Theory of Semiotics 0253359554, 9780253359551

Citation preview

A THEORY OF SEMIOTICS UMBERTO ECO

INDIANA UNIVERSITY PRESS Bloomington

London

Published by anansement with Bompiani, Milan Copyright© 1976 by Indiana University Prus All rights reserved No pan or this book may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopyin1 and recording, or by any information storaae and retrieval system, without permission in writin& from the publisher. The Association or American University Presses' Resolu• lion on Permissions constitutes the only exception to this prohibition. Published in Canada by Fitzhenry & Whiteside Limited, Don Mills, Ontario Manufactured in the United States of America Library of Conar- Catalo1ln1 in Publication Data Eco, Umberto. A theory of semiotics. (Advances in semiotics) Includes index. I. Semiollcs. I. TIiie. II. Series. P99.E3 74-22833 301.2' I ISBN 0•253-359SS-4 I 2 3 4 5 80 79 78 77 76

CONTENTS

Foreword

vii

Note on graphic convention, 0. ln1roduc1ion-Tow1rd I Loaic of Cullure 0.1. 0.2. 0.3. 0.4. 0.5. 0.6. 0.7. 0.8.

0.9.

Design for a semiolic lheory 'Semiolics': field or discipline? Communicalion and/or significalion Polilical boundaries: lhc field Na1ural boundaries: two definitions of semiotics Nalural boundaries: inference and signification Na1ural boundaries: lhe lower lhreshold Nalural boundaries: the upper lhreshold Epistemoloaical boundaries

I. Si1nific1tion and Communication 1.1. 1.2. 1.3. 1.4.

An elementary communic1tional model Systems and codes The s-code u suucture Information, communication, sipification

3 3 7 8 9 14 16 19 21 28 32 32 36 38 40

CONTENTS

vi 2. Theory or Codes 2.1. 2.2. 2.3. 2.4. 2.S. 2.6. 2.7. 2.8. 2.9. 2.10. 2.11. 2.12. 2.13. 2.14. f-1S.

The sign-function Expression and content Denotation and connotation Message and text Content and referent Meaning as cultural unit The interpretant The semantic system The semantic markers and· the sememe The KF model A revised semantic model The model .. Q" The format of the semantic space Overcoding and undercoding The interplay of c.odes and the message as an open form

3. Theory of Sian Production 3.·1. A g�neral survey 3.2. Semiotic and factual statements 3.3. Mentioning 3.4. The problem of a typology of signs 3.5. Critique or iconism 3.6. A typology of modes of production 3.7. The aesthetic text as invention 3.8. The rhetorical labor 3.9. Ideological code switching 4. The Subject of Semiotics

48 48 SO S4 57 SB 66 68 73 84 96 I OS 121 12S 129 139 ISI ISi IS8 161

172

191 217 261 276 289 314

Rtftrtnce1

319

Index of outhor,

347

Index of 1ubjtct1

3Sl

FOREWORD

A preliminary and tentative version of this text (dealing with a semiotics of visual and architectural signs) was written and published in 1967 as Appunti per una semiologia de/le comunicazion/ visive. A more theoreli· cally oriented version - offering an overall view of semiotics and containing a long epistemological discussion on structuralism -was published in 1968 as la sr"'rrura assente. I worked for two years on the French, Gennan, Spanish and Swedish translations (only the Yugoslavian, Polish and Brazilian ones appeared with sufficient speed lo reproduce the original Italian edition without any addition) re-arranging and enlarging the book - and correcting many parts of It lo take Into account reviews of the Orsi Italian edition. The result was a book half way between· la sr"'tn,ra assellfe and something else. This 'something else' appeared in Italian as a collection of essays. u forme delco111e11uto, 1971. As for the English version, after two unsatisfactory attempts 11 lranslalion and many unsuccessful revisions, I decided (in 1973) to give up and lo re-write the book directly in English - with the help of David Osmond-Smith, who has put more work into adapting my semiotic pidgin than he would have done if translating a new book, though he should not be held responsible for the results of this symbiotic adventure. To re-write in vii

FOREWORD another language means to re-thi11k: and the result or this truly semiotic experience {which would have strongly interested Benjamin ue Whorl) is that this book no lonaer has anythina to do with La stru1111ra asse111e - so that I have now retranslated it into Italian as a brand-new wo,k (Trattato di

semiorica gc11erale).

Apart from the different {but by no means irrelevant) organization or the material, four new elements characterize the present text as a partial critique of my own precedin3 researches: (i) an attempt to introduce into the semiotic framework a theory of referents; (ii) an auempt to relate pragmatics to semantics; (iii) a critique of the notion of 'sign' and of the classical typologies or signs; (iv) a different approach to the notion or icon­ ism - whose critique, developed in my preceding works, I still maintain, but without substituting for the naive assumption that icons are non-coded analogical devices, the equally naive one that icons are arbitrary and ruuy analyzable devices. The replacement or a typology or signs by a typology or modes or sign production has helped me, I hope, to dissolve the umbrella· notion of iconism into a more complex network of semiotic operations. In doing so, the book has acquired a sort or 'chiasmalic' structure. In its first part, devoted to a theory of codes, I have tried to propose a restricted and unified set of categories able to explain verbal and non·verbal devices and to extend the notion or sign-function to various types of significant units, so-called signs, strings or signs, texts and macro-texts - the whole attempt being governed by the principle of Ockham's razor, no11 su111 multiplica11da entia praeter 11ecessitatem - which wpuld seem to be a rather scientific procedure. In the second part, devoted lo a tl,eory of sign productio11, I relt obliged to proceed in an inverse direction: the categories under consideration {such as symbol, icon and index) were unable to explain a lot or different phenomena that I believed to rall within the domain or semiotics. I was therefore forced to adopt an anti-Ockhamislic principle: emia su,11 m1dtipli­ t:t111da propter necessitatem. I believe that, under given circumstances, this procedure is also a scientinc one. I would not have arrived at the results outlined in this book without the help or many rriends, without the discussions that have appeared in the first six issues or the review VS-Quuderni di s111d/ semiutiL"i, and without confrontations with my students at Aorcnce, Bologna, New York University, Northwestern University, La Plata and many other places around the WOfld. Since the list of rererences allows me to pay my debts, I shall limit myself to warmly thanking my rriends Ugo Volll and Paolo Fabbri, who have helped me

Fo,�word

ix

throughout the various stages of the research - mainly by merciless criticism - and whose ideas I have freely used in various circumstances. Milan, 1967-1974.

