According to Pareyson, all human life is the invention and production of forms.
— Umberto Eco, The Open Work
I. INTRODUCTION
There is an eccentric, theoretical object in the work of Umberto Eco. It occupies the third section of the Trattato di semiotica generale and bears the name of “theory of modes of sign production.”1 Its position within Eco’s A Theory of Semiotics is not marginal. While the first half of the book—largely occupied by the “theory of codes”—explicitly aimed at providing a theoretical systematization of a whole historical period (both in the author’s research and in semiotics), the second half (the “theory of sign production”) proved to be so advanced that it would not be read anymore for the next thirty years.2 The third section no longer discusses the static elements of cultures, such as the crystallized semiotic objects that they produce (the “codes”), but, rather, their dynamic ones. That is, instead of taking into account the products of the semiotic work, the theory of sign production directly investigates the modelization of this work, that is, the processes that lead to those products.
There are at least three problems that such a theory of sign production should try to address: (i) how to introduce a historical dimension into the crystalline building of the “system” as described by the theory of codes; as a consequence, sign production paves a theoretical way to studying semiotic practices as specifications of a more general semiotic “labor”; thus, a phenomenology of semiotic production becomes possible; (ii) how to define the articulation of the sign-function procedurally and not systemically, so that it becomes evident that the same sign-function that relates expression and content is only a momentary stabilization of a restless movement; (iii) how to provide a definition of some semiotic forms of subjectivity, according to the assumption that “semiotics is entitled to recognize these subjects only insofar as they manifest themselves through sign-functions, producing sign-functions, criticizing other sign-functions and restructuring the pre-existing sign-functions.”3 As precisely observed in Semiotics and the Philosophy of Language: “We recognize ourselves only as semiosis in progress, signifying systems and communicational processes. The map of semiosis, as defined at a given stage of historical development (with the debris carried over from previous semiosis), tells us who we are and what (or how) we think. The science of signs is the science of how the subject is historically constituted.”4
In the theory of sign production such a semiotic constitution of the subject is theoretically described by a typology of modes of sign production. The following contribution aims at providing a critical introduction to Eco’s theory of modes, trying to show that, though hidden, it is a key point in Eco’s semiotic thought.
II. TYPOLOGY OF MODES OF SIGN PRODUCTION: PRELIMINARY NOTES
As noted, section three of A Theory of Semiotics is dedicated programmatically to a “theory of sign production,” and yet “production” is a notion that is not entirely obvious in Eco’s masterpiece. An examination of the concept requires us to discuss at least three oppositions: the theory of codes versus the theory of sign production, semiotic versus physical, and production versus interpretation.
1. The Theory of Codes versus the Theory of Sign Production
The opposition between the theory of codes and the theory of sign production is based on the axis “signification” versus “communication.” Signification concerns the “socially conventionalized possibility of generating sign-functions,” while production deals with the work required “in order to physically produce expressions for many practical purposes.”5 A definition of a theory of sign production assumes that the latter takes into account “a large range of phenomena such as the common use of languages, the evolution of codes, aesthetic communication, different types of interactional communicative behavior, the use of signs in order to mention things or states of the world, and so on.”6
In this sense production indicates when history becomes relevant for a theory of codes. Compared with Eco’s model of the semantic space as the magnetization of marbles in a box,7 “a semiotics of the code is only interested in the results of this game, after the intervention of the magnetization,” while “a semiotics of sign production . . . is interested in the process by which a rule is imposed upon the indeterminacy of the source.”8 Consequently, “whereas a theory of codes was concerned both with the structure of sign-function and with the general possibility of coding and decoding, a theory of sign production will thus be concerned with all the problems outlined in Table 31.” The table in question “concerns the kind of labor required in order to produce and interpret signs, messages or texts,” and lists a plethora of semiotic phenomena and activities.9 With regard to the relation between systems and processes, Eco initially states a theoretical priority of signification (that is, of the system) over communication (that is, over the process): “It is possible, if not perhaps particularly desirable, to establish a semiotics of signification independently of a semiotics of communication: but it is impossible to establish a semiotics of communication without a semiotics of signification.”10
And yet, exactly at the end of the theory of codes, Eco states that “a semiotics of the code is an operational device in the service of a semiotics of sign production.”11 It cannot be otherwise, given the “indeterminacy principle” at work in semiosis, for which every act of interpretation, including the ones that stem from theoretical practices, inevitably changes the conditions of observation.12 But if the defining condition of semiotic interpretation is not that there is a sender, but, rather, a human recipient,13 then any interpretation is a production: as a consequence, semiotic production is in principle superordinate to signification. So, in truth, the priority of a theory of codes over a theory of sign production is just a methodological artifact, the latter being the point of arrival of the former, so that the theory of production is not only a correlate, but properly exceeds the theory of codes. As often happens with the literary genre of the treatise, a symmetrical organization hides the linearity of a theoretical path moving from some initial conditions to final conclusions. In relation to this point, the reader can consider an “example” of sign production ante litteram, the recognition of the stone as a tool: in this case, Eco initially states the priority of the system over the process, of signification over communication.14 However, what is shown in the description provided by the author is exactly the phenomenon of “code institution,” based on the technical mechanism of “recognition” as described by the theory of modes of sign production that will be presented later in the book.
The relevance of production is also apparent in the semantic models that de iure should belong to a theory of codes. In Eco’s semiotic theory the first definition of the “sememe as encyclopedia” appears in A Theory of Semiotics.15 Here the compositional nature of each sememe is said to reveal a “synchronico-diachronical spectrum,” having to account for the contradictory multiplicity of uses.16 This definition is very far from the atemporal purity of structuralist models.
2. Semiotic versus Physical
Production is therefore the “labor required in order to produce and interpret signs”: it exists through an “effort” that is both “physical” and “psychological.”17 The “physicality” to which Eco refers is related to the energy required to create the signal. At several points Eco insists on this aspect, which potentially bears semiotically paradoxical implications. As an example there would be a “privileged status” of the expression with respect to content: “Any expression unit can be defined in itself, not only independently of its combinational possibilities but also of its material qualities as a functive.”18 A consequence of such a status is that the typology deals with the “labor performed on the expression continuum in order to physically produce signals.” Signals are “produced” or “selected among pre-existing entities.”19 Thus there is at least a twofold meaning of production: a physical production (which does not concern the recognition of the signal as a possible sign, being the “expression” in this case independent of the “content”) and a semiotic production (which in fact includes recognition as a selection of expression with respect to a content). Eco’s semiotic theory originates from the separation—stated for the first time in La struttura assente (1968)—between the sign and the signal as defined by information theory. Even more, “the criterion of interpretability”20 states that what is relevant to semiotics is constitutively semiotic. This means that uniplanar signals are to be taken into account by semiotics only as they could enter into a biplanar relation with content units, thus resulting in sign-functions. The insistence on this “physicality” therefore remains somewhat obscure. But if no one has raised the accusation of physicalism to Eco, it is because, as we shall see, there is a peculiar rhetoric of physics in A Theory of Semiotics. In any case, the “material” autonomy of expression is indisputably a key point.
