Encyclopedic Dictionary of Semiotics 0899251374

Citation preview

Encyclopedic Dictionary of Semiotics General Editor Thomas A. Sebeok, U.S.A. Editorial Board Paul Bouissac, Canada Umberto Eco, Italy Jerzy Pelc, Poland Roland Posner, W Germany Alain Rey, France Ann Shukman, United Kingdom

Tomel A-M

Mouton de Gruyter Berlin · New York · Amsterdam

Approaches to Semiotics 73 Editorial Committee

Thomas A. Sebeok Roland Posner Alain Rey

Mouton de Gruyter Berlin · New York · Amsterdam

Mouton de Gruyter (formerly Mouton, The Hague) is a Division of Walter de Gruyter & Co., Berlin.

Library ofCongms Cataloging in P11blication Data

Encyclopedic dictionary of semiotics. (Approaches to semiotics ; 73) Contents: t. I. A-M - t. 2. N-Z - t.3.Bibliography. I.Semiotics -Dictionaries. I. Sebeok, Thomas Albert, 1920 . II. Series. P99.E6 5 1986 001.51'03 86-18714 lSBN 0-89925-137-4 (alk. paper)

C!P-K11rztitda11f11ahmt dtr De11tschtnBibliothtlt

Encyclopedic dictionary of semiotics / gen. ed. Thomas A.Sebeok. Ed. board Paul Bouissac ... Berlin ; New York ; Amsterdam : Mouton de Gruyter (Approaches to semiotics ; 73) lSBN 3 -1 1-010559-4 N E: Sebeok, Thomas A. [Hrsg.]; GT T. 1. A - M. - 1986.

Printed on acid free paper. © Copyright 1986 by Walter de Gruyter & Co., Berlin. All rights reserved, including those of translation into foreign languages. No part of this book may b e reproduced in any form -by photoprint, microfilm, or any other means - nor transmitted nor translated into a machine language without written permission from Mouton de Gruyter , a Division of Walter de Gruy­ ter & Co., Berlin. Typesetting: Satz-Rechen-Zentrum, Berlin.- Printing: Ratzlow-Druck, Berlin.- Binding: Liideritz & Bauer GmbH, Berlin. - Printed in Germany.

Preface

The embryonic idea for some kind of reference guide to semiotics came to me in 1973, while I was a Fellow of the Netherlands Institute for Advanced Study in the Humanities and Social Sciences.As Klaus Oehler later insightfully espied, I was animated to develop a project of this sort as I browsed, then ruminated, over Peirce's ethics of terminology and the relevance thereof for "all areas of lexicographic practice," particularly "the preparation of philosophical and scientific dictionaries" (Oehler 1981 t:;,. :348) 1• Far from envisaging this labor as a monumental "cooperative summing up" (Gerber 1967 /:;,. :199), the last words carved into granite, I thought of it as a modest further step "in the work of clearing and opening up ... semiotic," in the memorable locution of the self-declared pioneer, or backwoodsman, of the field, Charles S. Peirce (cf. 5.488). Initially, I had an amorphous conception of a reasonably comprehensive, one-volume venture-that, on the one hand, would attempt to answer relatively specific, narrow questions, but that would also, on the other hand, encourage its users to distend their horizons. In short, the handbook I imagined would aim to satisfy both demands of the mind: for direct problem-solving, as well as for the harvest of its peripheral vision. From the beginning, I firmly rejected the notion of constructing a work representing any particular "school of semiotic thought," thereby promoting the normative criteria of a single position "into a principle unifying all other positions," for, as pointed out in Oehler's preview (/oc. cit. 352-353), such an imposition to elevate any individual point of view as preeminent "would result only in a fictitious unity which for the purposes of the lexicon would be not merely useless but damaging." This renunciation had as its corollary the requirement to seek for the widest possible support throughout the interna­ tional community of semiotic practitioners and representatives of workshops located in various parts of the world.When I started, I knew only of a few lexicons devoted to semi­ otics: Maldonado's (1961 t:;,. ), for instance, contains fewer than one hundred terms in the compass of 21 pages. The dictionary of Bense and Walther (1973) had just ap­ peared, consigned, however, all but exclusively, to Peirce's semiotic. Later, other lexi­ cons were to appear, the most ambitious among them being Greimas and Courtes (1979 t:;,., 1982 t:;,. ), a polemical piece of work which has been judged "far from being a survey of the terminology in use in semiotics ..." (Serge 1984 t:;,. :269-270). An unprecedented opportunity for me to air my first private scenario came in June, 1974, in the course of the First Congress of the International Association for Semiotic Studies, held in Milan. I had proposed to Umberto Eco, the organizer of that Congress, that he set aside a period for a presentation of "my plan" and for a thorough discussion of it. This request fit well into his quasi-archeological scheme for the Congress, which had as one of its overall tasks to consider the history of semiotics, and, as another, to explore "the possibility of providing the discipline with a unified methodology and uni­ fied objective" (Eco, in Chatman et al. 1979!:;,. :v). About one hundred colleagues showed up at my special, informal session-not all, of course, at the same time-at

References followed by this symbol ( t:,.) are listed in the Supplement to the Bibliography, pp. 449 ff.

VI

Preface

which I set forth my thoughts for an Encyclopedic Dictionary of Semiotics, to comprise three interlaced categories of interalphabetized articles: (A) Entries tracing the historical background and range of the present usage of terms-seme, for instance-with recommendations, where appropriate, for stan­ dardizing current convention; (B) Evaluative biographical sketches of leading figures in semiotic studies, such as Charles Morris; and assessments of aspects of the work of others, such as Aristotle, who have made pivotal contributions to semiotic studies, yet are not commonly thought of in this context; (C) Expositions of the impact of semiotics on various traditional arts and sciences, say, Architecture, Mathematics, Music, and the like; and of the penetration of se­ miotic methods of inquiry into the study of established academic fields of study, such as the Philosophy of Language, Logic, and the like. After several hours of debate, the basic blueprint for the project was unofficially en­ dorsed, and thus carried to the concluding Business Meeting of the Congress, where it was formally adopted by a unanimous vote of the assembled members of the IASS, and subsequently officially codified by its President, Cesare Segre. I was appointed Editor­ in-Chief, with three immediate assignments to carry forward: (1) To form an International Editorial Board; (2) To procure the required financing; and (3) To commit the eventual book to a suitable publisher. The Editorial Board-whose membership is listed on the title page-was created mainly to help with the internal organization of the EDS and to assist in securing the best contributors available. These responsibilities were eventually carried out in part through very extensive correspondence and via numerous long distance phone calls, and in part in the course of three meetings: the first in West Germany, held September 1722, 1979 (hosted, by courtesy of Roland Posner, by the Technische Universitat Berlin); the second in Italy, held June 12-25, 1981 (hosted, by courtesy of Pino Paioni, by the University of Urbino); and the third in Canada, held June 10-15, 1982 (hosted, by cour­ tesy of Paul Bouissac, by Victoria University, at the University of Toronto). During our last get-together, members of the Editorial Board were able to examine, or, at least, to scan many-although, unfortunately, by no means all-lemmata in manuscript form, and to propose structural and individual modifications where required and feasible. Too, gaps were spotted and subsequently filled when possible. Users may well ask: why did thirteen years elapse between the conception and the realization of this work? The reasons are several, but by far the most arduous chore proved to be obtaining the funds necessary for even the most frugal operating expenses. I had prepared, at the outset of 1975, what I thought was a remarkably austere budget, comprising no compensation at all for any of the editors (myself, of course, included), providing for the salary of merely a single clerical helper and for minimal supplies, as well as for modest honoraria for invited contributors (this last sum estimated from comparable figures paid in the 1970's by the Encyclopaedia Britannica). The palpable­ perhaps the only-public agency to apply to then was the United States National En-

Preface

VII

dowment for the Humanities. Notwithstanding that semiotics never fell within the En­ dowment's legal purview, its officers were-as they are now-kindly disposed toward this roughhewn parvenu to the American academic landscape, and tried to help out with­ in their own fiscal limitations. After considering my submission for almost a year, I was advised that my proposal was approved "in principle," but proved fundable only to the tune of not quite 50% of the monies needed, this on condition that I raise the remainder from private sources. It was now the middle of 1976, so, if the plans were ever to be real­ ized, I was obliged to commence anew to reconnoitre for glimmers of interest in semio­ tics in the most unlikely novel venues for this (relatively) large scale support. The ensuing rummage had the makings of a picaresque novella. The exploration was, in any case, onerously time consuming, lasting, as it turned out, for over two addi­ tional years. In the end, the funds to match-indeed, to somewhat exceed-the NEH Challenge Grant were pieced together from generous gifts on the part of The Rocke­ feller Foundation and The Wenner-Gren Foundation for Anthropological Research, supplemented by donations from the Indiana University Foundation, the Indiana Uni­ versity Division of Research and Graduate Development, the Indiana University Press, and through the good offices of the Technische Universitat Berlin. After protracted and convoluted negotiations with a variety of major publishing houses, located both in the United States and the United Kingdom, a contract was at last signed with the foremost Continental firm that has the longest and most distin­ guished commitment to semiotic studies-Mouton de Gruyter-and that also has amply demonstrated its experience with the production of substantial reference works, com­ parable, more or less, with this one. From the original Project Description, I now quote, for the record, the description of the Editorial Board's duties: (a) "Select the entries to be included, i.e., the technical terms, names of major contribu­ tors to semiotics, fields with which semiotics interacts and the aspects of that interac­ tion." (b) "Assign priorities to the topics, determining the relative number of words or amount of space to be devoted to each." (c) "Arrive at a consensus concerning the best possible contributor for each entry, and a list of possible alternates. Collaboration by two or more authors may be considered for more complex entries." (d) "Outline the technical specifications, that is, produce a style sheet, with the advice of the publisher's representative."

As work began, the members of the Editorial Board were obliged to make certain decisions of principle: for instance, that entries in and environing linguistics would be minimized, since there already existed quite a few good independent, and comparable, works encompassing that well-established field-namely, the branch of semiotics which is devoted to verbal signs. It was also resolved to exclude biographical sketches of living persons. Too, in specific cases it proved impossible to obtain satisfactory entries dealing with certain topics, especially on semiotic theory and practice outside the Western tradi­ tion, and on some other subjects we deemed important but for which we failed to secure suitable contributors. The usual impediments interfered with the smooth flow of corre-

VIII

Preface

spondence in and out of countries where censorship prevails, occasioning further delays and other familiar difficulties.2 Thomas A. Sebeok

Bloomington, May 1986

At his express request, dated January 22, 1986, I must, in conclusion, cite the following statement: "Jerzy Pelc accepts responsibility for the scientific content (but not for the editorial side involving cross-referencing, asterisks, typographic differentiations, etc.) of entries or frag­ ments of entries signed B.S. (Babara Stanosz, Warsaw University), F.St. Fra:nciszek Studnicki, Jagiellonian University, Cracow), J.J. Gacek Jadacki, Warsaw University), J.P. Gerzy Pelc, Warsaw University), L.K. (Leon Koj, Maria Curie-Sklodowska University, Lublin), M.Cz. (Mi­ roslava Czarnawska, Warsaw University, Bialystok), M.H. (Michal Hempolinski, W.S.P. Szczecin), P.Br. (Piotr Brykczynski, Warsaw University), R.P. (Rostislav Pazukhin, Lodz Uni­ versity), T.E.H. (Teresa Hol6wka, Warsaw University), U.N. (Urszula Niklas, Peirce Edition Project, Indiana University-Purdue University, Indianapolis), and W.M. (Witold Marciszews­ ki, Warsaw University, Bialystok)."

Contributors

A.B. A.Bo. A.F.R. A.G. A.Gu.

Andrea Bendandi Andrea Bonomi Anthony F. Russell Alain Goldschlager Arnold Gi.inther A.H. Andre Helbo A.H.-L. Aage Hansen-Love Alice Jardine A.]. A.Je. Ann Jefferson A.Jo. Andre Joly A.K. Adam Kendon A.L. Adrienne Lehrer A.L.-S. Annemarie Lange-Seidl A.M. Abraham Moles A.M.P. A. M. Piatigorsky A.P. Augusto Ponzio A.Ph.-L. A.-Ph. Lagopoulos A.R. Alain Rey A.R.L. Andre Roch Lecours A.S. A. Scheffcyzyk A.Sh. Ann Shukman A.V. Alain Viau A.Wi. Alan Williams A.W.R. Allen Walker Read A.Z. Alessandro Zinna B.B. B.D. B.O. B.S.

Baron Brainerd Bernard Dupriez Boris Oguibenine Barbara Stanosz

C.B. C.E. C.H. C.J.F. C.N. C.P.

C.T. C.Ti.

Claude Bremond Christopher English Christoph Hubig Charles J. Fillmore Christopher Norris Claude Panaccio Cesare Segre Christian Thiel Catherine Tihanyi (translator)

D.B. D.d.K. D.G. D.G.C. D.J. D.K.

Dwight Bolinger Derrick de Kerckhove Dietfried Gerhardus Daniela G. Camhy David Joravsky David Kahn

c.s.

D.L. D.M. D.M.M. D.M.-S. D.P. D.Pr. D.P.V. D.R.-L. D.S.

David Lidov Dean MacCannell D. M. Mackay Dietland Muller-Schwarze Daniel Patte Donald Preziosi Donald Phillip Verene Daniel Rancour-Laferriere David Savan

E.A. E.B. E.B.F. E.F.M. E.G. E.N. E.P. E.S. E.Sp. E.St. E.V. E.v.S. E.W.B. H.-L.

Edna Aphek Eugen Baer Edmund B. Ford Eugene F. Miller Erhardt Gi.ittgemanns Eduardo Neiva Eva Picardi Eric Schwimmer Emilio Speciale Edward Stankiewicz Eugene Vance Eike von Savigny

F.A. F.D.B. F.E. F.G. F.M. F.R.-L. F.S. F.St.

Friedmar Apel Frances D. Burton Francis Edeline Franz Gi.inthner Francesco Marsciani Ferruccio Rossi-Landi Frits Staal F. Studnicki

G.B. G.Be. G.G. G.M. G.Me. G.Mo. G.P. G.T. G.W.H.

Gunter Bentele Gregg Beasley (translator) Gunter Gebauer Giovanni Manetti Georg Meggie Georges Mounin Giorgio Prodi Gideon Toury Gordon W. Hewes

H.E. H.G.S. H.K.

Herbert Eagle Henry G. Schogt Hermann Kalkofen

Ernest W. B. Hess-Li.ittich

X

Contributors

H.P. H.S.S. H.v.d.B.

Herman Parret H. S. Scheffler Holger van den Boom

I.B. I.E.-Z. I.K. 1.0. I.R.

Ira Buchler Itamar Even-Zohar Ivan Karp Ivo Osolsobe Irmengard Rauch

J.B.C. J.B.S. J.C. J.D.J. J.D.M. J.E. J.H.M. J.L.H. J.-L.N. J.M. J.M.H. J.-M.K. J.N.D. J.P. J.Pe. J.P.H. J.Pi. ].Pin. J.P.K. J.R. J.R.-D. J.S. J.S.P. J.U.-S. J.V. J.V.W. J.W. J.Wo. J.Y.K.

John B. Carroll J. B. Scoggin Jonathan Culler J0rgen Dines Johansen James D. McCawley Johannes Engelkamp John H. McDowell J. Jadacki Judith Lynne Hanna Jean-Luc Nespoulous John McClelland J. Michael Holquist J. M. Klinkenberg John N. Deely Jerzy Pelz Jean Petitot Jack P. Hailman John Pier Jan Pinborg Jane P. Kaplan Joseph Ransdell Josette Rey-Debove John Sturrock Janos S. Petofi Jean Umiker-Sebeok Josef Vachek James V. Wertsch Jacques Waardenburg John Woods Joan Yess Kahn

K.C. K.D. K.E. K.H. K.L. K.M. K.O. K.W.H.

Kvetoslav Chvatik Kevin Dunbar Keir Elam Karen Hansen Kuno Lorenz Karl Maroldt Klaus Oehler Klaus W. Hempfer

J.J.

L.F. L.Fo. L.G.K. L.J.P. L.K. L.R.W.

Lawrence Fine Lia Formigari L. G. Kelly Luis J. Prieto Leon Koj Linda R. Waugh

M.A. M.A.B. M.B. M.C. M.C0• M. Cz. M.D. M.d.M. M.D.Me. M.G. M.Go. M.Got. M.Gr. M.H. M.He. M.K. M.L. M.Lu. M.M. M.Mo. M.P. M.R. M.R.K. M.U. M.W.

Maryann Ayim Massimo A. Bonfantini Manfred Bierwisch Marc Chenetier (translator) Maria Corti Miroslawa Czamawska Marcelo Dascal Marco de Marinis Marx De Mey Martin Gardner Myrna Gopnik Mark Gottdiener Murray Grossman Michal Hempolinski Michael Herzfeld Martin Krampen Monique Leon Manfred Lurker Marina Mizzau Massimo Modica Mark Pastin Monica Rector Mary Ritchie Key Maria Ulkau Mauro Wolf

N.B.C. N.M. N.N. N.S.B.

Nino B. Cocchiarella Norman Martin Neal Norrick (translator) Naomi S. Baron

O.C. O.G.R. O.H.

Omar Calabrese Olga Grigor'evan Revzina Otto Haxthausen (translator)

P.A. P.B. P.Br. P.B.S. P.D. P.F.-D. P.G. P.J.S. P.M.

Peter Abelard Paul Bouissac P. Brykczynski Philip B. Stafford Phillip De Lacy Pierre Fresnault-Deruelle Pierre Guiraud Philip J. Stone Patrizia Magli

Contributors P.Mi. P.P. P.S. P.St. P.V.

Phillipe Minguet Paul Perron Peter Schmitter Peter Steiner Patrizia Violi

R.A. R.Ar. R.C. R.C.S. R.d.B. R.D.M. R.E. R.H. R.J.S. R.K. R.L.L. R.M.B. R.M.M. R.P. R.Pos. R.S.W. R.T.

Raimo Anttila Robert Armstrong Ranjit Chatterjee Roger C. Schank R. de Beaugrande Roger D. Masters Rudolph Engler Robert Harnish Robert J. Smith Roberta Kevelson Richard L. Lanigan Ruth M. Brend Richard M. Martin Roscislaw Pazukhin Roland Posner Rulon S. Wells Rene Thom

S.A.T. S.E.L. S.H. S.J.H. S.M. S.P. S.R.

Stephen A. Tyler Svend Erik Larson Shuhei Hosokawa Steve J. Heims Solomon Marcus Susanna Petrilli (translator) Stephen Rudy

S.S. S.T.

Stephen Slade Stephanie Toteff (translator)

T.A.S. T.E.H. T.E.Ho. T.H. T.v.U.