NOTE ON GRAPHIC CONVENTIONS Sin&le slashes indicate something intended u an expression or a sign­ vehicle, while guillemets indicate something intended u content. Therefore /xxxx/ means, expresses or refers to «xxxu. When there is no question of phonolo&Y, verbal expressions will be written in their alphabetic form. How­ ever, since lhil book is concerned not only with verbal signs but also with objects, imases or behavior intended u signs, these phenomena must be ex­ pressed through verbal expressions: in order to distinguish, for instance, the object automobile from the word automobile, the former is written between double slashes and in italic. Therefore lau1omobilcl is the object conespond­ in& to the verbal expression /automobile/, and both refer to the content unit «automobile». Sin&le quotation marks serve to emphasize a certain word; double marks are used for quotations. llalic denotes terms used in a technical 1ense.

INTRODUC TION: TOWARD A LOGIC OF CULTURE

O. I. Design for a semiotic theory 0.1 .I. Aims of the research The aim of this book is to explore the theoretical pouibility and the social function of a unified approach to every phenomenon of signification utd/or communication. Such an approach should take the form of a general semiotic 1heory, able to explain every case of sign-function in terms of underlying systems of elements mutually correlated by one or more codes. A desillJI for a general semioticsJ a result lhe notion of 'meaning' as 'cultural unit' becomes applicable not only to the cotr,ortmotic terms but also to J)l11cotqurt11UJtic ones. Instead of putting into structural relationship names of intellectual

76

A THEORY OF SEM IOTICS

qualities, names of colors, or tenns of kinship, Apresjian (1962) Indicated fields which place pronouns in opposition (pronouns which designate animate things vs. pronouns which designate in11nlmott things : also for instance the place occupied by /you/ in English compared with the place occupied by /tu/ /lei/ /IIOi/ In Italian) or fields of verbs which designate different operations within the same sphere of operations (for example to advise, to assure, to convince, to inform etc., all of which belong to the sphere of transmission of information). This enables us to race the problem or the po11ible content or syncategorematic terms and or the so-called 'functional monemes' (see for instance i-h, 1969). See on this subject 2.1 1.S., in which some suggestions for a componentill analysis of syncategorematic terms are proposed. Naturally structural semantics hopes to establish the Semantic Space (as the Form of Content In Hjelmslev's sense) in its totality. But this aim, which can constitute a general hypothetical framework for research, comes up against two obHacles; one empirical and the other inherent in the semiotic process. The first obstacle is that until now such studies as have been undertaken only arrived at a structuring of very restricted subsystems, such u for example that or colors, or botanical classifications, or metereological terms, etc. The second obstacle is due to the racl that the lire of semantic fields is briefer than that of phonological systems where the structural models attempt to describe forms which remain unchanged for long periods or time within the history of a language. Since semantic fields give shape to the units or a given culture and establish portions of the world vision belonging to that culture, movements of acculturation and critical revisions of knowledge are enough to upset a semantic field. If Sau11ure's metaphor or the che11board is accurate, the movement or one piece will suffice to change all the relationships of the system. Therefore it is enough that as the culture develops, the term /Kunst/ be given areas of application which are much wider than usual, for the whole system of thirteenth-century relationships studied by Trier to be changed, thus depriving the term /l/11/ of its value. 2.8.3. The segmentation of semantic field• In what sense does a semantic field show the world vision belonging 10 a culture? Let us go back to one or the classic examples of the theory of semantic fields and examine the way In which a European civiliution analyzes the color spectrum by assigning names (and therefore establishing cultural units) to various wave-lengths expre11ed In millimicrons.

n

11,eory of Cod,i

a, b. c, d. e. r. g.

red orange yellow green blue indigo violet

800-650 l11ll 640,590 mµ 580-550 l11ll 540-490 l11ll 480-460 rnµ 450-440 rnµ 430.390 rnµ

A preliminary and naive Interpretation might propose that the spectrum, divided into wave-lengths, constitutes the referent, the object or experience to which the names or the colors refer. However, we, know that the color was named on the basis o f a visual experience (which the simple speaker would define as 'pe rceptual reality') which is only tran�ated into wave-lengths by scientific experience. But let us assume that the wave,Jengths are something absolutely 'real'. There is no difficulty in stating that the undifferentiated continuum of the wave-length constitutes "reality'. Yet science comes to know that reality after having divided ii into pertinent units. Portions or the continuum have been cut out (and as we shall see, they are arbitrary) so that the wave-length d (which goes from 540 10 490 millimicrons) constitutes a cultural unit to which a name is assigned. We also know that science has divided the continuum in such a way as lo justify in terms of wave-length a unit which simple experience had already cut out or its own accord and given the name /r,een/. The choice based upon naive ex perience ...,notorbitrory , in the sense that the exigencies or biological survival probably forced 1h31 unit to be tenned pertinent rather than another Gust as the fact that the Esltimos divide the continuum o f experience into four cullural units in place of the one which we call /snow/ is due to the facl that their vital relationship with ,now Imposes distinctions on them that we can disregard without suffering any notable damage ). But it wa.r arbitrary in the sense that 111other cuhure divided the same continuum in a different way, which means that the continuum is a content-stuff which can be cut into different fonnal systems. We are not lacking in examples: for the portion of continuum � (blue) Russian culture has two di fferent cultural units (corresponding to /,ol11buj/ and /si11i;n, while the Greco-Roman civiiiz.otion probably had only one cultural unit for the vvlous names U1lo11c111/ /co,n,/111/) to indicate the purlion d , • and the Hindus combine under a simple term (and d1u1 a single cullur.tl unit) rht portion 11 • b. We can therefore say thal a aiven culture has divided the