3. Production versus Interpretation
In the previous discussion “production” and “interpretation” have been considered as synonyms according to the assumption, made explicit in A Theory of Semiotics, that “production” is actually the complex term that, in fact, subsumes both.21 For example, in the section dedicated to “overcoding and undercoding” Eco observes that the passage between semantic marks is an interpretation, in the form of an abductive inference that “would seem to enter into any act of decoding”: “it is the more evident instance of production of a sign function.”22 So at the very lowest level of encyclopedic functioning, the two notions of production and interpretation get blurred. And yet, as we shall see shortly, in the typology of sign production an opposition between the two terms is still at work: it seems that interpretation and production identify two different levels that Eco’s typology flattens by means of the graphic artifact of Table 39 in A Theory of Semiotics.
To summarize, the concept of production (i) is near and homogeneous with respect to that of enunciation; (ii) concerns the “physical” creation of the signal; and (iii) maintains an ambiguous relationship with interpretation, both of hypernymy and contrariety. It is urgent at this point to examine in depth chapter 3.6 of A Theory of Semiotics, “A Typology of Modes of Production.”
III. ON THE PRODUCTION OF SIGNS
I have already mentioned the emphasis that Eco places on the physicality of sign production. The “productive labor” is a real “physical stress”: “the labor of producing the signal; then the labor of choosing, among [a] set of signals . . . those that must be articulated.”23 Eco here finds an important reference in the Marxian approach to semiotics by Ferruccio Rossi-Landi. Rossi-Landi developed the idea of a “scheme of homological production” in which the labor performed by semiotic subjects (intended as historical subjects analogous to workers in the Marxian framework) is the pivotal notion for semiotics, prior to the materials on which the labor is exercised, for example, in the first place, language.24 “Labor is always labor”: labor is the invariant that makes it possible to build up homologies among different practices, including—of course—semiotic ones.25 Rossi-Landi, while declaring the exemplary bidirectionality of commodities as messages and messages as commodities,26 in fact considers messages as commodities. In contrast, at the opening of A Theory of Semiotics, Eco restores the semiotic centrality of the notion of value, maintained as superordinate to its economic assumption.27 That is, value is a semiotic primitive that receives specific determinations in different domains, for example, in the economic one. And yet in A Theory of Semiotics the Marxian reinterpretation of semiotics by Rossi-Landi is clearly present (and discussed in depth): not by chance, Eco proposes a notion of semiotic “labor” and not of “activity.”
The typology of sign production concerns the way in which the expression is produced: “the table records the way in which expressions are physically produced and not the way in which they are semiotically correlated to their content.” Therefore, the content in itself is not included among the relevant dimensions. Eventually, the way in which the content is correlated to expression may be “implied by two decisions,” which pertain respectively to the rationes (and in particular to ratio difficilis) and that bring into play the epistemic subject with respect to temporality, since they can be taken “either before or after the production of the expression.”28 This is a crucial and delicate point, and will be discussed in depth later. The independence from the content is stated explicitly since the table “speaks of physical procedures and entities that are ordered to the sign-function but that could subsist even if there were no code to correlate them to a content.”29 The passage is very complex and difficult to unravel. It postulates a principle of autonomy according to which an expression may exist per se as a consequence of its physical status: this feature can be expressed as “ordinability to function” (the term is indeed a hapax legomenon in Eco’s writing). It seems to indicate a possibility that characterizes the expression as what can be “put in order” in the semiotic function. In this sense it indicates what is entitled to be “ordained” as an expression of content, in the sense of a holy ordination. Thus the typology of sign production is offered as a theory of expression theoretically independent from the content (except for the oblique implication of the ratio): the semiotic map that Eco offers in A Theory of Semiotics reconnects the expression not to a phenomenology of perception but to a physics of production. Even if the purely functional definition of the sign-function by glossematics (Hjelmslev) is restated by Eco in many places, expression must be studied by its own proper physics. A specialized physics is needed because expression is a “natural” phenomenon, extraneous both to the passionate subject of phenomenology and to the natural object of physicalist reductionism: such a physics of expression seeks the most general laws that rule the interaction between semiotic bodies, as they exemplarily manifest themselves in the mechanics of imprints, but also in the reconfiguration of a syntax of interaction between subjects and objects as requested by clues, as well as in the selection for fragmentation implied by the sample, or even in the Pavlovian mechanics of programmed stimuli “intended to elicit an immediate response in the receiver.”30 This figurative framework does not at all imply a behaviorist or physicalist option. Rather, it proposes a theoretical landscape populated by subjects and objects playing the role of operators and operands respectively: sign production being precisely the physics of their interactions. So at the end the rhetoric of “physics” that abounds in A Theory of Semiotics is a strategic move that keeps dormant the censors of historical materialism from charging semiotics with the infamous accusation of idealism. As stated at the very end of the book: “There is sign production because there are empirical subjects which display labor in order to physically produce expressions, to correlate them to content, to segment content, and so on. But semiotics is entitled to recognize these subjects only insofar as they manifest themselves through sign-functions, producing sign-functions, criticizing other sign-functions and restructuring the pre-existing sign-functions. By accepting this limit, semiotics fully avoids any risk of idealism.”31
By means of this rhetoric of physics, Eco casts a spell on any Marxist censors, and continues to publish his encyclopedia files in the Monist. The following passage is still more revealing: “On the contrary semiotics recognizes as the only testable subject matter of its discourse the social existence of the universe of signification, as it is revealed by the physical testability of interpretants—which are, to reinforce this point for the last time, material expressions.”32
Here the interpretant is in itself a “material expression.” And it could not be otherwise, as in Eco’s interpretation of the Peircean dynamics of unlimited semiosis the interpretant is the expression of some content that in turn can be an interpretant, and so on.33 In this way, the purely functional definition of sign-function still holds, for which an expression is anything that opens the interpretation of a content, according to the “law of total reversibility” discussed by Eco both in A Theory of Semiotics and in I limiti dell’interpretazione.34 In fact “the matter segmented in order to express something expresses other segmentations of the matter.”35 In this model the sign-function acts as a sort of jumper cable on the matter, connecting two poles that, as a consequence, take the name of expression and content. To sum up, the theory set out in section 3.6 of Eco’s A Theory of Semiotics is not, then, so much a theory of expression sub specie materiae as it is a theory of matter sub specie expressionis: it describes the ways in which matter can be used as expression, classifying “types of productive activity that can give rise, by reciprocal and complex interrelations, to different sign-functions.”36
Once the problem of matter is discussed, it is possible to introduce the four dimensions whose combination yields the typology presented in Eco’s Table 39 (see Figure 1, below): “physical labor required to produce expressions,” “type/token ratio,” “continuum to be shaped,” and “mode and rate of articulation.”37 In this way it also becomes possible to disentangle the knot between production and interpretation.