Thomas A. Sebeok Theresa E. Holowka Thomas E. Hope Terence Hawkes Thure von Uexktill

U.E. U.M.Q. U.N.

Umberto Eco Uta M. Quasthoff Ursula Niklas

V.M.Z. V.P.G. V.R.

V. M. Zivov Volney P. Gay Victor Raskin Vilmos Voigt

W.A.

W.St.

Werner Abraham William C. Stokoe Winnie Frohn William F. Klein Wendy Holmes W. John Smith Wolfgang Lenzen W. Marciszewski Winfried Noth Wulf Rehder Werner Sauer Wendy Steiner

Y.T.

Yishai Tobin

v.v.

w.c.s.

W.F. W.F.K. W.H. W.J.S. W.L. W.M. W.N. W.R.

w.s.

XI

A Abduction The technical sense of the term abduction appears to be almost completely confined within the writings of Charles Sanders *Peirce. Peirce attributes the origin of the term to *Aristotle's Greek word anayroyT) 'apagoge,' claiming that an editorial blunder in the original text of the twenty-fifth chapter of Prior Analytics led to a mistranslation of Aristotle's term into abduction; the correct term, which most closely captures Aristotle's meaning, is retroduction, according to Peirce (1931-66: 1.65, 2.776, 5.144). In his own writing, however, Peirce appears to use the terms abduction and retroduc­ tion interchangeably. The word presumption is also used occasionally, and hypothesis more frequently, especially in Peirce's early writing (1931-66: 2.776, 2.623). Peirce employed the term abduction (or one of its synonyms listed above) in a variety of contexts, but always as contradistinguished from deduction and induction. In his early writing, for example, he saw these terms as representing different forms of argumenta­ tion, where hyp othesis inferred from the *rule and the result to the case, deduction from the rule and the case to the result, and induction from the case and the result to the rule. His classic example illustrating these differences is the following: Deduction. Rule. -All the beans from this bag are white. Case. -These beans are from this bag . . · ., Result. -These beans are white. Induction. Case. -These beans are from this bag. Result. - These beans are white. . · ., Rule. -All the beans from this bag are white. Hypothesis. Rule. -All the beans from this bag are white. Result. - These beans are white. . ·., Case. -These beans are from this bag. (1931-66: 2.623).

In his later writing, Peirce was less inclined to see these differences as adequately accommodated in this formal schema and he turned instead to the scenario of scientific inference. In this broader context, abduction was characterized as "a method of forming a general prediction" (1931-66: 2.270). It "consists in studying facts and devising a *theory to explain them" (1931-66: 5.145). Abductive inference carries no guarantee of certainty with it, but "merely suggests that something may be" (1931-66: 5.171); it is what Peirce calls "an originary Argument" (1931-66: 2.96). Thus Kepler's hyp othesis concerning the elliptical orbit of Mars is an example of an abductive inference; for Kepler realized that "the observed longitudes of Mars, which he had long tried in vain to get fitted with an orbit [Kepler had first tried to reconcile these observations with a

2

Abduction

circular and then with an ovoid orbit] were . . . such as they would be if Mars moved in an ellipse" (193 1 -66: 2.96). In contrast to abduction, deduction "is the only necessary reasoning. It is the rea­ soning of *mathematics" ( 193 1-66: 5.145). The theorems of Euclid thus constitute clear examples of deductive inference as do certain conditional predictions based on Kepler's elliptical orbit hypothesis ; that is, on the assumption that the elliptical orbit hypothesis is true, certain consequences regarding latitudes and parallaxes may be inferred deductively (193 1-66: 2.96). Induction is the last cog in the mechanism of scientific progress. It "is the experi­ mental testing of a theory. . . . It sets out with a theory and it measures the degree of concordance of that theory with fact" ( 193 1 -66: 5.1 45). To follow through with the same example, induction begins with the abductive inference regarding the elliptical orbit of Mars as well as the deductive inferences regarding predicted latitudes and paral­ laxes ; it (induction) consists in the performance of relevant experiments and in a *judg­ ment about the *truth of the abductive inference on the basis of an assessment of the experimental results. This judgment, however, is always "held subject to probable modi­ fication to suit future experiments" (193 1-66 : 2.96 ). It is only through abduction that new or original ideas are devised and made accessible to science. The scientist's use of abduction is justified in that scientific discovery would be unthinkable without it. Deduction and induction alone are insuffi­ cient to ensure scientific progress ; these traditional methods of scientific inference are incapable of formulating hypotheses, and hence if divorced from abduction they would have no hypotheses to explicate or test and would themselves be utterly useless ( 193 1-66: 2.270, 5. 1 45). Peirce claimed that historically abduction and induction have been characteristically confused even though they in fact "face opposite ways": abduc­ tion explains and induction classifies. Peirce acknowledged that in his earlier writings (in particular, his essay "A Theory of Probable Inference," published in 1883) he was himself guilty of confusing these two modes of inference (193 1-66: 2.632, 2.755, 2.636, 2.102). As Peirce characterizes abduction, it is based on instinct (193 1-66 : 5. 1 73, 6.475), in particular, on a "natural insight into the laws of nature" (Ayim 1974 :35) captured by the phrase ii fume naturale 'the natural light.' In spite of its instinctive base, abduction is clearly classified by Peirce as a method of reasoning. As such, it is subject to the condi­ tions of reason, and must be deliberate, voluntary, critical, and controlled (Peirce 193 1-66: 2. 1 82). Abductive inference is also linked to perceptual judgement by Peirce; perceptual judgments are extreme instances of abductive inference, "from which they differ in being absolutely beyond criticism" (193 1-66 : 5.1 8 1 ). The service which abduction renders to science is basically one of economy. Abduc­ tion generates small numbers of hypotheses (infinitesimally small when contrasted with the numbers of logically possible explanations) which appear to carry with them a cer­ tain antecedent probability of success ( 193 1-66 : 6.477). In the absence of such a vehicle for drastically paring down the number of hypotheses to be tested, it would be impossible to account for the historically verified success of science in generating true M.A. hyp otheses.

Abelard

3

Abelard, Peter ( 1 079- 1 1 42 ) Abelard developed a theory o f logic as scientia sermocinalis 'science o f discourses' : while metaphysics concerns things and their properties, logic is strictly concerned with nomina •�·names' or 'linguistic terms,' and it is not interested in the '�truth of things but with the truth of *propositions. In the course of his logical works (commentaries to Porphyry's Isagoge and •:•Aristotle's De Interpretatione, the treatises called Ingredientibus, Nostrorum, Dialectica, Abelard developed a theory of *meaning, *definition, propositions, dis­ courses or *texts. Many of his ideas sound absolutely original from a semiotic point of view, even though his philosophical lexicon was still in progress, so that he frequently used in two different texts the same term with significant shifts in its technical meaning. Abelard distinguishes between natural voces 'sounds' and sermo or dictio (which is the *sign as a bifacial entity) established as a significant sound. There are many forms of signification: per determinationem (when terms refer to a given substance as a subject), per generationem (when words generate an internal concept), per remotionem (as in "non man" that still signifies something), secundum adhaerentia vel comitatione, as it happens in inferences (when by saying that someone is father of someone else, we elicit the infer­ ence that the latter is the son of the former) and in connotations (for instance the expres­ sion latratus canis 'bark' suggests the idea of rage). However, the only signification inter­ esting for the logicians is the one per impositionem 'by convention.' The linguistic con­ vention is considered as inventio 'invention' (the activity of the first man imposing the right names to the things), as institutio 'institution' {the intentional aspect of a lan­ guage), and as imposition, which specifically concerns the correlation between vox and intellectus. Abelard seems to reserve the term significatio only for the relationship between a sound and the internal concept it elicits. A significant sound signifies an intellectus (or an internal representation); nominat or appellat 'names' or 'refers to' individuals; and designat or denotat both the corresponding sententia 'meaning' and the de/initio 'defini­ tion' : the double difference between signification and designation or •�denotation, on one side, and between definition and meaning, on the other, depends on the Abelardian notion of *universal. Abelard is neither a nominalist nor a realist, but rather a "conceptualist." The human intellect considers various aspects of the individuals and by *abstraction produces a confused common image ( conceptio communis confusa), a /igmentum 'fic­ tion,' which is expressed by the name and does not correspond to any real essence; it is possible to conceive of it even when the corresponding object does not exist. In order to define this sort of conceptual "third world" (in the sense of Popper 1 972 ), Abelard says that universals are status 'states' or modes of being {this notion being very similar to the Stoic idea of lekt6n). A status is an aspect of the thing, or the thing viewed under a cer­ tain description : it represents the thing aliter quam sit and not alia quam sit (in a different way, not as a different thing). Even though to signify (in a strict sense) means to definite designare 'to designate by definition,' usually significare means constituere intellectum 'to produce a mental state.' In this sense intellectus is everything that can be understood of a given word in a given context. Words can have multiple signification and Abelard insists on the importance of the determinatio as contextual disambiguation. Names express their universal *content by their nominative declensions but the other declensions-as

4

Abstraction

well as the different verbal forms-are matter for contextual determination (the theory of signi/icatio is different from the theory of modi signi/icand1). The intellectus can also correspond to the many possible descriptions of a given thing. Something can be described according to its difference (e.g., rational and mortal), to its genus (e.g., a body can be described as a colored substance), to its species plus some accidents (e.g., Socrates can be described as a white man, son of Sophroniscus), to its accidents only (e.g., Socrates is something white and son of Sophroniscus). On the con­ trary, a definition is provided (according to the Aristotelian tradition) only by genus and specific difference. Thus, while a word can signify everything that can be described and understood, it designates or denotes only its sententia. However, the sententia or meaning is not the definition, but the whole of the possible differences that can be predicated of something. Thus the word as name is determinativum 'can determine' of all the possible differences of the thing; its meaning contains (potentially) all these differences; the definition posits only the differences indispensable to isolate without ambiguity a sense of this word. Abelard seems to suggest a principle of minimal contextual disambiguation through the correct defini­ tion. A man can be intended as rational, mortal, two-legged, able to walk, and so on; there are infinite differences and the definition is the way of identifying those that are strictly necessary to establish the sense of a term. In this sense the sententia is richer than the definition, because the sententia potentially contains many differences and many possibilities of different definitions. U.E.

Abstraction

Abstraction is the process or result of developing a concept through concentration upon certain parts or characteristics of an object which are relevant to the special interest of the moment and a resultant disregard of those which are not relevant. More specifically, a total or precisive (due to C.S. *Peirce) abstraction disregards particular characteristics of its carriers and moves toward more universal concepts, as, for example, man, red, chair. A formal or hypostatic (C.S. Peirce) abstraction disregards the carriers of the abstraction by separating out its form-giving essence, thus humanity from human beings. Sometimes this is a simple linguistic shift as in the case of redness from red objects. In a physical formal abstraction, the relation to the carrier or carriers is pre­ served ("roundness" as of a ball) while in metaphysical formal abstraction the mind retreats from every relation to a carrier concentrating instead on form itself (God or �-Frege's notion of a cardinal number). An abstraction becomes subjectal (C.S. Peirce) when it becomes a subject of thought and logical manipulation, as when we pass from the notion of cardinal number to the algebra of cardinal numbers and prove theorems about it. When an abstraction becomes a permanent possession of the intellect and achieves a measure of independent substantial existence, it becomes an abstract idea. This aspect of the abstraction process is embodied in the linguistic distinction, abstract vs. concrete, as in the case of abstract noun (humanity, courage, happiness) as opposed to concrete noun (horse, chair, sea).

Abstraction

5

When used as a scientific or philosophical tool, abstraction may embody idealization. An ideal model of an empirically apprehensible situation may be constructed. Thus in Euclid's geometry points and lines are "ideal" objects, a point for example being an object with location only, as distinguished from its empirical manifestations as blobs of ink on a sheet of paper. When used as a tool, an abstraction may possess both an intensional and an exten­ sional aspect: intensional through its definition (the notion of group in mathematics is defined by a set of axioms) and extensionally through its carriers, the objects that pos­ sess its defining characteristics, (the instances of groups}. The extension of an abstrac­ tion may be potential as well as actual, as the extension of the phoneme b, intensionally defined in terms of its *distinctive features. Purely extensional abstractions are also possi­ ble: consider membership in set S constructed by drawing a finite number of objects at random from a population. Membership of a particular object in S can only be decided extensionally. Some abstractions have no empirical extensions (as for example "uni­ corn") while others have only mentalistic extensions (as infinite sets in *mathematics or deep structure in linguistics). In mathematics at least, the carriers of an abstraction may themselves be abstractions. The set of all integers under addition forms a group, i.e., it is a carrier of groupness but is itself an abstraction. B.B.

Semiotic Aspects of Abstraction The semiotic problem of abstraction is involved in the question how general words mean. If it is assumed that there exist only particular concrete things then the question arises on what ground a general word, to which nothing corresponds in the world, applies to many things that fall under it. A radical nominalistic answer is that generality is a mere convenience of speech: a general word applies to many things in virtue of nothing, that is, purely arbitrary, as the word vegetable applies to carrots, peas, etc. Under that interpretation, however, the problem of generality reemerges at the level of language; it is left unexplained on what ground particular occurrences of a word, e.g., inscriptions of the word the which appear on this page, are regarded as particular occur­ rences of one and the same word, that is, as tokens of the same *type. The classical British Empiricism offers a conceptualistic account of abstraction. According to �·Locke words signify things only indirectly; what words stand for are ideas in the mind. For him, then, the question how general words mean is how abstract ideas are formed, and his answer is that they are formed by the mental process of abstrac­ tion. "The mind makes particular ideas, received from particular objects, to become gen­ eral; . . . This is called abstraction, whereby ideas taken from particular things become general representatives of all of the same kind" (1961 : Book II, ch.1 1: 9). The operation of abstraction consists in retaining a quality, e.g., "white," we find common in many particulars (in chalk, snow, milk), and leaving out other qualities which are peculiar to them. It is based, then, on a recognized resemblance among things without implying, however, the existence of a quality in respect to which particulars are similar (they may be held to stand in the relation of resemblance to a selected paradigm particular). It has been pointed out that even if qualities as abstract entities are eliminated in favor of the

6

Abstraction Operator

relation of resemblance among things, the relation itself has to be regarded as a universal (*Russell 1940). Another way to approach abstraction is to introduce classes of objects to which gen­ eral words apply. The operation of abstraction is thus regarded as construction equiva­ lence classes of things which have certain properties; the method is called �·definition by abstraction (Kling 1967). An abstract property, such as a color, is identified with an equivalence class of things of the equal color. A disadvantage of this method is that that there are co-extensive properties, i.e., there may be two distinct properties, e.g., "three­ sidedness" and "three-angledness," while the operation of abstraction will yield one and the same class of triangles. Further, introducing classes is often motivated by a nomi­ nalistic tendency to get rid of abstract properties ; although this method provides an exact formulation of the relation between particular things and classes abstracted of them, it leaves open the problem of ontological characterization of classes. The problem of abstraction is related in many ways to the controversy over the nature of �·universals, nonetheless the distinction between abstract and concrete does not coincide with that between universal and particular. Ordinarily, a physical object, such as a red ball, is regarded as concrete, whereas its properties (redness, roundness) are regarded as abstract. However, a concrete object may be interpreted, as in the empir­ icist tradition, as a bundle of qualities ; the sentence This ball is red would be taken to mean "Redness and roundness are co-present here." The abstract/ concrete distinction appears to be relative to an epistemological constitutional system; in a physicalistic system physical things are treated as basic concrete individuals, whereas in a phenomenalistic system sense data are taken to be concrete and basic (Goodman 1966 ). U.N.

Abstraction Operator An operator, i.e., a logical symbol binding a variable, is called the abstraction operator if it transforms a sentential formula into a ''name of the set of those things that satisfy that formula. Let the formula be represented by Q(x). Then the name of the things satisfying Q(x) is written as (.i) Q(x), where the cap over the variable enclosed with pa­ rentheses plays the role of the operator. The definition, restricted here to unary predi­ cates, is as follows : ZE(.i) Q(x) = Q(z), i.e., z belongs to the class determined by " Q(x) " if and only if z satisfies " Q (x)." The expression "(.i) Q(x)" is called an abstract, while the operation of forming an abstract is called abstraction. For instance ZE(.i) (x :s; 3) = x :s; 3, provided that the variables range over the nonnegative integers and the object denoted by "(.i)(x:s; 3)" is the set 0, 1, 2, 3 . The abstraction operator is not indispensable, being avoidable by a definition like that mentioned above. But it is a very useful device for simplifying formulations that refer to sets and relations. Thus, e.g., the universal set can be conveniently defined as (x) (x= x) , and the empty set as (x)(x+ x). Furthermore, the other operators can be defined in terms of the abstraction operator. For example, the definition of the universal quantifier runs as follows : (Vx) Q(x) = (x) Q(x) = (x) (x = x)). From the linguistic point of view, the abstraction operator can be regarded as a grammatical device that forms a name of an abstract entity (a class, a property, a rela­ tion) out of a predicate or a sentential formula.

Actant

7

The abstraction operator can be treated as a special case of a more general operator that operates on any expressions whatever, not just predicate expressions. This device has been called the lambda-operator by Church. A lambda-expression, i.e., an expression of the form (Ax)( . . . x . . . ) is either a predicate expression or a *function expression. Both are defined by the *schema: (Ax)( . . . x . . . )a iff . . . a . . . ; that is, the definiendum is equivalent with what results from substituting "a" for "x" in the operand, i.e., in the expression that is within the scope of the operator. What is called the lambda-operator is the part of the lambda expression consisting of the Greek letter and a variable, enclosed within parentheses. W.M.

Actant The concept of actant-originally proposed by Tesniere-was adopted and developed by A.J. Greimas, who has broadened its scope to a narrative semiotics. Basically, the notion of actant is similar to one of �·case in Fillmore's case grammar : a case can be considered as an actantial position. Greimas stresses the difference between enonciation and *enonce. By enonciation he means the act of ,;utterance in so far as it is represented by language (enonciation enon­ cee). By enonce he means any meaningful entity of a linguistic chain (be it spoken or writ­ ten) . Greimas and Courtes (1979: 123, 125) propose to translate enonciation and enonce by the English neologisms enunciation and enunciate. An enunciate is analyzable into a predicate (*function) and actants, as the terms resulting from such a function. Greimas (1973b : 162) lists two forms of narrative enunciates : the function Subject - Object and the function Destinateur ('Addresser' ) - Object - Destinataire ('* Addressee'). This analysis of the enunciate into actants and predicates is linked to two concepts inherited from �-Propp ; the notion of narrative function (reformulated in terms of narra­ tive enunciates) and that of "dramatis persona," partially substituted by the notion of actor. From a narrative point of view, within a corpus of story variants, a single actant can be manifested by more than one actor and a single actor could be a syncretism of many actants. The distinction between actants and actors depends on the generative proc­ ess: actants are units of the "narrative �·syntax," while actors belong to the discursive structures. At this more superficial level the actants receive a "semantic investment" and are manifested as actors. In his more recent works (since 1976a and Greimas and Courtes 1979), Greimas has developed a broader typology of actants: ( 1) actants of narration (or of the enunciate) such as subject/object, addresser/addressee, helper/opponent; (2) actants of the •:•com­ munication (or of the enunciation). He distinguishes between : narrateur 'narrator'/ nar­ rataire 'narratee' (for instance when a narrating "voice" is manifested by a personal pro­ noun) ; enonciateur 'enunciator' / enonciataire 'enunciatee' (as the virtual and perhaps lin­ guistically unexpressed poles of every discursive unit) ; interlocuteur 'interlocutor'/ interlocutaire 'interlocutee' (for instance in represented •:•dialogues). A.Z.