78

A TIIEORY OF SEMIOTICS

continuum of experience (and it does not matter whether the conlinuum Is seen in terms of perceptual experience or defined by means of oscillographs and spectographs), making certain units pertinent and understanding others merely as variants, 'allophones'. Thus to single out a shade such as «light blue» and another such as «dark blue» means for an English speaker isolating a free variant, in much the same way as when two idiosyncratic pronuncia­ tions are sing)ed out from one phoneme which from the 'emic' point of view is considered a pertinent unit of the phonological system. All this leaves unsolved a question which will appear clearer when the units of two different semantic fields are compared in two different languages, Latin and English (Table 10), Table 10 Mouse Mus

Rat

which can be rendered as: "to the Latin word /mus/ correspond two different things which we shall call x 1 and x 2 " (Table I I). Table 11

IE] LG

On the other hand, since the existence of x 1 and x2 is only made evident by the comparison of the two semiotic systems, can we say that x 1 and x2 exist independently of the names which a language has assigned to them and which establishes them as cultural units and therefore meanings of a certain sign-vehicle? If we tum to colors the answer is simple. There is no reason why there must be a physical entity which begins at the wave-length 640 millimicrons and ends at the wave-length 590 millimicrons. In fact in Hindu culture the segmentation or the continuum occurs not at 640 but at S90 millimicrons. But why should there not be a cultural unit (and a unit of experience) which

77,,ory of Cod,i

79

goes from 610 to 600 millimicrons? Actually a painter with an extreme sensitivity to colors who possesses a more carefully graded system would answer that such a unit exists and is present in his own special code, where a specific name would correspond to that portion of the wave-length continuum. The problem concerning /mus/ is a different one. The zoologist would tell us that the x, and x, which correspond in English to /mouse/ and /rat/ exist as specific objects and that they can be analyzed in tenns of properties and functions. But what Foucault (1 966) has written on the 'epistemes' of different epochs and the variations in their segmentation of the universe, or what Uvi-Strauss ( 1962) has written on the taxonomy of primitive people, should suffice to make us aware that even on these points it is wbe to proceed with caution. Since, after all, a study of codes should not be concerned with x 1 and x2 , which are referents, it should be enouda to confinn that there exislS in English a semantic field governing rodenlS, which is more analytic than its equivalent in Latin, and that therefore for the speaker of English there exist two cultural uniu where for the speaker of Latin there exists only one. All this brings the problem of semantic fields back to the so-ailed Sapir-Whorf hypothesis and to the question of whether the fonn of communicative systems determines the world vision of a certain civilization. At this stage it does not seem appropriate to broach the question: it is enough to assume that (at least on the level of the segmentation of the experiential continuum into the form of the content) there exists a fairly close in1eraction between the world vision of a civilization and the way in which it makes iu own semantic units pertinent. Given the elements in play - Y (material conditions of life), X (units of perceived experience), U (corresponding cultural units) and SV (the sign-vehicles which denote them) - it is not necessary to know at this point whether Y determines X, which generates U, assigning to it the name SV; whether Y strives to elaborate SV in order to segment the experience to which U corresponds; whether semiotic activity on a deeper level leads man to base his thoughts on SV, which not only produces U and X but directly condilions human being, to experience Y and so on. These are still extra·semiotic problems. It would be more interesting, from a semiotic point or view, to understand within which civilizations a semantic field functions and at what point it begins to dissolve in order to make room for another; and how, in the same civilization, two or more semantic Oelds can coexist although in opposition, when different patterns of culture are superimposed.

80

A THEORY OF SEMIOTICS

A typical example is provided for us by the series of definitions which Aulus Gellius gave to colors in his Nocus Allica� (ii, 26) in the second century A.D.: he, for example, associated the term Jrufu.1/, (which we would translate /red/) with fire, blood, gold and saffron. He stated that the tenn /xanlhos/ (color of gold) was a variation of the color red, just like /ki"os/ (which in the chain of interpretants reconstructed by philology must be understood as equivalent of our /yellow-orange/). He furlhermore considered, as altemalive names for the color red, /flavus/ (which we are also used to associating with gold, with grain and with the water of the river Tiber), and //ulvus/ (which is usually the color of a lion's mane). But Aulus Gellius calls the eagle, the topaz, sand, gold /fulva/ while he defines Jflavus/ as a "'mixture of red, peen and white"' and associates it with the color or the sea and of olive branches. finally he states that Virgil, in order to de£ine the color ..greenish" of a horse, uses the term /caerulus/, which is commonly associated with the color of the sea. The extreme confusion which strikes the reader in this one page or Latin is probably due not only to the fact that Aulus Gellius' field of colors was different from ours, but also that in-the second century A.O., in Latin culture, alternative chromatic fields coexisted owing to the influence of other cultures. Hence the perplexity of Aulus Gellius, who does not manage to amnge the material, which he takes from the works or writen of different epochs, into strict fields. As we have seen, the •actuar experience which the author could feel from looking at the sky, the sea or a horse is here mediated by recoune to given cultural units, and his world vision is determined (in a nther incoherent way) by the cultural units (with corresponding names) which he finds at his disposition. We could therefore state that: (a) in a given culture thtre can exist contradictory semantic fields: this is an aberrant cultural occurrence which semiotics must take into consideration rather than try to eliminate it; (b) the sam• cultural unit can itself become part of compltmtntary semantic fields within a given culture. Carnap (1947, 29) gives the example of a double classification according to which animals are divided on the one hand into aquatic, aerial and terrestrial, and on the other hand into fish, birds and others. A cultural unit such as «whale» can then occupy different positions in the two semantic fields without the two classifications being incompatible. One must thus admit that the user of a language possesses within his 'competence' the possibility of coupling a given system of sign-vehicles with various systems of meanings: {c) within a given culture a semantic field can

disintegrate with extreme rapidity and rtstructure itself into a new fi•ld.