Figure 1. Classification of modes of production (from Umberto Eco, A Theory of Semiotics)
IV. LABOR
The first dimension that enters into the typology is labor. The term indicates “the physical labor needed to produce expressions (ranging from the simple recognition of a pre-existent object or event as a sign to the invention of previously non-existent and un-coded expressions).”38
This first parameter can be articulated in four modes: recognition, ostension, replica, and invention, distributed on a continuum ranging from the simplicity of recognition to the complexity of the invention. This sort of progression is indicated in Figure 1 by the positions of recognition and invention on opposite ends, while the continuity between modes is sanctioned by the co-belonging of two examples (fictive samples and programmed stimuli) to two modes, in a transition zone (respectively between ostension and replica and between replica and invention). To place recognition as a production labor is a policy statement for Eco’s semiotics, which considers interpretation as a form of production. However, it is not clear how recognition can be considered as a labor, that is, as already mentioned, “the labor of producing the signal”:39 in what way does to recognize a symptom mean to produce it physically? Yet the script that Eco has in mind when he speaks of labor is well demonstrated by other determinations that labor itself receives. For example, programmed stimuli are “intended to elicit an immediate response in the receiver”: “A flash of light during a theatrical performance, an unbearable sound, a subliminal excitation, and so on, are to be listed among stimuli rather than signs.”40
Thus they are “not semiotic” for the receiver. It is clear that the semiotic quality depends entirely on the sender: the receiver’s point of view is decisive for establishing the nature of the work since there is sign-function (for the receiver) when “the stimulus is the expression plane of a supposed effect functioning as its content plane.”41 Such semiotic artifacts are “productive operation[s] that are not usually considered as semiotically definable”:42 an obvious result since, as stated in the discussion on the boundaries of a semiotic theory, “we are . . . witnessing a process of signification—provided that the signal is not merely a stimulus but arouses an interpretative response in the addressee.”43 In the short reprise of the modes of sign production in Semiotics and the Philosophy of Language there is a compact enunciation of the problem: the programmed stimuli are those able to “produce or elicit a blind reaction,” and which have to be considered significant only for the sender, not for the receiver.44 It is clear that the semiotic point of view to be considered here is that of the sender: he or she is the subject of the semiotic labor. And yet unluckily things are not so clear, as Eco adds: “they are instead a case of weak sign that allows one to infer the actual cause from the possible and variably probable effect.”45 Here the problem becomes how to interpret the expression as a cause of an effect, its content: this operation has nothing to do with the preparation of the signal as it has no relation with the “physical” production of the expression. To sum up, in the typology of sign production (both in A Theory of Semiotics and in the following discussions by Eco) there is a constant oscillation between the points of view of the receiver and the sender.
Let’s start from the beginning, that is, from recognition. First of all, it should be stressed that recognition is to be fully considered as a mode of sign production: there is a production of recognition that takes place when the subject views “a given object or event . . . existing in a world of facts” as the “expression of a given content.”46 Recognition is a production because it is really a “re-constitution,” in a phenomenological sense, of the semiotic nature of the world: “In order to be considered as the functive of a sign-function the object or event must be considered as if it had been produced by ostension, replica or invention and correlated by a given kind of type/token ratio. Thus, the act of recognition may re-constitute the object or event as an imprint, a symptom or a clue.”47
The previous definition is absolutely dense. To unravel some of the issues: (i) there is a transcendental semiotic subjectivity, weak enough to be reduced to the impersonal subject of a “consideration,” an operative subject that constitutes the world as a biplanar semiotics; (ii) the semiotic quality of the world implies an “as if” that assumes an instituting mediation, which is, in turn, the very possibility of semiotics as a “theory of the lie”;48 (iii) this mediation is a labor—to be able to reconstitute the object, the subject must “consider as if” rather than passively receiving it; (iv) finally, the definition outlines a clear asymmetry between recognition, as the first mode of the productive labor of the expression, and the other three, ostension, replica, and invention. Indeed the goal of recognition consists exactly in a reconstitution of the object as if it had been produced by one of the other three possible labors. Therefore an ambiguous, double movement starts here. On the one hand, recognition is thought as a semiotic labor among others, as Figure 1 declares by placing it within the same series of the other three labors: recognition is therefore a full-fledged semiotic labor. But, on the other hand, recognition is intended as a semiotic metalabor, one would say a “metaproduction” as a “production of production.” This metaproduction is superordinate to the other three, as it can be intended as the presupposition of a semiotic quality that allows the passage to ostension, replica, and invention: such an interpretation of recognition may be confirmed by the circumstantial evidence of its initial position in the series of the four labors. The set of definitions given by Eco about the four productive labors emphasizes a real actantial asymmetry: on the one hand, in the definition of recognition the recognizer is the receiver of the object-sign that comes from the world, in turn recognized as the sender (producing ostensions, replicas, and inventions); on the other hand, the producer of ostension, replica, and invention is the sender that constructs the object-sign for a (theoretically) distinct receiver. In short, a two-level structure emerges. At the first level the subject of recognition is a metaoperator that establishes the semiotic quality of the world, thus practicing a semiotic metalabor. The metaoperator occupies the place of the receiver of semiosis. Second, assuming the set of positions given at the first level, the metaoperator installs at a second level an operative subject in place of the world: this subject is the subject of the other three semiotic labors. Thus recognition is a production because it is an interpretation (according to the equation between interpretation and production in Eco’s semiotics), while in the other three modes there is production in relation to an interpretation (presupposed by their definition). In these terms, even if recognition as production shares the same nature of the other modes of production, it is at the very same time separated from them. Hence the constant oscillation between the sender and the receiver while discussing examples of ostension, replica, and invention. The best example, as it includes all types of labor, occurs with invention: “A mode of production whereby the producer of the sign-function chooses a new material continuum not yet segmented for that purpose and proposes a new way of organizing (of giving form to) it in order to map within it the formal pertinent element of a content-type.”49
This definition, in which the relevance of the producer as the sender is quite explicit, is followed by the definitions of the other three previous labors of production, according to a rhetorical strategy of contrast. The first concerns recognition: “Everybody recognizes an expression produced by recognition because a previous experience has linked a given expression-unit with a given content-unit.”50 The definition explicitly connects recognition to a figurativeness of the world that precedes the subject, as I will discuss later. The two subsequent definitions are dedicated to ostension and replica: “Everybody recognizes an expression produced by a choice made on the basis of a common mechanism of abstraction, such as the acknowledging of a given item as a representative of the class to which it belongs. Everybody recognizes an expression produced by replica, because the replica replicates an expression-type which has already been conventionally correlated with a given content.”51 Now, a double perspective clearly pervades these definitions because the subject of recognition evidently subsumes the subjects of ostension and replica, as pointed out by the repetition of the phrase “it is clear that we recognize an expression produced.” The definitions of ostensions and replica are moved back to the first level, that of the semiotic metalabor of recognition: it is possible to recognize the expressions produced by someone else by means of the labor of ostension and replica.
