Semiotic and Linguistics

JOSEPH RANSDELL

INTRODUCTION

A semiotic linguistics would differ significantly from linguistics as usually conceived in recent decades. The extent and nature of this difference became increasingly clear to me as I transformed the oral address I presented at the conference into the paper now before you, in accord with a growing understanding that the semiotic conception cannot be of real use to the linguist—or to any other student of the meaningful and the significant as such—until the methodic basis of this approach is understood. This involves assumptions about what the empirical objects of such studies are, and how such objects are to be regarded, which differ radically from what has usually been assumed in linguistics, and the attempt to understand the semiotic perspective within the context of the usual assumptions can only result in confusion. Consequently, I focus here primarily on what seems necessary (in the present intellectual situation) to bring about an understanding of the relevant assumptions, rather than on the development of the analytic conceptions of semiotic. I do hope, though, to provide a way toward a more authentic and useful understanding of the icon/index/symbol distinction in particular than has been available heretofore.

In the first part of the paper I focus on the contrast between a semiotic approach and what I call “hermetic” linguistics, the term hermetic being my own term for a certain way of conceiving the science which I believe to be common to what is usually called “structuralist” linguistics, to Chomskyan linguistics, and perhaps to some others as well. I argue that linguistics, so conceived, cannot be a genuinely empirical science. (Hence such knowledge of language as linguists of these schools have achieved—and I do not question any substantive achievements, but only a certain conception of linguistics as a science-has been in spite of serious confusion about the nature of their enterprise.) The formal validity of my argument is one thing, however, and its application to this or that particular conception of linguistics quite another. I do not adduce textual evidence here that it applies to this or that school of thought—though I see no reason not to make clear to whom I think it applies—because this would invite responses in which the two distinct questions were confused. Thus you must regard what I say as an “if-the-shoe-fits” argument, and make your own decision as to whom it fits, if you think it valid. The second part of the paper focuses on the assumptional basis of the semiotic approach. In the third part, I try to convey a working sense for a few of the basic analytic conceptions, particularly the icon/index/symbol distinction. The conception of semiotic explained here is due to Charles Peirce, who discovered it more than a century ago in an attempt to provide a radically new basis for the science of logic.¹ My representation of Peirce’s theory can fairly be said to be an informed one; however, it is not a textual exegesis but rather an attempt to restate some of his ideas in a way appropriate to the present audience and to the contemporary intellectual context generally.

I

During this century, the dominant conception of linguistics as a science has been that the primary task of the linguist is to describe languages, that is, to write grammars and dictionaries. Many linguists have also held that a further task is that of developing a single canonical form for such descriptions (a “universal grammar,” with the term grammar used in an extended sense to include principles of dictionary writing, and perhaps still other principles not usually thought of as grammatical). Languages are thus regarded much as if they were Aristotelian “primary substances,” that is, individual entities (albeit inferred rather than observed individuals) having form or structure; and the task of the linguist is to describe, by inference from linguistic phenomena, the structural properties of these individual entities at particular points in time, and perhaps also to observe and describe their changes across time (much as one might write biographies of individual persons). Then if one believes in “universal grammar,” one attempts to describe the “essence” or “secondary substance” they all have in common that makes them the kind of thing they are. Whether the inferences from linguistic phenomena to the individual language are inductive or hypothetical in character, whether the properties consequently to be assigned to the phenomena are supposed to be at a “surface” or “depth” level, whether the inferred structures are of this sort or that—all such disputes fall within the general agreement that the description of languages is the primary task of the linguist, if not the final one.

In view of this basically Peripatetic conception of the science, it is surprising that so much has been made of the difference between what John Lyons calls “modern linguistics” and the previous linguistics traditions.² Lyons specifies five supposedly differentiating features of “modern” linguistics: (1) recognition of the priority of the spoken to the written language; (2) a nonprescriptive approach to language description; (3) recognition of the priority of synchronic to diachronic description; (4) recognition of the langue/parole distinction; and (5) the adoption of a “structuralist” view, defined as regarding a language as a system of relations “the elements of which—sounds, words, etc.—have no validity independently of the relations of equivalence and contrast which hold between them.”³ But these differences are more apparent than real.

Feature (1) would betoken a methodological shift of some importance if it meant that the differentiating characteristics of the spoken word (e.g., kinesic factors) had been integrated into linguistics; but I see little evidence that the focus on the spoken language has in fact made any great difference in this respect. It has given rise to an extensive and sophisticated development of phonology, and this is surely important, but it is no more than an extension of the traditional concern, constituting no fundamental reconceiving of the science.

Feature (2) is of no significance: a prescription is simply a description functioning prescriptively, and linguistics of the sort in question here has had little concern with such functional differences. Moreover, language descriptions are always relativized to the speech habits of some specially selected class of persons in any case, and it is of no consequence that linguists now say they are not prescribing the “best” usage. Whether their works actually function prescriptively is a matter of how they are responded to by those who consult them, not of the linguists’ pious avowals and intentions.

Feature (3) is misleadingly stated. Anybody who has accepted the conception of “synchronic” and “diachronic” description could hardly have failed to realize that any “diachronic” description must be based on (or imply) at least two “synchronic” descriptions. What differentiates so-called “modern” linguistics is rather the formulation and acceptance of the distinction itself. I have argued elsewhere that the conception is bogus, and I cannot repeat that argument here.⁴ It will have to suffice to say that its apparent plausibility depends upon a pseudo-analogy with state-descriptions in the physical sciences, and that it is illegitimate because it overlooks the fact that a physical state-description does not describe laws—much less rules (which are the linguist’s pseudo-analogue to laws)—but rather singular states-of-affairs. If this is correct, then “modern” linguistics differs from traditional linguistics in this respect only in having fallen into a pretentious and unhealthy confusion.

Feature (4) is formulated by Lyons in such a way that it does not affirm the priority of langue to parole, perhaps because not all modern linguists have accepted that priority. But if it is only a matter of the recognition of the distinction itself (vaguely conceived, as it must be if he is not referring only to some particular school of thought), then it is questionable whether it has ever gone unrecognized. In any case, its importance is a matter of where the priority is conceived to lie.

Only feature (5) seems to point to a substantial difference, unless (4) includes an affirmation of the priority of langue. It is apparently formulated with an eye to the Saussurean “associative” and “syntagmatic” relations. “Associative” relations, which include relations of logical contrariety, are doubtless of great importance; but concern with these has been far more prominent in Europe than here, and it is still not clear precisely what these relations are supposed to be, inasmuch as contrariety relations of quite different types, with importantly different relations to empirical fact, are cited as examples (e.g., color contrariety would appear to be quite different from familial contrarieties). Saussure’s “syntagmatic” relations are analogous to the “syntactical” relations discussed in North American linguistics, but these are, of course, relations of traditional linguistic interest. However, the force of this supposed differentiation probably lies in what Lyons is trying to capture in saying the elements have no “validity” independently of the relations, etc. I take it that by validity he means linguistic identity (validity perhaps being his translation of Saussure’s valeur), and that he means the relations are “unmotivated,” that is, logically ungrounded in the terms related. Lyons may be confusing or conflating several distinct things here, as is indicated by his citing both words and sounds as examples of elements. Thus he might mean: (a) that relational linguistic predicates are ungrounded in nonrelational linguistic predicates, or (b) that relational linguistic predicates are ungrounded in nonrelational nonlinguistic predicates, or (c) both of the above, or (d) that no linguistic predicates of any sort (relational or monadic) are grounded in any nonlinguistic predicates.

I am not certain that everything with a right to be called “modern” linguistics is committed to any of these doctrines, or that there is any sense whatever in the idea—something of a shibboleth in Continental structuralism, in particular—that a “structure” can be constituted on the basis of the relations of identity and otherness (“equivalence” and “contrast”) only. However, the ungroundedness of linguistic predicates in nonlinguistic predicates (the fourth alternative mentioned above) seems to be an important feature of the dominant schools and tendencies in contemporary linguistics, and it would seem to be another way of expressing the dictum that langue is prior to parole. For I understand this dictum to mean that linguistic properties are indefinable in nonlinguistic terms; consequently a linguistic phenomenon is linguistic only because of the de facto application to it of a coexisting rule of language (langue) which assigns a linguistic property to it thereby. The term is assigns, rather than attributes, because there are no perceptible properties a phenomenon could possess which would justify the application of the rule, not even if the phenomenon should happen to conform to some perceptible pattern specified in its antecedent clause. (The reason is that it would otherwise be possible for something clearly nonlinguistic—a sound made by an animal or nonliving being, or a randomly occurring visual pattern, for example—to conform to it accidentally.)

Let me say this another way. On the view in question, a langue is a set or system of rules, and a langue rule applies itself to a particular phenomenon such that, if the phenomenon to which it applies itself satisfies its antecedent (which presumably describes a perceptual configuration), then the phenomenon ipso facto has whatever linguistic property is described in the predicate in the rule’s consequent. An individual person, on the other hand, can only believe or assert or judge that the phenomenon has that property, presumably on the basis of a perception of its configurational type and knowledge of the langue rule that describes that type in its antecedent, plus the assumption that the rule has already been applied to that case to constitute its linguistic identity. For if we do not suppose that the rule applies itself to the case, thereby assigning or bestowing the linguistic property on the phenomenon, then no given person’s judgment (“intuition,” as some prefer) that the phenomenon has that property could be either true or false. Or to put it the other way around: were the individual person the applier of the rule to the case, that person would be simultaneously bestowing and recognizing the presence of the property bestowed, which is not only nonsense but also collapses the langue/parole distinction. So, on this view, we have to suppose that the langue has the power to do its own applying. What sense that makes I do not know; but since it is not my theory I suppose I am not obligated to explain such a mystery. (One is tempted to say that a langue is conceived as a sort of superior and benevolent Mind which goes about giving things linguistic properties so we can have words with which to talk!) Such appears to be what the dictum of the priority of langue to parole comes to, and I understand it to be another way of saying that linguistic properties are not definable in terms of nonlinguistic properties.

In any case, I refer to any linguistics so conceived as a “hermetic” linguistics, since it regards linguistic properties as “sealed off” from extra-linguistic properties (i.e., recognizes no extralinguistic implications of the words for such properties), and I am primarily concerned here with contrasting the hermetic conception—which appears to be shared in common by the leading Continental schools, the Chomskyans, and at least some of those whom the Chomskyans characterize as “taxonomists”—with the semiotic approach to linguistic phenomena. I do not address myself to behaviorist linguistics (except insofar as there are any behaviorists who are also hermetic linguists, which may be a contradiction in terms), other than to distinguish Peirce’s approach from a behavioristic one.

Any hermetic linguistics could be called “structuralist” (notwithstanding the Chomskyans’ aversion to the term as applied to themselves) because the logical effect of sealing off linguistic predicates—treating langue as prior to parole—is to satisfy an essential condition for regarding languages as individuals which have an internal and independent structure. Without such a logical seal, the linguistic properties of phenomena would be grounded in part in something external to the langue, which would mean the relevant “structures” would not be langue structures proper, but rather relations between the langue and something else. The question would then arise as to whether there is any such thing as a langue, considered as a describable individual; for if a langue is a bearer of relational predicates, it must have some monadic (nonrelational) properties: one cannot ascribe relational predicates to a blank Ding an sich. But what monadic properties could be ascribed to something which is unobservable, and about which the only things supposedly known are precisely the relational properties in question? We can see, then, that the idea that the basic task of the linguist is to describe languages, that languages are individual entities with structures (i.e., are Aristotelian “primary substances”), that langue is prior to parole, that linguistic properties are hermetic (i.e., indefinable in non-linguistic terms)—we can see that all of this is a package; and I suggest it is simply the traditional but vague conception of linguistics as language description brought to its logical conclusion, rather than the radically innovative or “revolutionary” new conception of the science it has been widely touted as being.

In fact, it would seem that, apart from phonological developments, “modern” linguistics is basically just traditional grammar and dictionary writing, minus any latent tendencies to explain what it means for something to be linguistic in character: the “Copernican revolution” in linguistics was nothing more than a decisive and intentional rejection of any possible demand upon the linguist to explain how linguistic things, as such, function or act (unless the inscrutable operations of the langue, as it goes about “arbitrarily” bestowing linguistic identities on things, is to count as a functioning). Thus it was, in effect, a formal reduction of the possible scope of linguistics, at a time when it might instead have conceived itself more broadly than before. Moreover, instead of subjecting the vague (and confused) traditional idea of grammar to close critical scrutiny, perhaps refining this conception or even replacing it by something more sophisticated, the very possibility of explaining the term grammar was definitively ruled out. Why? Because if the predicate is grammatical were defined in empirical terms, directly or indirectly, the logical “seal” would therewith be broken—and, indeed, I believe no such definition will in fact be found in the writings of such linguists (though of course one may find it defined intratheoretically by other nonempirical terms). Thus the traditional idea was simply formalized and protected from critical scrutiny without being analyzed. I do not find it surprising that some of these “scientists” have subsequently extended the use of the term grammar to everything with which they wish to concern themselves, so that “semantic” and even “pragmatic” considerations are now also said to be a part of “grammar.” Who can object when there is no way to know what it is to which one is objecting?—unless, of course, one objects in principle to the forging of a carte blanche for the vagaries of a priori “theorizing.”

In contrasting the structuralist (hermetic) approach with a functional one, I do not mean by function the uses to which linguistic entities may be put. This would imply that linguistic entities are tools or instruments, ready-formed and awaiting their use for whatever jobs one finds their structure fits them to perform; and a functional approach so conceived would in fact imply a prior structuralist theory, rather than being an alternative to it. Thus I agree with the structuralist rejection of the idea of linguistic entities as being basically of the nature of tools (though I should point out that the metaphor of langue as a machine nonetheless occurs with remarkable frequency in their writings). Words can of course be put to use, in some sense of use, but it is precisely the unlimited variety of their possible uses which makes the conception of use improper for explicating what linguistic entities are as such. In view of this, I understand why such linguists insist on regarding language as intrinsically functionless. But when I speak of the “function” or “functioning” of linguistic entities, I mean what words do as words, not the extrinsic purposes which their doings may happen to serve. More on this later, however. For the moment let us keep our focus on the peculiarities of the type of linguistics I am criticizing.

If a linguistics is hermetic in the way it treats its concepts, this will show definitively in its evidential methodology. We owe it to the Chomskyans, in particular, to have made it clear that such a linguistics will have an “intuitive” evidential base, though I cannot think much of their attempt to palm off intuition as an empirical (I don’t mean “empiricist”) base. Moreover, I find their pretension to homologize hermetic linguistics with a science like atomic physics reprehensible: this smacks more of academic strategizing than of serious philosophy of science, and it does a disservice to their linguistics colleagues by fostering a misleading self-conception that discourages clear thought about what linguistics actually is—and could be. Let me now try to explain why hermetic linguistics must be “intuitional,” and why such intuitions cannot be regarded as providing an empirical scientific base.

Since languages (langues) are unobservable entities, all langue-descriptions are the result of inference from their observable manifestations, insofar as such descriptions can lay claim to empirical grounding. Let us call such observable manifestations “parole-examples.” A parole-example must be described as having some linguistic property if any inference is to be drawn from it: let us call that “the linguist’s parole-description.” The terms in this description (i.e., the predicative terms) must have conceptual relationships with the terms in the langue-description if any inference is to be made from a ptfrtf/e-description to a langue-description; for the rules which the langue-àescYvptAon describes should be such as to generate logically the linguist’s parole-descriptions, though of course the linguist’s inferential task is, as it were, to go “upstream” to the generating source. This means the linguist’s parole-description must use theoretical predicates, which means in turn that this description is not itself empirically evidential, except insofar as it is in turn evidentially grounded. Thus we must distinguish between (a) the inference from the linguist’s parole-description to the langue-description, and (b) the inference to the linguist’s parole-description from the evidential base.

These distinct inferential moves have frequently been conflated and confused, especially in connection with the topic of hypothetical inference. It is clear enough that an inference of type (a), from a case to a rule, or from a rule to the rule which generates it, may fairly be construed as a hypothetical inference. However, it is far from clear that an inference of type (b), to the linguist’s parole-description, is also a hypothetical inference, in the case of hermetic linguistics in particular. My claim is that there is not only no possibility of a hypothetical inference here, there is no inference possible at all for the hermetic linguist. Hence, such a linguistics has no empirical evidential base: just “intuitions” without evidential force. Let us see why.

The problem, then, is the logical movement from a parole-example to the linguist’s parole-description. No one supposes that linguistic properties are monadic or nonrelational. Hence, the parole-example must be regarded as relative to something else if there is to be a parole-description with any pretense to an empirical base. A behaviorist, for example, will put the parole-example in relation to a person (or persons), as an element in a stimulus-response complex in which both participate. We need not specify exactly how; the point is that it will be the descriptions of that complex which will constitute the behaviorist’s evidential base. A hermetic linguist will also put the parole-example in relation to a person (or persons) and look for a response. However, the only response of interest in this case will be a response which is an assertion about a linguistic property of the parole-example by some person (possibly the linguist in another capacity) regarded as representative of the langue being investigated. Thus somebody will be designated as a “fluent” or “native” speaker, will be presented with the parole-example, and the assertional response will be recorded as evidence. The evidence is thus of that special type called “testimony.” Let us refer to this as ^uparole-testimony,” bearing in mind that this is quite distinct, in terms of logical function, from the linguist’s parole-description, even in the case where the linguist is his own informant and his parole-description is verbally identical with his own parole-testimony. For without this distinction all claim to an evidential base for the langue-description is thereby nullified. (As I pointed out above, the hermetic linguist’s parole-description cannot itself have any empirical evidential value, since it must be formulated in theoretical rather than empirical terms if inference to the governing rule is to be possible.)