Points (a) and (b) will be dealt with in 2.8.4,since they are matters fora theory of codes. But they also have direct consequences as far as sign production is concerned; mainly in the rhetorical and ideological treatment or diJcourse. So they will also be more deeply considered in 3.9. Likewise, point

Thtory ofCodts

81

(c) is the concern of a theory of code-changing, which is a branch or the theory of sign production. For an example of this see 3.8.5. 2.8.4. Contradictory semantic fields As for a suitable example or contradicting semantic fields, I shall consider the problem of antonymous terms as pairs of oppositions constitut• ing a semantic axis. Lyons ( 1968) classifies three types or antonymy: (i) complementary antonyms such as •masculine vs. feminine'; (ii) properly called antonyms, such as •small vs. large'; (iii) antonyms by convcrseness, such as 'buy vs. sell'. Katz ( 1972), on the other hand, subdivides antonyms into: (a) contradic· tories, such as 'mortal vs. immortal', which have no possible medial ion between them; (b) contraries, such as 'superior vs. inferior', and 'rich vs. poor' which have ,.some possible mediation between or beyond them ; (c) converses, such as 'husband vs. wife' or "buy vs. sell', which, like the converses in Lyons, imply syntactic transformations and entail an inferential relation of the type 'if . . . then._ 1

Even a superficial glance at some pain of antonyms 1nta), use the followingds and cs when the sememe in question is contextually associated with the sememe cu";(c/rc) are clrcums1anlial selections giving instructions of the type: "when you find (circa) use the followins ds and cs when the sign-vehicle corresponding to the sememe in question is cin:umstantially accompanied by

106

A THEORY OF SEMIOTICS

the event or the objecl la/, to be understood as the sign-vehicle belonging 10 another semiolic system". A composilional rree like this one shows that: (I) Syntactic marker, along with the subcaregoriz.allon rules lhal they imply, pertain to the expression, not to the contenl; thus a sentence like /a gloop is a bloop/ is synlaclically correct, even if you don't know what It means, provided that /gloop/ and /bloop/ are equally marked as Noun+Masculine+ Singular (in rhe same sense, a Oagpole with a green square Hag In which three yellow circles are inserted, is syntactically correct, outside verbal signs, even though no regisrered nation, stale or association can be identified with this type of symbol); (ii) A sememe may have (a. i) denotative markers which remain unchanged in every possible contextual and circumsrantial selection, such as d 1 and d2 (wilh some added connotation c1 and c2 , depending on d 1 ) or may have (b, i) differen1 ds and therefore different cs according lo diverse contextual and circumsranlial selecrions. Thus a /bachelor/ is convenUonally «young, only if a con1ex1ual selection specifies (con/chivalry), the denotalion «young» releasing coMolations such as «chastity»; in the above diagram a case like this is represented by the contextual selection (conrb ). The case of [circ J is 1 instead one in which, granted the same circumstances, there can be different contexrual selections; the case of (conta) is one in which, granted the same context, there can be different circumstantial selections; (iii) Contextual selections record other sememes (or groups of sememes) uSUDi/y associared with the sememe in question; circumsrantial selections record other sip,-vehicles (or groups of sign-vehicles) belonging to different semiotic systems, or objects and events taken as ostcnsive signs, usually occurring along with the Sip,-vehicle oorresponding to the sememe In question; both act as amalga1t111tion switchers. In this way, contextual and circumstantial selections do not require a specific type of instruction, for they are none other than cultural units or expressions, constituting the main node or other sememic representations or the elements of the compositional analysis of another sememe. Thus the same elements serve both as markers or as 'selection reslriclions': the same kind of entity performing a double role according to its stralegical position within the tree, an economical result has been attained (tntia non fUnt mu/tiplitanda pratttr n«essitatem). (iv) Selecrion restrictions are eliminated along with dislinguishers. The former are in fact anlicipated by both contextual and circumstanllai selection, the lauer diaolve into a complex network of semantic markers. Thus /bachelor/ oould be approximately analyzed as «man+youngtfulfillment+college+ • . ••

Th,ory of Cod,.

107

This is a very rough suggestion about what a compositional analysis (able to eliminate distinguishe") should be, and the suggestion can be improved only after a more painstaking analysis of the underlying semantic fields. The postulation of such semantic fields is an indispensable requirement, even in the case of other semantic approaches such as the one proposed by Bierwisc:h (I 970), who analyzed lexemes in this way: /father/ s X parent of Y + Male X + (Animate Y + Adult X + Animate Y) /kill/ = X1 Cause (Xd Change to (-Alive Xd)) + (Animate Xd) (v) When analyzing verbs a particular series of denotative marke" should represent the arguments of an n-places predicate, according to an inventory of roles or 'cases'. Those cases are semantic actants (in the sense of the "a1111ly1e actantiell," proposed by Greimas) rather than morphological cases (see also the suggestions of Fillmore, 1968 and 1971). In short, an action is accomplished by an Asent (A), by means of an Instrument (I), to reach an aim or a Purpose (P) and affecting an Objecl (0) - where Object still is an umbrella-category covering semanlic roles implied by different morphological cases. such as 'dative' or 'accusative': therefore the Object should be more finely analyzed as Addressee, Experience,, Object physically modified by the action, and so on. Let us assume that, when the verb is a locutionary one, there is a Topic (T) (de te ff) fabull1 (A) 1,ar,atur (locutionary)). Such an approach should take into account even the semantic presuppositions directly entailed by the sememe, without introducing new semanlic categories such as 'focus' and prcsupposition' (PS). Obviously, in order to elaborate this kind of representation one should of all distinguish between the various senses of the word /presupposition/ that in philosophical and logical literature sends back to radicaJly different phenom­ ena. R,ferential presuppositions concern a theory or mcntions(cf. J.3.)and ue lhe ones.studied by Frege ( 1 892): "if anything is asserted there is always an obvious presupposition that the simple or compound proper names used have a reference". Contumal presuppositions are the ones studied by a text theory and concern both textual inferences and rules of overcodinJ (cf. 2.14.J.). Hiz (1969) calls these mutual textual occurrences 'referentials" and says that, given the text /Two roads lead to John's house. One way goes through the woods. The other is shorter. Both are paved and he knows them very well/, /he/ refers lo the occurrence /John/, /them/ refers to /two roads/, /way/ refers to /roads/, and so on. 1

