedom, but only to peg one set of questions to another. When the bees dance their nectar-dance, do they give their signs freely? I suppose the general opinion would be negative; a common way of putting it, influenced by computers, would be to say that they are ‘programmed’ to do what they do; we find no occasion to say that one bee might refuse to participate in the dance if (like Achilles) he was angry at the other bees, or was engaged in a campaign to show that he was his own master. Of course as we get to know the bees better we may start to find occasions to say these things, in which case the nectar-dance would cease to seem an example of unfreedom.

Freedom has indefinitely many manifestations; one of them is the capacity for arbitrary decisions. Leibniz^13. was fond of saying that ‘liberty of indifference’ was a fiction and that where there is a perfect balancing of motive forces, a being will be moved by neither, and in particular where there is a perfect balancing of reasons, a reasonable being will do neither. There is one case he didn’t sufficiently consider. He did amply consider such cases as that of the traveler at the crossroads who weighs the reasons for going left against the reasons for going right and, finding them equal, goes neither to the left nor to the right, i.e., comes to a standstill. But he didn’t sufficiently consider the case where the traveler weighs the reasons for keeping going against the reasons for coming to a standstill and finds them equal; he considered that case, to be sure, but not sufficiently, because he contented himself with saying that it couldn’t happen. The case where the alternatives are to move and to come to a standstill, i.e., to move and not to move, is of interest because it is protected, as it were, i.e., it is assured of existence, by the Law of Excluded Middle, which means that it is the case where it is logically impossible to say,‘‘Neither alternative will be followed.’’ Leibniz did not propose to repeal the Law of Excluded Middle, but he did propose to withdraw challenges to it from the arena of choice; in such cases one could not say,‘‘Neither alternative will be followed,’’ but one could say, and he did propose to say, ‘‘Neither alternative will be chosen. ’’ His philosophical opponent, Samuel Clarke, defended the possibility of choice under these conditions—arbitrary choice, and without going into the deeper philosophical issues of the controversy I may say that at the very least Clarke’s contention fits prima facie appearances better than does Leibniz’s. When high-speed vehicles came into use and it became advisable to establish a convention, either that vehicles keep to the right side of the road or else that they keep to the left side, it was immediately apparent that the choice would be arbitrary. And in empirical fact it frequently proves to be the case that, when some group of human beings is confronted with an arbitrary choice between A and B, one subgroup arbitrarily chooses A and the other subgroup arbitrarily chooses B. In the case of automobiles, it is inconvenient that some nations made the left-side choice and some the right, but whichever choice was made, it was an arbitrary choice, in defiance of Leibniz’s Principle of Sufficient Reason, and having no reason in human right-handedness or in anything else. Units of measurement exhibit, most of them, the same arbitrariness; the English yard may be based on the human body, and the metric-system meter may be based on the aliquot division of the circumference of the earth, but though these applications suggest the choice of unit it could fairly be said of them that they were not (pace Leibniz) reasons in any respectable sense, but rather were admittedly extrinsic and strictly irrelevant appeals where genuine good reasons were lacking.

All this bears on ‘l’arbitraire du signe’. Human beings are not, as a rule, happy to make arbitrary choices; in fact, it is their last recourse. Now of course Saussure didn’t mean that at the first parliament human beings considered what name to give to the tree and some said arbor, some said arbre, some said Baum, some said tree; it was the other aspect of arbitrariness that he was focusing on. It was exactly because the choice on the part of some speakers to say arbor was arbitrary that their successors continued to say arbor for no better reason than that their forebears had; and so when because of externally caused sound-change their successors replaced arbor by arbre, there was nothing to stop them.

The willingness to exercise arbitrary power admits of degrees. The word ‘arbitrary’ suggests ‘despotic’, ‘tyrannical’, ‘dictatorial’, and so there is the further suggestion that nice people would not exercise arbitrary power. But a fuller discussion of the arbitrary would divest the word of these suggestions and would treat as a mark of intelligence the recognition that an enormous number of decisions, major and minor, are arbitrary. That is, the more intelligent one is, the more will one, in facing or in considering situations that call for arbitrary decisions, recognize them as of this type.

That someone recognizes the need for arbitrary decision does not entail that he acts on his recognition. It is for this reason that we posit another factor in the psyche besides intelligence, namely personality. I should accordingly reformulate my characterization of freedom as including ‘the capacity for arbitrary decisions’. Since this capacity includes a factor of personality as well as a factor of intelligence, and since only intelligence is relevant as a criterion for semiosis, I reformulate my proposition of two pages above as follows: Freedom has indefinitely many manifestations; one of them is the capacity to understand which decisions are arbitrary.