Before seeing why the hermetic linguist’s evidence—or what is supposed to be such—must be testimonial in character, let us note the prima facie oddness of appealing to a testimonial evidential base for a nomic theory. I know of no true parallel to this in the history of nomological science, not even in the case of introspective psychological experimentation. For in such experimentation the testimony is treated as indicative of facts which caused the experiences to which the testimony testifies, and the latter facts are those which provide the evidential base for inference to an explanatory theory; such facts, however, are not themselves testimonial in type. For example, Peirce himself conducted (with Joseph Jastrow) one of the first psychological experiments in this country (published in 1884).⁵ The test was of the Fechnerian hypothesis that there is an Unterschiedsschwelle or Differenzschwelle (i.e., a threshold of perceivable difference) for sensation, and it consisted in the production of controlled differences in sensory stimulation to which the subjects responded by testifying whether they could detect any difference. Their testimony—including responses to requests to guess whenever they could seem to perceive no difference—was then systematized and correlated with the facts of actual stimulation differences. The correlation between actual and testimonially attested differences indicated no such threshold. (Peirce also regarded this as evidence for awareness of perceptual differences of which one is not reflectively conscious.) For our purposes, the point is that the testimony itself functioned as only one element in the process of acquiring the evidential base, but was not regarded as having evidential value relative to the hypothesis insofar as it was just testimony. I suggest that testimony will never be found to have functioned as an evidential base per se in any reputable experimentation in a nomological science. Its proper role as evidence is in the law courts and as a basis for historical inference, not for inference to laws or regularities.

The reason why the hermetic linguist as such must nonetheless make appeal to such “evidence” is that such a linguist cannot define the predicative terms in his parole-description in extratheoretical terms (since this would break the logical “seal” which makes the theory hermetic), and therefore can establish no basis for a hypothetical inference. For it is an indispensable requirement of all hypothetical inference that the conclusion hypothetically inferred deductively imply that from which it was inferred, given appropriate instantiating conditions, which means that the parole-example must be described in terms which are present (by implication) in the theory or hypothesis inferred.⁶ But this is precisely what is precluded by the hermetic linguist’s refusal to provide extratheoretical definitions for his theoretical terms. Nor can any relevant sort of inductive inference be made. The only thing induction could do here would be to generalize the scope of application of a predicate; but what the hermetic linguist requires at this point is the replacement of the predicate in the evidential parole-description by a theoretical (and hermetic) predicate in the linguist’s parole-description, and only hypothetic inference could do that. This is why the hermetic linguist must instead appeal to testimonial evidence, the idea presumably being that since the testimony is already in terms of distinctively linguistic properties, there is no need to replace the testimonial predicates by the linguist’s theoretical predicates through an inferential move. Thus the appeal to testimony is an attempt to circumvent the need for a hypothetical inference from evidential facts.

But this is, of course, only logical sleight-of-hand. If the informant’s testimony is already in the linguist’s terms (these terms being understood in the same way), then the parole-testimony is on a logical par with the linguist’s parole-description as regards its lack of extratheoretical implications, and can no more be regarded as evidential than can the linguist’s description. Hence it cannot be regarded as evidential relative to his description. If, on the other hand, the testimony is not in his terms, or if the terms are apparently the same but not understood as the linguist understands them, then the linguist must establish some determinate relation between what his terms mean and what the testimonial terms mean in order to infer from the testimony, taken as true, to his description. But this the linguist cannot do without defining his terms so that they imply (directly or indirectly) the predicates in the testimony. For no matter how reliable the informants are supposed to be, the most that can be inferred from their testimony is that the facts about the parole-examples are as they say they are. But what they say about the facts is not what the linguist says about them in his parole-descriptions (and cannot be if they are to have any evidential weight). Nor is there any use in saying that the linguist does not infer from their testimony as true, but only from the fact that they testify as they do. For whatever the linguist infers from such facts cannot be the description needed without some conceptual relation being established between that description and those facts (however they be construed). Hermetic linguistics is thus essentially devoid of any empirical evidential base.

I believe it can now be seen that this peculiar talk of “intuitions” is just logical misdirection. An appeal to an intuition could only establish that someone is certain a parole-example has this or that property. But such certainties are not to the logical point. Let the informant’s intuition be felt by him as powerfully as you like: it cannot make any difference as regards the linguist’s logical right to infer from their testimony to his description. And let the linguist’s intuitions about his own descriptions be as powerfully felt as you like: it cannot add an iota of evidential force to a description which can have no such force except as derived from something which can count as evidence for it, and that cannot be a feeling of certainty. The idea, promulgated by the Chomskyans in particular, that intuitions are evidential is surely a debasement of the concept of evidentiality to such a state of confusion that anything resembling a scientific method in linguistics is out of the question as long as that conception is operative in the thinking of linguists. It is not that there is no place in scientific inquiry for intuitive perceptions and certainties, provided this only means that there is always an evidential starting point in something not doubted. Unquestioned (though not unquestionable) beliefs are at the basis of any science. But the logical problem is not whether anybody is certain of this or that, but whether there is any way in which those certainties, or rather the expression of those certainties in assertional form, can function evidentially. I see no possibility for that in any hermetic linguistics.

I submit, therefore, that such linguists have simply been attempting to devise a new and more general way of doing what grammarians have always done, namely, identifying and systematizing certain patterns of speech and writing recurrent in the discourse of people whom they regard as linguistically proficient (including themselves, needless to say). But what is proficiency? The traditional grammarian meant by a proficient speaker one who speaks in accord with the patterns of a reference class, and who therefore can communicate effectively and without embarrassment in such a milieu. But the kind of linguistics I have been criticizing does not differ significantly in this respect, except that the contemporary linguist disavows interest in this use of his work and focuses instead on trying to develop standard methods of representing such patterns for all languages. I have no objection to such an enterprise, but I cannot think it a worthy conception for a science of linguistics. A truly “modern” linguistics—one which would address itself to what is important about the linguistic dimension of our lives—would surely have set as its first task not that of systematizing and universalizing the description of prevailing grammatical patterns, but rather that of asking what grammar is. Grammatical patterns have at least two distinct functions. On the one hand, they provide indications of what one is talking about, what one is saying about it, how this is being qualified, and so on; on the other hand, they present esthetically pleasing and rhetorically effective forms: in short, they have both logical and stylistic functions. Any writer or teacher who reflects on what he or she is doing will have experienced conflicts between the imperatives of logical rigor, on the one hand, and communicational effectiveness, on the other, and will have noted the ambiguous role of grammatical form in resolving these conflicts. It is possible that grammatical functions could be so conceived as to reconcile these two prima facie distinct functions—and perhaps others as well—in a unitary conception of what grammar does. (I think myself that a suitably reconceived idea of rhetoric could do this; that is, that what is now called “grammar” would reappear in an adequate theory of rhetoric as a standardized rhetorical function.)⁷ But, however that may be, hermetic linguistics—structuralism, as I prefer to call it—definitively and formally rejects the task of sorting these things out openly, and instead relies on the fogginess of the notion of “intuition”—and the fogginess of intuitions—to permit itself to systematize with maximum freedom, and with maximum irresponsibility when it comes to explaining what they are systematizing and why they are doing so, anyway.

To me, this requires explanation, and I have a hypothesis. In introducing this hypothesis I may offend some of you, who may think it logically shoddy to move to motivational matters at this point. But I do not rest my case against the kind of linguistics I have been criticizing on this hypothesis,⁸ and, more to the point, I think that we—as students of human life—need to think far more about the consequences of what we are doing in our theoretical work than we usually do, particularly in view of the technology which our work generates insofar as it is successful. The atomic physicists—some of them—were brought (temporarily) to the realization in 1945, and B. F. Skinner’s blatant claim that he and his like should be allowed to rule the world surely makes it clear that, whatever our personal motivations may be, we are engaging in researches of a sort which, if successful, will be fraught with practical moral implications of the first importance. So I ask vou to take a look at something here.

The type of linguistics I am criticizing begins with the laying down, a priori, of a general theory of language (“universal grammar”) which is, in effect, a logic. I mean the term logic in a very broad sense, according to which it is a general theory of the human mind, at least insofar as the mind is linguistic in nature, as such theorists usually do suppose it to be. I say they begin with this because, in spite of all the sober-sounding talk about how “linguistic universals” are something to be inferred from a basis of descriptions of various languages, the fact is that the universal grammar is present already in the rules laid down for language description. For example, Chomskyan linguistics-linguistics conducted as Chomsky says it should be—is the universal grammar which he nonetheless keeps talking about as if it were something to be discovered. But what sort of “discovery” is it that consists in “inferring” the general rules implicit in the very cases which have been described in open accordance with precisely those rules to begin with? Had the name of Chomsky’s first book been Universal Grammar instead of Syntactic Structures it would have been obvious to anyone that the “discovery” was in fact made by Chomsky himself before he ever wrote the book, and that what purports to be a theory of universal grammar is just a set of prescriptions for writing the very grammars which are supposed to be its evidential base. Linguistics is thus conceived from the beginning as a priori cryptophilosophy.

I have no objection to cryptophilosophy insofar as it is good philosophy, but I object to it as cryptic because this is a way of avoiding intellectual responsibility for what one is doing; nor do any of the cryptophilosophical “linguistics” with which I am acquainted persuade me that I am mistaken in thinking that taking full and open responsibility for what one is doing is essential if one is to do it well. Moreover, I do not think the ambitions of such “linguists” are limited to doing irresponsible philosophy of mind. (I mean what is implicit in the tendencies of their work, not how they consciously regard themselves.) To the contrary, it seems obvious that it is the development and defense of a metaphysics, a doctrine of the basic nature of reality in general, which lies beyond—or perhaps rather within—this movement. For one need only adduce the premise that the real is the knowable to convert a theory of mind into a metaphysics; then if one adduces the further premise that the ultimate form of the real is divine in nature one has a theology (and such a premise is nearly axiomatic in theological thinking). Thus I suspect that linguistics has been unwittingly put in the service of an unacknowledged religious drive which has been unable to satisfy itself within the confines of traditional religious forms and is in this respect much like, say, Marxist “social science.”

One of the forms or guises such an unsatisfied impulse may adopt is political, and it is perhaps here that what I am trying to bring to your attention can be most clearly seen. The European semiological movement, with several decades’ head start on Chomskyanism, has already extended structuralist principles—or at least what purport to be such—to a number of paralinguistic areas (e.g., fashions, aesthetics, anthropology). The question is, when is it going to be—or has it already been—extended to the political sphere in the form of a structuralist “political science”? And what kind of political understanding would result from such an extension? (Bear in mind that, from the very beginning of speculation about the nature of language in the West, that is, from the time of Greek Sophism and the Socratic/Platonic response to it, theories of language and political doctrines have gone hand in hand: in fact, one of the most fruitful strategies for understanding philosophers is to identify the isomorphisms holding between their political/ethical models and their logics and theories of meaning.)⁹ Consider, then, the hermetic or structuralist model for linguistics reinterpreted as a political model.

The langue is, of course, analogous to the law of the land. Parole is citizen behavior. Being a citizen is behaving in accord with the law (which, like a langue, is definitive or constitutive—not merely regulative—of what it is to be a citizen). Failure to so behave is to be a political deviant or perhaps even a political nonentity, outside the law altogether and devoid of legal identity (including rights). Whether or not one is behaving in such a way as makes political “sense” is to be determined by qualified citizen informants who “intuit” the facts about one’s behavior relative to the law and assess its “acceptability.” The informant class is selected by the legal expert as exhibiting “obvious” proficiency in nondeviation, and of course the expert may function as his or her own informant. The legal expert—the analogue to the linguist—is the one who definitively determines whether the behavior is in the State as deviant or normal, or is not a part of the political reality at all. If my criticisms earlier are correct, this “scientist” will necessarily do so on the basis of no empirical evidence. It is true, of course, that the State—the law, the langue of political behavior, as it were—tends to change across time in virtue of deviant acts (though how this is logically possible I do not know), and that persons formerly classified as deviant, or perhaps even as beyond the pale of law altogether (wild animals, in effect), may subsequently be recognized as fully “acceptable” after all, and their corpses (or reputations) duly exhumed for belated public admiration. The experts who originally “misidentified” them because of “faulty intuition” or the testimony of “unqualified citizen informants” will be replaced by new experts (such is the usual fate of such “scientists”), though how the new experts are to avoid the same sort of error is not clear. But I need not elaborate this obvious isomorphism further. If you think it is a fanciful picture ... I hope you are right! (But before you so conclude, reflect upon the present political role of the “science” of psychiatry in certain countries.)

These linguistics theorists do not, of course, regard what they are doing in this light, and they would doubtless protest—quite sincerely, I should think—that it is a mere accident that their linguistics model is so remarkably isomorphic with the familiar type of authoritarian political order just sketched. But then Oedipus—I mean the Greek rather than the Austrian version—didn’t know who he was looking for either when he looked about for the source of pollution. In any case, I intend no personal criticism, and perhaps it would be appropriate if I offered in justification of mentioning these matters some words from Chomsky himself:¹⁰

I would add that the ideology involved might be in radical conflict with the ideology consciously held by the theorist.

II

In conveying to you something of what I understand of Peirce’s basic semiotic conceptions, particularly as they bear on the analysis of linguistic phenomen, I will go about it in my own way, for the most part, rather than follow the letter of Peirce’s writings. The course of discussion in the past fifty or sixty years-much of it highly idiosyncratic in Yerbal usage-on such topics as meaning, significance, sense, signification, reference, and so on, has so thoroughly· anarchized the terminology in this area that Peirce’s own way of talking about and explaining these things would almost certainly be misunderstood unless I were to interpret his words to you at far greater length than would be possible or appropriate here. Another reason for attempting a somewhat different way of explaining his ideas than is to be found in his own writings is that due account must be taken of reigning metaphysical conceptions which make some things seem to us “obvious” or “self-evident” or “natural,” and some other things “unintelligible” or “odd” or “confused,” though such responses may only indicate compatibility or incompatibility with questionable metaphysical assumptions so totally taken for granted that we are quite unaware of them and do not realize they are unnecessary and perhaps even undesirable.

The perspective Peirce articulates in his semiotic is an attempt to achieve, on a theoretical level, something equivalent to a metaphysical innocence: a premetaphysical way of thinking about the things with which semiotic is concerned. It is not a matter of actually being innocent, of course, but rather of intentionally thinking of things in the most innocent way possible. Such a stance is artificial and difficult both to achieve and to maintain. Yet only to the extent that such a stance is achievable in theoretical work of this sort can we escape the stultifying effects of the ethnocentricity which hidden metaphysical commitments necessitate. Whether Peirce’s semiotic actually succeeds in staying free of such commitments is another matter, as is the question whether I have understood the basic isues well enough to do justice to what he attempted. But, in any case, since I must myself try to subvert any metaphysical assumptions that might hinder your understanding of his ideas, I think it best if I take a somewhat free approach and present his ideas in a somewhat unusual way.¹¹

The icon/index/symbol distinction concerns what I call the “sign powers” of objects, and any object which has such powers is ipso facto a sign. Such terms as meaningfulness, significance, sense, intelligibility, and so forth, are vague nontechnical equivalents, in some contexts, for what I call “sign power,” and I occasionally use some such terms as an informal reminder of what sort of thing we are talking about. Bear in mind, though, that these are not technical terms (nor are they usually so used by Peirce), and I only use them here because of their suggestiveness.

The major difficulty you may have in getting a reliable sense for the semiotic approach lies in the fact that it is the signs themselves—not people who use signs—which are regarded as the primary agents, so that the introduction of reference to people (or any other entities which use or respond to signs) in a semiotic analysis means that a shift in analytic level has occurred, from concern with sign properties and actions as such to some more specialized investigative concern. In particular, the occurrence of terms such as mind, interpreter, user, and use usually indicate that such a shift has occurred from the basic level to a more specialized perspective, such as that of, say, psychology or sociology. (The shift to a more specialized perspective is, of course, legitimate, as long as one is aware of what one is doing).¹² Perhaps it will help, then, if we define the term use at this point. For present purposes, let us understand the use of signs as being the intentional presentation of signs in accordance with what one thinks or hopes the signs will do. (Do not construe the term intentional as equivalent to or as implying reflective or selfconscious.) All sign use—as distinct from sign behavior, which is what use exploits and therefore presupposes—involves, then, an exercise of skill-knowledge, comparable to the employment of any other sort of object in accordance with one’s understanding of its powers and of the conditions under which these powers are actualized.