That said, if there is clear dual focus, it is rather curious to note that the subject of recognition is played against the subject of invention, when it is rather obvious that two different levels are implicated. Still, a prominent example is that offered in the discussion of Gainsborough’s Mr. and Mrs. Andrews as an example of “moderate invention” in which two points of view are explicitly compared:
From the sender’s point of view, a perceptual structure is considered as a coded semantic model (even though nobody else would yet view it in this way), and its perceptual markers are mapped into an as yet unshaped continuum. . . . But from the addressee’s point of view the result is simply an expressive structure.
Using the painting as an imprint, he makes his way backward inferring and extrapolating similitude rules, and finally re-constitutes the original percept.52
To consider Gainsborough’s painting as an imprint means by definition to recognize (to reconstitute) it as an expression of a content given by the “class of all possible imprinters,”53 that is, to view it “as if” it were produced, in this case, by invention. The three labors of production that can be associated with “active signification” thus describe the syntax of production to be inferred by recognition (as a “passive signification”). This theoretical two-level device is optically flattened by Figure 1.
V. RECOGNITION AND ENUNCIATION
The asymmetry between recognition as a metaproduction and the other three forms of production could raise the suspicion that the common emblem of production hides an irreducible difference, an unbridgeable gap between levels. Put in these terms, the work of recognition seems to be the first step that follows a “postulate of enunciativeness,” where the barbarism indicates not enunciation but the condition of its possibility. This postulate can be described as follows: meaning is based on the labor of production—recognition—that applies to other productions. Something makes sense (that is, is semiotic) if it is recognized as the product of a semiotic labor. A similar assumption has interesting consequences with regard to a theory of communication.
The latter can be intended as the a posteriori reconstruction of an intentionality that ensures the presence of meaning (shown as a logic of production). Communication is the production of meaning through the recognition of someone else who literally “makes sense.” Thus a theory of communication follows a theory of enunciation. What is recognized (in the technical sense) as the product of a production is semiotic. In other words, recognition would be a framing labor that allows one to retrieve the other modes of production. However, although implicit in A Theory of Semiotics, it is perfectly possible that the labor to be recognized is not a replica, an ostension, nor an invention (the three forms of production tout court) but a further recognition. As it is possible to recognize the labor of setting up a replica or choosing an ostension, so it is possible to recognize the labor of recognition, which in turn will be the recognition of another labor. The situation can be formalized by proposing a very simple rewriting system, composed by the four terms (the metalabor Rec and the set of labors L: Ost, Rep, Inv) and by a unique “production rule” (to speak in the parlance of rewriting systems):
Li → Rec+Li, Li ∈ L
The rule states that, given as an initiator a labor Li belonging to the set L of ostension, replica, or invention, then an iteration process may start by applying recursively the production rule to its output. As an example, if Li is ostension:
n = 0: Ost
n = 1: Rec+Ost
n = 2: Rec+Rec+Ost
. . .
and so on. The first string (with the iteration n = 0) represents the theoretical possibility of a labor from the point of view of the producer, while the framing labor of recognition enters at the first iteration. In such a recursive schema the operation of recognition reapplies (potentially ad infinitum) to the result of a previous recognition.
Some preliminary remarks on such a recursive nature of recognition are possible. (i) Recognition certainly has a special status. First of all, because it allows access to the other labors; second, because it triggers the recursion. But it also shares the same status as the other three labors, as it can even be found on their same level (it can be recognized). (ii) This recursive formalization allows an explicit definition of representation. A representation is in fact a semiotic object or status that includes within itself the conditions for its readability: according to the etymological argument, a representation is the staging (repetition) of a “presentation,” that is, of the way in which the meaning has already been produced by a previous act of recognition. Thus the difference between a footprint and a photograph of the same footprint can be expressed in terms of the difference between the subjects of their respective productions: in the first case, an imprinter that may be described with respect to the productive labor of imprint; in the second case, an observer that in turn recognizes the labor produced by an imprinter. Instead of being reduced to a referential framework, representation is described as a specific form of semiotic production. (iii) Such a recursive operation can apply more times, and recognition may give birth to a chain of recognition acts (for example, a photograph showing a photograph). In short, modes of sign production can be nested in series. But at the same time, they can operate in a parallel fashion. As noted by Eco, empirically every text is the result of multiple labors. It would seem that the only labor capable of triggering a series of labors is recognition: in any case, for each recognition, an internal semiotic object is produced that in turn may trigger the discovery of multiple, parallel labors. The simultaneous presence of various semiotic labors, be they in series or in parallel, cannot be taken into account by the classic model of “text” that has been widely generalized—for example, in Greimasian theory—as a general model for semiotic objects. The main fallacy of the text model lies in its etymology that refers to fabric. In other words, this model assumes that the sign-text is the result of a single labor, weaving, performed by a single subject, the weaver. The notion of modes of production allows us to move to a different model that is not forced into the homogeneous nature of the sign-text. Here a semiotic object is not a text, but, rather, a territory. On its surface a plurality of traces of different subjects performing different operations (according to the modes of sign production) can be found.54
VI. ON MATTER SEMIOTICALLY RECONSIDERED
The third parameter of the typology concerns “the type of continuum to be shaped; this continuum can be either homomaterial or heteromaterial; a continuum being homomaterial when the expression is shaped within and by the same material stuff as that with which the possible referent of the sign-function could be made. . . . All the other cases imply a heteromaterial continuum which is arbitrarily selected.”55
To some extent paradoxically, the typology is a theory of expression that does not consider content as pertinent, but that assumes among its constituent features, by means of the “matter,” the possible referent of an expression without a content. Why mention the referent after tireless warnings about the risk of a “referential fallacy”?56 The discussion about the referent occupies chapters 3.2 and 3.3 of A Theory of Semiotics, where, according to the notorious “prudence” of Eco, the problem of facts is introduced by postulating the existence of “extra-semiotic circumstances” lying (in a Kantian fashion) “in the background of any phenomenon involving sign production”: it is exactly “the fact that semiosis lives as a fact in a world of facts [that] limits the absolute purity of the universe of codes.”57 Eco’s move consists in considering the referent as a cultural mediation, transforming the index-sensitive judgment into a metasemiotic relationship between two “semiotic entities” mutually related by means of toposensitive “pointers” that can make use of “marker[s] of proximity.”58 In the index-sensitive judgment “/ this is a cat /” two semiotic objects have to be compared, “the content of a linguistic expression with the content of a perceptual act.”59 The “perceptual act” as a whole indicates a sign-function, technically defined by the correlation between an expression and a content. As a final consequence, the act of referring shows a double relation with the same sememe: “the elements of the content plane of a code coincide with the elements of the content plane of another code.”60 The referent is brought back to the content, that is, to a description internal to semiotics. Or, rather, referentiality is a relationship between semiotic objects: in the case of the cat, between natural language and perception.