The open-ended or infinite power of the human mind has been claimed but never adequately analyzed. The infinity claimed is a so-called potential infinite. It involves the unknown future and the concept of non-actual possibilities, in the following way. A description of a past event need not involve non-actual possibilities. Thus, a description of all the objects that a certain human being has signified, up to a specified date, will be a finite list and yet will do justice to the facts. The same cannot be said of the task of being prepared to describe the future, not-yet-signified significations, whether of a person still living or of one not yet even born. Whether this is because the future is inherently indeterminate or because, though determinate, it is not known to us, is a perennial metaphysical question which does not concern the would-be describer. In either case, the list of one who would describe future human significations, and who wants to be sure that his list will not prove incomplete, must be a list of possibilities, including non-actualized possibilities, and each possibility must be either included in the list or precluded from it, and even if there are infinitely many possibilities that the lister is in a position to preclude, still—such is the claim—infinitely many possibilities will remain.

To assess this claim would be too detailed for the present paper, but it is evident that there is some great difference between the list a describer must have at hand if he is to be prepared to describe everything signified by a human being and the list he must have if he is to be prepared to describe everything that could be signified by a bee. It is true that if we were content with high probability, rather than certainty, our list would not prove incomplete; a list much more like the list adequate for a bee would be adequate for a human being; in any case a finite list, rather than an infinite one. And it is quite possible that there is some finite list that would in fact never prove incomplete. But what we are trying to avoid is not just a list that would prove incomplete but a list that might prove incomplete.

My discussion of infinite power has also, unavoidably, discussed mere possibility. In finishing my comment, then, on my list of powers that in semiosis par excellence accompany the semiotic power, it remains only to speak of negation.

The power to express negation marks the highest achievement of semiosis; without it, even human semiosis lacks the power to express abstract thought. Philosophers had made this point from time to time, but it was most forcibly brought to the world’s attention by Freud (Traumdeutung, especially Section VI-C, and ‘‘Die Verneinung’’ of 1925, translated in Volume 19 of the Standard Edition). The dream-language lacks any word for ‘not’, with enormous consequences. One consequence is ambiguity as regards wishing and fearing; to fear X is to wish for non- X, and since (whatever X may be) X and non-X are depicted in the same way in dreams, they are distinguished only by the feeling-tone, i.e., by the difference in the affect—wish or fear, respectively—accompanying the dream. The connection between negation and possibility is prescribed by the Law of Excluded Middle, p v ~p, together with the Law of Non-Contradiction, ~(p&~p). The connection is that possibilities come in pairs, such that at least one (so as not to violate Excluded Middle) and at most one (so as not to violate Non-Contradiction) of each pair is actual. I particularly call attention to the point that only human language has a general sign for negation, though various special cases can be expressed nonlinguistically (e.g., prohibition can be expressed by a conventional gesture such as an upraised hand).

It is sometimes maintained, e.g. (among moderns), by Descartes and by Cassirer, that the difference between brute and human is a difference of kind, not of degree. The position I have taken in the present paper may seem to make the same commitment. The next point I want to make, and the final point of my paper, is that a difference in kind between A and B does not preclude evolution from, A to B. To see this, all we have to do is to understand the logic of ‘change’.

It is common parlance to speak of one thing changing into another, and it is a common philosophical move to contend that change is really replacement, or succession. When the boy changes into the man, the boy as such ceases to be and in his stead a man comes into being. This philosophical viewpoint may be captious when applied to individuals, i.e., to things having what philosophers call continued identity, but when applied to abstract beings it makes perfect sense. Three cannot change into four; as Plato put it in the Phaedo (102-5), when evenness advances, three retreats. Nor can two change into three; but yet a family of two can change into a family of three. Numbers cannot change; two cannot change into three. But numbers can change; the number of people in a family can change from two to three. The number two (which is a number) cannot change, but the number of people in a family (which is also a number) can change. The example proves that the word ‘number’ is ambiguous as between two different kinds of things. It is part of the ‘logic’ (in the sense of Wittgenstein, Waismann, and Wisdom) of phrases signifying numbers of the second sort that they have continued identity, as individuals do, and are subject in part to the same conditions.

Now for an example that concerns organic evolution. The human being is in process of evolution in various ways. One of the ways is that he is losing teeth. This is the evolutionary meaning of our wisdom teeth. Evolution being gradual, a species that had thirty-two teeth does not all of a sudden turn into, i.e., become replaced by, a species that has twenty-eight teeth. There are various distinct ways in which this gradual loss of four teeth could occur; one of these ways, the way we are actually following, is that the four rearmost teeth are neither fully normal, healthy teeth nor merely vestigial teeth, but are defective teeth which erupt much later than the others and for which there is not sufficient room.

The example is, then, that though thirty-two cannot change into twenty-eight, yet thirty-two teeth can change into twenty-eight teeth, and moreover can do it gradually.