r,,.,

108

A 1liEORY OF SEMIOTICS

Cireu,m1a111id presuppositions concern what both the sender and the addressee know or are supposed to know about coded or uncoded entitles and events. Contextual and circumstantial presuppositions can be also called p,a,matic. Smumtic presuppositions strictly depend on the format of the sememe: if one says Uiat /John is a bachelor/ everyone understands Utat John is a male adull human being. But. insofar as they are directly entailed by the semantic orpnization (i.e., are analytically "included" as a necessary part of the complete meaning of a given expression, ef. Katz, 1972, 4.S.), Jet us define them rather than 'presupposition', as Jtmiolic tntailment (see the difference between semiotic and factual judgments in 3.2.) ("). Therefore only semantic presuppositions (i.e., semiotic entailments) directly concern the theory or codes and must be recorded by the semantic representations as parts or the meaning. Once this granted, let us translate In terms or the Revised Model the representation of two verbs studied by Fillmore (197 1 ), /accuse/ and /criticize/. FUimore asserts that, as far as 'meaning' and •presupposition' are involved, one verb asserts what the other presupposes and vice versa. In terms of the Revised Model this difference should be completely manifested by a series of denotations. Let us assume that /accuse/, syntactically marked as JvI

ntMu u

l

I

g

!�

Th,ory of Sip Production

219

The table records the way in which expressions are physically produced and not the way in which they are semiotically co"elated to their content� the latter is implied by two decisions that must be made either before or after the production of the expression. For instance, in the case of recognition of symptoms, there ii undoubtedly a pre-established motivation due to a preceding experience which has demonstrated that there is a constant physical relationship between 1 given agent and a given result; it has therefore been decided, by con,.ntion, that these resultant objects must be correlated with the notion of that agent under any circumstances, even when one cannot be sure that an existing agent has ,eally produced the result. In the case of wo1ds (which may be classed among 'systematically combinable units') the correlation is posiled after the production of the physical unit and is in any case independent of ils form (this assumption being valid even if by unverifiable historical chance the origin of words had some sort of imitative motivation). For this reason such non-homogeneous objects as a symptom and a wo1d are posited in the same row; every object listed there can be produced according to its pre-existing expression-type (ralio fad/is) and this happens irrespective of the 1easons for which these objects were selected as the expression of a given content. All of them could be produced by a suitably instructed machine which only 'knows' expressions, while another machine could assign to each expression a given content, provided II was instructed 10 correlate functives (in other words, two expressions can be diffuently motivated but can function in equally conventional fashion). On the other hand, all objects ruled by a ratio dl/Ttcilis are so motivated by the semantic format of their content (see 3.4.9.) that ii is irrelevant whether they have been correlated with ii on the basis of previous experience (as in the case of footprints, where die semantic analysis of the content has already been performed) or whether the content is the result of the experience of 'inventing' the expression (as in the case of painting,). Therefore the motivated way in which they have been chosen (see the further analysis of Imprints and projections below) does not affect their mode of production according to a ntio diff,ci/i1; they are corrdated lo certain aspects of their sememes - thereby becoming expressions whose features are also content-features, and thus projected ..,,,anlic nuzrk,n (•• ), In this sense a machine instructed to p1oduce these objects ,hould be considered to have also received semantic instructions. One might say that since It Is instructed to produce expressions, ll is bein1 fed with schematic semantJc representations (2 • ).

220

A THEORY OF SEMIOTICS

The items recorded in the row corresponding to the parameter 'type/ token-tatio' may look like 'signs', since to some degree they recall pre, existing sign typologies. Bui they ,re not; they are short-hand formulas Iha! should be re-translalcd so as 'to produce imprints', 'lo impose I vectorial movement' or •10 replicate combinable units' and so on. 'Imprints" or 'examples' musl, at most, be understood as physical objects which, because or certain or their characteristics (nol only the way in which they are made, but also the way in which they are singled out) become open 10 a significant correlation, i.e. ready to be invested with dignity or functive. In 01her words they are po1enliaJ expression features or bundles of features. According to the system into which lhey are inserted, they may or may not be able I� convey by themselves a portion or content. So thal although they can also act as signs, they will not necessarily do so. I t must be clear that the whole or Table 39 speaks or physical procedures and entities that are ordtred to the sign,func1ion but lhat could subsist even if there were no code to correlate them to a content. On the other hand, they are produced in order 10 signify and the way in which they are produced renders them able to signify in a given way. A ready-made expression like /cherry brandy/ Is the result or two procedures depending on a double type/token-ratio; it Is constructed rrom two combinational units ordered by a vectorial succession� likewise a pointing finger is both a vector and a combinational unit, while a road arrow is both a stylization and a vector. Therefore items like 'vectors' or 'projections' arc not types or signs and cannot be equated with typological categories such as 6indices' or 'icons'. For instance both 'projections' and •imprints' could appear to be icons but the former would imply an arbitrarily selected expression-continuum and the liner a molivatedly established one, while both or them (equally governed by a ratio di/Trci/is) would be motivated by a content-type (though imprints are 'recognized', while projections are invent· ed'). Imprints and vectors look like indices, but are in rac1 dependent on two different type/token-ratios. Moreover, certain categories (e.g. 'fictive sam, pies') come under two headings: they are the result or a double labm, since something must be replicated in order to be shown (ostension). All these problems will be dealt with fu rther In the following parag,aphs. I have only anticipated some examples In order to slress the ract that one must not look at Table 39 in order to find types or signs. This table only lisu types or productive activity 1h11 can give rise, by reciprocal and complex interrelation,, to different sign-runctlons, whether they are codtd units or codini texts.