So deeply entrenched in contemporary theorizing is the assumption that people—whether individually or socially—somehow give meaning, significance, and the like, to meaningful, significant, and sensical things, particularly as regards linguistically meaningful things, that it may seem strange at first that sign powers are not conceived, from a semiotic point of view, as essentially constituted by their relation to people or other sign users. (Nor, I might add, by relation to a langue, which is a queer sort of group mind, nor by relation to the “linguistic competence of an ideal speaker/hearer,” which is a fictitious individual or group mind.) According to the semiotic view, nobody gives meaning to any sign of any sort, if this is supposed to be a transfer of intended meaning from people’s minds to objects, as a sort of infusion or transfusion through intentionality, will, stipulation, fiat, or any other sort of direct psychic injection. There is a sense in which people can and do create meaning, particularly as regards linguistic signs, but it is not the sense usually involved in discussions of the “conventional” or “arbitrary” or “unmotivated” character of linguistic meaning. Sign powers are in the signs themselves, and any changes in these powers, or the accruing of such powers to objects not previously having them, are due primarily to the signs themselves and their actions, not to people’s actions (though the action of people is usually contingently instrumental in this respect).

Let me show you a sign which will, I am sure, be willing to perform for you if you will watch it for a few moments:

It is difficult to stop some of these things once they get started: watch it some more and see whether you can stop it from “going through its act.” (Persistent, isn’t it?) Linguistic signs are similar in this respect: concentrate on any of the linguistic signs on this page and try to stop them from “doing their thing,” that is, try to see them as meaningless marks. (They are rather persistent, too, aren’t they?)

You may say this is due to the innate and/or acquired perceptual dispositions of the perceiver. But we are not concerned with the psychology of perception here, and the fact is that it was the signs, not your dispositions, that you just observed to be doing something: the role of your dispositions in this is something inferred—and obvious only as a familiar inference—to account for what the signs were doing, and that inference is certainly somewhat less certain than are the observed facts about the phenomena for which it is supposed to account. Words as such do what they do differently than what the line-figure above does, of course, and it is to get clear on just such differences that the icon/index/symbol distinction (among others) is developed; but my purpose at the moment is simply to introduce you experientially to the idea that signs themselves do things that are empirically observable, while at the same time making it clear to you that this is not to say that signs “stimulate responses” in people. They do that, too: one can observe signs stimulating people (and other responsive beings as well), and one can hypothesize that somehow this sort of thing is involved in every case of observed sign behavior (i.e. behavior of signs). But that is a hypothesis about—and thus presupposes—the observable behavior of the signs themselves. (Semiotic is not behavioristic, except in the sense that it regards signs as behaving. The behavior of people can perhaps be explicated in terms of signs, but the behavior of signs is not to be explicated in terms of people on the basic level of semiotic analysis.)

It is also important to understand that Peircean semiotic is “phenomenologically” based. It is not essentially a part of psychology or sociology or dependent upon them (though semiotic conceptions can enter into psychological or sociological theories or analyses). It involves no “introspection,” either, unless this just means paying close attention to what one experiences, that is, unless it simply means observing. In Peirce’s sense of the term phenomenology,¹³ to say that semiotic is phenomenologically based is to say that the objects studied by semiotic are not limited to any particular existential category (including the category of nonexistence or unreality). They may be things sensed or they may be only imagined or dreamt or hallucinated or fictionally created. They may be something arrived at only by a complex and lengthy inferential chain or they may be as immediately present as any object ever is. They may be in the physical world or the mathematical world or the mental world or not referred to any metaphysical order. The reason is that the power of a sign to act as a sign is not contingent upon its existential status; on the contrary, we must draw upon our knowledge of an object’s sign powers in order to know how to classify it existentially: hence, the property of being a sign is attributable to an object simply as such, prior to any qualification or delimitation of the sort indicated.

“I have a dream,” said Martin Luther King. That which he dreamt (the object in the dream), and tried to describe and to actualize in the existential order, acted by producing (among other things) his descriptions and actions, which were in turn capable of producing, and did in fact produce, replicas of what he dreamt in the experience of others (apparently in the form of nightmare for the head of the secret police). Did King act upon his dreams or did his dreams act upon him? From the semiotic perspective, the latter is the more basic description. In fact, we might say that it was his dream as a sign which produced Martin Luther King as the sign that he was—or rather is. (A sign of what? Well, interpretations vary, of course.)

Now let us consider linguistic signs in particular, because the idea that meaning is due to human agency seems particularly compelling in their case. (That is, it seems that way to the contemporary mind: the ancient Greeks—like archaic peoples generally—did not regard logos as man-made until the rise of egocentric humanism, which achieved its definitive expression in the conventionalism espoused by the linguistic entrepreneurs of fifth century Greece whom we call “the Sophists.”) Consider, for example, the present communicational process: I mean the one occurring right HERE and right NOW. I probably don’t know you, but even if I do I have no way of knowing that it is indeed YOU to whom I am presently talking. (By “presently” I mean HERE and NOW.) So to achieve my purposes I can only select, combine, and take steps to present such visual configurations as I think may behave for you, whoever you are, as I would wish them to behave in your presence. I know many of them will not behave as I wish. They can if they “want” to. I am sure of that. But, like living things,¹⁴ they just do not always do what you want them to do. I can’t even know which of them are unlikely to behave as I wish. (If I did I wouldn’t bother you with them.) So I try to compensate for that by selecting, combining, and presenting them in a variety of ways in hopes of getting the right sort of sequences started for you, if not in virtue of one combination then perhaps in virtue of another. Now in the combinatorial presentation of them to you I may indeed achieve the creation of a more or less coherent and unique whole of resultant meaning in your experience; and it may be that certain of the signs—such as, say, Peirce, semiotic, phenomenology, icon, index, symbol, linguistics—will undergo some modification in their sign powers or meanings in virtue of the whole of meaning to which they are contributing here. I am not referring principally to a possible change in your understanding of their meaning as a result of your understanding of this entire paper, but rather to a change in the meaning or sign power of the words themselves. For as the words interact in combination here, some of them may undergo permanent changes in their sign powers as a reflexive result. In this sense I may indeed create meaning here, as we all create meaning in this way in much of our communication. (This is, in short, the locus of sign mutation or language change, which is a product of sign interaction in the co-operative creation of a unique communicational whole, always involving some spontaneity or unpredictability, which results reflexively in the change in sign powers of some of the signs responsible for that whole.)

The sense, then, in which I am (or may be) creating meaning here—which is not different in principle from any other linguistic transaction in which meaning change occurs—is roughly the same sense in which a breeder of animals can be said to create the progeny which results from the breeder’s mating experiments: one arranges a promising conjunction and hopes for the best. Thus to the extent that I do have some measure of success in this it is a matter of chance and skill, not will: I cannot willfully infuse any of these configurations with my intentions or hopes, like a god breathing life into lifeless matter. Moreover, even though my intentions are responsible in part for the overall whole of meaning which is the matrix of meaning change (since it is I who have selected the parents, thereby bringing about indirectly whatever offspring may be their issue), the result almost certainly will not be the result I vaguely intend, though perhaps it will be enough like it to have been worth the trouble. This is true even in the cases where I stipulate that certain signs are to be understood in certain ways for the purposes of this paper. Such a stipulation is no absolute fiat but rather a proposal, nothing more: man proposes; the sign disposes.

The chances are that you know little about me beyond the fact that I am the one who selected this batch of configurations which you are presently trying to understand. From where you are positioned it is even possible—wildly implausible, I hope, but possible—that I do not exist. Perhaps the editors of this volume are in collaboration with some psychologists who have taken seriously the old story about how the roomful of monkeys pounding away at typewriters produced the complete works of Shakespeare, and you are now reading (reading?) the results of a similar experiment. Suppose this were indeed the orzhlttt..-%va¹/2q Suppose this wethrddddddf4xx#8,nmh Suppppppppppppppppp etaion shrdlu! Suppose this were indeed the origin of this paper, and that you knew it. Would these configurations then be only meaningless marksmarksmarksmfksmmmppppzz;89rt? Colorlessgreecceenideasszzz.... (What is this? Some kind of monkey business?) Would these configurations cease to do the sort of thing they are doing now, as you peruse them? And if they just kept right on “doing their thing,” would this be a sort of pseudomeaning, rather like the pretense put up by a check written on a nonexistent account? Suppose further that, if one were nevertheless to read them as if they were intentionally and “intelligently” produced, they would not only make sense (i.e., be as if they made sense) but would also turn out upon investigation to be true, valid, cogent, and so forth (i.e., be as if true, valid, cogent, and so forth). What would be the force of the as if in that case? Does it signify a diminishment in value, or even an annulment of it? And if it were subsequently discovered that these configurations really were, after all, a “product of intelligence,” would this in turn result in an elevation or reinstatement of their cognitive worth? I pose these as rhetorical questions because one cannot settle such matters in brief fashion; but what I am suggesting—or those monkeys may be causing you to think—is that, whatever the role of intentionality in discourse may be, it is not that of giving meaning to things which otherwise would not have it.

Human agency in the production of words is, from a certain point of view, no different in principle from the skillful manipulation of materials with their own energetic propensities that is typical of any craft. Human creation is never ex nihilo, whether it be the creation of a pot, or a piece of music, or a painting, or a discourse; nor is it a peculiar sort of emanation from a mental substance. Voluntaristic conceptions of language—by which I mean both intentionalistic and conventionalistic conceptions—are perhaps just displaced theological doctrines decked out in the form of humanism. “Let there be light!” God says (in Hebrew?), and, Lo!, light is. “Let there be langue!” the conventionalist imagines people to say, and, Lo!, langue is. But this cannot account for the fiat itself, which is already linguistic. “Archaic” peoples—those closer to the basics (a̓pxaí) than we moderns—regard crafts, including speechcraft, as a natural process in which the craftsman participates and co-operates, but does not dominate: the craftsman gives direction, to be sure, but then so do the materials through their own propensities. In fact, it is “nature” (cf. the Greek φv́σις) which is the true craftsman because craft is only a special case of growth. Thus nature itself makes receptacles, for example; and such things as holes in the ground, caves, nests, baskets, pots, tombs, houses, and so on, are just so many products of nature’s various propensities for receptacle making. Is this a naive and inferior way of regarding such matters? It is no doubt true that, for certain purposes, it is profitable to conceive the human craftsman as a manipulator of materials which are passive relative to the craftsman’s agency; but do we reach a still higher stage of “sophistication” when we go so far as to elevate the craftsman to transcendent status and invest him with the powers of a god or magician? There is surely a fine irony in the fact that our “advanced” theoreticians of language, who talk so earnestly of their belief in “human universals,” proceed ab initio as if from a completely unquestioned assumption that the key to the nature of language is to be found in one version or another of a seventeenth century metaphysical doctrine according to which meaningfulness is something “arbitrarily” bestowed upon or infused into brute matter by the inscrutable will or wish of transcendent individual or group minds. But if any true human universality is to be discovered, perhaps it would be advisable to abandon the idea that the role of human agency in the production of signs is fundamental, without thereby denying that there may be special reasons in special contexts of inquiry for recognizing a relatively basic role for intentionality and will.

It is equally important to understand that human agency in the interpretation of signs is not fundamental either, as regards the constituting of significance or meaning. For if signification is not essentially relative either to sign producers or sign interpreters, then a semiotic linguistics need not be conceived as a part of psychology (as is usual in the United States) or of sociology (as is usual on the European continent), though of course its conceptions and theories may enter into those sciences, much as mathematics enters into the natural sciences. The interpretation of linguistic signs—the sort of thing you are doing right now—is first of all an intelligent and active experience of them, similar in many ways to sense-perception experience of any sort. Thus as your eyes scan a line of type, you experience what the signs mean, that is, you experience them as meaningful; and although this involves perceiving them as visual configurations, it involves more than that. In fact, what you will “carry away” from the total experience of the text will probably contain little memory of the sign configurations themselves, but will instead consist mainly of what you retain from the experience of the object(s) of the signs. In the case of the present text, these objects would include such things as Peirce’s theory in general, Peirce himself, me, linguistics, certain conceptions of linguistics, and certain things in Peirce’s theory. The experience of the object of the text will be the result of the actualization of its sign powers, along with those of the other signs you will have provided as your own context (the ever present, ever unique, and ever changing accompanying text). Now these things or objects—I use the term object here as synonymous with thing, in the broad, vague, and colloquial sense of thing—are what are presented phenomenally or experientially in and through the actualization of the powers of these signs, so that your experience of reading this text will have been an experience of things or objects which vou will have observed in virtue of the powers of the signs to present them to you for your observation. The signs themselves are among those things experienced, and how they are experienced is also a result of sign powers; but the focus of your experience will be primarily on objects not present on these pages which the objects on these pages—the signs or words—will have made present to you.

It is of little importance what this is like as any particular person happens to experience it in the flow of their consciousness. Concern with that sort of thing arises only for special reasons, and it is not essentially pertinent to the question of your understanding of what you have read. For the answer to this question would not be ascertained by introspection, but rather by having you describe the object(s) you have experienced. Any attempt to reproduce the actual flow of experience would result largely in a fictitious account, anyway, because one would be unable to remember most of it and would have to reconstruct what it might have been and put that forth in place of what it actually was. But then there is rarely any reason for such a description. Generally speaking, many of the objects experienced in the course of reading will occur (phenomenally), function as signs by producing other objects, and then disappear after they have fulfilled their semiotic roles; however, some of them will keep recurring, will combine with others to form more comprehensive objects, and so forth, and may finally appear as constitutent parts in the resultant object or constellation of objects. Usually, it is only the final result—the object (or objects) which the entire reading results in experientially—which is of concern to us. But what I want particularly to stress here is that the experience of the meanings of words (the interpretation of them) is an experience of the same general sort as, say, the experience of sensory objects by alert sensory observation of them.

Let me explain something at this point about the idea of a “phenomenal object” and its relation to the idea of a “sign” which may help in understanding what I am trying to convey about sign interpretation being a kind of observation. A sign is a phenomenal object, by which I mean an object in the phenomenological sense of the term object, as I explained earlier.¹⁵ The actualization of a sign’s power consists in its presencing a phenomenal object. To presence is a transitive verb which I use as a synonym for to present, when the latter is first construed in its most generic sense and then delimited in application to only those cases where the presenting is a making present of something to some responsive being (not necessarily human). It is roughly equivalent to the Greek verb φaívw, which has no equivalent in common use in English, so far as I am aware. The idea of something being present to someone is quite familiar to you, and it could be operationally (that is, extensionally rather than intentionally) defined as being something to which any response of any sort is made, provided the concept of response is so modified as to permit phenomenal objects in general to be stimuli (rather than being so conceived that only objects in the spatiotemporal order can qualify as stimuli), and provided the stimulus-response process is so conceived as to take adequate account of the contribution of the directional word to, as it occurs in the phrase response to something. In that case, to be presenced would be the phenomenologically extended analogue to the word to stimulate. (Note that I say “analogue” rather than “equivalent”; for being presenced is only extensionally and not intensionally correlative with the idea of a stimulus.) But, so far as I know, no attempt has ever been made to so extend the idea of response (in its theoretical use), in spite of the fact that we all know perfectly well, on the common sense level, that responses are often to things which are not in the spatiotemporal order, such as objects merely dreamt, conceived, imagined, hallucinated, and so forth.

What has just been said does not imply that the powers of a sign as such are fundamentally dependent upon responses to it. The somewhat subtle conceptual connection to responses is necessary in order to insure that signs are conceived as publicly available, but the basic powers of signs are not per se powers of generating responses to the things presenced, but are rather powers of presencing things. In other words, nothing can be a sign which is not publicly available; but what makes a sign a sign is not that, but rather its power of presencing. Thus there is no contradiction with the thesis that human agency, whether it be in the form of the sign producer’s intentionality or the sign interpreter’s response, does not account for meaning properties.

Signs are themselves presenced by phenomenal objects acting as signs, and the phenomenal objects they presence will in turn act as signs. A sign-interpretational process is thus an ordered procession of phenomenal objects, directly or indirectly related to one another in certain ways, in virtue of what I call “presencing.” To identify a phenomenal object as a sign is to impute to it the power to presence a phenomenal object. To identify it as specifically an iconic sign, or an indexical sign, or a symbolic sign is to say something about how it relates to other objects in the procession. A distinction must be made between what a sign can do and what it actually does. A semiotic description of an actually occurring sign process (i.e., of an actually occurring phenomenal sequence regarded as semiotic in character) would be a description of what each sign did or is doing or will do, and of how it relates to others in the process, the result being a description of a complex sequential process, usually involving many branchings and joinings of subsequences. A semiotic description of a particular sign text would be an account of some of the processes that could occur, starting from that text or any part of it. (Such descriptions of possible readings are infinitely extendable, even in the case of the simplest text; hence a semiotic analysis of any text will always be only partial.) To refer to a sign as a “text” is simply to refer to it as a starting point (usually composed of many subtexts) relative to some possible or actual sign process. Since one of the theorems of Peirce’s theory is that every phenomenal object has sign powers (not necessarily known powers), every phenomenal object is a possible text. Moreover, it is at the same time something that can itself be regarded as semiotically derivable from a prior text. A semiotic theory of the Peircean type is basically a systematically developed set of distinctions for use in the analysis of sign processes and texts, actual or possible. In other words, semiotic is a method of describing phenomenal objects as functioning or functional units in actual or possible sequential processes which are structured according to certain principles implicit in the idea of “presenting” power.