8

Action

Action

Although the term action may be applied to things or states of affairs in general (cf. the problem of action at a distance in physics), its main reference is to the doings of persons or to their deeds. It is action in both of these latter senses (act vs. actum) that empirical/ normative/rational theories of action (e.g., psychology and '}sociology/ethics/decision theory) are interested in. Clarification of the terms used in talking about actions and metatheoretical justification of the different kinds of principles used in the nomological components of these theories are the primary tasks of a general theory of action.

Elements of Action Description ,:•Descriptions of action may involve the following features : agent, act-':•type (of which the particular act is a token), ,:•modality (of manner or means), the temporal/spatial/ circumstantial context, and the rationale of action ; some reference to causal/ regular/ conventional/'}rule-governed/ normative '}presuppositions/ results/ consequences has to be included in any more detailed specification. Approaching the question What is an action ? obliquely, via the question How is an action to be described?, is a (good) first-step policy followed especially by ordinary lan­ guage philosophers and linguists. Proving the connection between actions and verbs to be a fallacious one ("to know"-the most famous example-is not a verb expressing an act of knowing; Ryle 1949), taking account of such basic distinctions as act vs. activity (Vendler 1967: ch.4) and attempt vs. achievement (Ryle 1949), and exploiting in particular the application criteria of action-modifying adverbs, their findings have made us aware that ordinary action language is much more fine-grained than traditional philosophers and grammarians had led us to believe. Austin's method of classifying actions according to what might go wrong with them led to his well-known distinction between different dimensions of *speech acts (Austin 1962a).

Basic Concepts and Logics When developing a general theory of action, however, more systematically constructed *explications become necessary. As our (rational) actions depend upon our *beliefs and preferences, there are three groups of concepts to be considered primarily. Praxeological Concepts The most fundamental example is acting itself. The main point of all explicative proposals is the distinction between action and mere behavior. The most favored differ­ entiating characteristics are consciousness, purposefulness, and *intentionality (Ans­ combe 1957), but all are highly controversial. The approach most fruitful in systematic respects is : Action = Behavior we are able to forbear. This accounts for the fact that we associate acting with freedom and responsibility and gives us, even if circular, at least a reliable criterion for adequacy of any further definition. It is problematical whether mental acts may be subsumed (Geach 1957). In many cases to act is to bring about some­ thing. This idea was first axiomatized in von Wright's sketch of a logic of action (1963 ). It was the starting point of Danto's (1965) concept of basic acts as being acts not brought about by doing something else (paradigmatic case : raising one's arm). Further praxe-

Action

9

ological concepts include: X lets it happen that p; p is avoided by X; X can bring it about that p (different meanings of can, Ryle 1949). Doxastic and Epistemic Concepts

Subjective probability is the fundamental doxastic term, on the basis of which X's believing that p can be defined as p's being regarded by X as more probable than not-p (weak belief) or as maximally probable (strong belief), respectively. Logics for rational­ ized versions of both of these belief-concepts and of the related concept of knowledge (strong belief plus *truth) have been developed in the frame of intensional *semantics by Hintikka ( 1962, improved and in fact replaced by Lenzen 1980). Voluntative Concepts Fundamental here is the notion of Subjective preference, its logic founded by Jeffrey, with extensive application in decision theory. Further concepts include : X wants that p, of which intention is a special case. By doing f, X intends to achieve p, which is roughly equivalent to: X does f, wants that p, and believes that by doing f he brings it about that p. In this case, p may be referred to as the aim of X's doing f. Fully developed logics of "wants," etc. do not yet exist. First approximations may be found in Kutschera 1980 and Lenzen 198 1. A logic of action as the core of a general theory of action will have to contain concepts of all of these groups. The most promising approach up to now is Aquist 1974, modified and extended in Kutschera 1980. For a list of preliminary logics and future prospects, see Apostel 198 1. The application most relevant for semiotics is : Action-theoretical semantics (Meggie 1986, following the lines of Grice 1968 and D. Lewis 1969).

Explanation of Action With actions, causal explanations (using some deterministic or statistic laws; Hempel 1965) are to be distinguished from rational explanations (using principles of rationality explicated in decision theory ; Luce and Raiffa 1957). Whether the propositional atti­ tudes mentioned in rational explanations (beliefs and preferences as constituting the agent's relevant reasons) do function as causes or not, is the question most debated in this context. Practical inferences-being a type of rational explanation schema (Meggie 1978 )-and their connection with understanding of actions are discussed in von Wright 197 1 (cf. Kutschera 198 2 : ch.2).

Ontology and Metaphysics of Action The problem of individuation of actions has been rightly approached via action *de­ scription (Anscombe 1957; Goldman 1970). In giving the logical structure of action sentences, actions as particular events are introduced in Davidson 1967; for actions as �·universals see Landesman 1971. I stressed above the essential bond between "action" and mentalistic terms-a topic highly controversial in all the discussions relating to an explanation of action as well as to the following problems pertinent in contemporary phi­ losophy: the mind-body problem, the problem of other minds (Putnam 1975a), the problem of the freedom of will, and, of course, all the problems generated by reductive behaviorism (Scriven 19 56). G.Me.

10

Addressee

Addressee The party toward which an act of ,;communication is addressed or directed. �-Buhler's model of linguistic communication (1934) represents the addressee (Adres­ sat) as one member of a triad of components, together with the speaker and the external situation; Buhler defines as one of the three possible functions of language a vocative orientation toward the addressee (Appell). In �"Jakobson's expanded version of the model (1960) - adapted by many commentators to nonlinguistic communication the addressee is seen in relationship with five other elements (addresser, or ,;sender; ,:•message; ,;code; context; and �·channel); Jakobson terms the orientation toward the addressee the conative function. More recent theory has suggested, however, that the single category of the addressee in fact brings together at least three complementary but distinct factors: (1) the actual receiver of the message; (2) the intended addressee of the message as conceived by the sender; and (3) the virtual addressee immanent within the message itself. The noncoinci­ dence between factors (1) and (2) has been noted by Lotman (1971a), who emphasizes the autonomy of the actual receiver and his codes with respect to the sender's *inten­ tions and codes. Factor (2), instead, has been defined by various commentators as a set of ,:-presuppositions on the part of the sender regarding the semiotic *competence, social status, social roles, and attitudes of the addressed party (see Wunderlich 1971a; Schmidt 1976; Lyons 1977). Such presuppositions will normally influence the precise form of the message, particularly its indexical elements marking the presence of factor 3 : in the case of linguistic discourse, these chiefly comprise deictics (second-person pronouns and pos­ sessive adjectives, verb inflections indicating the vocative case or a symmetrical or asym­ metrical relationship between sender and addressee, etc.) (see *Benveniste 1966a). The distinct constituents of the addressee function are further multiplied in the case of artistic, and especially literary, '�texts. Theorists of ,;literature have identified, in addi­ tion to the actual ,;reader, a model or ideal addressee presupposed by the �-author, an internal addressee (or narratee in the case of narrative [Prince 1973]), and an encoded or implicit addressee toward whom the text is apparently oriented (Booth 1961b; Genette 1972; Culler 1975; Corti 1976; van Dijk 19766; Eco 1979a; Pagnini 1980; Suleiman and Crosman 1980; on the question of the addressee of a visual text, see Marin 1980a). K.E.

Advertising Advertising has been studied extensively from various points of view. Following upon the previously dominant approaches from psychology, '�sociology, ,;economics, publish­ ing, *rhetoric (Galliot 1955), and linguistics (Leech 1966; Romer 1968), the semiotic analysis of advertising adds a new perspective. This semiotic expansion of advertising . analysis affects the object of investigation as well as the method. In its approach to adver­ tising, semiotics attempts to explore, within a unified framework, not only the verbal *message of the advertisement ,;text but also the nonverbal ,;code, along with the entire communicative situation in advertising (see Koch 1971a; Lindemann 1976). In regard to the method of investigation, the general theory of ,;signs provides appropriate instru-

Advertising

11

ments for the exploration of the various messages in advertisements. The several direc­ tions in semiotics have developed different approaches to advertising.

,:•Denotation, Connotation, and Codification in the Advertisement The approach characteristic of semiology (cf. Victoroff 1972) was first introduced by ''Barthes (1964a) in his exemplary analysis of the "rhetoric of the picture." On the basis of ,:•Saussure's dyadic model of the sign (signi/iant + signifie = signe) and ''Hjelmslev's further distinction between denotation and connotation, Barthes recognized three mes­ sages in the illustrated advertisement text he investigated. A first linguistic message con­ sists of the product name and the linguistic commentary. The two additional messages inhere in the picture (a photograph): one is the uncoded iconic message, in which the pho­ tographic image analogously denotes the "real" objects in such a manner that the *signi­ fier and the signified are "quasi-tautological," and the other is a coded iconic or symbolic message. The latter includes the connotations of the picture that form the "image" of the product. These connotations are coded, since they depend upon the "cultural knowl­ edge" of the viewer. This implies that an advertisement text, a "form of speech" (lexie), can be subject to different interpretations or "forms of reading" (lexiques). In advertising and in the mass media generally, connotations in fields of association amal­ gamate into systems of connotations, in which the ,:•ideology of the society represents itself. The totality of the "connotators" that form an ideology Barthes calls a rhetoric. Barthes's semiological approach stimulated a series of advertisement analyses (e.g., Ehmer 1971 ). An important work for the semiotic analysis of advertising is also Barthes' comprehensive study Systeme de la Mode (1967), in which ''fashion journals, a text ''genre related to advertising, are subjected to a detailed structural analysis.

Messages and Functions of Advertising A further step in the semiotic analysis of advertising is taken by Peninou (1966, 1968, 1970), who analyzes advertising as a multilayered system of messages embedded within one another and serving various �·functions. Peninou (1968) distinguishes five primary messages in advertising: (1) the message of genre classification, with the function of dif­ ferentiating advertising texts from other types of texts; (2) the linguistic message; (3) the message identifying the sender (the advertising agency) ; (4) the denotative message; and ( 5) the connotative message predicating the value of the product. Peninou ( 1970) distinguishes further, dependent upon the type of assertion about the product, three types of advertisements, each with its own characteristic use of pic­ tures : (1) the "appearance advertisement," which introduces a new product; (2) the '' constative advertisement," which refers to the existence of a product; and (3) the ' 'pre­ dicative advertisement," which intends to emphasize the value of a product. In the analysis of the functions of advertising, Peninou (1970) goes on to apply �Jakobson's ( 1960) model of the six ''functions of language, recognizing three functions as dominant in advertising: ( 1) the appellative (conative) function, directed at the audience; (2) the referential function, denoting the product; and (3) the poetic func­ tion, for the signs in an advertising text have effects beyond their purely semantic func­ tion.

12

Advertising

The Rhetoric of the Advertising Code

Following the approach of French semiology, Eco ( 1 968) extends the rhetorical analysis of the advertisement from the verbal into the iconic dimension. Eco speaks of advertising codes with double registers, one verbal and one visual. He distinguishes five levels of visual codification : ( 1 ) the iconic level, similar to Barthes's uncoded iconic messages; (2) the iconographic level, based upon historical traditions or publishing conventions (cf. Barthes's coded iconic message); (3) the tropological level, with the visual equiva­ lents of rhetorical figures (cf. Bonsiepe 1 965; Durand 1 970); ( 4) the ''topic level, with the premises and topoi of ''argumentation; and (5) the enthymematic level, with the actual structure of the visual argumentation. Structural ,:•semantics of the Advertisement

The attempt to investigate semantic deep structures and even universals of advertising is undertaken by Langholz-Leymore ( 1 97 5 ) in a comprehensive study. Her semiotic approach follows Levi-Strauss' model of the structural analysis of *myths and Greimas' structural semantics. (See also Lindekens 1 975 on the application of Greimas' approach in the analysis of advertising). According to Langholz-Leymore, advertising is a mediator between the concrete (the product) and the abstract (the *symbols and values). The semantic characteristics of the advertisement appear as a binary structure in which an *opposition between the positive properties (c) of the product in question (a) and the negative properties ( d) of competing products (b) are explicitly or implicitly contrasted. The properties (c and d) as well as the products (a and b ) form a twofold sign in which the products represent the signifier and their properties the signified. Within this twofold sign an equivalence relation holds of the form a : b =< c : d (e.g., butter : margarine =< expensive : cheap). The analysis of these binary oppositions leads to the conclusion that advertising, just like myth, is concerned with finding answers to universal human problems such as life/death, happiness/misery, war/peace, and hate/ love. These universal themes are realized with the same regularity in the deep structures of both text types. But because advertising works with simpler means, it appears as a degenerate form of myth.

* Sign Types in Advertising The relevance of Charles S. *Peirce's triadic sign model for the analysis of advertising was first shown by Bense ( 1 965b ). Every type of sign Peirce classifies can occur in advertisements: the sign in itself (representamen) can be qualisign (e.g., a color), sinsign (e.g., a product sample), or legisign (e.g., a maxim); depending upon its relation to its object, it can be an icon (a pict(¾re of the product), a symbol (the name of the product), or an *index (to the degree that it points to qualities of the product). Finally, with respect to its *interpretant, it can be rheme (as a single word), dicent sign (as assertion about the product), or argument (e.g., in the form of a logical conclusion). For the pur­ poses of advertising analysis, the sign types of the object relation are of primary impor­ tance, since they concern the semiotic connection between the advertisement and the product.

Advertising

13

Textual Deep Structure and Indexical Feature Trans/er in Advertising An advertisement exhibits three distinctive semantic units in its textual deep structure : (1) the •�name of the product, (2) a reference to its positive qualities, and (3) an appeal to purchase (Noth 1975). These three entities need not always be realized in the textual surface structure: while the name of the product must always be mentioned, a direct pur­ chase appeal is often missing. The latter, however, remains intact in what Peninou calls the "message of genre classification." If the type of the "constative advertisement" (see above) does not always mention the positive qualities of the product in its surface struc­ ture, it is only because it presupposes their familiarity to the audience. In advertisements for mass consumption, articles whose positive features preclude visual representation, e.g., beverages and cigarettes, advertisers employ a special semiotic strategy, of indexical feature trans/er (Noth 1975, 1977). The product whose positive features cannot be shown is represented in contiguity with valuable objects (gold, antiques, etc.), famous people (actors, athletes and so forth), or similar things whose qualities (valuable, famous, etc.) are known. By means of this contiguity relation a transfer occurs, associating the known features of the "valuable" or "famous" with the less well known product. This feature transfer takes place by means of an indexical sign relation: the transferred features become part of the set of features of the product and thereby refer as an index to it. The process of indexical feature transfer explains the process of connotation formation in advertising, for the connotations of the product are just the transferred features. In the genesis of sign usage, indices appear on the earliest level. Even the learning process in the conditioning of animals through stimulus-response chains is based upon the use of indexical signs. Indexical sign use is thus a genetically primitive semiotic pro­ cess (cf. Noth 1977). Because of the special importance of indexicality in advertisements for mass consumption articles, these prove to be of a primitive text type, which can explain their psychological effectiveness on deep levels of consciousness in semiotic terms. W.N./N.N. The importance of the indexical sign relation in advertising had been earlier dis­ cussed by Veron (1970). He held that indexicality defined, according to Peirce, by the contiguity between the representamen and the object it stands for, entered as one of two basic sign relations into every coding process. Jakobson, following Saussure's (1916) dis­ tinction between associative and syntagmatic sign relations, had described the two coding operations as selection of units from a repertory and combination of the selected units into a message. Selection is made from a stock of similar units which may be one substituted for the other. Combination results in the contiguity of units. Whereas a meta­ phor is based on substitution of a new but similar unit for a habitual one, a *metonymy is constituted by taking units which are linked by contiguity as indices for each other (as in the rhetorical figure of pars pro toto). Klopfer ( 1977) has emphasized that the discovery of metonymic indexicality between the units of partial systems or entire figural constella­ tions which the advertiser plants into his messages affords the fascination and effective­ ness of them especially with an intellectual public. It could be maintained that the fascination of the *hermeneutic interpretation of advertisements in the semiotic literature (e.g., Barthes 1964a; Bonsiepe 1965; Eco 1968 ; Ehmer 197 1 ; Goffman 1976; Haug 198 1 ; Kerl et al. 1973; Klopfer 1977; Noth 1975; Umiker-Sebeok 1979, 198 1, 198 5a, to name some outstanding examples) has exerted a

14

Advertising

similar fascination because it rests exactly on the same process of discovery. Some of these interpretations go, however, beyond the single example, referriQg, on the basis of large collections of advertisements, to specific anthropological problems such as stereo­ types of sex roles (Goffman 1 976; Kerl et al. 1 973) or age groups such as children and old people (Umiker-Sebeok 1 979, 198 1 , 198 5a). Posters Posters are advertisements of large format applied to outside or inside walls of buildings if no particular billboards are provided for them. They are geared to larger reading dis­ tances and frequently employ an eye-catching device to attract the attention of pas­ sers-by. What has been said about advertisements in general applies, mutatis mutandis, also to posters. There are posters based on printed texts only and posters featuring a combination of image and text. The image is often used as the eye-catching device. If a combination of text c·symbol) and image (';icon) is used two basic types of inter­ action between symbol and icon may occur. In the first type the symbol is autonomous, i.e., understandable without the icon. In that case the icon may, for example as an eye­ catching device, constitute a redundant illustration or a metonymical index featuring selected properties of elements represented by the symbols (text) . In the second type of interaction, the understanding of the symbol depends, at least in part, on the icon. This is the case if, for instance, an icon is substituted for a symbol in the text (e.g., a picture of a house for the word "house") or when the icon disambiguates a polysemic text. In the latter technique, the icon is either coupled with a sentence rendered ambiguous, for instance, by a personal pronoun without mentioning the subject ("it's fun" ), or clarifies a "double sense" in a play of words. As to all advertisements, Jakobson's (1 960) six functions of language apply also to posters. Where posters are employed for commercial and political reasons the appellative (co native, persuasive) function tends to dominate, followed by the referential function (presentation of the merchandise or candidate, etc.). But posters are frequently applied to announce cultural events. They may even become an integrated part of the organiza­ tion of such an event, as Bouissac (197 1 , 1 976a) has shown for ,:•circus posters. The less commercial or political the task of a poster is, the more the expressive and artistic (*aes­ thetic) functions of language are emphasized in its composition. This implies that post­ ers become more experimental as far as the parallelism of typographic and iconic ele­ ments is concerned. Icons do not serve merely to attract attention, but become meta­ phors; typefaces are not only used to render a message legible, but also as a pretext for formal de-automatization. In short, the poster becomes a work of art. This tendency has been particularly pronounced in traditional Polish posters (Krampen 1 982). In the semiotic investigation of advertisements and posters, methods of quantitative empirical research have begun recently to replace the single case hermeneutic interpreta­ tion. These methods are based on the free sorting of larger amounts of advertisements or posters into groups of similar ones, where the criterion of similarity is chosen individ­ ually by the subjects in the sample and indicated only after they have completed their sorting (Krampen 1982). Similarity is taken in this case as an index of common *meaning (G.A. Miller 1 967, 1969b) and the semantic structure of a corpus of advertise­ ments is produced by cluster analysis over the sortings of all subjects in the sample. M.K.