Thtory of Sip Production

221

3.6.2. Rec0g11ilion Recognition occurs when i given object or event. produced by nature or human action (intentionally or unintentionally), and existing in a world or rac:1s as a fact among facts, comes to be viewed by an addressee as the expression or a given content, ellher through a pre-existing and coded correlation or through the positing of a possible correlation by its addressee. In order to be considered as the runctive or a s;gn.runction the object or event must be considered as if ii had been produced by osJension, replica or invention and correlated by a given kind or type/token-11110. Thus the ,ct of recognition may re-constitute the object or event u an imprinl, a symptom or a clue. To interpret these objecls or events means to conelate them 10 a possible physical causality runctioning as their content, ii having being conventionally established that the physical cause acts as an uncorucious producer or signs. A,; we wUI see, the inrerred cause, proposed by means or abduction, is pure content. The object can be a fake or can be erroneou!ly interpreted as an imprint, a symptom or a clue, when in fact it is the chance product or other physical agents: in such a case the 'rec0g11ized' object expresses a content although the teferent does not exist. In the recognition of imprim,, the expression Is ready-made. The conlily falsifiable. Any naive interprete, of a projection 'reads' it u an imprint, that is, u the direct mapping from the actual upects 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. So it is always possible to projectfrom nothinc or from contents to which no referent corresponds (u in a clmical painting representing mythological heroes). The existence of social conventions in projections (so that is Is possible to map from a perceptual or a semantic model) make easy the reverse procedure, that is, to map front the projection to an unexlsling and supposedly projected entity. What reinforces our criticism of naive iconlsm is that since it is possible to draw ralse iconic signs, lconism is a matter of a highly sophisticated semiotic convention (> • ). When considered as mentions, projections are frequenlly false; they uy to assert that something exists, which actually looks likt the expression item, when this Is not the case al all; they can thus display images of Julius Caesar as well as of Mr. Pickwick, irrespective of the differing ontological st11usof the two. It is in cases of projections thal the so-called 'scales of iconism' can be accepted u heuristically useful. Thirdly, there are graph• or topok>t{cal tra11sfonnatioru (u ), in which spatial points in the expression correspond to points of non-toposensitive relation; such Is tho cue In Peirce's existential graphs (see 3.S.3.): a spatial

A THEORY OF SEMIOTICS

258

expression displays infonn11ion about I correla!lon which is not spatial at aU but which inslcad concerns, for instance, economic relationships, u when one uses the graph or Table 4S Toblo 45

@

to express the following relationship: "every dependent worker belongs lo the class or exploited and alienated prolet1rians". In any case this whole range of inventive representations, from congruences or cuts to graphs, displays not signs but texts; when these texts lint appear there is u yet no distinction between ptrtintnr and lrrtlev11111 features. It is only in the cour>e or decoding them that pertinent features emerge, and they begin to produce signs (and thus their own mannerisms). Because of the difficulty or isolating the content-type to which they refer (by a procedure or tatlo dif/icilts), lhese texts are not easily replicable. To NCCessfully copy a painting is no mean feat, and to farce 1 Rembrandt may wdl be considered a par2-artislic achievement, for ii Is very difficult to detect the pertinent properties on which the significant power or the expression relies, and only remarkably skilled counterfeiters have a capacity for isolating and reproducing them . When only one person in the world is able lo falsify a mode oflnvemion (i.e. not to copy a pven painting. but to paint according lo the same type or invenlive procedure) the code proposed by 1h11 painting has not yet been accepted by a culture; when ii becomes possible lo paint ii la manlae dt, then the invention (as a code-making proposal) has succeeded semiotic.Uy; a new convention exists. But it is clear Iha! the present discussion Is continuously shifting from the problem or code-making and or the acquisition or new conven lions, lo the problem or the aesthetic: use or a language. Arty discussion or invention inevitably opens up the problem of the ambiguous, ,elf-focusing and idiolectal use or a code, and compels us to return, once more, to the discussion on aesthetic texts. 3.6.10. Productive features, signs, texts 11le typology or modes or production or the signal, outlined in' this

2S9

111,ory of S;,n Product/on

section, has definitely clarified the fact that what one usually calls 'signs' are the result of many intertwined modes of production. For instance, a perfume of incense, if smelled in a church, is only a case of m:o,nilion, that is, a symptom by which one recognizes that a liturgical ceremony is taking place. When produced, it is at the same time the nplica of a Jty/ization and a prof'Qmm•d 1timu/11s. When used during a play in order to suggest a mystical situation, it is both a proirammtd 1timulu1 and a ficti'lt (the incense for the whole ceremony). A smile can b, a ,ymptom or the r,plico of a sty/izarion, and sometimes eYen a •tctorialization. A musical melody, when quoted in order to recaU the entire symphony from which it ha, been extrapolated, is a .ramp/,; but it can be the r,p/ico of a text composed by combi1u,tional units and, sometimes, even a complex of proirommtd Ulmuli mixed with ps,udo-combi1111tional units. And it is usually all these things together. A geographical map is the result of a previous transformation (half way between a projtction and a iraph) which has definitely become a uyliution, and as such is the result of a r,p/ico . Clothes in general are replicable ttylizations with intertwined

.,,.p1,

pseudo-combi1,a1ioMI u11i11 and programmtd stimuli.