Since a phenomenal object is any object,¹⁶ regardless of existential status, it will not necessarily have properties which it will have if it is rightly classifiable as an individual existent in the historical order of nature, occurring at a certain place and a certain time (or existing continuously through time and moving through space). If it is classifiable as an individual within this existential order, then it follows, for example, that for any given predicate (within the class of predicates categorially relevant to it), it either has that predicate or does not have it. Or perhaps it would be more accurate to say that it is our practice to expect this particular “logical law” (the law of excluded middle) to hold true of any object thus classified, though it may be that some exceptions have to be made (as perhaps in the case of certain entities on the subatomic level).¹⁷ On the other hand, a merely imagined object, for example, need have only a few properties and may be highly indeterminate—I mean indeterminate in itself, not merely as known to us—in respect to innumerably many others that would be either true or false of it were it in the space-time order instead. And this is true of any phenomenal object, unless it is especially classifiable as being an object of the sort which must meet such “laws” (the force of the must deriving from what we assume in so classifying it). Thus, like a denizen of Hades, a phenomenal object can be a “mere flitting shade”—or, indeed, a far more insubstantial thing than that—though it must be capable of being referred to (in order to be an object) and be minimally describable (in order to be phenomenal).

For example, I am at present writing with a cheap black pen. This pen is a phenomenal object in what we call “the real world,” and thus it is a determinate individual, possessing innumerable properties of which I am not aware and for which I have no concern. But your idea of my pen (that is, the object of which you are now aware in virtue of my description of my pen), which is also a phenomenal object, is highly indeterminate; it is black, cheap, and a pen, but it neither is nor is it not made of a certain sort of plastic. (My pen either is or is not made of that kind of plastic, though, because it is an entity in “the real world.”) Shakespeare’s Brutus was an honorable man—or at least an “honorable” man (scare-quotes used to pacify the shade of Shakespeare’s Antony), but it is neither true nor not true that he had a stiff drink on the morning of that fateful day, before setting out for the forum; for Shakespeare left him indeterminate in that respect. The “real” Brutus, however, may not have been so honorable and he either did or did not have a stiff drink the morning of the assassination. (The relation between one phenomenal object and another which is signified by saying that the one is an “idea of” the other is not a topic we can pursue here, though I should remark that nothing precludes a certain limited or partial identity of the two in the case where truth is involved; thus, for example, your idea of my pen, assuming I told you the truth about it, is my pen as it appears to you now, notwithstanding the fact that we can also speak of them as two phenomenal objects. This is because identity relations are far more various in type and are capable of being construed with far more flexibility than is usually realized. Moreover, the identity predicate has as many different ways of applying to things as there are differences in the ways of counting things.)

Such is the idea of an “object” or “thing” which you must keep in mind in what follows, and not suppose that a phenomenal object is always a “full-bodied” object of the sort we encounter when we are dealing with objects in the space-time order we call “the real world.” But mere flitting shades though some of these objects may be, they all have sign powers nonetheless. Equipped with this initial understanding of the terms phenomenal object and sign, let us now return to the fact that the reading of a text of any sort is to be construed, for semiotical purposes, as the observation of phenomenal objects as presenced through the action of signs, which are themselves phenomenal objects.

When the subject matter of a text is a theoretical object (involving innumerable subobjects, also mostly of a theoretical nature), such as in the case of the present text, we tend to think of interpretation or under-standing as something quite different from sense perception. In the case of reading a novel or a poem the difference is clearly not so great: thus no one would deny that in reading a description of a landscape, for example, the linguistic signs are so acting as to presence objects “in the imagination,” and that this experience is, in many respects, quite similar to an experience of sense-perceiving them. A skilled reader, deeply immersed in the reading, will often have little significant awareness of the linguistic signs which are mediating the experience, and it will seem, experientially, very like “being there.” Indeed, an unsophisticated cinematic version of a child’s fantasy story—the sort of trash children watch on television now—is often experienced by them with a lesser sense of direct observation than the same story is experienced by children who read it or hear it told to them, the mediating images on the television screen being less powerful semiotically than the word-signs, even with regard to pictorial “realism.” I find in my own case that the mediating images on a television or movie screen rarely have as much power to presence the objects depicted as do the word-signs in the corresponding novel or short story. This no doubt varies greatly with individuals, but the point is that we can and do in fact observe “through” words just as we do “through” any other sign configurations presented to us. Thus the semiotician will regard all of these cases—the reading of a theoretical text, the reading of a novel or poem, the seeing of a cinematic representation, and the direct sensory perception of something—as fundamentally alike, in that in all such cases (and in innumerable other kinds of cases), phenomenal objects functioning as signs produce other phenomenal objects which are themselves signs, which in turn function to produce still further phenomenal objects which are signs, and so on.

There are important differences, too: differences in the relative roles of the various kinds of signs; in the kinds of phenomenal objects; in the patterns and the sequences of interactions of signs as they actualize their sign powers; and still others. But in common would be the fact that some entities functioning as signs are actualizing their powers such that objects are being observed by the interpreter, whether the objects are imaginary or real, and whether they are correctly or incorrectly perceived.

— Well, if reading a theoretical text, such as the present one, is supposed to be basically like, say, watching a football game from the grandstand, then tell me this: What does Peirce’s conception of semiotic look like (or smell like, or whatever)? I don’t see how you can liken the sense perception of a physical object to the intellectual awareness of a theoretical conception, unless you mean by the phenomenal object the words themselves. And that is what a theoretical conception is, isn’t it? A sequence of word-signs? But if that is what you mean, then it is highly misleading to say that these words are in any sense producing or “presencing” a phenomenal object which is that theory: the words are the theory, that is, they are the theory as put in your own words. That is what you are saying, isn’t it?

No, that is not what I am saying. The words in the present paper are not Peirce’s theory (or my conception of Peirce’s theory), nor are Peirce’s words Peirce’s theory either.¹⁸ Nor is the atomic theory of matter the same as the word-signs in which it is expressed (or the class of all such word-signs), anymore than the history of England is the word-signs which the historians have used to describe it. Peirce’s conception or theory of the sign-interpretational process is the signinterpret ational process itself—as Peirce conceived it. And to the extent that he conceived it correctly, we can drop the qualification.

Think of it this way. There is (1) a phenomenal object which consists of all sign-interpretational processes that ever were or will be. Then there is (2) a phenomenal object which is exactly like the one just mentioned but which is highly abstract, lacking concrete details and including only those properties which would be specified in a true description of (1), if one were describing only the types of sign powers involved in (1). Then there is (3) a phenomenal object which is Peirce’s idea (conception, theory) of (2). Then there is (4) a phenomenal object which is my idea of (3). Then there is (5) a phenomenal object which is your idea of (4), and perhaps also of (3), if you are hearing of Peirce’s theory for the first time and put some belief in my account. I could go on, of course, but this is enough to make my point. The extent to which these various phenomenal objects are identical or different is another matter. For example, to the extent that Peirce is right, and I am right about Peirce, (4) has some partial identity with (3), (3) with (2), and (2) with (1).

My imaginary interlocutor asks what Peirce’s theory looks like. My answer is that one of the things it looks like is a set of Chinese boxes, with no innermost or outermost box. That is, this is one of several appearances of the sign-interpretational process itself, as I conceive that Peirce conceived it. (As it happens, I don’t perceive Peirce’s theory in such a way that smell or sound or taste or tactile qualities have any prominence, though I suppose others might possibly so perceive it.) That is, there is a certain isomorphism between the form of the signinterpretational process, as Peirce conceived it, and the form exhibited by a series of such boxes (though Peirce’s own favorite isomorph of the process was an onion with no outermost or innermost skin), so that if I were teaching a class on Peirce’s theory I might so present it and explain the isomorphism, much as a teacher of atomic physics might use the old Bohr solar system model at an early stage in his pedagogical explanations.

— Now wait a minute! You surely aren’t saying that an atom actually looks like Bohr’s pedagogical image, in the sense that it actually appears in that form?

Yes.

— But everybody knows that atoms don’t look like anything, for the simple reason that they are imperceptible.

I don’t “know” that. Speaking as expositor of Peircean semiotic, I say that neither atoms, nor electrons, nor quarks, nor any other such entities are imperceptible. If, for example, the Bohr model—inadequate as it may be for the sophisticate—still has some genuine use as an isomorphic representation of the reality it represents, then that which it represents does indeed look like that. This is one of its appearances, just as, say, a very poor photograph of a person may still be a genuine likeness of the person in some respect, even if not in a very useful respect. Thus if all one can discern from the photograph is that the person has a more or less human shape, it certainly wouldn’t be of much use; but it would still be true that that person really does have such an appearance. Similarly, the Bohr image is a genuine appearance of an atom in certain superficial respects, even if its use is now quite limited.

— But not all theories have accompanying models, if that is what you are getting at with this idea of the phenomenal object which word-signs “presence.” Quantum mechanics was mistakenly criticized by some scientists on just this ground, namely, that every theory has to have a pictorial model accompanying it; but quantum mechanics has no such model. And in general, although such models may be helpful, when available, they are not necessary—and are not in fact constructable for some theories.

I didn’t say the model must be pictorial, if that means visual, and the role of models vis-à-vis theories in scientific representation is merely one special case of what I am talking about (though an exceptionally interesting one because of the sophistication of the critical literature on this topic). But with regard to your critical point, one answer would be that mathematical representations also function in such theories, and can be regarded as having a modeling function; for mathematical signs are not in general identical with word-signs, though mathematical representation includes word-signs.

— Not if the formalist theory of mathematics is correct, according to which all of mathematics is reducible in principle to word-signs.

That is a theory about mathematics, but it was not Peirce’s theory and it is not the only theory held. Peirce was himself an accomplished mathematician, working at the leading edge of that science, and he was a master logician as well, whose contributions to the latter science are without parallel, as regards both scope and depth. Thus he was at least as qualified to analyze the nature of mathematical representation as anyone else, now or then, and he decisively rejected the formalist analysis. However, the reader should understand that there is nothing approaching the consensus opinion on this topic which would have to obtain to make it reasonable to accept any theory of mathematical representation as well established: in fact, it is precisely the lack of a generally agreed upon theory of meaning (semiotical or otherwise) which precludes any substantial weight being put upon any such theory at this time. Do not mistake the confident tone with which some philosophers of science and logicians pronounce their dicta as a sign of their authority in these matters. There are no authoritative opinions on these matters, even in the weak sense of authority according to which a scientist who is highly respected by his colleagues in a well-established science might be said to speak “authoritatively.” I stress this because those not well acquainted with logic and philosophy of science often mistakenly believe that fundamental questions have been settled in this area, whereas in fact the situation in this area is not different than in any other part of philosophy.

In sum, then, the semiotic approach considers all sign interpretation—including the reading of linguistic signs—to be the observation of objects produced phenomenally by signs, which are themselves phenomenal objects produced by others and producing others still. Thus, interpretation is not regarded as a mental process distinct from what the signs themselves do in presencing phenomenal objects. In other words, what we ordinarily think of as a flow of thought (including feelings and the like) which is produced in us by the signs is regarded semiotically as an objective procession of phenomenal objects (even “feelings” being regarded as appearances of objects).¹⁹ The mind, semiotically construed, is not something within which the flow occurs, nor is it a sort of spectator of the flow; it is the flow. (The operation of the interpretational process is a reflexive operation within the process itself, which can itself be the object of observation in virtue of another reflexive observation, and so on.)

This means that the concept of mind is gratuitous or redundant, strictly speaking. The sign-interpretational process, the processional flow of phenomenal objects functioning semiotically, is that in terms of which such terms as mind, thought, thinking, feeling, and so forth, are to be explicated. Such terms explain nothing and add nothing; they merely indicate something about how the process is to be regarded for some special purpose. So also with regard to the idea of the interpreter: to introduce this idea is either, as Peirce wittily put it, to “throw a sop to Cerberus,” because one finds it expositionally convenient to avoid crossing existing theoretical prejudices in order to make a point which is not affected thereby; or it is to speak in shorthand of something which would be too complicated for exact expression in a given context (which is how I have used the term interpreter here); or else it signifies that one has shifted from a basic semiotic level of analysis to a special context wherein the concept of an interpreting mind or body is especially introduced because it has some useful meaning for that context.

In short, there is no place in the semiotic analysis (except as just qualified) for human minds as agents apart from or in addition to the agency of the signs themselves, either with regard to the sign user or the sign interpreter. There are only signs produced by signs and producing signs, with no absolute starting point and no absolute stopping point. Alternatively, there are only phenomenal objects presencing and presenced by phenomenal objects, such objects often becoming subobjects within more comprehensive objects in various ways. The purpose of a semiotic analysis is to articulate the ways in which this can and/or does occur.

I said earlier that the phenomenological stance is an attempt to attain a sort of metaphysical innocence. I say “attain” rather than “regain,” for I see no reason to suppose there was ever a time when human beings were metaphysically innocent, any more than there ever was a time when human beings were not social beings. But many conceptions we now take for granted (e.g., those expressed in such terms as mind, feeling, imagination, dreaming, sense perception, and so on) are actually of quite recent origin, relative to the length of time during which distinctively human life has existed, and they are intimately bound up with metaphysical conceptions of the real and the unreal which we organize our lives around. But the contemporary Western ways of organizing experience are significantly different even from those of the Greeks of the late classical period, and to move back only a few centuries earlier in that same culture is to move to ways of experiencing often astonishingly different from our own. This point has, of course, been stressed repeatedly by anthropologists in connection with contemporary “primitive” cultures. In fact, it has often been overstressed, to the point of implying that such people are so radically different in their Weltanschauungen that they are radically incomprehensible to us. We forget that the identification of them as human beings already supposes that they are not radically incomprehensible (and that, conversely, to regard their lives as incomprehensible is to dehumanize them in our thinking). But the difficulties in translating their ways of experiencing into our ways is due largely to our refusal to abandon the use of the kind of terms which preclude any such translation because they are essentially bound up with our taken-for-granted metaphysics. Yet it is surely possible to free ourselves, at least for theoretical purposes, from this sort of ethnocentricity, given the fact that some of the most basic elements in our present (and far from coherent) metaphysics have been incorporated not only within recorded Western history but within the past few centuries. The phenomenological point of view assumed by semiotic can be regarded as an attempt to establish a nonethnocentric theoretical base upon which to stand in the process of coming to understand both our own metaphysical presuppositions and those of others. Yet the difficulty in adopting and maintaining such a stance is shown in the difficulty we have in grasping what is meant in saying that thoughts, ideas, feelings, emotions, and so on, are to be regarded, one and all, as phenomenal objects, quite on a par with physical tables, chairs, and trees, in the basic phenomenological description.

The difficulty is not merely that we take it for granted that certain things are real and others unreal, some substantial and some insubstantial, such as people with other metaphysical assumptions would disagree with us about. A further difficulty lies in our habit of treating the idea of an unreal object as identical with the idea that there really is no object, that is, in our tendency to regard the phrase real object as equivalent to really is an object, so that unreal object is implicitly treated as a contradiction in terms. There is, of course, no possibility of understanding things phenomenologically insofar as this gratuitous equivalence is taken for granted, and there is, therefore, no way to step back, as it were, from our own metaphysics without stepping into another. Thus arises the characteristic twentieth century theoretical belief that we cannot truly understand another culture “from the outside,” but if we understand it “from the inside” we lose the “disinterested” position from which we could assess and compare our findings. And on a less sophisticated level, the effect is that, while we may be willing, in a superficial sort of way, to look with a tolerant eye on persons or peoples who recognize “different realities,” we cannot really make any sense of it because we assume all the while that their experience of these unreal objects (as we regard them) is an experience of nothing at all. Being unreal, they are not objects; not being objects, they cannot be referred to; not being referrable to, they are not there to be experienced; hence the experiences of them are just experiences of nothingness! Notice how odd it seems to us when we first learn that “primitive” peoples—and not only they—usually have no difficulty in recognizing that there are gods other than their own, and often assimilate them to their own pantheon enthusiastically. Perhaps this seems odd to us because a nonexistent god is, for us, no god at all; consequently our “tolerance” of religions and world views other than our own is the sort of tolerance we pay to lunatics, who are similarly prone to experiences of nothingness which are somehow at the same time not just experiences of nothing. I cannot here illustrate the way this stultifying assumption works in much contemporary logic, psychology, sociology, anthropology, and other scientific fields, nor consider the deleterious effect it has had on all esthetic criticism (imaginary objects being no objects at all); but I do want to underscore that it must be dropped if the semiotic point of view is to be utilized effectively.

III

Given this understanding of what is meant by phenomenal object, we need only consider one further aspect of the conception in order to take up directly the nature of sign action and distinguish the various types of signs. A phenomenal object must be capable of appearing. Considered as appearing, a phenomenal object is a phenomenon. A phenomenon is thus not itself an object but rather an appearance of an object.²⁰ But the English word appearance has two distinguishable (though related) meanings; on the one hand, it is synonymous with appearing, as, for example, in the sentence

John’s appearance at the party—after he told us all that he definitely

wasn’t coming—was quite a surprise.

On the other hand, it also refers to a perceptible Gestalt of the thing which is appearing, as, for example, in

John’s appearance at the party—with all that scraggly long hair

trimmed off and in his nice new suit—was quite a surprise.