Aesthetics

15

Aesthetics Aesthetics (Greek ato-SaVE:o-Sm; Latin sentire 'perceive'), in a broad sense, is that branch of philosophy or that philosophical theory that deals with actions leading to per­ ceptual knowledge as opposed to those leading to conceptual knowledge. It is its aim to provide theoretical foundations for perceptual cognitive processes so that it becomes possible to improve them rather than letting them go on as special kinds of naturally evolving processes. The proper object of aesthetics as well as the meaning of the terms needed for its construction and the criteria of validity for its propositions are still open to scrutiny; only in few and restriced areas is reliable textbook knowledge available. In the course of recent studies within semiotics and the �-philosophy of language, man as "animal symbolicum" (�·cassirer) has acquired the position of a guiding princi­ ple : not in some epistemological a priori but in our semiosic activity "wherein our deal­ ings with things are sedimented and become manifest" (Gadamer 1 976b : xi); and, fur­ thermore, "the study of specific *action in a specific situation" CMorris 1964 :5) be­ comes necessary in order to understand the semiosic activity. Aesthetics, in a narrow sense, denotes theoretical activities concerning the arts, which during history have been split into specialized fields like theory of (fine) arts, theory of �-literature, theory of '�music, and others. From these subtheories the corresponding dis­ ciplines that now exist have in turn been derived. Within a (general) theory of art it is common to distinguish between aspects of pro­ duction and reception (aesthetics of production, aesthetics of reception) and aspects of formation (aesthetics of works). Production and reception refer to the use of works of art (i.e., to the processes of producing and experiencing them), whereas formation refers to the works themselves. The (general) theory of art is concerned not only with the genesis of actions leading to an artistic product, but also with its materials, its media, its procedures (its matters and its manners, Goodman [ 1 968 ] ), as well as its uses. Actions of production, which may be performances of an action (*sign as process) or traces of an action or marks (sign as thing), both of which stimulate perceptual cog­ nitions, are called artistic actions. Actions of reception, which are actions of judgment that differentiate either perceptions or values - following the Greek usage of alPort-Royal tradition on the gen­ eral theory of language sign ; sometimes they are original and drawn from his personal experience as a philosopher of mathematics and other sciences. They are often contro­ versial reactions against the official trends in French philosophy of the time (e.g., eclecti­ cism), more often self developing reflections, where '>pragmatics and ethics are deciding factors. We can cite, as examples, the premonitions about *zoosemiotics (Rey 197 1 :61 -62), or the theory on the semiotic origin of language as a combination of two signifying sys­ tems issuing from two biological sources, one expressed through the optical ,:•channel (gestures, then '>mimicry, then the visual arts), the other through the auditory channel (cries or shouts, then '>music, then poetry, first mixed with music). Resulting from this analysis, vocal expression is derived from poetry, itself a product of music, and written expression is on the same genetic level as '>painting, sculpture and mimicry (from the more recent and elaborate to the more archaic). Evidently, the philosophical tradition of *Rousseau and '>Vico has many points in common with Comte's, as Benedetto Croce later also had some, but the founder of positivism is original in many other respects. For example, he always gives his simple and sometimes strange historical patterns - gov­ erned by the famous "law of the three stages" - a logical and semiotical counterpart. Thus, after the primitive "Logic of Emotion" ("logique des sentiments"), the "Logic of Images" fits with the polytheistic stage of mankind. It is only with the "Logic of Signs", synthesizing emotions and images, that man is able to link artificial, social (arbi­ trary) signs with �·abstractions and '>concepts. Peirce's �·symbols are, for Comte, the con­ ditions for the monotheistic stage of man, along with many other social characteristics. By such ideas, Comte is strongly opposed to �·nominalism and formalism, and even to nineteenth century logic which, he writes "concentrates all [its] attention on the most voluntary, but the less powerful, of the three main modes composing mental combina­ tion" (185 1-54, vol. 2 :241).

1 44

Conceptualism

Despite its many shortcomings, Comte's main work gave birth to a new concept of ,;anthropology, with its own philosophy of logic and semiotics. That this should date back to the mid-nineteenth century deserves special notice and should encourage scholars to study thoroughly a work characterized, among other outstanding qualities, by a consciousness of its own logical, linguistic, and semiotic nature. A.R.

Conceptualism Conceptualism (Latin conceptus 'notion,' conceptus animi 'thought,' 'idea') or moderate *nominalism : a view in ontology, semiotics, and methodology advanced in the con­ troversy about universals, permitting only subjectives from among the ontic categories other than individuals, and opposed to *nominalism, *conformism, and *realism. In semiotics the view that every ''name has only one object : the object of a singular name is an individual; the object of a universal name is a •funiversal, that is, a subjective, that is to say either a conception (according to terminism) or a notion (according to sermonism). For instance, the author of "Symposium" is the only object of the singular names "Pla­ to" and "the founder of Academy" and respectively the conception (or the notion) of philosopher is the only object of the universal name "a philosopher," Some contemporaneous semioticians take the view representing a middle way between conceptualism and conformism. According to them every concrete name desig­ nates an individual and every abstract name designates an abstract. Every universal (con­ crete as well as abstract) name denotes an abstract, that is, a set (of individuals) and con­ notes a subjective, that is, conception or notion. The modern form of conceptualism in semiotics had its most outstanding representa­ tives in John Stuart *Mill (being rather a terminist) and Franz �-Brentano (being initially a sermonist). J.J.

Concretism Concretism (Latin concretum 'concrete'): one of the variants of �·nominalism in ontol­ ogy, *semiotics, and methodology. The only ontic category is identified here with that of things. From the point of view of reistic concretism - or, simply, reism - things are regarded as bodies (spatiotemporal and inert objects) or souls (feeling objects). According to pansomatistic concretism - or, simply, pansomatism (somatism) - things are simply bodies, souls being a kind of body. The creator of concretism in its modern form is Tadeusz Kotarbinski (see •:•Lw6w-Warsaw School), following in Stanislaw �-Les­ niewski's footsteps. A position congenial to concretism is also held recently by Willard van Orman Quine. J.J.

Condillac

1 4S

Condillac, Etienne Bonnot de (1715-1780)

Condillac devoted a great part of his philosophical work to ,:•language and ,:•signs. It is possible to reconstruct how language gradually attained its full importance in Condil­ lac's thought, beginning with his criticisms of ,;Locke's empiricism. Though he had made himself a disseminator of Locke's ideas in France, he nevertheless criticized what he believed to be a trace of Innatism: in particular the fact that in Locke's reconstruction of the mental processes a kind of break occured between the primary moment of sensa­ tion, which acted as the basis for all knowledge, and that specifically human faculty of reflection, from which, in Locke's opinion, developed the possibility of true human knowledge. Condillac endeavoured to reduce all mental activity to one fundamental principle - thus eliminating any risk of Innatism - and at the same time he used explan­ atory hypotheses to account for the spiritual life in its entirety without having to resort to external and unverifiable postulations. It is here that language makes its first appear­ ance, in Condillac's attempt to base reflection, the superior activity, on the primary sen­ sation: We can only aspire to discover a first experience which none may call into question and which is sufficient to discover all the others. It must distinctly show us the source of our knowledge, its materials and the principle on which they are organized, the instruments used for this pur­ pose, and, lastly, the manner in which we must make use of them. I think I have found the solu­ tion to all these problems in the connection of ideas with signs and in their interconnection. (Es­ sai sur l'Origine des Connaissances Humaines, 17 46, Introduction).

The study of language, its �·rules and characteristics thus becomes fundamental for an empiricist philosophy that attempts to reach an exhaustive explanation of human knowledge by going right back to its origins. From this moment on (the Essai is Condil­ lac's first work), language will almost constantly be at the center of his philosophical interests. He sees in it one of the prime movers of the mental processes and their essen­ tial formal principle, and as such the key to understanding the way in which ideas are formed and interconnected. And so, bearing in mind his close attention to the develop­ ment of thought from its origins, it is not surprising that he took up the discussion on the problem of the origin of language. But clearly the problem was understood not as historical but as functional, and for Condillac hypotheses on the origins of language were of no value as historical proofs but served instead as explanatory hypotheses on the logical and genetic level. Using this perspective, he classified signs into three categories: ( 1) Accidental signs, or objects which particular circumstances have connected to some of our ideas, with the result that they can effectively arouse them; (2) Natural signs, or cries which nature has established for feeling of joy, hate, pain, etc.; (3) Institutional signs, or those which we ourselves have chosen and which have only an arbitrary connection with our ideas. (Essai, p. 19). Condillac places the birth of language proper in the second category of signs. With the natural cries tied to the expression of feelings and emotions, both language and sociability develop simultaneously; in a word, we have here the origin of intersubjective ''communication.

146

Conditional

Condillac calls this early developmental phase the "language of ''action," from which one can trace the development of mankind and, analogically, the development of each personality. From this language of action, composed of gestures and cries, starts the process of institutionalizing signs; these, which were previously natural, become arbitrary and conventional thereby bringing into existence "institutional language." It is within this last category that the human spirit, freed from the conditioning of nature, can fully develop its own faculties. The particular way in which this occurs relies on the main characteristic of articulated language, which is that of arranging into a linear chain the propositional elements that the language of action had condensed into cries or ges­ tures. This makes possible that analysis which properly constitutes the method of knowl­ edge and the guarantee of the correctness of our associations. Analysis is a central concept of Condillac's semiotics and his famous and modern­ sounding statement that "Every language is an analytical method, every analytical method is a language" (La Langue des Calculs, 1798, p.419) is based on it. The language which has remained most subject to analysis is algebra, and so �·mathematics proves to be the model science in so far as every mathematical *proposition lends itself to being analytically broken up into its constituent elements. The procedures by which the organi­ zation of these elements can produce complex propositions must be founded on the prin­ ciple of analogy and identity which is the guarantee of scientific correctness on every dis­ course. However, this ''function of language, which we may call ''cognitive, is not the only one which Condillac attributes to signs. From the same language of action there also starts the development of another of language's important functions, which we may define as the expressive function: it concerns the role played in any spoken language by all those elements such as ''intonation, intensity and ''style which are connected to that natural and emotional foundation which the gestures and cries of the language of action brought into play in primitive communication. In Condillac's opinion, even these aspects of language contribute as determining factors to the specific forms with which, throughout ''history, the particular character, genius, and individuality of each people have developed. Other than in the works quoted, Condillac extensively discusses the problems of lanF.M. guage in La Logique ( 1780) and in Cours d'Etudes (1775).

Conditional The discussion concerning applications of logic to natural language raises the problem of logical analysis of conditionals. The material ,:•implication cannot be considered to be a logical representation of conditionals, since in common parlance we would never say, "If all lemons are black, there will be no wars in Europe" with the feeling with which we state a ''truth, because the antecedent is false. A form that is closer to conditionals is formal implication, i.e., a universal statement in which a variable, appearing both in the antecedent and in the consequent, occurs in the scope of the same quantifier, viz. ('v' x)(A(x) :::i B(x) ). Many conditionals fall under the schema of formal implication. Note that necessity is replaced here by universality, which is a weaker requirement. But due to the fact that a bound variable is a link between

Content Analysis

1 47

the constituents of a conditional, formal implication is closer to the requirement of •·relevance, which is regarded as characteristic of conditionals. Another approximation is Lewis's conception of strict implication. This does not prevent, however, such para­ doxical cases as "If 2 = 2 and 2 2, then all men are mortal," which must be recognized as true according to the criteria for strict implication. The greatest difference between material implication and conditionals appears in the case of counter/actuals (contrary-to-fact conditionals), i.e., statements in the subjunctive form "If A were the case, then B would be." For instance : (o) If this body were immersed in water (A), then it would dissolve (B). (p) If this body had been immersed in water, then it would have dissolved. We use (o) when we feel that the fulfillment of the condition is improbable; we use (p) when we know that the condition was not fulfilled. The difference between (o) and (p) is so insignificant from the point of view of formal logic that we can take the simpler form (o) as representative, though the adjective "contrary-to-fact" is more appropriate for (p). Suppose that the body referred to in (o) was never immersed in water. Then, if we interpret (o) as a material conditional A :::) B, this conditional proves to be true on account of the false antecedent. It might be suggested that (o) should be interpreted instead as the conjunction ~ A&(A :::) B), but then a paradoxical result occurs as well : ~ A&(A :::) B) proves to be equivalent to ~ A. Such difficulties are removed when the "if . . . then" relation in (o) is interpreted as A&(A-+ B), where the relation symbolized by the arrow involves deducibility of B from A. Since the fact that a body dissolves in water cannot be deduced from an arbitrary false ,:-proposition, the undesirable consequence disappears. W.M.

*

Conformism Conformism (Latin con/ormare 'to form') is a view put forward in the controversy about •·universals permitting only abstracts from among the ontic categories other than indi­ viduals, and opposed to ''realism, ''conceptualism, and ''nominalism. In semiotics, the view that every •·name indicates an individual (in the case of concrete names like "a man") or a universal, namely an abstract (in the case of abstract names, like manhood) . In this century, views approximate to conformism i n semiotics where held by Bertrand ,:-Russell in the later period of his life. J.J.

Content Analysis Given the importance of ,:•language and language productions to the understanding of ''cultures, it is not surprising that several quite separate scholarly traditions have de­ veloped that study textual records. Each such tradition has evolved to serve somewhat different objectives; each has defended itself against encroachments of neighboring

148

Content Analysis

traditions. Content analysis is one of these traditions, drawing upon some of the particular objectives of *communication research. Although Krippendorff (1980) claims that the term content analysis is only about fifty years old (and was indexed for the first time by Webster's dictionary in 1961), sev­ eral excellent content analysis studies were done before the turn of the century and at least one detailed comparative investigation (of unorthodox ideas in Swedish church hymns) was done in the eighteenth century (Dovring 19 54). For the most part, the main growth of the content analysis tradition has been American, as developed in the writings of Lasswell (1927), de Sola Pool ( 1951), Berelson (1952), Osgood (1956), Stone et al. (1966), and Holsti (1969), especially as it emerged from propaganda analysis during World War II and the study afterwards of mass media �·symbols. Consequently, the Eng­ lish term has been a standard referent, with Germans, for example, making an uneasy choice between the literal translation of Inhaltsanalyse versus Kommunkationsanalyse (e.g., Mochmann 1980). While communications research continued to set the paradigm of content analysis, applications rapidly expanded to a variety of fields, so that compendiums such as Pool 1959, Stone et al. 1966, and Gerbner et al. 1969 include studies in psychology, political science, psychiatry, ''anthropology, ''literature, *history, education, and �·sociology. The range of procedures varied from simple studies of word usage to rather complex thematic analyses. What then are the essential features of the content analysis tradition, in contrast to other approaches to the study of *text productions ? Some earlier definitions placed content analysis in the domain of *semantics, such as Kaplan's (1943) definition of con­ tent analysis as "the statistical semantics of political discourse." The restrictions of both political discourse and semantics are unwarranted. Berelson (1952) described content analysis as "a research technique for the objective, systematic, and quantitative *descrip­ tion of the manifest content of communication." But this restriction to "manifest con­ tent" is either ambiguous or overly limited to denotative content. A joint definition de­ veloped by Stone et al. (1966) and Holsti (1969) defines content analysis as "any tech­ nique for making inferences by systematically and objectively identifying characteristics of •:•messages." As amplified by Holsti, the direction of the inference may be ''sign/*ref­ erent, sign/sign, ''sender/sign, and sign/receiver, in these ways projecting to major domains within semiotics. The communications research context, however, imposes a rather different proce­ dure from that usually reported in semiotics journals. Often faced with vast amounts of text, the communications analyst devotes considerable attention to issues regarding sys­ tematic sampling. Statistical inference generally presumes independence between the cases sampled, a condition violated when a subset of the materials comes from one source, a chain of commonly owned newspapers, or perhaps journalists from a common background in journalism. Much effort thus goes into developing explicit sampling rules to yield an unbiased representation of the universe being portrayed. A few content analysis studies have been performed on an entire universe of documents (such as a case study of a diary), but even idiographic research tends to be sampled (e.g., Paige 1966). Similarly, an emphasis on "objective" procedures for making inferences has led to a preoccupation with defining coding rules so that the resulting content analyses can be reliably replicated by other researchers. This of course results in the use of fixed catego­ ries that do not anticipate all the special circumstances that merit scholarly attention.