Back to typology. By defining the referent in this way, the “referential fallacy” is entirely avoided. It is not by chance that, in the typology, the referent is defined in terms of a “relation of referentiality.” The parameter is binary as it allows only two values: “homomateriality” versus “heteromateriality,” whose definition is, in fact, relational: there is no definition of matter per se but only as a relationship between two semiotic entities, expression and referent. However, if the referent is attributed to the content, then the latter again plays a relevant role in the construction of the typology, in contrast with what was said about the independence of the classification from the correlation between expression and content. It will be a matter of making a comparison between expression and content, designed as a selection of the features of materiality displayed by a possible producer (referentiality).
What then is the theory of sign production? It is a peculiar definition of interpretation (intended, as already discussed, as a “production”) as an abductive inference that correlates an expression to a content, in the semantic context—an isotopy—of “production.” It is possible to speak of isotopy because such an interpretation selects a starting location and a successive path on the encyclopedic graph that includes the semantic features related to the construction of the expression. In this sense, and only in this sense, can it be said that “the table records the way in which the expressions are physically produced and not the way in which they are semiotically correlated to their content.”61 It is not, for example, the whole of the semantic spectrum of the sememe “imprint” that is to be discussed, but only its “physical” component. The modes of sign production therefore seem to explain a “general formativeness” (to speak with Eco’s philosophical mentor, Pareyson),62 semiotically rearticulated and extended so as to include both the paintings by Gainsborough and the footprints of the hare and the rabbit (with basically the same range of phenomena that Pareyson included in his discussion of natural beauty).63
While approaching the dimension of the material continuum, two disjunctions emerge. First, “heteromaterial” is opposed to “homomaterial.” In the latter case the object is “viewed as an expression which is made with same stuff [sic] as its possible referent.”64 The definition allows us to understand what matter and referent mean: the latter is the unit of content conveyed. Being a sememe, this content will be equipped with a mark of “materiality” that, in the case of homomateriality, also belongs to expression. Homomateriality is opposed to heteromateriality, in which the componential representations of expression and content are assumed to be independent. In the case of heteromateriality a second opposition emerges between “motivated” and “arbitrarily selected,” detaching the first three regions from the last six. This opposition needs to be discussed further.
For example, let us consider symptoms and the combinational unit: in both cases there is a complete independence of the expression from the referent, and the prevalence of an encoding/decoding mood. On the one hand, “the best known kinds of replicas are phonemes and morphemes; expression-units constructed according to ratio facilis, using a continuum completely alien to their possible referents, and arbitrarily correlated to one or more content units.”65 On the other hand, medical semiotics defines a symptomatic alphabet whose deciphering seems to follow the model of verbal language. Now, in “recognition of the symptoms. . . . the content is the class of all possible causes (organic alterations),” while in the discussion of replicas Eco said that the expressions are “arbitrarily correlated to one or more content units.”66 Such a remark is particularly interesting because it should not concern the modes of production (which are by definition independent of the content, as mentioned several times). On the other hand, it could be said that the verbal language may be recognized “symptomatically” as a production activity that generates replicas of discrete units (a model of recognition that is at the base of motor theories of speech understanding). To add another element to the discussion, one can take into account the relationship between imprints and congruences or projections. An imprint “says: if a certain configuration on an imprintable surface, then a certain class of imprinting agents.”67 Take the case instead of congruences (or casts) and projections that are arranged along a “continuum of transformations.” The prototypical example of a congruence is a death mask in which “a point in the physical space of the expression corresponds to each point in the space of a real object.”68 In projections “points on the space of the expressive token correspond to selected points on the space of toposensitive perceptive or semantic models.”69 In contrast to what happens in the congruences, projections are based on a selection of points and on the absence of metrics. In the end, what is the difference between imprints and congruences/projections? Eco would say that the question is simply misplaced because to recognize an already produced artifact is one thing, to produce a not yet existing artifact another. However, this close relationship between imprints and projections is demonstrated by the fact that “any naive interpreter of a projection ‘reads’ it as an imprint, that is, as the direct mapping from the actual aspects of a thing! On the contrary, the projection is always the result of a mapping convention by means of which given traces on a surface are stimuli compelling one to map backward and to postulate a content-type where one only sees an expression-token.”70
In this definition of projections the focus is again on the interpreter, only in this context imprints and projections are comparable. And they are comparable because in both cases the problem is the inscription on a surface that requires a backwards transformation: the difference between imprint and projection consists, then, in the figurative properties of the “real thing.” In fact, “If, in Table 39 [Figure 1], imprints (even if accidentally replicated rather than recognized) were not classified as straightforward transformations under the heading of inventions, this was for good reason. In the case of an imprint the content-model already exists. . . . When replicating an imprint one is mapping from something known.”71
It is apparent that the difference consists only in the presence/absence of a preexisting, and thus established, “content [that is, cultural] model.” In the heteromaterial district, between the left and the right side of Figure 1, a kind of algebraic-geometric purification of the natural figurativeness takes place, so that the symptom is converted into a combinational unit and the imprint into a projection.