(The Greek words ϵȋδoϛ and l̓δϵ̓u are rough equivalents to the English word appearance in this second sense.) In order to avoid confusion, I will use the term quality for appearance in the second sense,²¹ and appearing for the other sense, and in neither sense are we to think of appearance as something “subjective,” as in the appearance vs. reality distinction. A quality is an aspect of phenomena, and of phenomenal objects thereby, and in appearing an object may exhibit more than one quality. Thus this phenomenal object has both the quality of squareness and the quality of blackness. Since many things are square and many are black (have squareness or blackness as qualities), a quality as such is in some sense independent of particular phenomena and phenomenal objects. (Neither a quality, nor a phenomenon, nor a phenomenal object are what some philosophers call a “sense-datum.” There is no equivalent in Peirce’s philosophy to such a concept.)

We are to understand a sequence of phenomenal objects construed semiotically as being a sequence of consecutive presencings, with each object in the sequence being a sign insofar as it presences, and an interpretant insofar as it is presenced.²² There is no assignable first or last member of the series. Hence, every member is at once a sign of the one which follows it and an interpretant of the one that precedes it in the series. The object of any given sign is the object of every member of the series qua sign. That is, there is a common and unifying reference throughout; and since there is no assignable first or last member of the series, the object is always assumed to have more signs referring to it than are represented in any given representation of the series. The object is not, however, anything other than the referential unity of the series.²³ (The incomplete representability of the object at any given time is, I think, essential to its being a genuine unit; but I cannot here go into the issues that arise in connection with the formal representation of such sequences.) Conceiving a sequence of phenomenal objects as a semiotic series is therefore tantamount to claiming that they can all be represented as in a single series, every member of which is a sign of the same thing. In other words, all of the phenomenal objects in a series are regarded as logical parts or components of the same comprehensive phenomenal object. (This does not preclude subidentification of phenomenal objects in the series as numerically identical or diverse; and, looking at the matter macroscopically, it does not preclude an entire series being related through numerical referential identity to other series, such that the comprehensive object which is the unity of one series becomes a subobject within a still more comprehensive object.) Construed dynamically, such a series of presencings is something like a “moving picture” of the object, which is itself the principle of unity of the series and is thus what appears in every member.

Strictly speaking, every sign of an object, regardless of the type of sign, is a presencing—an appearing—of the object, which thus essentially manifests itself (since it is simply the principle of unity of a sequence of appearings). Thus the words on this page are, as signs, appearings of Charles Peirce’s semiotic. But of course there is an important difference between types of appearings of an object. To take a more intuitively clear example: a physical description of a person—say Peirce himself—is clearly different in some important way from a photograph of Peirce or from him as “appearing in the flesh.” Semiotically, these are all equally Peirce’s appearings, but there is still an important difference, such as is brought out by the fact that we might say that Peirce looked like his pictures (or that they looked like him), whereas we would hardly want to say that he looked like the words on this page.

In semiotic terms, the difference is that some of the signs in the semiotic series are of the sort to which looks like the object—or more generally, is similar to the object—is a pertinent predicate, while some are not of this sort. Iconic signs are signs of this sort: they are signs regarded as phenomenally exhibiting qualities which are similar to (perhaps the same as) qualities of the object itself, in the sense in which we would ordinarily understand the idea of the qualities of an object. Word-signs of an object are not usually like that: the word-signs on this page are appearances of Peirce, in one sense, but he didn’t look like this page looks. (His semiotic theory doesn’t look like this page looks either, which is what I meant earlier when I said that the word-signs of a theory are not the theory itself.) Iconic signs of an object are, then, signs which are appearings of the qualities of the object itself, as we would ordinarily understand that.

However, the difference between an iconic sign of the object which we would ordinarily say is not itself the object (e.g. a portrait, photograph, sketch, diagram, etc., of the object) and an iconic sign the appearing of which we would identify as the appearing of the very object itself (as, for example, when we are in direct sensory contact with the object) is not such a sharp or significant distinction as common sense and some philosophical theory takes it to be.²⁴ For example, I recall that when I visited Venice briefly a number of years ago I could never shake the feeling that I was seeing the city only through a movie travelogue; somehow I could get no more sense of “presence” or “reality” from actually being there than I would have had by seeing it “indirectly” on a movie screen. An odd reaction, no doubt; but my point is that, even from an experiential point of view, the difference in question is not always a significant one. And if we turn to the criteria of physical mediation, with the idea that we can differentiate by saying that I was seeing Venice directly, in a physical sense of directly, when I was in Venice, but would have seen it only indirectly, through physical mediation (film, camera, projector, etc.), if I had seen Venice in a movie, we still find no generally significant basis for differentiation; for all sense perception is physically mediated: the human body with its sense organs, and so on, is not a fundamentally different sort of thing than such artificial organs as are involved in filming and showing the film of something. Thus the need for a fundamental logical distinction here is questionable, and there is no such fundamental distinction in semiotic theory; but insofar as there is any need for such a distinction, it would simply be a matter of mustering criteria of numerical identity and diversity for the phenomenal object which is the iconic sign and the phenomenal object which is its object. These certainly can be identified (as a subidentity in addition to the numerical identity constitutive of the unity of the entire semiotic series), but apart from a special context which would provide such criteria we can only say that such identification (or differentiation) can be made, provided such cirteria are actually available in a specific area of inquiry or concern.

Let me elaborate on this, because of its practical importance in semiotic analysis. From the perspective of this paper, where the topic is the basic semiotic theory itself, we have no substantive criteria for numerical identity and diversity available to us. Such criteria are not provided a priori, but rather by concrete contexts of semiotic analysis. In the various areas of our experience and concern we do have various criteria for identity and diversity. But what counts as one or many is a function of predicates used in a given area and in accordance with what they require, whereas we are talking here in abstraction from the kind of predicates which so determine unity and plurality. In practice, we are often far from clear on what these criteria are, and we are also often inclined to use criteria of this sort where there is no necessity to do so, because of our adherence (conscious or otherwise) to metaphysical theories. Suppose, for example, that you have ten living human bodies, all behaving intelligently. Are there therefore ten minds, one per body? Cartesian metaphysics says yes. Materialistic metaphysics of the crasser sort says there are no minds there at all. Semiotic theory ignores the question, insofar as there is no specific context that makes it necessary or desirable to use such criteria (though when such a context does arise no one imbued with the semiotic point of view is likely to find either the Cartesian or the materialist answer satisfactory).

So, in sum, although a distinction of the common sense sort can be made between an iconic sign which is its object (numerically) and one which is not, the actual criteria for making this distinction are not available apart from some specific context. Iconic signs are, therefore, all alike in being appearings of the object which exhibit some intrinsic quality of it. If the Chinese box image is a good iconic sign of the semiotic sequence, from a certain perspective, then that is what it looks like, just as Peirce as seen “in the flesh” in his later years looked like a man with a rather longish beard. (I know he looked like that because I’ve seen him—in a picture.)

Let us now consider symbolic signs, which are actually a species of indexical signs. For a symbol is not the same thing as a symbolic sign. A symbol is a rule the relation to which differentiates a special class of indexical signs, which are accordingly called “symbolic signs.” (Similarly, an icon is not the same thing as an iconic sign: an icon is the quality in the sign phenomenon which, because it is similar to or the same as a quality in the sign’s object, constitutes the basis for the sign being an iconic sign of it.) Any indexical sign which is a symbolic sign is a replica of the symbol which “rules” the sign, thereby making it a symbolic sign. To understand the symbolic sign we must, then, first understand the nature of indexical signs generally.

All occurring signs are indexical signs, because an indexical sign is simply any phenomenal object with the power of presencing another. Even an iconic sign—though not an icon proper—is an indexical sign. (This is a consequence of the expository approach I have taken here, and it should be qualified by saying that all signs which are phenomenal objects are indexical signs. Neither icons nor symbols proper are phenomenal objects, as the term phenomenal object has been explained here, though this does not mean that they are phenomenal objects but only that a deeper analysis of the concept of a phenomenal object-involving distinctions such as that between “qualisigns,” “sinsigns,” and “legisigns”—would be necessary in order to make clear the exact status of the icon and the symbol proper. It is not profitable to attempt to distinguish here between the terms index and indexical sign.) Now the following phenomenal object

is an iconic sign of a square. (What square? Any square you like.) But it also indexes (indicates). For example, it indexes a sign which looks like this:

square

Or if appropriate criteria for numerical identity and diversity were contextually available, we might, for example, wish to divide the phenomenal object above, which is an iconic sign of a square, into innumerably many different phenomenal objects, simply by regarding it as a numerically different object at each successive moment in time: we could then say that each such object is indexing or presencing a sign exactly like itself at each successive moment. Similarly with this one:

We could regard it as a new object every time a figure-ground reversal occurs, were it appropriate and desirable to do so, though of course we could not speak of an it which is doing a figure-ground reversal in that case. The general point is that iconic signs are also indexical signs, and they may be such in a variety of ways. But to regard them as iconic signs is not the same as to regard them relative to their indexical or presencing power, but rather in relation to their exhibitive or qualitative characters.

Indexical signs are not to be identified with “natural signs”: the latter are a species of the former. The term natural sign could be defined here as meaning an indexical sign which occurs in the natural order and which presences, directly or indirectly, another object in the natural order. (Any sign in a series can be said to presence indirectly any sign subsequent to the interpretant it directly presences.) But signs do not operate only in the natural order; for the power of presencing is a property of phenomenal objects primarily, and it is not contingent upon their existential status. Generally speaking, the usefulness of a sign lies in its power of presencing phenomenal objects (their interpretants), such that an interprétant of the sign which satisfies the purpose animating one’s interest in the sign is finally produced. I cannot here go into the formal structures involved in such series (i.e., the structures of representation involved in representing a phenomenal sequence as a semiotic series), but I should remark that the presencing power of a sign involves a kind of causation not the same as what is usually called “efficient causation.” Efficient causation relates to signs as occurring in the natural order, but semiotic causation—the sort of causation implicit in the term presencing, as I have used that here—is of a sort which I can only hint at by saying that a sign (or complex of signs) may finally presence the interpretant that satisfies one’s interest in the sign only long after one has lost all conscious interest in the matter, and by an interpretational chain which could not possibly have been predicted in advance. Yet long-range interpretational results of signs are by no means completely unpredictable in spite of that (which is why there is a basis for talking of causation in this connection).²⁵ For example, the revolutionary goes about saying “Oppression!,” “Injustice!,” “Freedom!,” and so forth, in appropriate situations, and can often be reasonably certain that these signs will, in due time, and in one way or another, result in interpretants of the sort he hopes for, even though it may be several years away. To deny some kind of causation here would be obtuse; to claim that it can (in principle) be explicated in terms of efficient causation only would be mistaken, if it is true that sign power is a feature of phenomenal objects primarily. In any case, though, bear in mind that what a sign presences, directly or indirectly, is, like every sign of an object, the appearing of the object itself (in the sense explained earlier).

Symbolic signs are differentiated by their relation to a “rule” which somehow delimits or defines a range of interpretants of them as symbolic, and as the particular symbolic signs they are. But how is the idea of “rule” to be explicated? Contrary to the belief of many theorists, the term rule is far too anarchic in its sign power to bear the explanatory burden usually put upon it. (Peirce himself occasionally uses the term, but not regularly, and I use it here only because it has some helpful familiarity to the reader.) The idea that langue is prior to parole is tantamount to saying that the rule gives the meaning to the sign. But this exactly reverses the situation. What converts an indexical sign into a symbolic sign is rather a delimitation (cf. definition) or restriction, a narrowing down of the meaning or sign power it already has. The problem is not “How do words get their meaning?” but rather “How do meaningful entities come under control such that their meaning is made finite or limited?” An attempt at a precise answer to this cannot be attempted here, but an intuitive sense for this can be gotten by realizing that words and other symbols are, as it were, moral/ethical entities. (I say “as it were” because it is probably best to treat this idea as a metaphor here, though in fact it is no metaphor.)

Consider what we would call a “social role,” such as that, say, of being a physician. Any person who purports to be such (e.g., by setting himself up in practice) thereby incites certain expectations about himself or herself: for example, it is expected that they have a certain knowledge and that they will act upon their patients only insofar as they are reasonably convinced that their actions will be conducive to health. The morality or ethics associated with social roles is not mysterious in origin: it is simply the standing expectations directed towards anyone identified as being in that role. We know, of course, that no individual will ever perfectly meet such expectations. But when they deviate sufficiently from such expectations we are inclined to say something like “That so-and-so is no surgeon, but just a greedy butcher,” notwithstanding the fact that the person has all of the relevant credentials, and perhaps even the knowledge, we expect of a physician. So also with words as such. A word is a word, and is the word that it is, because it “lives up to” certain expectations about its sign powers, and insofar as it deviates from these it is not counted as a word, or at least not as the word it is supposed to be relative to the expectations. Thus the word may well have—normally does have—sign powers falling quite outside the expectations that make it a word: insofar, it is merely an indexical sign. But insofar as an indexical sign has the expected powers it is a symbolic sign, the expectations being the symbol. (By expectations I mean that which is expected, not the psychological state of expectation. A definition formulates such expectations, and thus a definition might be regarded as the symbol proper relative to the symbolic sign defined. But insofar as a definition appears, as a phenomenal object, it, too, is a symbolic sign with a corresponding symbol, and is not itself a symbol.)

The term expectation has its own problems, needless to say, and I am not supposing that it is any more clear than the term rule. However, my purpose is to move us toward a more fruitful way of thinking about words and other symbols than is suggested by the term rule. Thus I suggest that we provisionally understand by the term symbolic sign an indexical sign which is given special treatment when it occurs, with only some of its presencing powers being regarded as properties of it as the particular symbolic sign it is, the rest being regarded as properties of it only insofar as it is a nonsymbolic sign. What this means is that the interpretants of a symbolic sign identified as such are treated in a special way in the process of our observation of the presencing powers of signs. I have laid great stress on the idea that the sign-interpretational process is best regarded as primarily an observational or experiential process: instead of thinking of the reading of a text as a mental process accompanying a sense perception of it, for example, I have tried to encourage thinking of it rather as a perception of what signs are themselves doing. But of course all observation—perhaps all experience whatsoever—involves an active element also. For example, when one looks at an object one is doing something: first one focuses attention here, then there, deriving information from each perspective, synthesizing that information, and so on, finally arriving at a seeing which satisfies one. So also in reading a text of any kind: one observes a sign and perhaps experiences a certain interpretant sequence, moves to a new sign and its interpretants, perhaps returns to the first one again and reexperiences its powers, and so on, synthesizing interpretants from both, and so on, just as in sense-perceiving. The difference between the symbolic sign and the merely indexical sign has to be understood in relation to this active aspect of observation. A symbolic sign is one which can function as such only because one knows how to observe a symbolic sign as such, that is, only because one knows that it must be treated in a special way in “reading” it if its special sign value is to be made accessible.

The sign-interpretant series is not a simple linear chain. It is rather to be thought of as a complex of initially independent sequences which join with one another, branch, join with still others, branch, and so on, such that by the end of the process all have ultimately contributed (like contributory rivers) to a final resultant interpretant. The inter-pretational process is a process of composition of a phenomenal object, such that one begins with a number of unrelated phenomenal objects, follows out the interpretational sequences they generate qua signs, and ends up with a composite object which is the resultant of the sign powers of all of the phenomenal objects that constituted independent starting points. Thus, all of the signs in the present paper, for example, were put together by me with the idea that the reader might be led, by following out the sign powers of the individual signs, to a resultant single phenomenal object which is the object of the composite sign which is the complete paper. (That object is, of course, Peirce’s conception of the sign-interpretant process.) The author’s intention is not what I am underscoring here: that intention is available to the reader simply as more signs—more context—which, along with the context each reader individually provides, adds still more starting points (i.e., starting points in addition to the signs in the paper itself). The point is rather that the sign-interpretant process, construed as one in which the observer is trying to observe intelligently, is one which involves strategies and techniques of observation of the signs and their sign actions, and the symbolic sign in particular has to be understood in terms of the appropriate strategy for the observation of it. In fact, it can only be differentiated from nonsymbolic signs with reference to that strategy. Thus while an iconic sign can be defined in terms of de facto qualitative likeness to its object, and an indexical sign in terms of the de facto power of presencing another phenomenal object, the symbolic sign can be defined only relative to how it is to be treated or regarded by the sign observer (interpreter, reader, audience). Its unique sign value as symbolic is wholly dependent upon its being treated in a certain way by the intelligent sign observer in the observational process. (This is a shorthand way of speaking: it does not mean that we must introduce an external mind as an independent factor; it means rather that a certain reflexive operation must be introduced into the analysis of the sign-interpretation sequence whenever a symbolic sign as such is supposed to occur.)

It is not possible here to explain that treatment adequately because it would entail developing the formal principles involved in construing a phenomenal process semiotically, that is, it would entail explaining in detail how to represent such a process in semiotic terms. I can only say that the representation of a given reading of a text would involve starting a new sequence with every sign which is not an interpretant in a sequence already started; following that with its interpretants until one of them presences two or more objects directly, which means a branching; joining sequences whenever certain signs occur which would warrant it (each such joining representing a further composition of the object); guided all the while by semiotic conceptions construed as principles dictating and/or advising branching and joining moves, given the aim of arriving ultimately at a resultant object exhibiting maximum coherence. (Such a representation, which would itself be an iconic sign of an idealized experience of a text, would involve the use of judgment continually, and is far from being a representation of a mechanical process.) Whenever a sign is functioning at once in a symbolic, indexical, and iconic way (as often occurs) there is ipso facto a three-way branching, but there are other causes for branching as well. There will also be level moves involved, that is, moves in which signs take signs as such as their objects. (In fact, the differentiation of iconic, indexical, and symbolic signs cannot be adequately explicated in a formal way without understanding these level moves and how they relate to the branching and joining aspects.)