Content Analysis

1 49

While it may represent a blessing for some, this "coding reliability" (considered essen­ tial for a "scientific" study) is a curse of insensitive inflexibility to others. Although content analysis research tends to be quantitative, it should be noted that this aspect of the definition does not necessarily involve assuming ordinal or ratio mea­ surement scales. Judgments may be nominal classifications, and fairly often are. Yet it is expected that even the simplest classification procedure will be described carefully enough so it can be replicated. Occasionally, procedures have been specified to an exactitude where the actual application can be carried out by a computer. The computer may simply identify the presence or absence of the message characteristics rather than be employed to carry out any elaborate quantitive procedures. This classification of words or themes usually involves a dictionary for assigning words to categories, or better yet for assigning word senses of homographs (Kelly and Stone 1975). Nondictionary procedures include such indexes as ''type/token ratios or the use of multivariate analyses to identify empirically clusters (e.g., Iker and Harway 1969). Krippendorff (1980) argues that the Stone/Holsti definition of content analysis is deficient in that while it "recognizes the inferential nature of identifying the forms of ideas, values, and attitudes to which content analysis attends, it does not make explicit the importance of relating the classification, categorization, and frequency counts of these forms to other phenomena." While it may be debated whether, as a prerequisite, "content analysis stands or falls by its categories" (Berelson 1952), it is clearly evident that the validity of content analysis often fails to be established for lack of attention to context. In examining research designs, this is partly reflected in whether content anal­ ysis measures of text are explicitly related to nontextual data, such as repeated survey measures (e.g., Stone and Brody 197 1 ) or economic indexes (e.g., McClelland 196 1 ). Isolated content analyses, no matter how well carried out, are likely to offer few useful insights and be difficult to interpret. One test of contextual ''relevance is whether or not the analysis holds up in court. Inferred motivation from content analysis has a poor court record, for even though one source produces significantly more messages of a certain type, this does not mean, for example, that the author was an agent of foreign propaganda (Lasswell et al. 1965 ). Authorship tends to be identified on the basis of minor stylistic frequencies (e.g., Mos­ teller 1964 ), rather than thematic content analysis. The argument that a textbook has a biased effect because it does not mention some minority groups as often as do other text­ books is simpleminded, leading to an absurd formula approach (literally of tokenism) as to what constitutes bias. One of the more successful acceptances of content analysis in court concerned a fed­ eral trial in which the plaintiff asked that the Mississippi state committee accept a second state history text in addition to the one already used. Inasmuch as one of the commit­ tee's own criteria was "to make the student aware of his rich inheritance from the past," the two books were compared in terms of their unqualified references to "Mississippi­ ans." Each time the term was used, a check was made to see if the referent could have included the ancestors of black children as well as white, thus fostering an awareness in both races of being Mississippians. Obviously a sentence such as "After the Supreme Court's desegregation decision of 1954, Mississippians took vigorous measures to re­ sist" does not elicit much identity from black students. In these ways, it could be shown that the implicit messages in the adopted text were contrary to the state's own explicit

150

Content Analysis

objectives (Stone 1979) ; the analysis contributed toward making a successful case (New York Times, 4 April, 1980). The importance of adapting text research to consider different facets of the context has been argued not only from within the content analysis tradition, but by some neigh­ boring traditions. For example, on the sociological side, Cicourel ( 1 980) argues for con­ sideration of the organizational context in which discourse takes place, while Bourdieu ( 1 977b) argues for contextual considerations, including the authorization of language, its profit of distinctiveness, and its symbolic capital. An emphasis on immediate context is particularly characteristic of different ethnomethodological traditions, particularly that version known as "conversational analysis." Other critiques of content analysis have often reflected a strong anti-ideological bias, especially from Marxists who view �·ideology as an epiphenomenon, justifying their stance on an interpretation of �·Marx and Engels' The German Ideology. While this bias continues in some schools of sociology, it has been tempered considerably by such Marxists as E. 0. Wright ( 1978), who consider ideology to be selective within, a range of opportunities provided by a socioeconomic setting, as well as reflecting that setting. Rather than attempt Hegelian explanations, many content analysts today are more interested in monitoring the heterogeneity of grass-roots ideologies in different subcul­ tures as they relate to other factors. Thus, there has been an interest in developing cultural indicators to complement the monitoring already provided by social and eco­ nomic indicators (Stone 1 972 ; Firestone 1972). In addition to the earlier examples of Sorokin 1 937 and Lasswell, Lerner, and Pool's Comparative Study of Symbols (1 952), recent cultural indicators include Gerbner 1979 and very longterm series analyses by Namenwirth ( 1973) and Weber ( 1 98 1 ) . Cultural indicators based on very large amounts of text may be aided by the recent availability of flexible optical readers and better computer data base management techniques (Stone 1980). With textual records going much further back in history than most other data for creating social indicators, various indicators based on mass media, school readers, trade literatures, and so on may be increasingly important as they become available. It should be noted that in the process of producing such cultural indicators the com­ puter at most assists in addressing problems of coder fatigue rather than providing any major innovative capabilities. The application of many categories to large amounts of text, so as to be able to record a significant absence (e.g., Johnson, Sears, and McConahay 197 1 ) as well as the presence of text characteristics, has made fatigue a major obstacle in the content analysis field. While it might be expected that the sophisti­ cation of content analysis would soon be enhanced by resources developed in the field of '�artificial intelligence, such as by Schank and Abelson ( 1 977b) or Lehnert ( 1 978 ), the more successful illustrations of computerized natural language analysis so far apply only to rather limited domains. There is a marked discontinuity from the world of blocks on a table (Winograd 1 972) or from disaster stories (Schank, Lebowitz and Birn­ baum 1980a) to a more comprehensive coverage of media. Obviously such large endeavors as cultural indicators could benefit from construc­ tive advice from several of the traditions concerned with text analysis, especially those traditions willing to entertain at least some validity to aggregate measures. In a very large academic environment, such as the United States, it has been convenient for the dif­ ferent traditions to ignore each other. As illustrated by Rosengren ( 1 9 8 1 ), the situation has been much different in Scandinavia, where not only are different traditions more

Co-text and Context

151

likely to be in contact, but where scholars are more fluent in reading different literatures and a society has been more progressive in underwriting cultural indicators. Contribu­ tors to the Rosengren volume, such as Lindkvist ( 1 9 8 1 ), attempt to supplement the com­ munications model of content analysis with lessons form the ,:•mythem clustering of Le­ vi-Strauss, the •:•semes theory of Greimas ( 1 966a), the semiotics model of Eco ( 1 968), the associations theory of '�Barthes ( 1 975b ), the ideologem theory of Kristeva ( 1 969), and the �-hermeneutic circle of Ricoeur (1 974b). Yet, such an integration is but a begin­ ning, for with the contributors' attentions primarily given to confronting Marxists, they give less attention to semiotics and little to ethnomethodology. The future interface of semiotics and content analysis can indeed be a helpful exchange if the differences in the backgrounds and goals of the two traditions are recog­ nized. With their concern for aggregation, including statistical sampling, content anal­ ysts must pay a price of a certain insensitivity, just as almost every economic indicator, such as the GNP or the price index, includes some very arguable and insensitive deci­ sions in order to obtain aggregate measures. Rather than attempt to bury this inherent paradox, the two traditions may together assist in addressing it realistically. P.J.S.

Co-text and Context The two terms, co-text and context refer, respectively, to verbal environment and situa­ tional environment. Adoption of the term co-text (by Petofi 1971 b) stresses the distinc­ tion beween those elements which are instrinsically textual (or intratextual) and extra­ textual elements : the latter being proper to extensional •:•semantics (':•possible worlds, etc.) and to *pragmatics (production, reception, interpretation of the *text). The distinction corresponds to two stages of research on Textlinguistik. The one turns its attention to the regularities of the text, to its coherence, the other, to what ties the text to its pragmatic context. In the first phase, what is defined, apart from grammatical and syntactical rules, is all the transphrastic techniques which guarantee the coherence of the text. The co-text is arrived at when the pattern of the relations which tie together the thematic links of the '�sentences has been achieved. It must be added that the co-text embraces, alongside a textual grammar, a non-grammatical com­ ponent wich is concerned with questions of meter, ryhthm, euphony, etc., all of them involved in the text. A second phase brings to the fore the implicit reference which the text, using to this end extensional semantic values, makes to cultural notions (the so-called "encyclopae­ dia"), to logically organized collective experiences (�·presuppositions and implications) and, in general, to the overall expectations of the interlocutors ; implicit reference of this kind has the added advantage of keeping text length within reasonable limits. It shall be added that the environment is not completely involved in the context, since "the con­ text . . . is a theoretical construct, in the postulation of which the linguist abstracts from the actual situation and establishes as contextual all the factors which, by virtue of their influence upon the participants in the language-event, systematically determine the form, the appropriateness or the *meaning of the •:•utterance" (Lyons 1 977 : 572 ). Prob­ lems relative to the production and the reception of the text also fall within the province of the con-text : they are problems of text pragmatics. It may prove possible to arrive at a typology of texts by combining a contextual analysis based on non linguistic criteria,

1 52

Counterexample

and co-textual analysis based on aspects of the text susceptible of theoretical description (Petofi 1975: 124-125). In the real situation of text production, it is not so much a question of succession as of interrelation between co-textual and con-textual aspects. Petofi is engaged in the elaboration of a partial text theory that is logic-orientated, and which has as its object the relationship that exists between the '�structure of the text and the extensional inter­ pretation (in terms of possible worlds) of the world, or worlds, a given text textualizes (Conte 1977). Petofi (1974) is convinced that, within his partial theory of the text, and in terms to the interrelations of grammatical (co-textual) and semantic-extensional (con-textual) elements, it should prove possible to analyze and describe those semiotic elements that �-Morris divided into *syntax, semantics and pragmatics, elements which, in practice, are always found intertwined. C.S.

Counterexample The method of counterexample is a proof procedure, combined with a '�semantic approach, that may be used for deciding whether an argument is valid or not. The method can be applied in *sentence logic and in �·predicate logic, though in the latter it must sometimes fail, as predicate logic does not possess the property of decidability. By a counterexample is meant a model that satisfies the premises but not the conclu­ sion of the argument in question. An argument is valid, i.e., has a demonstrative force, if it admits no counterexample. Suppose that Xis a set of premises and C is a conclusion. If the argument is not valid, then there must be a counterexample and we shall find it when supposing nonvalidity, i.e., taking formulas X and non- C as our assumptions. Then we transform these under suitable inference rules until we obtain all the simplest compo­ nents of them, i.e., either atomic formulas or the denials of atomic formulas. These ulti­ mate components refer to instances that are compatible with the supposition that X is true and C is false, hence they provide the required counterexample and confirm that supposition. It may happen, however, that the systematic search for a counterexample results in a contradiction, that is, a pair of formulas such that one of them is the denial of the other. Then, since the claim that there is a counterexample leads to the contradiction, no coun­ terexample can exist, and the argument thus tested proves to be valid. The procedure of splitting up assumptions into ever smaller subformulas gives rise to a tree structure in which the smallest components, i.e., either atomic formulas or their denials, are end points. The idea of constructing a proof in tree form goes back to Gentzen 1934-35, it has been adopted by Beth (1955), Hintikka (1955), and others. Beth provided a method he called the semantic tableaux, the name being adopted because of the occurrence of two parallel columns. The left column, labeled "valid," contains the premises and their successive consequences, while the right column, labeled "invalid," contains the denial fo the conclusion together with the consequences of that assumption of invalidity, both chains of consequences ending with the simplest formu­ las. The contradiction arises if the same formula appears in both the column for valid and that for invalid formulas. Then the tableau is said to be closed. If a tableau remains open after arriving at the simplest formulas on both sides, these formulas form a counter-

Counterexample

15 3

example, that is, they provide a valuation that shows the inference to be invalid. An example will be given below after another form of tableau is discussed. A tableau may be simplified by reducing it to one tree if each formula occurring in it is prefixed by a '�symbol corresponding to either the assumption of validity or the assumption of invalidity, e.g., T for "true" and F for "false. " A more transparent form is obtained when the prefix "false" is replaced with the negation sign and the prefix "true" with the empty sign; both of these forms are called analytic tableaux. The following analytic tableau is a proof of the formula ('f/ x)(Px=i Qx) =i (('f/ x)Px­ =i ('f/ x) Qx) , and simultaneously a proof of validity of the inference from ('f/ x) ( Px=i Qx) to ('f/ x)Px=i ('f/ x) Qx) : 1 2 3 4 5 6 7 8.1

(Vx)(Px -=,Qx)

assumption, premise assumption, negation of conclusion 2 by (~°J) in section 3 . 1 below 2 by (~-=,) 3 by ( V ) in section 3.2 below 4 by (V ) 1 by (V) 7 by (-=,) in section 3 . 1 below

~((Vx)Px -=,(Vx)Qx) (Vx)Px

~( vx)Qx

Pa ~Qa Pa-=,Qa ~Pa 8.2 Qa

The vertical space in 8 denotes the branching to 7 into alternation 8 . 1 and 8 .2 . The double horizontal lines mean that the tableau is closed, since 8 . 1 contradicts 5 and 8 .2 contradicts 6. The corresponding semantic tableau is to be drawn as follows (the numbering of lines is identical with that adopted in the analytic tableau above) :

Invalid 2 (Vx)Px-=,(Vx)Qx

Valid 1 (Vx)(Px-=,Qx) 3 (Vx)Px 5 Pa 7 Pa -=,Qa 8.2 Qa

4 6

(Vx)Qx

Qa

8.1 Pa

A set of rules is needed to specify the possible *transformations resulting in either closing a tree or finding a counterexample. The set of rules displayed in Figure 1 has been chosen as being particularly handy and almost self-explanatory (cf. Bell and Machover 1 977). Each rule will be presented in the form of a tree that shows the struc­ ture of that part of a proof resulting from the application of the rule in question. The lo­ gical connectives in parentheses are adopted as names in rules.

(~ ~) ~ ~ A A (&) A&B

I

A B Figure 1 .

(~&)

~(A&B)

/�

~A

~B

(v)

(~ v) ~(AvB)

I

~A ~B

1 54 ( ::>)

Cryptology

A ::::>B

I�

~A

(~ ::>)

B

Figure 1. ctd.

J

~(A =>B) ~B

)\

A B

~A ~B

The technique of branching has been devised to deal with disjunctions. Thus, for example, two alternative courses of reasoning are generated by ( ~ &) : if a conjunction is not true, then either the first member or the second member (or both) is false, and both alternatives are to be taken into account. ( :::) ) holds that either A is false or B is true (or both), hence there are again two possibilities to be explored. In the case of ( = ) and ( ~ = ) as well, two alternative courses of inference should be adopted. For predicate logic the following set of rules can be introduced : (E) (Ex)A(x) provided that the individual constant c, which is to replace X A(c) throughout A, does not occur in any earlier line of the proof (in a formula different from A). (~V)

~(Vx)A(x) ~A(c)

(V )

(Vx)A(x) A(c)

(~E)

~(Ex)A(x) ~A(c)

with the same proviso as above.

The two latter formulas do not require the restriction added to the first two. The reason for such a restriction is as follows. Suppose that in the argument in question it has been proved that there is an object x that satisfies the condition A, that is (Ex)A(x). We can then introduce an individual object c by stipulating "let c be such an x" and writing A(c). If we subsequently demonstrate that there is an object x satisfying another condition B, we are not allowed to say "let c be such an x," since " c" has been chosen to stand for some x satisfying A(x) and we do not know whether there is any x satisfying both A and

B.

For a proof of the semantic completeness of such a system see Smullyan 1968. W.M.

Cryptology

Cryptology is the science of secret *communications. As a social creature, man com­ municates. As a private one, he does not, fearing rightly that others' knowledge of his intentions may enable them to frustrate those intentions. This noncommunication is secrecy, and secrecy is the element that distinguishes cryptology from other aspects of semiotics.

Cryptology

155

The silence of an individual is a special case. Under consideration are the cases in which several people wish to limit to themselves knowledge of what they are saying to one another - to be silent, in other words, toward outsiders. They may also want to pre­ vent outsiders from deceiving them by injecting purportedly genuine *messages into their communications. The various means used by them to keep their communications secret from others or free from tampering are called ,:- signal security. The means used by others to gain '�information from these communications are called signal intelligence. Together they constitute cryptology. Signal security consists of communication security, which deals with written, oral, or pictorial communications, and electronic security, which deals chiefly with radar and telemetry. Signal intelligence consists of communication intelligence and electronic intelligence, whose division of work parallels that of the security function. Electronic security and electronic intelligence will not be dealt with further here. Communication security has three parts: (1) Preventing others from knowing of the very existence of a message. Nonelec­ tronic forms of this are called steganography. Technological steganography includes invisible inks and photographic reductions of a typewritten sheet to the size of a printed period (microdots). Linguistic steganography includes open �·codes, which replace the important words of the true message with innocuous-sounding phrases, and semagrams, which convey the secret message by innocuous-looking objects, such as the position of heads in a drawing. Electronic forms of concealment are called transmission security. An example is spurt or burst transmission, in which a message is compressed (as by a tape recorder) and radioed at high speed. (2) Preventing others from gaining formation from the routing and volume of messages. This is called traffic security and serves mostly for radio communications. Examples are changing the call signs of radio stations and sending dummy messages. (3) Preventing others from gaining information from the actual ,:•texts of messages. If the messages are written, either in ordinary languages or in those of computers, the methods are those of cryptography; if the messages are spoken, the methods are those of cryptophony; if pictorial or gestural, those of cryptoeidography. Cryptography converts the letters, the binary digits, or other elements of the original message (the plaintext) into secret form in two ways. One is transposition. In this, the plaintext elements (usually individual letters) are jumbled : street might become RETIES. The other is substitution. In this, the plaintext elements are replaced by letters, numbers, or '�symbols. Thus street might become TUSFFU or & + ) ?? + , or, in more complicated systems, 48 77 18 14 82 15. Substitution and transposition may be combined. These systems are ciphers. A cipher transforms plaintext elements of uniform length and is nonphonetic (a cipher will divide a t from an h in a th, for example). Cryptologists differentiate a cipher from a code, which is a special form of substitution. A code consists generally of a book-length list of plaintext elements of varying length, usually words and phrases, and their corresponding substitutes, normally four- or five-element code groups, such as 3964 or PEDIV. Thus "await further instructions" might be represented by 2 1 563. Substitution ciphers with long lists of equivalents merge into short codes, and no sharp dividing line may be drawn. Cryptology uses the term code as do linguistics and semiology, the only difference being that cryptologic codes are secret because their dis­ tribution is restricted.