When a first figurative, “immediate” recognition fails (where the heteromateriality fails to be “motivated”), it becomes necessary to switch to a second form of figurativeness, a more abstract one, understood as the reconstruction of a geometric operator. There’s something imprinted in the ground: if the interpreter does not recognize the imprinter, the latter is recognized at least as an abstract operator (“arbitrariness” of heteromateriality). Or, in the opposite case, a stereotypical imprinter is naively recognized, where a watchful gaze might reconstruct an operation that would put in question that projection’s status as an imprint.
VII. MODES OF ARTICULATION
This aspect seems to be confirmed by the third parameter of the typology, which concerns “the mode and the complexity of articulation, ranging from systems in which there are precise combinational units that are duly coded or overcoded to those in which there are texts whose possible compositional units have not yet been further analyzed.”72 The range suggested by the definition actually seems to define a partition between two regions, that of the “pre-established (coded and overcoded) grammatical units (according to different modes of pertinence)” and the “proposed undercoded texts.”73 The first region includes all the modes up to combinational units, the second includes inventions; in the middle, two “difficult” categories, pseudo-combinational units (shared between two modes of articulation) and programmed stimuli (which would be already part of the “texts”). Undercoding and overcoding are discussed in the theory of codes, their place in this significantly occupying the penultimate chapter. To a certain extent, they are explicitly derived from the concept of encoding:
Overcoding proceeds in a twofold direction. It may be that, given a code assigning meaning to certain minimal expressions, overcoding will assign additional meanings to more macroscopic strings. . . . But it may also be that, given certain coded units, overcoding will analyze these units into more analytical entities.74
Undercoding may be defined as the operation by means of which in the absence of reliable pre-established rules, certain macroscopic portions of certain texts are provisionally assumed to be pertinent units of a code in formation, even though the combinational rules governing the more basic compositional items of the expressions, along with the corresponding content-units, remain unknown.75
If encoding rightfully belongs to the theory of codes because it is the map of a state of semiosis, overcoding and undercoding are operations carried out starting from encoding, and as such are related to a theory of sign production: hence their liminal position in A Theory of Semiotics, when the theory of codes (given in section 2) is declared useful only in view of the theory of sign production (given in section 3). Extracoding, which subsumes undercoding and overcoding, does not describe a state but, more properly, a process: “overcoding proceeds from existing codes to more analytic subcodes while undercoding proceeds from non-existent codes to potential codes.” This double nature of extracoding is intrinsically ambiguous, as “the movements of extra-coding are the subject matter of both a theory of codes and a theory of sign production.”76 It is an ambiguity that is rooted in two possible meanings of the terms “encoding,” both used by Eco. In the classic semiotic meaning, encoding specifies a certain state that has been established between expression-units and content-units. But from the perspective of information theory, encoding is rather the operation that establishes the code. Such an operation should be properly described, based on the semiotic perspective that Eco develops, as an extracoding, that is, an interpretation that produces a new encoding with respect to a previous state. Here the concept of interpretation as production clearly emerges: there is only extracoding, and enunciation is in itself an extracoding activity. This ambiguity between the state and the process seems to be at stake in the definition of the “modes of articulation,” as they are organized along the axis opposing “preestablished” to “proposed,”77 that is, between what is already stored in a cultural memory, immediately available for withdrawal (encoding, and overcoding in the first sense, in fact as a decoding) and that which requires an “inventive step” (undercoding, including many ambiguous cases where it is not easy to discriminate between overcoding and undercoding).78
To sum up, it can be said that a tension emerges between a sort of encoded foreground and an extracoded background—a tension describing the variation of the interpretive effort—and confirms what has been previously observed about the close relations between symptoms and combinational units, and imprints and projections, respectively. Not by chance, the examples prototypically discussed in relation to undercoding are precisely inventions.79
VIII. RATIONES ET MACHINAE
The same tension emerges in the discussion of the remaining parameter: the ratio, which indicates the relation by means of which a token is accorded to its type, that is, the type/token ratio.80 Two cases can be considered as relevant. (1) In the case of ratio facilis “an expression-token is accorded to an expression-type, duly recorded by an expression-system and, as such, foreseen by a given code.” (2) In the case of ratio difficilis “an expression-token is directly accorded to its content,” either “because the expression-type is identical with the content-type” or “because the corresponding expression-type does not exist as yet.”81 In other words, there are two subcases of ratio difficilis: the first involves the identity of types between expression and content; the second the absence of expression-type. (a) In the first situation (identity, at least partial, of the types), the physical construction of the expression depends on the organization of the sememe. As an example, in “toposensitivity,” for instance, a pointing finger, the “spatial co-ordinates of the conveyed content to some extent determine the spatial properties of the expression.”82 The ratio here indicates a relation of “motivation” (elsewhere, “projection” and “mapping”) between semantic markers (the content) and syntactic ones (which in Eco indicate the expression). (b) In the second situation (absence of expression-type), “the expression is a textual cluster that should convey imprecise portions of content or a content-nebula”: in this case, “given a content-type that is in some way cognizable, its pertinent features must be ‘projected’ into a given expression continuum by means of certain transformational rules.”83
As already discussed, the table shown in Figure 1 “records the way in which the expressions are physically produced and not the way in which they are semiotically correlated to their content.” The type/token ratio is to be considered exclusively from the side of the expression. Yet the way in which the content is related to the expression “is implied by two decisions” that concern the ratio—of course—and that may be taken “either before or after the production of the expression.”84 In the case of symptoms (where the content is the cause), the decision concerning the arbitrariness of the correlation between expression and content has to be taken before, while in the case of combinational units it has to be taken after. Why? Because the point of view in the case of symptoms is that of the subject of recognition, which deals with expression starting from a set of cultural conventions, while in the case of replicas of combinational units the point of view changes, shifting to the subject of production. To speak with Eco, the latter associates an expression s-code to a content s-code. Again, two different levels are clearly at stake here.
In order to cope with the difficult task of deciding whether the ratio is facilis or difficilis, Eco proposes a sort of Turing test:
— In the case of ratio facilis objects “could be produced by a suitably instructed machine which only ‘knows’ expressions, while another machine could assign to each expression a given content, provided it was instructed to correlate functives.”85
— In contrast, in the case of ratio difficilis “a machine instructed to produce these objects should be considered to have also received semantic instructions. One might say that since it is instructed to produce expressions, it is being fed with schematic semantic representations.”86
Exactly as in the Turing test, the Eco test requires building a specific abstract machinery. A modelization for the machinae facilis et difficilis is proposed in Figure 2.