The general pattern of gradual transition will be this. At one stage, s _1’we will find instances that clearly answer to the description A; at some later stage, s_4’we will find instances that clearly answer to the description B, logically incompatible with A; in between, we will find a stage s₂ where there are instances which, though not clear instances A, yet fit description A better than B; and in between s₂and s₄ a stage s₃where there are instances which, though not clear instances of B, yet fit description B better than they fit A. And so on, without any preassignable limit to the number of intermediate stages that can be discriminated. As regards our evolution toward a species with twenty-eight teeth, I take it that we are now in something like stage s₂; it is apter to describe us as having thirty-two than as having twenty-eight teeth, and yet the description has to be qualified.

The logic of the word ‘change’ is the logic virtually used not only by those who speak of evolution but by those, e.g., historians, who speak of transition from one period of history to another. For example, Cassirer, as noted by Randall,^14. held that the Middle Ages were a distinct, i.e., a conceptually discrete, period from the Renaissance, while granting (as he could not help doing) that the transition from the one to the other was chronologically continuous. And, coming back to organic evolution, this same logic is applicable in particular to the evolution from non-human to human, and from the absence to the presence of semiosis.

I made a point of mentioning Cassirer, because I had mentioned him earlier in connection with signals and symbols. In the Essay on Man he is so emphatic in claiming that symbols are a different order (a different dimension, a different kind) from signals that transition from one to the other seems out of the question. But the knowledgeable reader will see that exactly the same logic makes a transition from the Middle Ages to the Renaissance out of the question. And in one sense this is true. The Middle Ages, defined by their nature and not by their date, can never change into the Renaissance or anything else. The Middle Ages do not change; when the Renaissance approaches, the Middle Ages simply withdraw. This is true, in the typological sense, and the only thing to add is that besides the typological there is the chronological sense of change. Ambiguity is part of the logic of the word ‘change’; change is impossible or is possible according as that which changes is characterized typologically (formally) or chronologically (materially), the latter being the characterization which warrants our speaking of continued identity throughout change.

Cassirer gave his life to the study of change; but I have searched his writings in vain for evidence that he ever grasped this distinction and brought it into focus as a theme. I must say the same of Humboldt. Indeed, Humboldt is at pains to display his confusion. He says,‘‘So natürlich die Annahme allmähliger Ausbildung der Sprache ist, so konnte die Erfindung nur mit einem Schlage geschehen.’’^15. The reason is that ‘‘der Mensch ist nur Mensch durch Sprache; um aber die Sprache zu erfinden, miisste er schon Mensch sein.’’ Now the phrase ‘‘mit einem Schlage’’ sounds chronological, as if it referred to time, but just there is where the confusion lies. The phrase ‘at one stroke’ is infected by the same ambiguity to which ‘change’ is subject; and the formal fact that humanity is lacking ‘until’ the power of speech is present and vice versa is independent of and has no bearing on the material question whether every living organism that is human is clearly human, or on the material question whether in every succession of generations from clearly non-human ancestors to clearly human descendants there is a last ancestor that is clearly non-human and the very next generation—the child (immediate descendant) of this last, clearly non-human generation—is clearly human. Humboldt’s doctrine is plausible if taken formally, groundless if taken materially.

The ambiguity, and the distinction which resolves it, are to be taken into account in all discussions of evolution, whether of man, or of human speech, or of anything else. I discussed the general pattern of change from A to B. In the special case where A= non -B (meaning here by ‘non-B’ not just any member of the complement of B, but some member that is similar to B), e.g., the case where some non-human brute changes into man, it will always be plausible to say, speaking formally, that non- B changes at one stroke into B. But this formal truth exercises no constraint on matter, which can evade the formal sharpness by presenting instances that are in between (on the borderline, indeterminate).

Linguists have instituted a concept that copes with this fact. It is the concept of performance, as contrasted with competence. The concept has been put to many uses, and these have not been clearly distinguished from one another, nor has it been proved that they necessarily go together. But one use, perhaps the original use, has been to reconcile the ascription of a certain infinite power—speech—to man with the fact that only a finite portion of this power is actually exercised. It would be quite in keeping with this established use to deal with organisms of which we were doubtful whether to say that they had speech by saying that they definitely have speech competence, rudimentary though their performance may be.

Throughout this paper my leading idea has been to judge the less clear cases by the clearer cases, taking speech as the supremely clear case of semiosis. It has not been my intention to reject any alleged case, but only to urge that the allegation needs to be supported. I can put my point in terms of a contrast introduced by geometers, highlighted by Kant, and signalized by Cassirer.^16. The very concept of a datum (a given) comes from Greek geometers; and they oppose the data of a problem to the quaesita as beginning and end, respectively. The contrast is expressed more neatly with German words than with Greek, Latin, or English ones. My point, then, has been that the unity of semiotics is not something given to us but something to be sought: it is not gegeben, but aufgegeben as an Aufgabe.