156

Cryptology

The distinction between codes and ciphers has an analogue in cryptophony. Using a secret language such as Pig Latin, or a foreign language where third parties do not understand it, parallels code. Using a scrambler to distort speech parallels cipher and the process is called ciphony. Scramblers operate upon the electric current from a micro­ phone. They can replace the high speech frequencies by low, or can record speech on a tape and pick off portions in mixed order for transmission. The same parallel exists in cryptoeidography. Tramps' ,:•signs form a code; television scramblers, used in subscription broadcasting, are the analogue of ciphers. The terminology used in communication security includes the following : putting a plaintext into secret form with the use of a cipher is "enciphering, " with a code is " en­ coding," with any cryptosystem is "encrypting." Restoring a ciphertext, codetext or cryptogram to plaintext legitimately (that is, with the authorization of the users) is "de­ ciphering," "decoding," or "decrypting." Communication intelligence consists of (1) intercepting messages and, for radio transmitters, locating their position; (2) analyzing traffic to determine *senders and recipients and volume; (3) solving enciphered or encoded messages (cryptanalysis). Cryptanalysis is made possible by the fact that the frequency of letters in a language is remarkably stable. If one counts the letters of several different texts (all in the same lan­ guage), each of some 200 letters or more, all will show about the same percentage of ds, of !Is, and so on. (Claude Shannon, in a paper applying *information theory to cryptol­ ogy, said that this regularity was due to the '�redundancy - of sounds, grammar, and relations expressed - that inheres in all natural languages.) If, then, in a simple substitu­ tion cryptogram, one ciphertext letter is seen to be more frequent than the others, it probably is the substitute for the most frequent letter of the language. More complicated systems require more complicated methods to reduce the cryptograms to where this method can be applied, but it remains the basic method of cryptanalysis. To restore a cryptogram to its plaintext without the authorization of the communi­ cators, often reconstructing the cryptosystem in doing so, is to "cryptanalyze," or "solve," "break," "crack," or "read" it. Specialists avoid the confusing practice of using "decipher," "decode, " or "decrypt" to mean unauthorized reductions to plain­ text. Some things that appear to be cryptologic are excluded from the science for a variety of reasons. *Kabbalah, gematria, and the "discovery" of the signatures of Francis *Bacon in Shakespeare's plays educe ambiguous messages: all persons will not achieve the same results, as they would in solving or decrypting a true cryptogram. Rebuses, rid­ dles, and crossword and other languages puzzles and games involve a secrecy that is intended to be penetrated. The "decipherment" of lost languages likewise is not cryptol­ ogic because no secrecy was intended (though cryptanalytic techniques may help in the decipherment). For the same reason, Morse code, semaphore, code flags, though they convert a message into a different alphabet, are not cryptologic because no secrecy is intended; the same holds for artificial languages. On the other hand, the use of language to obscure '�meaning, as in politicians' speeches and allusive references by confederates, may be regarded as cryptology, although, because of its lack of rigid prearrangement, as marginal. Voltaire's comment that "Men use . . . speech only to conceal their thoughts" is a statement - though barely - in cryptology. Cryptology's two great historical tendencies are ( 1 ) generalization and (2) seculari­ zation and quantification.

Culinary Semiotics

157

Cryptology is a child of or form of communication ; it has grown as communication has, and its great stages are those of communication. As a society becomes literate, cryp­ tology arises, apparently often spontaneously. At first it serves only modest private ends, such as concealing spells or *divinations in writings on magic. When voluminous govern­ ment correspondence through a hostile environment emerged in the West with the estab­ lishment of permanent embassies during the *Renaissance, cryptology became an instru­ ment of state. The telegraph, which engendered heavy military traffic, fostered the field cipher. Radio, with its easily intercepted signals, forced all military communications to be encrypted. But the enemy could sometimes break the cryptosystems and read the mes­ sages. World War II provides many examples of the advantages gained thereby. The bat­ tles of Midway and of the Atlantic were won to a considerable degree through the help furnished by the Allied solution of Japanese and German cryptograms. More recently, the vast expansion of telecommunications, including networks for computers, together with the decreasing cost of microprocessors and the growing fear of interception, has enormously generalized the use of cryptology. Cryptosystems pro­ tect business messages and data in transit and in storage more widely than ever before. Subscription television has put "decoders" into thousands of homes. The economic incentive of all this has, incidentally, drawn many mathematicians, engineers, and entre­ preneurs to the field and thereby shattered the virtual monopoly that governments have long had on cryptology. Following and accompanying these widespread and more general applications, cryptology, like other fields of knowledge, became first secularized and then quantified. It has, in other words, gradually replaced the occultism that long permeated it with the more efficacious rational modes of thought, and then made these more precise and more powerful by applying �·mathematics wherever possible. Though valid cryptosystems have always also served useful ends, cryptology was for centuries stained with mysticism because it concealed magical formulas, employed the same strange signs as astrology and alchemy, mingled in books with charms and angelic salutations, manipulated symbols like gematria, and resembled divination in extracting meaning from mystery. But as society divorced itself from religion, cryptology purged itself of occultism. One of the most important steps in the history of cryptology, the architect Leon Battista Alberti's invention of polyalphabetic substitution around 1466, adapted for profane use a device that the medieval philosopher �·Lull had intended for religious uses. The growing use of cryptology for mundane purposes secularized it, re­ moving most of its pseudoscientific overtones. Then, in the late nineteenth and the twen­ tieth centuries, mathematical tools increasingly improved both cryptography and crypt­ analysis. Problems that once were handled intuitively were quantified and treated mathematically, achieving results never before attained. Cryptology's use in and study by computers has enormously accelerated quantification. This long evolution has washed cryptology of all but its last tinges of mystery - and, some would say, of roD.K. mance.

Culinary Semiotics Semiotic investigation of food-related practices is of mainly anthropological and lin­ guistic focus. Since cooking performs a selective and categorical �·transformation of

15 8

Culinary Semiotics

naturally produced substances, it organizes the conceptualization of nature-''culture contrasts, thereby providing a source of tropes for the expression of social experience. Fundamental organizational principles of food habits, Levi-Strauss' (1958a: 99) "gus­ temes," reproduce the structural characteristics of society. Semiotic/semiological investigations of food habits treat these as encoding identity. Levi-Strauss' (1962a) explanation of totemic *symbols as bonnes d penser replaced the earlier functionalist ascription to them of basic alimentary properties as a primary moti­ vation; the symbolic organization of natural goods was thereby shown to provide a vehicle for categorizing cultural entities. This approach transcended earlier distinctions between primitive and civilized ; it allowed •:•Barthes ( 1957 :7 4-79), for example, to treat red wine and rare beefsteak as "totems" used as a means of signaling French national identity and symbolizing national character. Barthes (1957 : 128-30) also considered as significant the degree to which journalistic representations of food forms might represent the economically unattainable. In these formulations, food provides a •:•meta­ language for the projection of cultural and social *ideology. Louis Dumont ( 1966) suggests that cooking creates conditions of marginality, as a process which transforms the natural into the cultural, and that food is symbolically ren­ dered more permeable to pollution by it. Levi-Strauss' ( 1966a:390-4 1 1) "culinary trian­ gle" (raw/cooked/rotten) more systematically explores the semiotic properties of culi­ nary practices as a model for social ideology; the triangle is subject to modulation (de­ grees of rawness, etc.) to express complex transformations of social category systems (es­ pecially those governing the exchange of women). Such transformations may be apprehended diachronically. Levi-Strauss' (1968b: 396) absolute, •:•synchronic pronouncements on the difference between French and American perceptions of degrees of putrescence in cheese, for example, are comple­ mented by Aron's (1973) documentation of shifting perceptions of the same feature in France. Levi-Strauss' (1970) remarks about attitudes to mushrooms, furthermore, suggest the importance of historical experience for the retention of symbolic associa­ tions between edible forms and cosmological concepts ; Barthes ( 19 57 :208 ), by contrast, treats all "mythological" forms, including that of food, as a dispossession of historical consciousness (cf. also Levi-Strauss' [1964:24] view of •:•myths as machines for the sup­ pression of time). Food appears as a semiotic of sexual relations in at least two major ways. First, food production may furnish tropes of procreation. A notable instance is the "cheese anal­ ogy," discussed by Ott (1979) for modern Basque as well as ancient Greek thought. The blending of substances (here, milk and rennet) to produce a finished product allows men to reproduce women's reproductive role symbolically (cf. the couvade). Leach and Leach (1977) suggest that the residual etymology of Turkish kzsir links it to female bar­ renness within a larger system of food prohibitions and sexual associations. Second, roles of food consumption may replicate the social regulation of sexuality. Leach (1964) argues that degrees of edibility correspond to degrees of acceptability in classes of marriage-partners ; by contrasting English with Kachin usages, moreover, he shows that this correspondence is subject to an encompassing cultural �-rhetoric in each case. Similarly, Dumont's (1966) analysis of Indian pollution concepts shows that com­ mensality and endogamy, though linked, are not necessarily subject to identical restric­ tions or interpretations: castes are endogamous, but food-sharing rules regulate both internal relations and relations between the castes (see also below, on "gastro-poli­ tics").

Culinary Semiotics

159

The social implications of dietary *rules have been extensively analyzed by Douglas (1966, 1975), who argues that they are an epiphenomenon of the rules by which a society classifies and regulates itself internally. This argument is developed with particular subtlety in "Deciphering a Meal" (Douglas 1975 :249-75 ), where she also suggests that an external threat to group survival may lead to dietary conservatism even when the original ideology of purification has become attenuated. In an extension and modification of this approach, Galaty (1979) suggests that members of marginal catego­ ries may have alternative perceptions of the status of particular foodstuffs, and may follow a divergent reading of culinary symbolism for ideological reasons. Culinary rules are capable of considerable manipulation in social practice, as any shift in emphasis from formal •�structuralism to semiotic ethnography must recognize. Food may express a seemingly fixed classification of the social world; this is demonstrated in Aron's (1973) treatment of nineteenth-century French cuisine as expressive of the same formal properties as the Lamarckian evolutionary taxonomy, and of a social dominance hierarchy that excluded women as rivals to the pleasures of food. The diachronic perspective of Aron's analysis shows both individual resistance to this system and long-term change. The concept of food as a "metaobject" of social relations (Loveday and Chiba 1985) necessarily entails the possibility of such maneuverability (cf. also Chang's [1977 : 19] "food *semantics"). Important ethnographic support for this contention includes Chang's (1977 :16) discussion of food preparation as an expression of intimacy, against a background of strict ritual concerns of a culinary nature (1977 :17-19), and Appadurai's (1981) concept of "gastro-politics. " These approaches relate *structure to *action and choice. Much of the discussion of food as constituting a ' "�language" might be more effec­ tively framed in terms of "discourse : " (cf. also Monod-Becquelin 1977 :154-55). This mode of �-description incorporates the dynamic implications of formal structuralist anal­ ysis; see, for example, J. Courtes' (1973 :41-75) critique of Levi-Strauss' culinary trian­ gle. A focus on discourse also permits a critical dissection of the myth-science dichotomy : Detienne's (1972) analysis of Pythagorean and other ancient Greek con­ cepts of consumption and pollution shows a striking convergence in these two areas of discourse, and places the historical development of scientific concerns with ingestion within a larger semiotic complex. Aron's (1973) discussion of the partial mutual exclu­ siveness of food and conversation in nineteenth-century France justifies his characteriza­ tion of the eater as a "semiologist" - i.e., a decoder with a choice of •�codes (food, sexual dalliance, conversation). Culinary semiotics cannot be treated as a purely autonomous domain of inquiry if it is not to be reduced to a sterile formulation. The focus must remain ethnographic, to permit an exploration of the transformations of structure in verbal and other forms of discourse. Since food is not invested with identical significance in all cultures, more inclusive organizing principles of classification have been sought with some success; these, however, should not be divorced from the indexical uses of food and food-related practices in the management of social interaction and the diachronic transformation of the encompassing principles themselves. M.He.

160

Culinary Semiotics

Linguistic Aspects of Culinary Practices All cultures have cooking practices, and unsurprisingly, all languages have cooking words. There is a rough correlation between a number of paradigmatically contrasting words in a language and the complexity of the culinary practices.French and Chinese, for example, the languages of cultures reputed for their cuisine, have more words than the languages of Australian aborigines. However, the absolute number of words is not particularly important, since distinctions in a language can be made syntagmatically. English, which incidentally has more different words than either French or Chinese, paradigmatically contrasts roast, bake, fry, broil, boil, and steam, whereas Spanish uses cocer al vapor 'cook with steam' and cocer en horno 'cook in an oven' in place of steam and bake respectively. Languages may differ with respect to their hierarchical depth. English has four levels in its taxonomy (Figure 1), while Amharic has two (Figure 2 ) and Yoruba three (Figure 3 ). Figure t. English

J

cook steam boil1

broil� roast, bake

fry

simmer

I I

poach stew braise

boil2

deep-fry saute French- grill barbeque charcoal fry

From Lehrer 1974.Shaded areas indicate overlap of meaning. Figure 2. Amharic

besse/,e 'cooke' fella 'boil'

k?ek?k?el/,e 'boil'

gagger 'bake'

t?ebbese 'fry roast'

Data is from Susan Hoben. Angle bracketS indicate selection restrictions.

k?olla 'parch'

Figure 3.Yoruba

se1 'cook' se2 'boil' I

yan 'roast, bake'

din 'fry'

bp Data is from Bascom 1951.

ta 'roast'

sun 'roast in fire'

bu 'bake in ashes'

Culinary Semiotics

161

The cooking word taxonomies are not perfect ones since there is sometimes partial overlap in meaning (partial �·synonymy) and partial hyponymy. English roast overlaps in meaning with broil on one hand and with bake on the other. A large chunk of meat cooked under a broiler or on a rotisserie could be described as broiling or roasting, while the same chunk of meat cooked in an oven could be described as baking or roasting. This example brings up an additional property of cooking words : the choice of items depends to some extent on '�lexical collocation or selection restriction, that is, what food is to be cooked. If a ham is cooked in the oven English speakers use bake, while if a chunk of beef is so cooked, they use roast. In Japanese, niru is a general word for 'boil,' while yuderu is used for boiling rice. In Amharic (Figure 2) each cooking word is associated with a particular class of food, and in Yoruba (Figure 3 ) many words are lexically restricted. German toasten is used just for bread. English toast is similarly restricted, but perhaps collocations such as toasted grated coconut would be acceptable. A common phenomenon in semantic structures is that a word has both a general and a specific sense, and one finds this frequently in culinary word taxonomies as well. Eng­ lish boil (Figure 1 ) can mean roughly 'cook in water' (boil 1 ) , or it can contrast with simmer, where it refers to a full boil (boil2 ). A *sentence such as Cook the soup on a low flame but do not let it boil illustrates the second, more specific sense. A number of lan­ guages use the same word to mean both 'to cook' and 'to boil.' German kochen, Polish gotawac, Yoruba se, Hebrew bashel, and Serbo-Croatian kuvati all mean 'to cook,' thereby heading the lexical/taxonomy, and 'to boil', serving as a hyponym . It is also instructive to see how the cooking words are related to neighboring semantic domains. In some languages the word for 'cooked' also means 'ripe' : Jacaltec tahce 'to cook, to ripen,' Chinese sh6u-te-tung 'cooked, ripe,' and Walbiri yinmi 'cook, ripe.' The general semantic concept seems to be 'ready to eat.' Japanese yaku 'to bake or roast' can also mean 'to burn,' and Yoruba bu 'to roast in ashes' also means 'moldy.' Although cooking is a universal activity, there is considerable variety in cooking methods, and one would expect the words in each language to be culture-bound. How­ ever, a sample of two dozen languages from ten different language families reveals a number of generalizations, enabling one to state implicational universals (in the sense of Greenberg 1963b): ( 1 ) If a language has two or more cooking words that divide the semantic domain, one will refer to boiling. A ve ry few exceptions to this generalization can be found, primarily in the languages of cultures where boiling was not traditionally used as a cooking method, as among the aborigines of Australia and New Guinea. (2) If a language has three or more cooking words, one will refer to boiling and the nonboiling domain will be subdivided. However, the way in which the nonboiling domain will be divided is not predictable. Some languages Qapanese, Chinese, Polish, Navajo) combine cooking in an oven, over an open fire, or under a broiler, using the same word for all of these methods. (3) In general, if people in a culture cook food in fat, there will be a seperate word for this process, e.g., English fry, French frire, and Yoruba din. Chinese and Japa­ nese do not have a generic term for 'fry' ; instead they have two different words which distinguish deep-frying from stir-frying. ( 4) If steaming is a common cooking practice in a culture, the language is likely to have a separate term, where 'steam' contrasts with 'boil'. Chinese, Japanese, Eng­ lish, German, and Persian are such languages.

162

Cultural Unit

(5) A frequent contrastive feature found, but less generally than those above, distin­ guishes between direct and indirect heat: bake vs. broil (or grill in British English). Jacaltec and Amharic also make this contrast, as does Yoruba, but at a lower level in the taxonomy. The most important contrastive features, those listed above, are closely connected with the results - how the food tastes after it has been cooked. This fact may not seem surprising, but one can imagine other possible and plausible lexical structures which do not occur. For example, cooking in water and cooking in oil are never incorporated into the same word in contrast to other methods. Presumably boiled food and fried food are sufficiently different. Some languages (Navajo, Jacaltec) have several verbs of eating, and the appropriate verb depends on the shape of the food eaten. But such contrasts are not carried over to the cooking words. Although fasts and feasts are culturally impor­ tant, languages do not have different cooking terms for such occasions. Orthodox Jews strictly separate meat and dairy products in eating and cooking, but neither Hebrew nor Yiddish has separate cooking terms corresponding to 'cook (Meat)' versus 'cook (Dairy).' Other aspects of culinary semantics include words for aspects of food preparation besides the actual cooking, such as words for cutting ( cut, chop, grind, mince, dice, slice, mash, or quarter) and words for various kinds of utensils (pot, pan, grill, skillet, griddle, or Dutch oven). A.L.

Cultural Unit In every culture a unit "is simply anything which is culturally defined and distinguished as an entity. It may be a person, place, thing, feeling, state of affairs, sense of foreboding, fantasy, hallucination, hope or idea. In America culture, such units as uncle, town, blue (depressed), a mess, a hunch, the idea of progress, hope and art are cultural units" (Schneider 1 968:7). In this sense the notion of cultural unit is crucial in every structural approach to semiotic phenomena: a structure is a system of units selected according to certain criteria of *relevance and inserted within a system of oppositions. In phonology, phonemes and *distinctive features are "'femic" units. �-Hjelmslev ( 1 943) proposed to analyze the plane of the content as if it were organ­ ized as the plane of expression, and attempted to identify content units. In the framework of an intensional semantics in the format of a componential analy­ sis, all the semes, or semantic properties, or semantic markers composing a sememe, are cultural units. According to different authors, cultural units can be identified with the whole sememe (the expression cat corresponds to a given cultural unit "cat," as opposed, let us say, to "dog" or "tiger" in a given zoological system) or with the elementary compo­ nent of the sememe, semes (the expression cat stands for the cultural units "animal, feline . . . "). Cultural units can be isolated either according to a ,:-dictionary-like repre­ sentation, or to an encyclopedia-like representation. Eco (1976a) considers cultural units as components of the content of expressions, in an encyclopedia-like semantics, and identifies them with �-interpretants. P.M.