The problem becomes more difficult when one mus! defute a painling. In any case, a painting is certainly not •a sign': it is a complex text rcsuJting from the network of many modes of production. One might suppose that a porlralt of a given man represents a perfect case of 'proper name' necer.sarily sendin& back to a physical referent (while the verbal proper names have been demonslrated in 2.9.2. to have a content). One could belier say thal such a portrait is neither a sign nor a complex super-sign but rather a mt11tion Uthis is a man and he possesses these properties . . . /). On the other hand It might be said that the same portnit is equivalent to a dtscription. Goodman (1968: J.S.) remarks that there is a difference between the picturt of a mo11 (the portrait of Napoleon) and a man-picrur, (the portrait or Mr. Pickwick). In fact such a portrait ,mbodies different types of activities, praclically covering the entire range of types of semiotic labor outlined in Tables 31 and 39. It is a mmrion because, through prorrammtd stimuli, i t displays the surrogate of a percept and by means of some graphic devices attributes to it lhe markers of I possible corresponding sememe; it is an u,�,,tiOlt insofar as the perceptual model does not yet exist; it is a facmol j114lme111 (/there exists a man so and SD/) and a description U• man so and so/). Being still uncoded it is at lhe same time relying on a lot of already coded fo11ures. and the invenlion is made acceptable by the Intervention of coded lnrprin11, sryiizatio,u, .rampltt,

260

A THEORY OF SEMIOTICS

pst11do-eombinatio1111I u11its, Vtctorializations, and so on. Therefore such a portrait is a complex text whose content ranges from a coded detectable unit («Mr. So and So1) to an infinite discourse or a content-nebula. But II rar 11 the portrait is accepted and recognized by a culturo, it creates a 'type' (in the sense of a 'literary type", inlended as the representation of some 'universal' properties: the Hero, the Gentleman, the Beautirul Lady, 14 btllt dame ,an, mcrci, and so on). At this point it becomes the model for funher stylizations. So what In a given historical period may be viewed II an Inventive projection, In another period becomes a stylization. The same happens with the so-called 'architectural signs'. Even 1r many researches in semiotics of architecture have tried to isolate the existence of 'architectural signs' (n), It Is absolutely clear that even the most elementary architectural configuration is always a text. Let us consider for Instance a staircase. It is undoubtedly a semiotic device which signifies certain functions: but to compose such a device productive labor is requested to display the following features: (i) articulation or p1tudo-combiN1tio1111/ units; (U) •tetorill/ization, (the staircase indicates a direction making rocourse to toposensitive parameters); (iil) programmtd uimulations (the staircase in a certain way obliges one to move one's feet for climbing up); (iv)sty/izarions (the staircase corresponds to a precise typology); and so on. It ls not without sense to try to Isolate precise expression-units In architecturo, but It is indispensable to take into account the lot or producth< features that these units bring into play. AU this reminds us that, the more a text become complex, the more complex is the relationship between expression and content. There may be simple expression units that convey content•nebulae (see for instance many cases of programmed stimulation); expression-clusters that convey a procise content unit (a triumphal arch can be a very elaborate architectural text and nevertheless convey a strictly conventional abstraction such as 1victory1); precise p-arnmatical expressions, composed of replicable combinational units, such II the phrase /I love you/, that in certain circumstances convey dramatically a content-nebula; and so on. This must not allow one to neglect to isolate precise sign-functions when they aro detectable, but serves to romind one that In the semiosic process we are usually racing undercoded or overcodcd texts. When more analytical units are not detectable, it is not a case of denying the existence or a semiotic correlation; the presence of the cultural convention is not only witnessed by the emergence of so-called elementary sips. It ls first of all revealed by the detectable existence or modes of

'Tlr#ory ofSip Production

261

semiotic production (recognition, ostension, replica and lmention) that the present section has outlined and whose presence demonstrates that - even when there are not precise unit-to-unit correla1ions - there is. however, a posited sign-function. 3.7. The aesthetic text u invention 3.7. 1 . The semiotic purport or the aesthetic text The aesthetic use or a language deserves auenlion on a number Ii different levels: (i) an aesthetic text involves a vtry peculiar labo denotes «red» and cred» connotes «scarlell). This explains all the rhetorical distinctions linked to the phenomenon of synecdoche. As for metonymy, a satisfactory solution can be reached by inserting within the semantic representation n-places predicates according to a typok>&Y of ro/a or 'cases• {cf. 2. 1 1 .1.). In this way one can record relation, such as causa pro tffttto and vice versa. a po:s:sessore quod pol:Sidttur, ab to quod contintt quod continetur etc. Let u, examine the Aeneid, 10,140: vu/11tru diripre ti calomos 111111an �ntno

A THEORY OF SEMIOTICS

282

where /,ulntro diri�/ means «to lnOlct a blow (in order to cause wounds)• and stands for /dlri� ttlil/, /diri�t ictus/, /diriitre p/aps/ or /vu/11erare/. Let us suppose that it stands for /diri(tr, ttla/ (with /diriftrt lcn11/ the result would be the same). A tentative representation of /ttlum/, excluding many other possible selections and referring to a stDndard Latin, appears as follows (where R is the Result of the action exercised): Conto:+homo) d R:.,.,1ntU,lctv1,p,.,. < (cont A:+ltonto ) •t�lum» d""'"',.""''"' d.,_,. . . ./ ,

conlo:-llomo)

(conlA:-ltomo) dA•h , , · · • • • • • • • · · • Then f,u/nm, diri�/ is a metonymy of the type (I) - m11ker for sememe - and represents a case of substitution of the instNmental cause by the effect. If the same exp,ession stands for /,u/11,rare/ the rhetoric mcchlnism would not change, except that it would be a lilllc more complex: "'lntrort

d.-,to /ecrum mottu

dtrrlrw

�rn,t�re IN,.