Take the present text, for example. As a whole, this paper is a single (but highly complex) symbolic sign whose object is Peirce’s semiotic considered in relation to linguistics. In order to gain access to that object experientially through this sign (which I composed with that purpose in mind), it is necessary to interpret its constituent symbolic signs, though not in any strict order. Signs are not “logical atoms,” that is, ultimate units of analysis, and there are no absolutely “simple” signs; hence, what counts as a constituent sign is a matter of what it is profitable to regard as such, given one’s particular analytic aims (e.g., a word, a phrase, a sentence, a paragraph, or whatever). To simplify, though, let us suppose that the interpreter starts with the first word of the first sentence of this paper, which happens to be A. Now, this sign is iconic in visual shape with signs the interpreter already knows to occur in contexts in which it is customarily interpreted to signify symbolically that the sign which follows it in a certain customary order of interpretation is a symbolic sign which may signify any of a number of different objects (i.e., is a general sign); and the interpreter also knows that the sign A customarily signifies that in interpreting the sign it signifies—which is the sign semiotic linguistics, in this case—one may select any one of its objects in the process of interpreting it, at least until such time as the interpreter may encounter a further sign which qualifies or restricts this in some way. (Were the first word The instead of A, the interpreter would be alerted to expect that such a qualification would shortly be forthcoming.) When we say that A (or a or An or an) is an indefinite article we are saying something more or less equivalent to this, and it would add nothing interpretationally pertinent to say also that it is in the English language; for it is not what language it is in but what signification it has which is to the interpretational point.

That the sign A is an indefinite article, that is, has that kind of ideal signification (ideal because it is a symbolic sign), is something which the interpreter has to infer from the evidence. In this case, the evidence consists in such facts as that it occurs with a space between it and the next sign (that is, the sign sequence is A semiotic linguistics instead of Asemiotic linguistics), the fact that it occurs in a book whose primary readership will be people who, as the reader knows, will in fact construe it as having the signification of an indefinite article, and the fact that the author must have known this, too, and surely would not have arranged for such a sign to be in that place had he not assumed that such an interpretation would be put upon it. Such considerations do not deductively necessitate the conclusion that A would be best interpreted in the way described above (i.e., as an indefinite article), but the interpreter would normally—and reasonably—draw that conclusion. Thus in order to determine the identity of A as a symbolic sign it is necessary first to interpret it as an indexical sign; and there is no way to determine, on that basis, what symbolic sign it is without interpreting it as an iconic sign in the way indicated at the beginning of the analysis. This is true of symbolic signs generally: there is no access to their symbolic identity without prior interpretation of the facts of their occurrence, their perceptible form, and their likeness and spatial or temporal relationships to other signs, which involves interpreting them both indexically and iconically. Moreover, the signification of the special sort of symbolic sign which A is, namely, the sort we call an “indefinite article,” requires ascertaining in every case what it is in fact spatially adjacent to in that particular occurrence; for without that being ascertained there is no way to know which symbolic sign it in fact signifies in that occurrence. Because its signification can be determined only if this kind of fact is known, which involves construing it as an indexical sign in a special sort of way, indexicality plays a role in its interpretation which it does not play in the case of all symbolic signs, and that is why it—like definite articles and demonstratives—is regarded, rightly, as a special sort of sign relative to other symbolic signs. But, like all articles and demonstratives, it is still a symbolic sign, nonetheless, just as and are symbolic signs.

Before proceeding through a few more steps of this—which will be enough to indicate to you something of what is involved in such an analysis—let me interject that this is not to be construed as a psychological analysis, as if the aim were to hypothesize about what would actually go on in the mind of the interpreter. I am describing this in a shorthand way in order to convey what I deem essential in as short a space as possible, but it should be thought of as an analysis of the inferential steps one might go through if one had to think through the justification of a claim that a certain text has a certain possible interpretation. I am assuming that the signs in question are themselves acting so as to produce a range of interpretants, among which the interpreter has to identify those interpretants which are their interpretants as specifically symbolic signs, and as the particular symbolic signs they are. In other words, the interpreter’s task is not that of “reading meaning into them,” for if they do not already have meaning, and are not in fact presencing those meanings, the interpreter can do nothing. The task is rather that of isolating those meanings which will ultimately yield the best overall interpretation of the text; and any sign the interpreter considers may in fact be yielding many meanings with which he has no concern because they are not germane to his interpretational purpose, regardless of what that purpose it. If the analysis I am giving seems to be too complex to be plausible, given the fact that we normally go through no such conscious thought process, I can only repeat that the psychology of the matter is not at issue here. I might suggest, though, that a person who knew no English but was working from a translational handbook might go through some such steps quite explicitly. (One might also compare the sort of minute analysis involved in the decipherment of archaeological texts.)

I should also emphasize that the semiotic analysis of a text need not start on such a microscopic level as this. There is no a priori principle which could decide what are to be taken as the analytic units, as this is entirely a matter of the purposes of a particular inquiry. For example, I have been working for some years on an analysis of Plato’s dialogues in which I make use of semiotic principles. My hypothesis is that the object of the entire corpus of Plato’s work, taken as a single sign, is a thing called “ϕιλoσoϕίa”, this being effected through the symbolic construction of several other objects (the most important one being the figure of Socrates) which function as iconic signs of that object. In the process of analyzing the dialogues in this way I occasionally find it necessary to work with particular word sequences in a microscopic way, somewhat as illustrated here; for the Greek text is usually so badly misrepresented in translation that nothing can be taken for granted about the significance of most of the passages important for my purposes. But the extent to which I go into such minute detail is dictated solely by the need to confirm and support the hypothesis, and much of the analysis is in terms of relatively macroscopic objects yielded by the text which are themselves functioning as signs of objects which are in turn functioning as signs, and so on. For it is characteristic of literary works, myths, and so forth, that the objects ultimately signified are available only through several levels of signification, such that objects of signs are themselves signs (of one type and another) of objects which are themselves signs of objects, and so forth.

Let us return to the sample analysis. Having ascertained that the next sign to interpret is the sign semiotic linguistics (which is composed of the second and third words at the beginning of the present paper), the interpreter will then take account of the fact that this sign is composed of two signs, semiotic and linguistics, each of which is visually iconic with other signs which the interpreter already knows to be in customary use to signify things with which he or she has at least some familiarity (namely, semiotic things and linguistics things), and it will be reasonably assumed that the presence of signs of such shapes in this place are good evidence that they are indeed occurring with these symbolic significations. Moreover, their propinquity and relative spatial relationship, taken in conjunction with the fact that semiotic appears with no final s, whereas linguistics has one, is an indexical sign of the fact that the object of the sign semiotic linguistics is a linguistics of the semiotic type (rather than a semiotic of the linguistics type). Given the prior interpretation of the sign A, the reader can move from the sign A semiotic linguistics to any phenomenal object of his or her choice which has the property of being a semiotic linguistics. (This could be its only property, and of course it may be a merely possible object.) The next step, then, would be to go to the sign would, the interpretation of which would affect how the object of the sign A semiotic linguistics is to be represented in relation to the modalities of the possible and the necessary; it would therefore require the introduction of some signs of those objects—I mean the objects signified generally by the words the possible and the necessary—which would have to be situated appropriately within the interpretational sequence. I will not pursue the analysis further here.

In order to do this sort of thing carefully, a notation would have to be developed comparable in some ways to the notations developed for the purposes of deductive logic, though I am certain it would be rather more complex. It might even require the use of hints from topological mathematics. Such private attempts as I have made to work out the principles of such notation—and, so far as I know, no one else has attempted to understand Peirce’s semiotic in this way to date—have led to frustration due to the limitations of two-dimensional representation. Peirce himself did some experimentation with three-dimensional graphical representations in connection with his Existential Graphs (an ingenious system of notation he developed as his preferred alternative to the algebraic type of representation for deductive logic), but whether any help for the present problem is to be found in his work on that I do not know at this time. In the absence of any such notation, it is difficult to know whether one is making a blunder even on a relatively simple level of analysis, and the sample analysis I give above is intended only to be suggestive of what this would be like, rather than being put forth with confidence as a correct analysis.

I should make it clear, though, that I am not saying that a semiotic linguistics—or any other form of applied semiotic—would have as its basic task the constructing of such analytic representations, or of the general method of such representation. A linguist working from the semiotic approach might well develop and use a special method of representation, but this would be merely a technique in the service of other ends. It would always be possible in principle—though, I think, prohibitively difficult in practice—to work without any such aid, just as the late Medieval logicians developed a way of doing in highly controlled Latin some of the same things contemporary logicians do with their specially devised notation. Such notational systems are like mathematics in some respects, but their purpose is not primarily that of a calculus—even though some calculating can be performed with them—but rather that of a highly controlled way of talking about certain properties of things. Thus, for example, the logician’s form (x)(Fx→Gx) is a way of saying what All men are mortal says without saying anything about men and mortals or any other beings in particular, and without leaving any ambiguity as regards whether or not any existing things are being signified as existing. The primary purpose of developing such a notation is that of sharpening our theoretical ideas by developing what is, in effect, a highly controlled descriptive symbolism which can function as an organ of perception in the analyses of concrete phenomena. It would be quite wrongheaded, in my opinion, to see in such notational devices something comparable to the calculative role of mathematics when it enters into the natural sciences.

As regards the relevance of all this to linguistics, it seems to me to be something like this. A linguistic sign would be a special case of a symbolic sign. The differentiation of the linguistic sign from the symbolic sign in general, and thus the differentiation of linguistics itself, would be based on what it is convenient to regard as specifically linguistic, rather than on any fundamental principle within semiotic itself. Symbolism undoubtedly functions in all of the arts, for example, and the differences between the various arts are not differences of great importance at a theoretical level (or so I would guess), no matter how remote in type music may now seem from literature or painting. Still, there is sufficient prima facie difference between these sorts of things to make it reasonable to suppose they had best be regarded as importantly different areas of study at present, and I should think that linguistics, construed as the study of the principles and varieties of specifically linguistic symbolism, would in fact constitute a conveniently distinct scientific enterprise for some time to come.

As I hope I have made clear, semiotic is a general theory of the principles to follow in understanding how signs work, and it is thus essentially oriented toward the perfection of what the Greeks would have called the Tϵ̓xvn-craft, art, skill—of understanding signs. Given what is meant by sign, this is tantamount to saying that it aims at the perfection of the learning process. But it is a nonpsychological conception of what learning is; for it assumes—through its basic phenomenological stance—that learning is the acquisition of skill in the intelligent observation of signs, and that this depends upon knowing what signs themselves can do, and on developing both mandatory and preferential techniques for observing them (experiencing them, understanding them, interpreting them) in order to be able to learn the most we can from them. A semiotic linguistics would thus be a science which aimed at making us maximally intelligent when it comes to observing—and of course using—signs of the special sort it would be concerned with.

From such attempts as I have made at understanding the nature of words from the semiotic point of view, I am myself convinced that they have systematically distinguishable powers and functions beyond anything existing theories are able to articulate, and that, moreover, there are possible linguistic powers—I mean powers of signs, not powers of sign users—beyond any that have ever yet been discovered and used by even the greatest of word users. If the basic idea of semiotic is sound, or capable of being made sound, then linguistics might in due time find a way through it to the nature of words and their workings which would reveal linguistic properties—both possible and actual—comparable in their marvelousness to the physical properties which the natural sciences have discovered. To discover these properties, however, linguistics would have to abandon the idea that its basic task is that of interrogating “native informants” about the “acceptability” of simple sentences in order to compile grammars and dictionaries, and instead direct itself toward studying human discourse—poems, plays, treatises, novels, conversations, and so on—with the aid of an analytic technique for discovering what is going on in such phenomena, and hypothesizing about what further kinds of powers the logos might be found to have, if we could learn how to orient ourselves toward it properly.

NOTES

1. The foundational paper is the 1867 essay “On a New List of Categories” (CP 1.545-59). The abbreviated reference is to Volume 1, paragraphs 545 through 559, of The Collected Papers of Charles Sanders Peirce, in 8 volumes, C. Hartshorne, P. Weiss, and A. Burks, eds. (Cambridge: Harvard University Press, 1931-1958). This collection is but a small part of the total corpus of Peirce’s work, the rest being available in a variety of media, including some 80,000 manuscript pages now available in microfilm and xerographic form. See M. Fisch, K. Ketner, and C. Kloesel, “The New Tools of Peirce Scholarship, with Particular Reference to Semiotic,” Peirce Studies No. 1 (1979):I—I7, for a comprehensive and reliable description of the present state of availability of Peirce’s work.

2. John Lyons, Introduction to Modern Linguistics (Cambridge: At the University Press, 1969), sec. 1.4.

3. Lyons, p. 50.

4. J. Ransdell, “Semiotic Objectivity,” Semiotica 26(1979):261-88, esp. 28 if.

5. C. S. Peirce and J. Jastrow, “On Small Differences of Sensation,” Memoirs of the National Academy of Science 3, Part 1 (1884) 73-83, reprinted with corrections and later reflections at CP 7.21-48.

6. Peirce was the originator of the theory of hypothetical or “retroductive”—as distinct from inductive and deductive—inference. The Collected Papers contain so many discussions of this topic that the interested reader should simply consult the indexes to the volumes under the headings of hypothesis, retroduction, abduction, and, of course, inference.

7. This does not imply that logical form is a species of rhetorical form. Certain linguistic forms are indicative of (but never identical with) the pertinent logical forms, and such linguistic forms are themselves rhetorical forms. A logical form is something to which a linguistic entity is put in relation for the special purposes of logical assessment, and is “in” the linguistic entity only in that sense.

8. I present a second formal argument against the same type of linguistics in J. Ransdell, “Semiotic Objectivity,” Semiotica 26(1979):261-88, esp. 281-84, though my argument is relativized there to the conception of objectivity I develop in the earlier part of that paper.

9. The discussion of the moral/ethical nature of the symbolic sign in Part III of this paper may suggest why theories of meaning have so often been isomorphic with political doctrines.

10. Noam Chomsky with Mitsou Ronat, Language and Responsibility (New York: Pantheon, 1977), 89.

11. See Gérard Deledalle, Théorie et pratique du signe: Introduction à la sémiotique de Charles S. Peirce (Paris: Payot, 1979), for a sound and systematic account of Peirce’s basic semiotic conceptions, particularly in relation to linguistics. I was not acquainted with Deledalle’s book until after the present paper was written and I cannot make any helpful comparative comments here. I can say, though, that Deledalle has an informed and generally accurate grasp of Peirce’s basic ideas, and that, while he perceives clearly that no mere eclectic synthesis of Saussurian and Peircean conceptions is possible, he does indicate how one can move from the Saussurian to the Peircean conceptual framework. He remarks on page 40:”... [I]l est possible et même relativement facile de retrouver dans la sémiologie de Saussure quelques-uns des concepts fondamentaux de la sémiotique de Peirce. Ce que ne veut pas dire qu’ils soient assimilables. Leur signification contextuelle interdit au contraire en principe toute assimilation des uns aux autres.” Work in progress by David Savan will undoubtedly result in an especially penetrating account of Peirce’s ideas on the purely philosophical level, and philosophers may also look forward to a revised version of John J. Fitzgerald’s Peirce’s Theory of Signs as a Foundation for His Pragmatism (The Hague: Mouton, 1961). Douglas Greenlee’s Peirce’s Concept of Sign (The Hague: Mouton, 1973), is, in my opinion, so vitiated by certain basic misunderstandings of Peirce’s writings as to be worse than worthless as an introduction to Peirce’s semiotic, though it will have some value for those with an independent understanding. See J. Ransdell, “Another Interpretation of Peirce’s Semiotic” and Jarrett Brock, “Draft of a Critique of Greenlee’s Peirce’s Concept of Sign,” both of which are in the same issue of The Transactions of the Charles S. Peirce Society 12(1976):97-no and 111-26, respectively, and see also Greenlee’s reply in the same issue, at 135-47.

12. Peirce’s use of mentalistic terminology has been seriously misleading to his contemporary readers, who usually overlook the fact that Peirce began developing his semiotic theory prior to the existence of what we now think of as psychology, and who ignore his repeated statements in later years that semiotic is not psychologically based, as he conceived it. Peirce’s use of mentalistic terminology does not usually indicate a shift to a psychological perspective, because these terms did not bear the meaning for him that they usually bear for the contemporary reader. In his later writings, when psychology was beginning to flourish, Peirce sometimes uses such terms with the awareness that they might be misconstrued, but with the idea that others probably will grasp little of his meaning unless they first understand him psychologistically. The purpose of his special semiotic terminology was that of making it possible to break out of a psychologistic stance toward meaning-phenomena. His dilemma—on the horns of which he impaled himself repeatedly—was that if he defined his terms in a non-psychologistic way no one would understand him at all because his approach was too alien to the prevailing ways of thinking; whereas if he defined them psychologistically his readers might gain some idea, but a badly distorted one. One cannot read Peirce with any real understanding without a grasp of the desperate communicational situation he was in and the effect this had on how he expressed his ideas.