Culture

1 63

Culture The first formal theory for the semiotic study of culture was that presented by Ivanov, Lotman, Uspenskij, Piatigorsky and Toporov at the Slavic congress of 1 973. They coined the phrase semiotics of culture for a science where culture was defined as a domain of organization (*information) in human society, in opposition to disorganization (en­ tropy), i.e., an integrated hierarchical arrangement of *sign systems. The basic unit of culture, in this sense, would be the "'ftext, " vehicle of '}function and *meaning. A "cul­ ture" would be the sum of its texts. To each culture, there would correspond a type of chaos (non-culture) which, from the viewpoint of an observer immersed in that culture, appears as non-organized, though to an outsider would rather appear as differently organized. To this category belong : alien cultures, childishness, exoticism, the subcon­ scious, the pathological, etc. The term text is applied not only to 'fmessages in natural *languages but to any sub­ stratum of a significant whole such as a rite, a work of art, a musical production or a public performance of any sort. On the other hand, not all messages in natural lan­ guages are texts from the viewpoint of "culture," but only those pertaining to a definable verbal �·genre (such as prayer, 'flaw, novel) i.e., possessing a certain meaning and a common function. The authors of the theory recognize that no isolated semiotic system could be called a "culture," as this term implies a hierarchy of linked systems of meaning. They suggest that linkage could be mediated in large measure by the system of the natural language. Let us note, however, that the authors set up their theory to apply specifically "to Slavic texts." They do not claim that it applies elsewhere, but implicitly recommend semioticists working in other cultures to investigate principles of hierarchical ordering and "linkages" between texts and semiotic systems elsewhere. This challenge has so far been taken up to only a very limited extent. On the other hand, a century of tradition in the discipline of cultural *anthropology, in the United States and Western Europe, has yielded many theories of culture that are partially or almost wholly semiotic. It would therefore appear that in these countries the development of a "semiotic theory of culture" would inevitably occur in close dialogue between semioticists and anthropologists, and that the latter would at the outset have to be given considerable authority in view of their vast apparatus of theory and data. With regard to this anthropological input, semioticists will, however, have at least three major reservations about its value in yielding a viable "semiotic theory of culture :" (1) many anthropological theories of culture are not holistic or define "culture" to include elements that could not be called "texts" ; (2) even when they define culture as made up of "texts" in a semiotic sense, the ultimate end of their text analysis is often to set up "a sociological model of actual social relations and activity" or to suggest causal explanations "in terms of subsistence needs" (Galaty 1 98 1 :65 ) ; (3) anthropological the­ ories suffering from neither of these inconveniences (such as, to a large extent, the ,:•structuralism of Levi-Strauss 1950, 1962b, 1 964, 1 966b, 1968b, 1 97 1 a) are apt to avoid any serious reference to cultural boundaries and emphasize intercultural '}communica­ tion instead. Such differences in approach cannot, of course, be expected to disappear as long as anthropologists find their present theories scientifically profitable. These theories go back, historically, to a phase when it was their principal task to compile records of vanishing non-literate cultures. Until fairly recently, "culture" was

164

Culture

therefore, to anthropologists, "that complex whole which includes knowledge, *belief, art, law, morals, custom, and any other capabilities and habits acquired by man as a member of society" (Tylor 1871, 1878). Such an approach called for very systematic data collections, but the order or organization to be found in these data derives mainly from general evolutionary postulates (Tylor 1878). Boas moved the study of culture much closer to a semiotic perspective, for he saw cultural phenomena in terms of the imposition of conventional meaning on the flux of experience, and as determinants of our very perceptions of the external world (Stocking 1968). In order to study these phenomena, Boas compiled innumerable texts which make up almost the whole of his written work. He insisted that each culture was distinctive in the way it assigned meanings to phenomena (Boas 1896) but did not admit the possi­ bility to integrating these specific semiotic systems into an ordered whole. Culture was, he argued, a matter of shreds and patches (Boas 1966). The American configurationist school of the 1930's did admit such a possibility but had much difficulty in finding a valid ordering principle for cultural systems. Benedict (1934) believed that the ordering principle in a culture is its "dominant principle" as established by an external analyst, on the basis of data obtained in that culture. Several others (e.g., Kardiner [1939] ; Mead [1963]; Kluckhohn [1945, 1956] ) used the psychological concept of "basic personali­ ty" as ordering principle, but this was only one of several methods all aiming at discovering the "dominant principle" of cultures. Goodenough believed that cultures are organized by their "belief systems" (1963 ), while Turner (1967) looked for his ordering principle in ritual '�symbols, Schneider in a normative system of meanings (1968, 1976) and Geertz in •:•ideology (1973). They and their numerous followers increasingly developed semiotic methods of enquiry which led to much brilliant and ele­ gant analysis of cultural "texts." Is it possible to integrate these well-analyzed semiotic domains into a "semiotic of culture" by ordering these domains hierarchically without too many intuitive leaps? A number of suggestions to this effect have been made, including two rather impressive ones by *Bateson (1972c) and Dumont (1966) , but it is too early to say whether these will find wide acceptance. The reason why the suggestion of the Soviet scholars (to treat the natural language as the chief ordering principle of semiotic systems) has found few recent Western advo­ cates is probably that this suggestion tended to become identified with the ill-fated "Sa­ pir-Whorf hypothesis" (*Whorf 1956 ), implying that the grammatical and �·semantic •:•structure of one's own native language confines one to certain patterns of thinking. Quite free of such an implication is Keesing's recent (1972, 1979) theory of culture. He regards natural language as the primary sign system that articulates the individual with the shared knowledge of the culture. This theory deserves to prosper, but it rejects the possibility of a global ordering principle, i.e., the possibility that the cultural knowledge of a human being can be described in the form of a "cultural grammar" or analyzed as a self-contained system. A similar viewpoint has been cogently put forward by Sperber in polemical essays (1974, 1975, 1979) in which he challenges claims by Turner, Douglas (1975) and various others to have been able to establish such systems. One problem for theories of culture is to explain practices in conformity with a scientific or materialistic perspective. Many theorists have done this very directly, as though social effects can be shown to follow necessarily from natural causes, from tech­ nological or economic conditions. This was the assumption of L.H. Morgan, as well as

Culture

165

Steward (1955), Vayda (1967) and of Rappaport (1967, 1979). Here the principle of cul­ tural order coincides with the natural one of adaptive advantage. Or again, theorists have followed Durkheim's sociological *epistemology in which "the first logical catego­ ries" (ordering principles of the culture) are social categories: it was because men were grouped (as clans or moieties} that in their ideas they grouped other things (1903, 1912). Such are not, of course, semiotic theories of culture. Sahlins (1976: 1 16) rightly remarks that Durkheim formed a sociological theory of symbolization but not a symbolic (i.e., semiotic) theory of society. This means that he did not base his sociological findings on a systematic enquiry into the Aborigines' own perception of their universe. Yet a theory of culture acceptable to a semiotician would at least require that all information (sociological, environmental and other) should be presented in the form of a text couched in a secondary language of bearers of the culture. This requirement by no means excludes the consideration of sociological or envi­ ronmental (or technological) causes of cultural practices. It has, in fact, long been part of the tradition of British �-anthropology (Evans-Pritchard [1937, 1965 ], Leach [ 196 1 , 1972, 1976], and their many followers) to analyze very closely the structural patterns embedded in *mythology and ritual exegesis, and then to apply this text analysis to func­ tional explanation. This is, indeed, also the practice of Geertz and Turner, though their theoretical formulations are more purely semiotic, and the viewpoint of Rappaport since 1979. Conversely, we cannot class as "semiotic theories of culture" the vast array of the­ ories by often egregious authors who draw a firm distinction between "symbolic *action and "�·pragmatic or empirical action" (Firth 1973 :427) as though the latter were con­ ceivable without the former, for nothing is more pragmatic than to look before you leap, and you r�ad signs when you look (symbolic action). Finally, by no means all semiotically oriented theory accepts the concept of culture as a basic analytical category. Among those who accept it explicitly we may quote Geertz (whose "thick description" elicits a specifically cultural *code) or Turner, Sahlins, Schneider. Opposed to this view is the perspective in which "any two cultures are seen as superficially different representations of one abstract culture, human culture" (Se­ beok 1968a:4). In Mead's concept of "total communication" (1964) and even more in Bateson's *cybernetic epistemology, whose concept of "�·mapping" is surely semiotic, the chief emphasis is in the transmission of information accross boundaries. Levi-Strauss, who defines culture as "a set of symbolic systems within which lan­ guage, marriage rules, economic relations, art, science and religion have preeminence" (1950:xix), and who is deeply convinced of the blessings of cultural diversity, has nonetheless rarely conducted an inquiry confined within the limits of a single culture. His theory is based on the observation that cultural knowledge is ceaselessly being com­ municated across cultural boundaries, that it is transformed as it travels, and that the key to the working of the human mind lies in the pattern of these �-transformations. If we compare these anthropological theories of culture with the theory of the Lotman school discussed below, we note that these theories are bound to differ according to whether they are evolved from the perspective of research largely within one's own culture (the case of Lotman school) or from the perspective of research in exotic cultures (the case of anthropological theories). The next step in theory develop-

166

Culture

ment may well be that of rc::cognizing the complementarity of these two perspectiE.S. ves. Moscow- Tartu School

Semiotic Definitions of Culture According to the theory of culture elaborated by semioticians working in the Soviet Union, culture is the totality of the signifying systems by means of which mankind, or a particular human group, maintains its cohesiveness (its values and identity and its inter­ action with the world). These signifying systems, usually referred to as secondary ,:•mod­ eling systems (or the "languages" of culture) comprise not only all the arts C:·literature, *cinema, '�theater, *painting, �·music, etc.), the various social activities and behavior pat­ terns prevalent in the given community (including gesture, dress, manners, ritual, etc.), but also the established methods by which the community preserves its memory and its sense of identity (�·myths, �-history, legal system, *religious beliefs, etc.). Each particular work of cultural activity is regarded as a �·text generated by one or more systems. The foundation and central axis of this notion of culture is natural language. Besides providing the "raw material" for many secondary modeling systems, natural language is the only means by which all the systems can be interpreted, memorized, and taken into the consciousness of the individual and the group. Because of its primacy, language is referred to as the primary modeling system, the other cultural systems being termed "secondary. " The relationship between primary and secondary systems may also be considered ontogenetically (the child acquires natural language prior to its acquisition of other cultural languages), and analogously (secondary systems are constructed in the same way as natural language, or methodologically may be so regarded). Cultural systems and natural language are referred to as "modeling" systems: that is, they are the means by which man cognizes, explains, and attempts to influence the world about him. They are, to use other terms, the means by which man processes, exchanges, and stores information about the world. The notion of modeling thus implies both the reworking and the communication of information (information being understood in the widest sense to include not only knowledge, but also values and �-be­ liefs). All human cultures comprise at least two secondary modeling systems: most usually these are a verbal art (myth for instance) and a visual art (painting), i.e., a symbolic system and an iconic one. This universal binarism of human culture has been related by Ivanov to the binary structure of the human brain. Beyond this universal binarism, how­ ever, every culture hierarchizes its secondary modeling systems in a different way. Some cultures may give precedence to literature (for instance Russian nineteenth-century cul­ ture), others to a visual art (television and cinema in modern Western culture), yet others to music, and so on. Culture then can be regarded as a complex hierarchized structure of interrelating secondary modeling systems. Culture is further to be understood as the positive term in the ,:•opposition culture/ nonculture. If culture is the organized system of systems which preserves and renews information for a community, then nonculture is disorganized, destructured entropy (that which obliterates memory and destroys values). Particular cultures will have their own notions of nonculture according to their position in the world and their outlook: it

Culture

167

may be simply "them" as against "us " in all its historical and racial variations or it may be more sophisticated notions such as the unconscious vs. the conscious mind, nature vs. culture, chaos vs. cosmos, the world beyond ''signs vs. the world of signs (in each case the second term is the one endowed with positive value). On the importance of "noncul­ ture" as the structural reserve for the evolution of culture, see § 5 below. The semiotics of culture as outlined here was first formulated by members of the Moscow-Tartu semiotics group at the fourth summer school on secondary modeling sys­ tems (Lotman 1 970c) and was further elaborated in the Theses on the Semiotic Study of Cultures (Lotman 1 973d). The work of Lotman and Uspenskij on the semiotics of Rus­ sian cultural history since that date have given practical exemplification of these princi­ ples. Typologies of Culture Within the framework of the semiotics of culture, work has been done of the typologies of cultures (Lotman, Uspenskij ). Cultures may be classified and compared according to the way by which they hierarchize their secondary modeling systems, according to notions of nonculture, according to their concepts of time and space, or by attitude to ''semiosis itself. Some cultures show marked orientation toward origins, others toward ultimate goals. Some cultures operate with a circular (mythic, or cosmological) under­ standing of time, others with linear (historical) time. Different cultures variously con­ ceptualize themselves geographically, marking off "our world" from "the unknown" or alien worlds. These different orientations may be manifest in particular texts, or par­ ticular secondary modeling systems, or may be universalized within a culture as the dom­ inant ideology. By their attitudes to semiosis, cultures may be categorized according to whether they give prevalence to '"'expression" or to "content, " that is, whether the expression of an already given ,:•truth is accorded greater value than the process of discovery of truth. The former category may also be seen as "text-oriented" (giving value to established texts) and the latter as "*rule-oriented" (giving value to the *generation of new texts). Cul­ tures may be classified as "paradigmatic" (all phenomena are signs of some higher reali­ ty), or "syntagmatic" (meaning arises by relationships between individual phenomena, not by reference to a higher Meaning) : the high degree of semioticization in medieval culture is an example of the former, while the eighteenth-century Enlightenment, with its desemioticizing tendencies (preference of nature over convention), is an example of the latter. The Communicatory Functions of Culture It is an assumption of the semiotic approach to culture that culture is a mechanism for processing and communicating information. Secondary modeling systems operate with conventions (or codes) that are shared among members of the social group. Unlike nat­ ural language, however, where broadly speaking an identity of code may be assumed among speakers of one language community, the codes of secondary modeling systems are variables acquired to varying degrees (or even not at all) by the individual in the course of his maturation and education.*Noise (in the sense of any of a number of impeding linguistic, psychological, or social factors) may block or hinder the communi­ cations �·channel. So universal, indeed, is the fact of imperfect communication that it

168

Culture

can be seen as part of the very nature of culture. All cultural exchange involves therefore some act of �-translation as the �-addressee interprets the message of the addresser through his only partially shared code(s). The fact of partial communication, even at times noncommunication, within a culture stimulates the formation of an ever-increasing number of new codes to compensate for the inadequacies of the existing ones. This "proliferation" factor is the impetus to the dynamism of cultures. The Metalanguage of Culture The •:•metalanguage of a culture is the principle that organizes, hierarchizes, and defines a culture for itself; it is in this sense that ideology or set of values which, expressed though one or more secondary modeling systems, gives a culture its stability and its model of itself. As with any act of �-description, the metalanguage simplifies its object, rejecting what is destructed, extra-systematic, and thereby to some extent distorting its object. (It follows that no culture can be scientifically described solely from the point of view of its own metalanguage.) The metalingual tendency is thus a counter to the "proliferation" tendency of a culture to multiply codes in compensation for the inadequacies of cultural communication. The Dynamism of Culture The capacity of a culture to change and to adapt is a function of the interplay of the metalingual and the proliferation tendencies inherent in every culture. The tendency to proliferation is the result both of the need to compensate for inadequate communication and of the need to process and circulate the ever-increasing quantity of information acquired by the culture. However, if the proliferation of codes takes over, cultural cohe­ siveness is lost and communication in fact becomes impossible ; on the other hand, if the metalingual function takes over, culture atrophies, change is impossible, and communi­ cation becomes unnecessary. Changes in culture come about from the drawing into cul­ ture of elements from the destructured, noncultural "periphery" which acts as structural reserve and which is unacknowledged by the metalanguage. However, as it incorporates these changes, the metalanguage itself evolves. Within a culture, different secondary modeling systems evolve at different rates : since each system has its own metalanguage (the languages of criticism for the arts, •:•so­ ciology for social behavior, mythology for myths, and semiotics of culture for the overall functioning of culture), the overall pattern of change is repeated at different rates within each secondary modeling system. In cultures of a high degree of complexity, such as modern cultures, the role of the creative individual (the artist) in the invention and renewal of codes is particularly important. The greater the complexity of a culture, the greater the significance of each individual as structural node in the overall system. The fact of the essential dynamism of cultures gives greater weight to a diachronic description of a culture than to a •:•synchronic one that must be less adequate. A.S.

Cybernetics

169

Cybernetics Derived from the Greek kubernetes 'steersman,' the term cybernetics was first used by Andre Ampere to denote the science of government (la cybernetique) . It was redefined by *Wiener (1948) as "the science of control and �·communication in the animal and the machine." In this sense it embraces '�information theory and computer science, pointing particularly to their applications in neuroscience and in the theory of intelligent automa­ ta. Whereas physics is concerned primarily with the flow of energy and the determina­ tion of one force by another force, the central concern of cybernetics is the flow of infor­ mation and the determination of one form by another form. It thus finds common ele­ ments of interest across the whole field of telecommunications and human and animal communication. It also covers the theory of self-regulating and goal-pursuing systems of all kinds, from simple servosystems to the complex hierarchic or heterarchic '�struc­ tures that shape the behavior of a living organism, an ecosystem, or a human society. Two things unify this ramifying subject. The first is the concept of information as a quantifiable entity or commodity. For example, Shannon's (1948) proof that random errors can be automatically corrected by a system with sufficient *redundancy was soon shown to have powerful implications not only for the design of reliable telegraph '�codes and computing machines but also for our understanding of human language (Cherry 1957), genetic reproduction (Yockey, Platzman and Quastler 1958), and the organiza­ tion of the brain itself (Cowan and Winograd 1963). Following Wiener and Shannon, the information-content of a form-determining event is usually measured by the reduc­ tion in statistical uncertainty that it brings about. This depends on the range of alterna­ tive forms that the outcome was equally likely to have taken, and is thus formally analo­ gous to entropy as defined in statistical thermodynamics. It is a mistake, however, to equate the two concepts, which are dimensionally quite disparate (MacKay 1969). The second unifying factor is the concept of evaluative feedback. This means more than a simple, passive back-reaction, such as a floor offers to a weight, or a pendulum bob to a displacement. It implies a process whereby the state of the field of *action is compared or evaluated against an internal criterion or standard, and the resulting *signal is arranged to bring about an active adjustment of the field in a direction calculated to improve the evaluation. Unlike a simple Newtonian reaction, cybernetic adjustment may be much more powerful than the change giving rise to it. (This is so because cybernetic control normally relies on amplification, a process whereby the spatiotemporal form of a small amount of energy or matter determines that of a much larger amount.) The optimal adjustment also depends typically on the rate of change, and indeed on the whole past history of the system under control. In common usage the term cybernetics has suffered from associations with somewhat unrealistic claims for the powers of digital computing machines, the prospects of imitating natural mechanisms in automata ("bionics"), and the like. In the long run, however, especially in its adjectival form cybernetic, this term seems likely to meet an enduring need for which no other existing term is adequate. D.M.M.