,dA:homo,do:homo ,d p;.,uln111,d 1:1C'lum· , .c dJ,.ctto

In fact «vulnus• Instead of «vulneraru would be a substitution of the efT',cient caUM• by the effect, but there is also a partial substitution of the connotation of «direction» for the directional act of wounding: a very risky synecdoche indeed (that can work only when supported by the 'stronger' metonymy). Let us now suppose that, In order to indicate a friend of mine who is a bachelor, I say /that unlucky seal!/. Provided that my audience has read Katz, Fodor and Postal (and in the circle that I move In everybody has), the substitution is easily understood. The problem that arises here is whether this rhetorical figure is a metaphor or a metonymy. Since /bachelor/ as human male being and /bachelor/ as seal are both readings of the same semerne, one should speak of replacing a sememe with one of its markers and therefore of a type (i) metonymy. Nevertheless It is clear that the substitution was based on the 'identity' of the marker •unmated» (which is more general than tnever,married• and which in any case springs from «never-ma,ried• because

T/t,ory of Sip Production

283

of the redundancy n,Jes). Al this point only two solutions are possible: either the two •bachelors» are simply different but synonymous sememes, or one must speak of me.. phor even when dealing with a semic identity betw..n two different re•dings of the same sememe. In any cue, insofar as metaphor ls no longer considered as a similarity belwecn thinp, and metonymy is no longer considered as a contiguity between things, the categories of both semic identity and scmic interdependence are a matter or infra· or intn-Jtmtmle connections. In this perspective even semic identity is permillcd by the underlying texture of the semantic system, so that a son of s,,ucruml coml,uiry suppons and governs all these rhetorical interactions. Metaphon and metonymies are made po.,ible by the existence of a semantic global universe whose format is thal of Model Q. (• • ). 3.8.4. Rhetorical code-changing Al this point it becomes necessary to establish by what rule 'good' metaphors or metonymies are distinguished from 'bad' or �rivial' ones. A tentative solution might be as follows: a good metaphor occun when the 'identical' markers are comparatively periphel21 and particularly chuacteristic of the two sememes in question. To call a group of warrion /fflffl!/ is dearly an embryonic metonymy because all soldiers are men, but emu• is a marker shared by many other sememcs and consequently this kind of substitution does nol particularly characterize a warrior. But if one says /I have two thousand swords at my disposal/ in order lo convey that one has two thousand warriors at one's command, then we have a more successful metonymy since warrion are the only type or men lo have swords. A belier example of successful metonymy is offered by the inlcr· dependence established by the Romans between •gladiatou and ,ready lo die> or •dcath·seeken• (A., Caesar, moriruri " mlu111n1!). ln chis ialler cue not only does the metonymy seem more 1nventive' but ii increases one's awareness of the semantic entity •gladiator•. Suppose now that one substitules «warrior, by cgladialor», and «gladiator» by «morin,n,11. Not only are waniors seen in a less cus1oma,y light but they are also chancterized by a peripheral marker that is shared by other sememes thal might up to this point have been considered far mnovcd from the one under examination. For instance it now becomes possible to associate a warrior metaphorically with a «scapeaoat, (as a vnoriturun by dennilion), so that an army of warrioB may be defined as /the scapegoats of the King's ambitions/. Insofar as •scapeeoau has a marker of •innocence,

284

A THEORY OF SEMIOTICS

the way to a more complex network or substitution, Is open; the warriors can become /two thousand iMocent swords/. And 10 on. At the extreme point or thi1 substitutional shifting the way In which warriors are usually viewed has enormously chanced; the coMotations or «fierceness», «courage». «pride•, and «victory• do not disappear, bul merge with 11n1onymous coMotations such as «rear•. «sorrow,, «shame» and «defeat,. The rhetorical tracing or underlying connections In the semantic fields has revealed fertile contradictions. Since it has to take place between branches or the sememes, and since any node within these branches it the patri1rch or a new sememe (see Model Q), rhetorical substitution, by establishing further connections, runs the whole gamut or the Global Semantic Field, revealing its 'topological' structure. In this activity contextual and circumstantial selections are frequently switched and overlapped, and short circuits or all sorts create sudden and unpredictable connections. When this process is rapid and unexpected and joins up very distant points, it appears as a )ump' and the addressee, though confusedly sensing Its legitimacy, does not detect the series or steps within the underlying semantic chain that join the apparently disconnected points together. AJ a result he believes that the rhetorical invention was the product of an intuitive perceplion, a sort or "illumination', or a sudden revelation, whereas in fact the stndtr has simply caught a glimpse or the paths that the semantic organization entitled him to cross. What was for him a rapid but distinct look at the possibilities of the system becomes for the addrtssee something vague and indistinct. The latter attributes to the former a superior intuitive capacity, whereas the former knows that he had a more immediate and articulated view or the underlying structure or the semantic system. Both have, however, discovered a new way or connecting semantic units, so that the rhetorical process (which can, in some cases, equal the aesthetic one) thus proves itself to be a form or knowledge, or at least a w'!)I of upstrtitlf ocquirtd knowledge ( •• ), Suppose, reformulating Table IS in 2.9.6. in order to get an ad hoc example, that there is an axis containing two semantic units (u 1 and u2 ), that are usually considered mutually Incompatible, because their first respective denotative markers are units derived from an oppositional axis (a2 vs. a, ), but that, through o1 , have a connotation 'Yo in common (Table 48). Lei us now suppose that, through a series of rhetorical substitutions, a sememe can be named (and therefore rendered rhetorically equivalent: ;) either (i) by one or its markers (a case or metonymical substitution, represented by mltl, followed where necessary by the marker via which the

Tlltory of Si,n Production

285

connection is made), or {ii) by another sememe with which it shues a giffll marker (1 case of metaphorical substitution represented by mtf, foUowed by the marker upon which the Nbslitution relies), as shown in Table 49.

Tobie 48

Table 49

Provided that the rules not of formal logic but of rhetoric are in play, then u , (because of its equivalence to u2 ) acquires both mukert cr1 and cr, , which were previously seen to be antonymically incompatible (Table SO).