13. Neither the conception nor the term phenomenology was derived from Husserl, and Peirce subsequently changed to the neologism phaneroscopy in order to avoid confusion of his and Husserl’s conceptions. However, in the past few decades the word phenomenology has come to be used in a sufficiently indeterminate sense to make it reasonable to use it now with reference to Peirce’s theory, as an indication of its general type, though it would doubtless be advisable to use the word phaneroscopy whenever the differentiating features of his particular approach are under discussion. See J. Ransdell, “A Misunderstanding of Peirce’s Phenomenology,” Philosophy and Phenomenological Research 38(1978):550-53. (I should also remark that the word semeiotic is the preferred term in recent scholarship when one wants to differentiate Peirce’s semiotic from other possible instances at that general type of theory. But I prefer myself to use semiotic for the same reason I prefer phenomenology to phaneroscopy, namely, because I think it is rhetorically preferable at present not to reinforce the reigning tendencies to perceive Peirce’s thought as idiosyncratic.)

14. Peirce often used the word psychical as an equivalent of semiotical, drawing implicitly upon the meaning of the Greek word 𝜓vxή during the classical period, according to which whatever possesses psyche is ipso facto alive, as in Plato’s definition of 𝜓vxή in Phaedrus 245E. (Perhaps it is needless to say that psychical in this sense has nothing to do with the unfortunate meaning the word psychic has acquired in virtue of its occurrence in the popular phrase psychic powers.)

15. Presence is my term, not Peirce’s. His corresponding terms would be suggest (in a psychological context) and determine (in a logical context); but I could not use his terms here without extensive commentary. The term phenomenal object is also mine, and it corresponds roughly to his term phenomenon as he often uses the latter term in his later writings, and also to his use of the term object in many contexts. He sometimes uses the term idea in this sense, also; for he always understood by idea not a mental entity but that oj which one is aware. (Compare the Greek term ɩ̓𝛿𝜖́a, which was not a mentalistic term.)

16. Such an object may be “a single known existing thing or thing believed formerly to have existed or expected to exist, or a collection of such things, or a known quality or relation or fact, . . . [or] a collection, or whole of parts, or it may have some other mode of being, such as some act permitted whose being does not prevent its negation from being equally permitted, or something of a general nature desired, required, or invariably found under certain general circumstances” (CP 2.232). In short, anything you can think of.

17. See Alberto Cortes, “Leibniz’s Principle of the Identity of Indiscernibles: A False Principle,” Philosophy of Science 43(1976):491-505.

18. Except in a somewhat recondite sense, as explained in Part III of this paper.

19. “. . . [A]s for feelings, they are always referred to some object, and there is no observation of feelings except as characters of objects” (CP 7.376; see also 5.244-49, 462).

20. My distinction between phenomenal object and phenomenon is not systematically made by Peirce in these terms, who tends to use phenomenon indifferently in this respect.

21. Quality is a Peircean term, and he sometimes uses the word idea as an equivalent, though he sometimes uses it as indicated in note 15 above.

22. This is not a definition of the sign-interpretant relation. I do not define that relation here.

23. My basic objection to Douglas Greenlee’s Peirce’s Concept of Sign (see note 11 above) is that he does not understand that the semiotic object is not something logically external to the semiosis process itself.

24. I develop this point in a paper on the iconic sign and its function in reconciling the ideas of “direct” and “indirect” perception. See J. Ransdell, “On the Epistemic Function of Iconicity in Perception,” Peirce Studies No. 1 (1979):51-66.

25. Peirce called it “final causation,” which, as I explain elsewhere, is actually the idea of a self-corrective or “cybernetic” process, which is regarded by Peirce as involving efficient causation but not reducible to it. See J. Ransdell, “Some Leading Ideas of Peirce’s Semiotic,” Semiotica 19 (1977):157-78.

Thirdness and Linguistics

RULON S. WELLS

Peirce considers Thirdness to have many aspects: generality, universality, regularity and lawfulness (and their correlative, conformity to rule or to law), mediation, continuity, vagueness.¹ This paper considers the bearing of his doctrines about Thirdness upon language. It proceeds by first considering certain major facts of language, and then by asking how Peirce’s system takes cognizance of them. The system turns out to perform rather poorly in this respect, and the paper ends by inquiring into the reasons for this.

Concern with the facts of language occupies the first two parts of the paper. Part One compares rules governing language with rules governing art, and draws some conclusions about rules in general. Part Two compares rules with laws (laws of nature, that is), emphasizing their specific differences even though both are varieties of Thirdness. Part Three comments on Peirce’s treatment of rules and of laws, remarking the infertility of his system as a heuristic for ascertaining an adequate treatment of linguistic rules and analyzing the reason for this infertility.

I. PRELIMINARY REMARKS
ON RULE AND LAW

By the seventeenth century, the concept of law had developed to the point where one distinguished the laws of man, of God, and of nature. Laws of nature are descriptive, not prescriptive or normative. Later, people thought of the differences this way: if you break a law of man or of God, so much the worse for you; but if you break a putative law of nature, so much the worse for that law of nature.

The word rule is, as the word law once was, primarily normative, but with the difference of connoting conscious or deliberate or witting obedience. Dante in De vulgari eloquio c. 3 (quoted in Leo Spitzer, Essays in Historical Semantics [New York: Vanni, 1948], p. 19) defines the vulgar way of speaking as that which we, imitating our nurse, adopt without any rule. In prevailing current linguistic usage, set by Chomsky (e.g., “What is a rule of grammar?”), a linguistic rule is neither deliberate nor normative. If we ask, Why call it a rule, then?, the answer could be given: Such a rule is not an absolute norm, but it is a conditional one. Take, for example, the rule which states that in order to speak correct English (or, on pain of not speaking English), if you use a plural subject you must use a plural verb. These rules are conditions, and, moreover, they are necessary but not sufficient conditions. All the rules of English together will not tell you what to say; they will only tell you what you may say. In popular parlance, they are don’t’s, not do’s. In technical parlance, they impose constraints, or, as Bloomfield and Harris called them, limitations upon distribution. So it is possible to find normative character in Chomsky’s sense of rule, whether or not he intended it: linguistic rules are norms.

Under this conception, rules are constraints upon us; but (a) they only constrain us from, prohibit us from, enjoin us against saying certain things, and (b) their constraint is not total. The speaker has freedom within constraint.

At this point, we temporarily leave language and turn to an area where the relation between constraint and freedom is even more striking than in language, namely, art. Then we will come back to language.

Pope, in his Essay on Criticism 1.88-91, makes a point which he then illustrates by Homer and Virgil:

It will turn out that this point holds not only for art (whether verbal or non-verbal), but also for language (or rather, for speech, the actualization of language). As the artist is bound by necessary conditions, but within these bounds has his freedom, so, too, is the speaker. There are other analogies: the artist, but likewise the speaker, has (a) his materials on which and (b) his implements with which he operates, as well as (c) the circumstances in which he operates; and artwork, like an utterance, is typically used—or at least intended and adapted—for communication. Most important of all, in both kinds of communication there is an active and a passive role.

On the subject of freedom and constraint in art, the great theorist is Kant. (Peirce has nothing of his own to say about esthetics, and does not seem to understand what Kant has to say.) In section 46 of his Critique of Judgment, Kant relates art to genius. He makes these points: (1) there are rules for producing art (i.e., great, good art), but it takes genius to find them; (2) art is different from science in this regard; and (3) the rules are subjective, as taste is; there is no rule of criticism. Kant’s examples are: for a scientist, Newton; and for an artist (specifically, a poet), Homer and Wieland. In his second point—the difference between art and science regarding subjection to rule—Kant seems to be mistaken both in his conclusion and in the reason he gives for that conclusion (namely, that a genius is different in kind from a nongenius, whereas Newton differs from a tyro only in degree). But there is a consideration, unnoticed by Kant, that could be placed at his service; it involves something like the difference mentioned above between active and passive roles. There are many who can understand Newton though they cannot achieve anything like his achievement. They can follow but cannot lead, as Glaucon and Adeimantus in Plato’s Republic could follow Socrates’s insights but could not originate them. Just so, it is often remarked that in mathematics the ability to follow proofs is different from the ability to discover proofs. Genius, then—at least the kind we are interested in here—is genius at discovery. Any other kinds there may be (for example, genius at organization, genius at persuasion, genius as an infinite capacity for taking pains) are here left out of account.

Now let us begin to put this analogy between art and language to linguistic use. Both the speaker and the artist are both constrained and free within their constraints. Are these lines sharp? Is the constraint utter constraint, and the freedom utter freedom? It would please Peirce to find degrees and continuity.

In a first-approximation model, there is a sharp line between the grammatical and the ungrammatical. This is possible because utterances are atomistic, they are molecules built out of discrete atoms: the units of sound, the phonemes (or, according to prevailing contemporary theory, distinctive features; similarly mutatis mutandis for written language), and the units of vocabulary, the morphemes.

This capital point deserves some elaboration. Because there are units, we can with mathematical rigor speak of combinations. And because the number of units is finite, for any finite length, the number of combinations is finite. This gives us an enviable control. Even though the set of grammatical sentences is infinite, we can organize it, for example, alphabetically, into the one-unit sentences, the two-unit sentences, and so on. In our present discussion, the principal use of this control is that it lets us speak in an obviously meaningful way both of discovery and of creation as selection. It is a fiction, but one whose implications and presuppositions can be definitely stated, to say that it is as if all the possibilities are laid out for one to review, and all one has to do is to choose. We could analyze the difference between the genius and the duffer thus: that although each of them selects from a range or set of possibilities which he has considered, the set considered by the genius is larger than the set considered by the duffer.

When we are considering possibilities, it is advantageous, for certain purposes, to confine our attention to possibilities such that each possibility is sharply distinct from each other possibility. It is so advantageous in the discussion of creativity and of genius that we may now ask whether the possibilities of every art may be conceived after this fashion. We have already seen that it is possible with versification because of the discrete, atomistic nature of verse. So the only remaining question is whether the matter of every art other than versification is similarly atomistic. I believe that the answer is that in every art the matter either is naturally atomistic or else can, in an arbitrary but admissible way, be treated as atomistic. To give but a single and brief example: every painting, even the smoothest, can be conceived pointillistically; it is only a question of choosing points so small that adjacent points will not be perceived as distinct from each other (precisely as, in the cinema, adjacent frames are not perceived as distinct).

A slightly different way of characterizing creativity is, on the Greek (versus the Judeo-Christian-Moslem) model, an arrangement, a disposition of a préexistent matter. So far as this matter is discrete and atomized, one may speak of combination. At one time, the image of the kaleidoscope was used to suggest this model.

Let us now return to language and its actualization, speech. So far, we have the contrast between what one may (can, could, might) say and what one does say, with creativity placed inside the bounds of those possibilities. We have not challenged the sharpness of the constraining bound, nor of the freedom within the bound. Before we do that, I have in mind a last comparison with art.

On the model so far, one is entirely free to mean. If, given the constraint of grammar, one adds the constraint of circumstances, the constraints are increased, but the basic sharp contrast between constraint and freedom or license is not altered. And there is still room for creativity, for the selection which will make other people say, “I wish I’d said that.” According to our model, it would be more appropriate to say, “I wish I’d meant that”; having settled on a meaning, how one says it follows automatically.

In art criticism, we speak of effectiveness; also, as elsewhere, of success. When we speak of art, we have in our minds (whether in the back or in the front) the distinction between bad and good art, good art including great art (great meaning very good or the best). The great artist, in making his selections, takes into consideration not only grammatical and circumstantial constraints, but also the question of effectiveness; and here, no doubt, psychological laws enter the discussion. So far as that is true, it is even easier to think of the creator (artist or speaker) as discoverer. However, psychological laws are not relevant here as rules which govern the creative genius (whether they do govern him or not), but as one more set of constraints that bound his liberty.

So far, no difficulty has been found with the view that bounded freedom divides sharply into a component of sheer bondage and a component of sheer freedom. To find difficulties we must turn elsewhere.

An important vista has appeared before us in this discussion. We have not discovered anything new, but we have been reminded of something old that we had neglected: the distinction between simple (pure) and composite (mixed) cases. The situations in speech and in art where we found an element or factor of bondage and an element of freedom are composite cases, which break cleanly apart into a component of sheer bondage and a component of sheer freedom. The possibility then immediately suggests itself, that this cleavability is true of all such cases. This would be a heavy blow for synechism to withstand: a whole class of cases of apparent continuity would be done away with, not as illusory (i.e., as merely apparent), but as composite. Synechism, to be interesting, must be concerned with simple (fundamental, irreducible, original, primary), not with composite (derivative, reducible, secondary) continuity.

But there is a kind of rule, yet to be examined, that has a better claim to exhibiting simple continuity. Consider the following rule of poetics. It is common in verse to allow substitutions; for example, in English iambic pentameter, to allow a trochee to substitute for an iambus. This substitution is more acceptable in the first foot than elsewhere; this is sometimes expressed by calling the first foot the favored position. We have not yet accurately stated the rule, though. It is not a mere permission, because if it were, there would be nothing wrong with making every first foot a trochee. It is possible, however, for something to be wrong; there is such a thing as excessive substitution. (Similarly, in various matters of morals and of mores, there is such a thing as abuse of a privilege.) So a more accurate formulation of the rule is: a moderate amount of substitution is permitted. In this more accurate formulation, the rule can be seen at a glance to be vague.²

From considering such a rule we also gain a further insight. It is sometimes said that rules are made to be broken; and we are sometimes urged to be an untrammeled Walter von Stolzing rather than a score-counting Beckmesser. In that witticism and this romantic advice two quite different things are confounded. The point may be that such and such a rule is not binding on y, even though x says it is binding; for example, that contrary to Beckmesser’s claim, Walter is not subject to the Mastersingers’ rules. Or, quite differently, the point may be that rules should not be followed literally and slavishly, but with discretion. The latter interpretation holds that the rules as formulated are a good approximation—and so are by no means worthless—but still are not perfect, and so should not be followed literally.

Both interpretations are common, but it is only the second that is of present concern. Many who agree with the witticism interpreted the second way would go further and hold that, in general, it is humanly impossible to achieve perfect formulation of the rules that people follow. Michael Polanyi’s concept of tacit knowledge covers the case where someone knows something but cannot perfectly formulate or articulate it.

From the same example we can draw a further lesson. The class of rules that are difficult or impossible to formulate perfectly falls into two subclasses which are worth distinguishing from each other. Those of one subclass are difficult because of the amount of detail; those of the other subclass are difficult because of the subjective judgment that is expressed by terms like moderate. (Rules of this second subclass are like Aristotle’s mean.)³

A rule in the first subclass is precise, in that persons render categorical judgments and agree with each other in their judgment whether a given expression conforms to or violates the rule; in the case of a rule in the second subclass, by contrast, a person’s judgment will be gradational rather than categorical, or else one person’s judgment will differ from another’s, or else both situations will obtain.

Rules of the second subclass, such as the rule “Be moderate in substitution,” correspond roughly to what would commonly be called rules of style. In fact, the correspondence is close enough so that we may consider making it even closer by changing our use of terms slightly. To state my proposal using the labels that Rudolf Carnap proposed, can the concept rule of the second subclass serve as a precise explicatum for the vague explicandum rule of style? It seems to me that it can, and, if so, then we can divide the province of grammar from the province of style in this neat way: grammar comprehends the rules of language and of speech that are easy to formulate, and those hard rules that belong to the first subclass; style comprehends the hard rules belonging to the second subclass. Poetics therefore falls into a linguistic and a stylistic component: for example, in metrics, a part of poetics, the linguistic component includes the rule (hypothetical imperative) that, for English verse, sets up iambic pentameter as a type; the rule enjoining moderation in substitution belongs to stylistics.

Finally, we can say that stylistics belongs to art. Ernst Cassirer is right in contending that language and art are two distinguishable symbolic forms, distinguishable even though each is involved with the other.

II. LAWS VERSUS RULES

Let us now compare rules with laws. Peirce tries to assimilate them, but there are fundamental differences. It would take us too far into the realm of metaphysics—into what he calls realism; into his animistic, personifying conception that nature tries to conform to laws—if we were to examine his claim that conformity of instances to law is essentially like what Kant calls acting out of respect for a law. The following argument enumerates, rather than elaborates, points of difference, and therefore is conditional, not absolute. The essential difference is that which exists between fact and value. Peirce endeavors to disunite the category of sheer fact by splitting it into varieties of two of his categories, namely, Brute Fact (Secondness) and Law (Thirdness). My reply to him is an ad hominem argument, not based on personalities, but based on logical consistency: namely, Peirce reintroduces the familiar category of fact under the label of truth, without realizing what he is doing. I mean truth in the sense subsequently termed the semantic conception of truth, such that for any assertion p there is a necessarily equivalent assertion “It is true that p.” No doubt this conception of truth is fallible, as is every conception, including Peirce’s categories; therefore, fallibility does not suffice to reduce a conception (fact is the conception of present concern) to Peirce’s categories. Peirce makes a distinction between fact and truth that has some basis in ordinary use, but not enough to warrant the identification of fact thus distinguished from truth with Brute Fact, in his sense of Secondness. Peirce’s insistence on this distinction, however, is also insufficient.