The machinae are made up of two submachines, the expression- and content-generators respectively, receiving instructions from a type descriptor. A third submachine is the + operator that correlates the outputs from the generators.
In the case of the machina facilis the two generators are independent as they receive descriptions/instructions from distinct types, their resulting outputs (tokens) being correlated by the + operator into the sign-function. In the machina difficilis there are independent tokens but not independent types. There is only one type including instructions for both expression and content. Finally, the sign, whether produced by ratio facilis or difficilis, results in an “object,” a “fact in a world of facts.”
Figure 2. Models of machinae facilis and difficilis (copyright 2017 by Andrea Valle)
A question must be raised: if the Eco test is aimed at discriminating between rationes, who is in charge of submitting it? In the discussion about rationes, Eco again mixes different perspectives. The point of view of the producer is at the origin of another definition: in systems working according to ratio facilis the content is realized (attualizzato) “by manipulating the plane of expression following its own rules,” while in systems working according to ratio difficilis, “formal rules proper to the content determine the realization [attualizzazione] of expression.”87 But the sign producer does not need to submit the Eco test, as s/he does not have to understand which the ratio is: he or she simply produces according to its ratio. Rather, the Eco test is submitted by a specific metaproducer, the interpreter, that is, the subject of recognition: he or she is the subject that makes the “two decisions” mentioned before. So the test is a procedure that, starting from objects in the world, aims at recognizing them as semiotic products produced by two specific behaviors, the rationes. This production model (the ratio) must be inferred by the metaproducer in reverse, that is, from the object to the sign-function up to the ratio. The Eco test is a tool to be used in the recognition labor that tries to answer which kind of machina the producer is. In Semiotica e filosofia del linguaggio Eco suggests a method of reconstruction, noting that in the ratio difficilis “provided that the projection rule is constant, the results obtained by manipulating the expression are diagnostic or prognostic with respect to the content.”88
So the process of inferring the type of the machine may work like this:
(i) collecting the objects as expression-units;
(ii) abducting a production rule for expression (the modus operandi for the expression-generator);
(iii) manipulating the abducted expression-generator;
(iv) checking if there are or have been changes in the content of the resulting sign-function (prognosis/diagnosis);
(v) in the positive case: it is a machina difficilis, where the content-generator shares the same type with the expression-generator; otherwise it is a machina facilis, where content- and expression-generators have each their own type.
Such a procedure is coherent with Eco’s statement that the table in Figure 1 does not list objects, but, rather, abductible activities: “they are short-hand formulas that should be re-translated so as ‘to produce imprints,’ ‘to impose a vectorial movement’ or ‘to replicate combinable units’ and so on.”89
Because the ratio in the Eco test is not evident per se but must be inferred, the issue of information available to the interpreter can be relevant. The second version of ratio difficilis is precisely a case in which the interpreter can analytically determine neither the expression-type—appearing as a “textual cluster”—nor the content-type, presenting itself as “a content-nebula in some way cognizable.” In this sense, ratio facilis versus difficilis may be interpreted as the opposition between analytically known versus analytically unknown sign systems. Only in analytically known systems can the types for expression and content be exhaustively described by a formal notation. Such an interpretation echoes Goodman’s notorious proposal: ratio facilis would define allographic systems, to be opposed to autographic ones, operating with ratio difficilis. But while Goodman mostly focuses on a theoretical description of the systems (or, with Eco, a theory of codes), here Eco is interested in interpreting their outputs (the theory of sign production).
A trace in the mud (an object in the world) is semiotically nothing until it is recognized as the figurative investment of a machina difficilis that operates by means of a projection on a heteromaterial continuum (figuratively motivated): the footprint (expression) of a rabbit (content). The footprint’s specific orientation is also predictive of the path followed by the rabbit (toposensitivity). The interpreter faces a sheet filled with signs. He or she thus submits the potential sign-function to the Eco test. The interpreter may assume that the signs are the output of a certain expression-generator, thus a certain content-generator is selected. The interpreter manipulates—though, indeed, not necessarily in a physical way—the expression-units he or she thought to have found. By slightly modifying a cross-like sign (and abducting that it is a “t”) the interpreter verifies that there is no variation on the content plane. So the reconstructed producer is a machina facilis, an aseptic grammatological correlator. With Saussure: it is well known that “the same person can write t, for instance, in different ways. . . . The only requirement is that the sign for t is not to be confused in his script with the sign used for l, d, etc.”90 Or, in the opposite case, the manipulation of the same expression “t” allows one to record a change in the content: this time, the producer is a machina difficilis, a pulsion-driven correlator, like those investigated by graphological studies, which links muscle motor behavior in writing to a psychoanalytic causal framework. To summarize, the modes of sign production are based on a double abduction. The first assumes an isotopy of production, the second outlines a phenomenology of the possible resulting correlations.
IX. SOME FINAL REMARKS ABOUT THE MODES OF SIGN PRODUCTION
In the end what question does the typology of modes of sign production try to answer? At least three aspects seem relevant, and they all concern history.
First, interpretation—that is, semiosis—is at any level a production because in principle there is not a level higher than production. Following the “indeterminacy principle” at the basis of A Theory of Semiotics,91 each interpretation (including the epistemological, metalinguistic ones) is a production because it necessarily modifies the landscape that it crosses.
Second, the notion of production can be included in semiotics only through an isotopy of production that unfolds its own specific figurative dimension: matter, producer, labor, imprint, trace, and so on. This isotopy of production is a path on the encyclopedic graph that recognizes/reconstructs history as the sign production of a semiotic subject. The typology is therefore the way in which the notion of history can be thought internally to semiotics in terms of the reconstruction, necessarily a posteriori, of the strategies to be adopted by such a subject. It is a matrix of possible histories of semiosis as a process.
Finally, semiotics as a form of sign production can rely on the notion of production to think about itself as a historical discipline: it is a condition that Eco first stated strongly in La struttura assente. In the famous section D, Eco urges us to replace the (comfortably semiotic) question “who is speaking?” with a very different one: “who is dying?”92 Apart from a quite unusual existential pathos, the issue raised here concerns how to think about history from within semiotics. The answer of A Theory of Semiotics states that semiotics as an interpretive practice is not to be thought as the exploration of the sea, “where a ship’s wake disappears as soon as it has passed,” but, rather, as the exploration of a forest: this means that, technically speaking, the interpretation leaves “footprints” and “cart-trails” that “modify the explored landscape.”93
ANDREA VALLE
UNIVERSITY OF TURIN
DECEMBER 2015
NOTES
1. Umberto Eco, Trattato di semiotica generale (Milan: Bompiani, 1975). Literally “Treatise of General Semiotics,” this work was Eco’s Italian “retranslation” of his newly finished, English-language work, A Theory of Semiotics (Bloomington: Indiana University Press, 1976). All quotations come from A Theory of Semiotics. More generally, when I quote from an Italian text (if an English version is not available), I am responsible for the translation.
2. In particular, the modes of sign production are so eccentric that their relevance has been missed not only by other scholars but also by Eco himself, apart from two quick discussions, one in Semiotica e filosofia del linguaggio (Turin: Einaudi, 1984), 43–51, which is absent from Semiotics and the Philosophy of Language (Bloomington: Indiana University Press, 1984), and the other in Lector in fabula: La cooperazione interpretativa nei testi narrativi (Milan: Bompiani, 1979), 217n4; The Role of the Reader: Explorations in the Semiotics of Texts, trans. Anna Cancogni (Bloomington: Indiana University Press, 1979), 42n21.
3. Eco, A Theory of Semiotics, 317.
4. Eco, Semiotics and the Philosophy of Language, 45 (the final sentence does not appear in the English edition). See Eco, Semiotica e filosofia del linguaggio, 54.
5. Eco, A Theory of Semiotics, 4. Note the emphasis on the status of “physicality” in the expression.
6. Ibid., 3.
7. For this example, see Eco, A Theory of Semiotics, 124.
8. Ibid., 126.
9. Ibid., 152.
10. Ibid., 9.
11. Ibid., 128.
12. Ibid., 29. In A Theory of Semiotics, “principio di indeterminazione” is translated literally as “indeterminacy principle,” but here Eco is explicitly referring to the uncertainty principle of quantum mechanics, known in Italian as “Principio di indeterminazione di Heisenberg.” The notion has intrigued Eco since Opera aperta (Milan: Bompiani, 1962); see the discussion in the “eponymous” first chapter, “La poetica dell’opera aperta,” 53.
13. Eco, A Theory of Semiotics, 9, 16.
14. Ibid., 22.
15. Ibid., 112.
16. Ibid., 114.
17. Ibid., 152.
18. Ibid., 91.
19. Ibid., 153.
20. Eco, Semiotics and the Philosophy of Language, 43. The phrase translates the Italian neologism “interpretanza” from Semiotica e filosofia del linguaggio.
21. See Eco, A Theory of Semiotics, 142.
22. Eco, A Theory of Semiotics, 148n24, 132.
23. Ibid., 151.
24. Ferruccio Rossi-Landi, Il linguaggio come lavoro e come mercato: Una teoria della produzione e dell’alienazione linguistica (Milan: Bompiani, 1968), 201ff.
25. Ibid., 235.
26. Ibid., 236.
27. See Eco, A Theory of Semiotics, the section on “Commodities,” 24–26.
28. Ibid., 219.
29. Ibid., 220.
30. Ibid., 241.
31. Ibid., 317.
32. Ibid.
33. See ibid., 125–29.
34. See ibid., 23, and Eco, I limiti dell’interpretazione (Milan: Bompiani, 1990), 219.
35. Eco, Semiotics and the Philosophy of Language, 45.
36. Eco, A Theory of Semiotics, 220.
37. Ibid., 218.
38. Ibid., 217.
39. Ibid., 151.
40. Ibid., 241.
41. Ibid., 242.
42. Ibid., 241.
43. Ibid., 8.
44. Eco, Semiotics and the Philosophy of Language, 182.
45. This passage does not appear in Semiotics and the Philosophy of Language. See Eco, Semiotica e filosofia del linguaggio, 50 (my translation).
46. Eco, A Theory of Semiotics, 221.
47. Ibid.
48. Ibid., 6.
49. Ibid., 245.
50. Ibid.
51. Ibid.
52. Ibid., 252.
53. Ibid., 221.
54. Such a situation may be described formally in relation to the proposed rewriting system assuming that—in an empirical semiotic object—there can be potentially many initiators, determining a parallel application of the production rule.
55. Eco, A Theory of Semiotics, 217.
56. See ibid., 58–59.
57. Ibid., 158.
58. See ibid., 163–64.
59. Ibid., 168
60. Ibid., 165.
61. Ibid., 219.
62. The relevance of Pareyson’s theory of formativity cannot be discussed here in depth, but it is evident in just considering the following passage from it, which bears a striking resemblance to Eco’s formulations about sign production: “This movement [of interpretation] is a process of production, because it consists in figuring the images [figurare le immagini] in which to enclose the sense of things.” Luigi Pareyson, Estetica: Teoria della formatività (Milan: Bompiani 1954, 1988), 190 (my translation). Eco’s notorious first major work, Opera aperta, begins by declaring in the first footnote the author’s debt to Pareyson’s theory. Eco, Opera aperta, 33n1.
63. For Gainsborough’s paintings, see Eco, A Theory of Semiotics, 250. For the example of the hare and rabbit, see ibid., 222.
64. Ibid., 225.
65. Ibid., 228.
66. Ibid., 223, 228.
67. Eco, Semiotica e filosofia del linguaggio, 45 (my translation).
68. Eco, A Theory of Semiotics, 256.
69. Ibid., 257.
70. Ibid.
71. Ibid., 249.
72. Ibid., 217.
73. Ibid., 218, Table 39 (reproduced here in Figure 1).
74. Ibid., 134.
75. Ibid., 135‒36.
76. Ibid., 136.
77. Ibid., 218, Table 39 (reproduced here in Figure 1).
78. Ibid., 136.
79. See ibid., where reference is made to the section of chapter 3 titled “Invention.”
80. Ibid., 182.
81. Ibid., 183.
82. Ibid., 185.
83. Ibid., 184, 189.
84. Ibid., 219.
85. Ibid.
86. Ibid.
87. Eco, I limiti dell’interpretazione, 67 (my translation).
88. Eco, Semiotica e filosofia del linguaggio, 45 (my translation).
89. Eco, A Theory of Semiotics, 220.
90. Ferdinand de Saussure, Course in General Linguistics, trans. Wade Baskin (New York: The Philosophical Library, 1959), 119–20.
91. See Eco, A Theory of Semiotics, 29
92. Umberto Eco, La struttura assente (Milan: Bompiani, 1968), 357 (my translation).
93. Eco, A Theory of Semiotics, 29.