Discussion of fact is complicated by actuality. This complication bedeviled the Positivists’ conception of laws as descriptions. When it was said that if the laws of man or of God were broken, so much the worse for the breakers, but if the putative laws of nature were broken, so much the worse for those laws, it was meant that for laws of nature, unlike laws of man and of God, to be broken and to be invalidated are the same thing. The difficulties in moving from this formulation to the formula that laws of nature are descriptions arose because description was needlessly taken to mean description of actualities, thus omitting altogether Peirce’s element of would-be (Chisholm’s contrary-to-fact; Goodman’s counterfactuality). Peirce made the mistake of moving from recognition of would-be—recognition that, as he put it, general laws are really operative in nature—to an assimilation of this variety of Thirdness to the variety known as norms or values. This mistake of confounding formal identity with material identity is discussed in detail in Part Three below. At present, it is my concern to point out that Peirce, in his 1898 “Lectures on Vitally Important Topics” (Collected Papers 1.616-77), accords due recognition to those properties of norms which distinguish them from laws of nature. There results a hard, sharp dualism which—this is the inconsistency I spoke of—is glaringly at odds with synechism and is of the sort which, when readers called such anti-synechistic dualisms to Peirce’s attention, he defended with the airy observation that “the question cannot be decided in that way” (CP 2.116; see below p. 196).

Both rules of language and rules of esthetics are, therefore, subsumed under norms, and as such they can be distinguished from laws of nature. The distinction is probably partial rather than total; that is, there will be some properties that norms and natural laws have in common, and others that norms have but that natural laws do not, or vice versa. We may as well follow the convenient Aristotelian terminology and say that norms and natural laws are two species of one genus. The inconsistency of Peirce that concerns us is that he sometimes neglects—virtually denies—the specific differences, while at other times recognizes them. This paper assumes without further discussion that they must be recognized.

The circumstance that the same word, law, is used both normatively and factually—normatively of the laws of men and of God or the gods, factually of the laws of nature—is easily explained historically. The normative conception came first and the factual conception evolved from it. The transition must have been mediated by ambiguity in the word why. Why do people pay taxes? Because that’s the law of the land. Why do people rest one day in seven? Because it was decreed in the Decalogue. Why does water freeze when it gets cold? Because Nature has made it so. Very likely, the personification of nature contributed to the mediation. The change of meaning, described in current linguistic terms, would be a metonymy. It is a law of nature—of human nature—that, by and large, laws of men and laws of God (or of the gods) are followed. (There are large exceptions: one class of exceptions is often labeled dead letter; another is the class involved in the witticism about rule-breaking described in Part One above, as interpreted in the second way.)

The transition from normative to descriptive law occurred in ancient Greece. The transition is almost complete in Aristotle, but he retains one major vestige, namely, the doctrine that laws of nature hold always or for the most part; in other words, that there are genuine exceptions to laws of nature. For Stoicism, on the other hand, as Peirce rightly says, laws have no exceptions. This same doctrine accorded with the Judeo-Christian belief in God’s infinity and omnipotence. But on Aristotle’s view, unnatural things (things contrary to nature) happen, though not often and not for long.

Nature, personified as female, is not completely the mistress of the matter which she rules. The interesting steps that lead to Aristotle’s view may be retraced briefly. Cornford (From Religion to Philosophy) emphasizes the morality which pervades the archaic Greek view of nature. Destiny and Fate coincide with Right. Anaximander says that things perish because they are unjust to other things; Heraclitus (fragment 62 Bywater=80 Diels-Kranz) says, contradicting this, that the universal strife is not universal injustice but universal justice. And why does the sun follow its regular course? Because if it strayed, it would be found and punished by the Erinyes (fragment 29 Bywater=94 DielsKranz). In short, there is a Lawgiver who imposes his law; this echoes the phrase of A. N. Whitehead (Adventures of Ideas §§7.5ft.), who has especially Plato’s Timaeus 4iff. in mind.

These doctrines make it seem that things are a certain way because some personal, divine power has decreed them to be so; but this is only one side of the Greeks’ view. From the beginning, their view had an opposite side which retracted almost completely the first side. The first side tends to say, That is right which is decreed by the divine powers; the other side says, rather, That which is right is decreed by the divine powers. The way in which the Greeks accomplished this contradiction is by placing in the background, behind the personal gods in the foreground, impersonal divine powers such as the Lots (Moirai), to which the personal gods are subject. The two-sidedness already evident in Homer is clearly identified by Albin Lesky’s History of Greek Literature (1966, pp. 65-66). How will the Trojan War turn out? Sometimes Zeus decrees his will, but sometimes he lifts his scales to find which one comes down. In the former case, he determines events in the sense that he decides the outcome; in the latter, he determines them in the sense that he learns, finds out. As a last attempt to evade the issue, the Greek mind claims that you can act against the Lots, but that the Lots will get even with you for it (Lesky 1966:66; Cornford 12 and n. 3, on Iliad 16.433). ft ^a further step—and a Stoic one—to replace the Lots, which one can disobey, though with penalty, by the Fates, which lead the willing and drag the unwilling (Seneca Ad Lucilium 107).

Something should be said about the neutralization, the desacralization, the secularization of the concept of rule. ‘Law’ (nomos, lex) was desacralized earlier, and then, centuries later, ‘rule’ (regula) followed suit. But the route by which this took place was different for the two concepts. With rule, as is obvious, for example, in Chomsky (a major recent agent in the conceptual change), the change is mediated by the distinction (Kant Metaphysical Foundations of Ethics §2) between categorical and hypothetical imperatives. A rule entails a command, expressed by an imperative. Unless we are instructed to the contrary, we are to construe a rule as a categorical imperative. What Chomsky has done, in effect, without instructing us, is to call any imperative a rule; and in fact he only applies the term rule to hypothetical, not at all to categorical, imperatives.

III. RULES, LAWS, AND THIRDNESS

Having considered rules of language and of art, and having pointed out the distinction between rules and laws, we are ready to look at Peirce and ask what he can contribute to the discussion. The answer that I find myself forced to give is “Not much.” The observations that any adequate treatment of rules and of laws must include are compatible with, but not entailed by, Peirce’s system. It follows, as a corollary, that his system has no heuristic power in leading us to insights about rules of language.

These are adverse, even harsh, contentions, but I believe with Peirce that truth has nothing to do with being agreeable to reason, and, moreover, I believe that I can locate within Peirce’s system the reason for this impotence. I speak of the reason, in the singular, because there is only one, although it can be stated in several ways. The most startling way, perhaps, is: idealism; a more analytical way is: a failure to do justice to the difference between formal identity and material identity.

Peirce drew the distinction between the two clearly enough, but he failed to respect it. Indeed, he deluded himself that he was empirically sensitive and had taken due precautions against such hybris. But he slipped, and he slipped by believing in something that was agreeable to reason, namely, in continuity (mediation). He rejected any deductivism—any claim to deduce the singularities of the world from its generalities, such as Hegel’s notorious youthful indiscretion of coming down against the possibility of an eighth planet—as breathing the spirit of the seminary rather than of the laboratory. But he thought he saw a way lying between aprioristic deductivism and utterly passive empiricistic inductivism; namely, a logic of inquiry, a heuristic, which would ask though it would not answer; would suggest though it would not insist. His belief in the reality of Thirdness led him to think in such a way; it led him to slip.

I make a double claim: not only did Peirce slip; but the slip is one that he would at least sometimes acknowledge as a slip.

It is easy to charge that Peirce is a “triadomaniac.” Evidently Peirce has been chaffed about this, and is sensitive to the charge. What concerns us here is not his defense, but his exceptions, that is, the cases where he rejects a triadic treatment, and more particularly, the places where he rejects a mediating, synechistic treatment.

The most conspicuous exception is his mind-matter dualism. Peirce admits that at first glance mind appears to be distinct from matter, but he answers monistically that they are but two sides of one fact. (He gives the same answer to his dualism of inner and outer.) This answer seems triadic, because the thing of which mind and matter are two sides mediates, at it were, between them. But this answer backfires; neither side of x is closer to x than it is to the other side, so that the would-be mediation is unsuccessful; the third thing, which it sets up in order to unite the first two, is merely claimed to unite them. Showing that it is like the one in one respect and like the other in another does not necessarily show that it unites them. There are places where Peirce tries a genuine mediation (e.g., CP 4.87), but this attempt to establish that the difference between mind and matter is one of degree only is forced to the two concessions that (1) “the difference ... is certainly very, very great” and (2) there is a “remarkable absence of intermediate phenomena,” concessions which make the attempt silly.

On occasion, as in CP 2.116 (cf. 2.87), Peirce says that “one might antecendently expect that the cenopythagorean categories [i.e., Firstness, Secondness, Thirdness] would require three modes of being. But . . . the question cannot be decided in that way. Besides, it would be illogical to rely upon the categories to decide so fundamental a question.” So, purporting “to make an entirely fresh investigation,” he arrives in forty lines at the conclusion that there are three modes of being.

The dualism that this paper has focused on is the dualism between law and rule, between fact and value. Can Peirce’s system do justice to this? In a superficial sense, yes. Peirce himself emphasizes the dualism in the 1898 “Lectures on Vitally Important Topics” (CP 1.616-77). the question immediately arises, Was he true to himself in doing so? Isn’t the position of these lectures at odds with other, and perhaps deeper, principles of his? He connects this dualism with three others: that between theory and practice, that between reason and instinct, and that between the time being and the long run; and, of course, each of these connections is a mediation in some sense, so that each of the dualisms is mediated—subordinated to Thirdness. But Peirce is less sensitive to the equally cogent consideration that the Thirdness in question is simply the general truth that there are radical, ineluctable dualisms, some of which lead to the others.

It is not hard to find a category for ineluctable dualisms: each member of a dualism is a Second to the other. It is hard, however, to find a heuristic. Let us return to the original question. What the linguist primarily wants from Peirce is a heuristic—a system, or a framework—that will help to discover significant facts, or to describe them. In the present instance, Peirce’s system would justify itself if it directed the linguist’s attention more forcefully than other writers do to the fact-value (or law-rule) distinction, or if it offered the linguist a more lucid exposition of the distinction. But the system does not do either of these things; and, moreover, its exposition, in spite of setting up a dualism at odds with his synechism, does not compensate for the inconsistency by being notably apt. The values Peirce deals with are ethical and logical values, and his purview is limited to these by the restriction to deliberate self-controlled conduct. Behavior that is unconscious, and largely innate, thus falls outside the scope of normative science.

It might not be hard to emend his system so as to abolish the restriction and thus enlarge the scope, but that would hardly support claims of adequacy and fruitfulness.

The question before us is whether Peirce’s system helps us to characterize linguistic signs. A negative answer is harder to prove than a positive one, because proof of the positive need only prove that there is a place in his system which is helpful, whereas proof of the negative must prove that there is no place at all in his system which is helpful. The preceding argument may prove that the two places in Peirce’s system, his treatment of norms and his treatment of laws, are not helpful in treating linguistic rules, but this obviously does not prove that no places are helpful.

In this paper, however, I am offering the difference between laws and norms not as a sample but as an example. I cite it not as evidence for an induction, but in the same way a diagram is employed in geometry. And just as the diagram by itself is not sufficient, so an example needs to be accompanied by an explanation of what the example exemplifies.

Briefly, the reason why Peirce does not adequately distinguish rules from laws is that he treats both of them as Thirdnesses. If we could say that rules = Thirdness and laws = Thirdness, his identification would be valid. But we could only make that statement if we forget, as Peirce tends to, a distinction that he draws between formal and material categories. In the case before us, there is formal but not material categorial identity between rules and laws. Thirdness itself is only a formal category.

Described in Peirce’s own terms, Peirce’s mistake is to move from formal to material categorial identity. The temperamental penchant that leads him to this mistake is, to give it his name for it, idealism. The reason why we can describe the mistake in his terms is that he was not always subject to it; he occasionally acknowledged that his usual idealism was a mistake.

Peirce even reveals the history of the mistake. In the 1860s, when he addressed himself to the topic of categories, he spent two or three years working at the problem, his work taking the external form of a critical study of Kant. He came to think (i) that there are two kinds of categories: formal and material; (2) that the formal categories are also aptly characterized as universal, and the material as particular; (3) that he could make no headway in dealing with the material, particular categories, but succeeded in establishing that there are three formal, universal categories; and (4) that his three formal categories—eventually labeled Firstness, Secondness, and Thirdness—substantially agree with Hegel’s three Hauptstufen der Gedenkens (Enzyklopädie§83): Sein, Wesen, and Begriff. “On account of their connection with numbers,” Peirce had by 1902 (CP 2.87) come to call his categories Ceno- or Kaino-Pythagorean, alluding (CP 1.521, of 1903) to Aristotle’s remark at the beginning of his De caelo about Pythagorean Three. “They agree substantially,” he says in the former place, “with Hegel’s three moments.” (Hegel does not use the term moments for his three Hauptstufen; the three moments of §79 are something quite different.) (The above is a synthesis based on §§1.284, 288, 300, 561; 4.3; 5.37—40, 43; and the passages already cited.)

It is important to note that while Peirce is ready to agree that there are material categories in addition to the formal ones, his lack of headway with them (CP 1.284, 525; 5.38) makes him neglect them. His semiotic, in particular, is developed deductively, in the main (as he acknowledges, in a limited way, in a letter of 1909⁴), from universal categories alone. Therefore, it, too, is merely formal and lacks content or substance. I imagine that if we were to ask Peirce why he is not disturbed by this limitation, he would give a twofold answer: (1) the formal categories are more important (CP 1.288), and (2) semiosis not only falls under one of them, Thirdness, but almost is that one.

Indeed, the word almost is an afterthought. His first published statement of the categories (1867) identified Thirdness and Semiosis outright, by calling his third category representation; when he came to see the inadequacy of that conception (two essays of three decades later, CP 1.564-65 and 4.3), he tried to repair it by admitting that representation (alias representability, intelligibility, reasonableness, destiny, regularity, law; see Note 1) is only one of two varieties of Thirdness, the other being growth (alias continuous growth, life, continuity). (In “The True Nature of Peirce’s Evolutionism,” p. 321, I argue against admitting disjunctive characterizations.⁵)

The formal, deductive origin of Peirce’s semiotics is unable to guarantee its fruitfulness. There are taxonomies that prove fruitful; in such cases, people are apt to say that they are more than mere taxonomies. The periodic table of the elements is a fine example. Employed heuristically, it suggested the hypothesis that the apparent gaps were not real gaps; in other words, that empty cells in the matrix of chemical elements represented ignorance, and that missing elements could be discovered. (Both Mendeleev and Newlands predicted gallium in 1871, and Boisbaudran discovered it in 1878.) One wants to employ Peirce’s classifications of signs similarly, but efforts to do so have not proved fruitful. This paper has exposed a complementary deficiency. Instead of showing that a possibility presented by Peirce’s taxonomy remains empty, it shows that a possibility not presented by Peirce’s taxonomy is full, is actual. One might think of developing Peirce’s semiotics so that places appeared to accommodate hitherto neglected phenomena, but what concepts would generate these places? Where would they come from? They cannot escape the dilemma of being either internal or external to the system as it stood before. If they are internal, then their previous apparent absence was a mere defect in the exposition of the system, not in the system itself; if they are external, then their introduction requires us to extend the system. In either case, the point remains that the system was not heuristically effective.

What does it mean to say that concepts are internal to the system? All concepts are internal to every system, in the respect that they may be introduced by the law of excluded middle. For instance, we could introduce the previously absent concept of stylistic rule at the point of legisigns by saying that every legisign is either a stylistic rule or not a stylistic rule. This would treat the legisign as a genus with two species. But the question raised by Aristotle in his discussions of taxonomy remains pertinent: is the specific difference that divides the genus into species internal to the genus? In the present example, we would have to answer no, or at least say that it has not been shown to be relevant. Internality on the basis of excluded middle is trivial, because excluded middle says that every concept is internal to every system. Any stronger claim that the concepts needed to describe linguistic rules are internal to Peirce’s semiotic remains quite unsupported.

NOTES

1. Not to mention representation, intelligibility, reasonableness; destiny; growth, life. See below p. 199.

2. It is vague in the ordinary sense in which we speak of a vague boundary, as opposed to a precise one; it may not be vague in Peirce’s sense, in which a vague assertion grants license to the hearer, in contrast with a general assertion which reserves license to the speaker. In Peirce’s sense, every existentially quantified proposition is vague.

3. Nicomachean Ethics 2.6:1107a1; 2.9:1109a24, b19; 4.5:1126b3; 6.1:-1138b29; 6.8:1141b29-1142a31.

4. To Lady V. Welby, 14 March 1909; complete letter in C. Lieb (ed.), Charles S. Peirce’s Letters to Lady Welby (New Haven: Whittock, 1953), 34-40. This passage (Lieb, p. 36) is also quoted in C. K. Ogden and I. A. Richards, The Meaning of Meaning,, 5th ed. (New York: Harcourt Brace, 1938), 288. Peirce happens to make his remark in connection with his concept of interpretant, but it applies just as aptly to all of his semiotic concepts.

5. In E. C. Moore and R. S. Robin (eds.), Studies in the Philosophy of Charles Sanders Peirce, Second Series (Amherst: University of Massachusetts Press, 1964), 304-22.