D Dance Dance, as generally found among Homo sapiens, has presages in evolution. Dance is composed, from the dancer's perspective, of purposeful, intentionally rhythmical, and culturally patterned sequences of *nonverbal *body movements other than ordinary motor activities, the motion having inherent and •·aesthetic value. This conceptualization encompasses common characteristics in the kinds of phenomena different groups through historical time and across geographical space consider to be dance. Human dance is a form of thought and •·action. Compared with other animal non­ verbal *communication, dance serves a wider variety of purposes with greater complexity in open •:•lexical, ''semantic, and *syntactic systems. Humans have a greater potential for motor variety, control and learning, and also for creative manipulation of patterns within ''rules, so that individuals and groups can select a variety of ''styles and •·structures. Although nonhumans may be trained to perform human-like dances, there is no evidence that they can create •:•meaning and transmit to other animals dance sequences that are physically, affectively, or symbolically complex. Nor is there data that nonhumans reflect upon dance and leave something representing it that lingers beyond their lifetime, e.g., drawings, sculpture, notation, or ''film of dance. The human body, the instrument of dance, is the locus of •:•transformation of both inner and outward experiences. The dancer draws upon innate •·cognitive structures, acquired memories, physical maturation, and social and cultural ''message exchanges to externalize inner thoughts and feelings. Consequences follow from self-perception and the response of others to the performance. Thus dance is a social fact embedded in human evolution. Given that meaning derives from social interaction, shared knowl­ edge about the dance and experience in its use or observation affect the degree of under­ standing a dance performance. The efficaciousness of dance to move information is due to its use and capture of the sensory modalities : the sight of performers moving in time and space, the sounds of physical movement, the smell of physical exertion, the feeling of kinesthetic activity or empathy, the touch of body to body and/ or performing area, and the proxemic sense. The extraordinary nature of dance often enables it to be arresting or seductive com­ munication. Communication may involve a dancer sending messages to him or herself through kinesthetic proprioceptors, to other dancers, and to observers. For communication to be most effective there should be shared knowledge about the dance form and experience in its use ; shared notions about when, where, how and why messages are sent; and information sufficiently lucid to be perceived through distractions or impediments. Understanding depends upon the interplay between skilled dancing and sensitive percep­ tion. Meaning in dance can be found in three interrelated domains : the ''pragmatic, the semantic, and the syntactic. The pragmatic domain encompasses antecedents, •·ideology, and consequences as well as the encounter between the dancer and other members of society. The dance reference is primarily to its social environment (structure, irregular

Dance

171

relations, and individual and group aspirations). The purpose of dance may be to com­ municate in any realm of human life: *culture, economy, polity, religion, sex, or aesthet­ ics. The consequences of dance fall into the broad categories of moving people to social action (conservative, conciliatory, or innovative) and altered states of consciousness. In the domain of semantics (the relation of ,:•signs to what they signify, information content, and the substantive nature of motional patterns), there are at least six modes or devices for conveying meaning that may be utilized in dance. Each device may be con­ ventional (customary shared legacy) or autographic (idiosyncratic or creative expres­ sion of a thing, event, or condition). ( 1 ) A concretization is a device which produces the outward aspect of a thing, event, or condition, for example, mimetically portraying an animal. It is an imitation. (2) an ,;icon (used in Redfield's sense) represents most properties or formal characteristics of a thing, event, or condition and is responded to as if it where what it represents, for example, dancing the role of a deity which is revered or otherwise treated as the deity. (3) A stylization encompasses somewhat arbitrary gestures or movements which are the result of convention, for example, pointing to the heart as a sign of love. ( 4) A metonym is a motional conceptualization of one thing for that of another, of which the former is an attribute or extension, or with which the former is associated or contiguous in the same ,:•frame of experience, for example, a war dance as part of a battle. (5) A *metaphor expresses one thought, experience, or phenomenon in place of another which resembles the former to suggest an ,;analogy between the two, for example, dancing the role of a leopard to denote the power of death. (6) An actualization constitutes an individual dancing in terms of one or several of her or his usual statuses and roles, for example, Louis XIV dancing the role of king and being so treated. The devices for encapsulating meaning seem to operate within one or more of eight spheres : ( 1 ) the sociocultural event and/or situation; (2) the total human body in action; (3) the whole pattern of the performance; (4 ) the discursive aspect of the performance ( the sequence of unfolding movement configurations - this may include syntactics) ; ( 5) specific movement; (6) the intermesh of movements with other communication *media (for example, dance meaning is inseparable from song, *music, costume, accoutrements, and/or speech) ; (7) dance movement as a vehicle for another medium (for example, dance is merely background for a performer's poetry recitation) ; and (8) presence. A dance may employ single or multiple devices and spheres. Using the analogy of verbal language, dance generally conveys meaning more like poetry than prose. There are several pervasive signs. Those of identity refer to ascription, e.g., sex, age, gender, and family, as well as achievement, e.g., marking rites of passage and social dramas with their conflict, mediation, and resolution. Dance signifies roles, world view concepts, and transformations of wished for results in everyday life. Some analysts impose meaning upon dance, others seek the perceptions of the dance participants, and yet others rely on what participants do, what they say, and the relationship of reports and actions to other semantic systems. The domain of syntactics focuses on the rules dictating how signs in dance may be combined, the grammar of how the realm of movement style and structure is related to the realm of meaning. Syntactics is critical to aesthetics viewed as notions of appropri­ ateness and competency. In this context, meaningful motor units are based on the archi­ tecture and function of the body. A discursive ,;description and analysis to identify dance structures would comprise dimensions and their values (amounts) of space (direc-

1 72

Dante Alighieri

tion, level, amplitude, focus of eyes and body, grouping of dancers, shape or contour of movement design of body), rhythm (tempo, duration, accent, meter), dynamics (straight lines or curves in space, unimpeded or bound flow of energy, locomotion, pro­ jectional quality produced by combination of elements), and characteristic use of body (posture and movement that activate or are largely supported through the whole body, locomotion, gesture) . Units of meaning or structural units may be found in any move­ ment or phrase - a wavelike motor pattern marked by alternations of activity and rest which cluster distinguishing traits - and the rules of combining motor units within a phrase and with others. Syntactics governs not only the patterns of moving but those of rest and those exhibiting the intention to move. Like verbal language, there is no universal dance language. There are several kinds of structural analyses of dance : notational, stratificational, •:-distinctive feature, and distinctive feature (underlying structure) with surface transfor­ mations. A notator's score records many important features of dance but does not identify the underlying structure or constitutive rules of a given work. Generative de­ scriptions purport to describe all possible elements that go toward making specific or nonspecific movements. Syntactical accounts are often devoid of semantic analyses. Structural analysts rely on their own categories, seek to elicit the dancer's categories, or a combination of both. Studies of dance semiotics usually focus on the pragmatics (the *context), syntactics (the *text), or semantics (the •:-symbol). While a single study must limit its scope, it is nevertheless valuable to view the interrelation of the three domains of meaning relying on participant perspectives and analyst observations as well as both insider and outsider views in relation to other aspects of the dancer's social life. Although the semiotic study of dance is new compared to the application of this perspective to many other subjects, there are already several different conceptualizations and methodological approaches. References to - and evaluations of - these studies of dance appear in Gell 1 979, Hanna 1 979a, 1979b, 1 980, and Royce 1977. J.L.H.

Dante Alighieri ( 1 265-1 3 2 1 ) Dante was familiar with the theorizations o f Modistae logicians concerning the theory of speculative grammar. It is along these lines, as has been recently shown (Corti 1 9 8 1 ), that he formulates his *sign theory in spite of his Augustinian background. In Dante we need to distinguish between theorization, which almost exclusively concerns the verbal sign, and the use that the artist makes of the processes of polysemy and symbolization. His theories are presented in De Vulgari Eloquentia, where the sign is defined as being rationale signum et sensuale ("sensual" in so far as being a significans, and "rational" because it is a significatum) ; it is ad placitum, that is, conventional, arbitrary. The verbal sign is solely human as neither animals nor angels dispose of it, the former lack ration­ ality and the latter are devoid of materiality. In creating the first man, together with the soul, God gave him a forma locutionis: this is not a concrete language but rather the formal cause and general principle according to which language is structured, exactly like the /orma substantialis in specula­ tive grammar. It forms, that is, the set of linguistic *universals. After the episode of the

Dante Alighieri

173

Tower of Babel, God only leaves man with accidental forms which enter into the historical languages. The notion of linguistic universals is left to laborious reconstruc­ tion on the part of the inventoris grammaticae Jacultatis, who distinguish themselves as philosophers and logicians; they are not simply grammarians for whom the term posi­ tores grammaticae is used. Dante develops the need of restoring linguistic universals thanks to the influence of thirteenth century Modistae logicians and, perhaps, to ''Boethius. On the basis of these universal principles, they also offer him the notion of ' "'rule-governed language". Uni­ versal linguistic principles are the modi significandi, the signa of the various expressions of reality. Dante also finds here the necessary theoretical background for his theory of high vernacular which in substance is a theory of the universality of poetic verbal signs. High vernacular gets closer to the ideal of "rule-governed language" than any other ver­ nacular. To use Dante's words, it gets close to those simplicissima signa of language that, together with customs and behavior constitute the measure of the universal within the particular. The question is not merely linguistic in nature but is also sociocultural as the "very simple signs" of poetic language or high vernacular constitute the unifying element within the diversity of spoken Italian; they form the consciousness of national linguistic unity. These very simple signs are also nobilissima signa; such a situation fits into a �·cul­ ture with a very high level of semioticity in which we have identification between the initial stages of any one thing and the sign. In such a civilization, things have value at an existential level while only signs count at a cultural level. The relationship between these two elements is not unidirectional, as there is sometimes transition from things to semiotic structures ( modi essendi et significandi through the modi intelligendi) and from structures back to things. The two-directionality of the process also emerges in Dante's thoughts on poetic language. Stimulated by the needs and experience of the writer, Dante arrives at the simplicissima signa locutionis; this "rule-governed" language is a vivid mental mechanism which in its turn generates poetic ''texts in the same way that speculative grammar generates "rule-governed" languages. In contrast to verbal signs of historical languages, the poetic verbal sign is not ad placitum, that is, it is not conventional, arbitrary. It answers to the principle nomina sunt consequentia rerum, and because of this is not translatable from one language to another, though it does have the characteristic of being a necessary sign. This "necessity" is also due to the fact that the poetic sign is always polysemic. In Epistola XIII to Can Grande della Scala, Dante defined the sensus of the Commedia as non simplex, but polisemos, a word very rarely used in that time and which Dante takes not only from Uguccione da Pisa, author of Magnae Derivationes, but also from Servio's Comment in the first verse of Canto 1, Eneide. The polysemy that Dante attributes to his poetic work is the "pluri-stratified ''semantics" that Lotman and modern semiotics, particularly the Russian and Italian schools, identify as the dominant structure in most literary texts. According to Dante, the verbal signs of poetic texts in general and of the Commedia in particular are encoded several times with *meanings organized at different levels - literal, allegorical, ethical, anagogic; the text is therefore a hypersign carrying maximum information. Dante uses the sign theory of Modistae logicians for his syllogistic reasoning in both the Convivio and the Monarchia. Furthermore we need to emphasize that the whole semantic-terminological field of the sign is present throughout all Dante's works ; not

1 74

Deconstruction

only does the word "sign" occur fifteen times in the Convivio, but the terms "to signify, signification, significance," occur alongside it while being very rare, almost non­ existent, in contemporary literary texts in vernacular. Not only is Dante ahead of his time because of his semiotic theory, he also distinguishes himself by his genial sensibility to the contemporary cultural mechanism in its globality, that great res signi/icata that we know as culture. See: Dante 1979. M.Co.

Deconstruction

Deconstruction is Derrida's term for a critique of metaphysics that demonstrates how the

hierarchical •:•oppositions of philosophy undo, subvert, or "deconstruct" themselves. Since one of Derrida's principal targets has been the conceptual hierarchies in theories of �·language and in structuralist extentions of these theories, deconstruction has a bearing on semiotics. It has also given rise to a type of �·literary analysis : deconstructive criticism is the "careful teasing out of warring forces of signification within the •:•text" Qohnson 198 1 :5). Focusing on the way of literary work subverts the principles of signi­ fication on which it depends, deconstructive analyses (see Brenkman 1976: Chase 1978 ; Jacobs 1978 ; J.H. Miller 198 1 ) elicit subtle and paradoxical accounts of signification by interpreting works as "allegories of reading" (De Man 1979; see Culler 1982 ; Gasche 1979). Derrida's writings undertake a critique of the "logocentrism" of Western meta­ physics - the postulation of an order of �·meaning (thought, �·truth, reason, logic, the Word, the logos) conceived as foundation. Philosophical oppositions such as meaning/ •�form, intelligible/sensible, essence/ accident, serious/ nonserious, transcendental/em­ pirical, soul/•:•body, literal/figurative, are hierarchies in which the first term is prior, independent, and fundamental and the second term is treated as a complication or mani­ festation of the first. Analysis has thus been a matter of returning in idealization, to an origin or to a "priority" seen as simple, intact, normal, pure, standard, self-identical, in order then to conceive of derivation, complication, deterioration, accident, etc. All metaphysicians have proceeded thus . . . : good before evil, the positive before the nega­ tive, the simple before the complex, the essential before the accidental, the imitated before the imitation, etc. This is not just one metaphysical gesture among others ; it is the metaphysical exi­ gency . . . (Derrida 1 977a :236)

To deconstruct these hierarchies is first to invert them by showing that the supposedly independent first term can be treated as a special case of the second. For instance, in discussions of figurative language the terms in which the figurative is defined turn out to apply to literal language as well, thus permitting one to characterize literal expressions as �·figures whose figurality has been forgotten (Derrida 19726 ; De Man 1978, 1979). A major instance of logocentrism is the phonocentrism of theories of language: the devaluation of writing as the representation of a representation and the attempt to con­ ceive of language in general on the model of speech. Derrida argues that to set writing aside as derivative and parasitic is a way of excluding certain aspects of language. If writ­ ing, with its impersonality, distance, need for interpretation, and possibilities of misun­ derstanding, is treated as a derivative or fallen version of language, then one can take as

Deduction, Natural

175

the essence of language an ideal associated with speech : in particular, the experience of hearing oneself speak, where hearing and understanding are inseparable, *signifier and signified seem joined in a ,;sign, and the expression seems bound to the meaning it arises to express (Derrida 1 967a, 1 976b ). One deconstructs the hierarchical opposition of speech and writing by showing that the qualities attributed to writing in the attempt to demote it apply to speech as well. ,;Saussure, for example, denounces writing as a defor­ mation of speech, but his account of language as a system of differences without positive terms leads him to adduce, as the best illustration of the differential identity of linguistic units, the example of writing, thus in effect treating speech on the model of writing (Der­ rida 1 967a). Saussure's concept of a system of differences is powerfully antilogocentric, but the phonocentrism which contradicts this principle is, Derrida argues, essential to Saussure's linguistic and semiotic project. A science of signs depends on the possibility of grasping signifiers and signifieds as positive presences; it needs them as points of departure for analysis of the system. The Saussurean concept of the sign, with its rigid opposition between signified and signifier, is fundamentally metaphysical, though it is deconstruct­ ed by the principle of the purely differential nature of signs (Derrida 1972c). Pursuing that radical insight, Derrida coins the term dif/erance 'difference,' 'differing,' 'deferr­ ing' : the differences Saussure treats as the conditions of meaning are not givens but effects of differing or deferring. In its oscillation between ,;structure and event, diffe­ rance is an impossible object for a science, yet it indicates, Derrida claims, the nature of signification. "Dif/erance is the systematic play of differences, of traces of differences, of the spacing by which elements refer to one another" ( 1 972c:38). Derrida's works also contain deconstructive analyses of ,:-Husserl's theory of signs ( 1967b ), semiotic constructions of ,;Hegel, �-Austin, and ,;Benveniste (1 972b ), and the theory of the arbitrariness of the sign ( 197 4). The self-deconstructive movements identi­ fied in sign theory are not errors that might be avoided or corrected but inescapable correlates of the impossible project of mastering signification. The theory of signs is shown to lead to principles that must be suppressed if analysis is to take place. In addi­ tion to the works cited above, the following treat deconstruction's engagement with ,:•semiotics : Culler 1979; Frank 1980; Weber 1976, 1 978, 1 980. J.C.

Deduction, Natural Any deductive system can be defined as a pair (A, R), A being a set of axioms and R being a set of �-rules of inference. If A is not empty, then axioms and derived theorems are used in proofs as premises ; such a system is called axiomatic, and if, in addition, it is formalized, it is called a logistic system. In contradistinction to the traditional axiomatic approach, originated by ,:-Frege, the method of natural deduction is defined as yielding a system of logic in which the set of axioms is empty and the only tools of deduction are inferential rules. The first systems of natural deduction were devised independently by G. Gentzen and S. Jaskowski in 1934. Jaskowski's investigations were motivated by a problem posed in 1 926 by J. ,:•Lukasiewicz, who called attention to the fact that in practice, ,:• mathema­ ticians do not appeal to logical theorems but make use of other methods. In natural

176

Definition

deduction systems, inferences are drawn from suppositions (assumptions) that are not asserted as theorems but are operating, so-to-speak, as temporary axioms assumed locally within the proof in question; hence the term suppositional proof. The original systems of both J askowski and Gentzen are too involved to be discussed in an elementary account. A simpler version of Jaskowski's system is found in Slupecki and Borkowski 1 967, where the procedures of both direct proof and indirect proof ( re­ ductio ad absurdum) are adopted. Gentzen created two methods of natural deduction (though he called only one of them "natural" ). The method known as sequent calculus proved especially fruitful and evolved into perspicuous and elegant systems adopting the procedure of indirect proof. See also Counterexample, the Method of. W.M.

Definition To define an expression is to introduce it into a language by specifying its �·meaning or its use in terms of those expressions that are already available in the language in question. This formulation covers all varieties of definitions with the exception of the so-called axiomatic and ostensive definitions. Definitions in a more restricted sense are singled out by calling them normal. A definition is said to be normal if it enables us to eliminate a newly introduced expression, by replacing it with older ones, in any '�sentence of the given language. A normal definition is the genuine definition in the traditional sense, hence, when we speak of a definition without any adjective we mean a normal definition. Normal definitions take the form of equivalences or identities. The expression being defined, say E, occurs on the left-hand side, called the de/iniendum, and the terms in which E is defined form the expression on the right-hand side, called the de/iniens. The choice of the form of a definition depends partly on the '�syntactic category of the expression being defined. Some ,:•names and name-forming functors can be defined by identities, while the definitions of sentence-forming functors (sentential connectives and predicates) take the form of equivalences. The following definitions serve as examples ( 1 ) of a name-forming functor, the successor of x : S(x) = x + 1 ; (2) a name: l = S(O ); (3) a name-forming functor (where the definiendum takes the form of an equiva­ lence): x - y = Z = z + y = x; ( 4) a sentential connective : (p � q) = ( ~ pVq); and (5) a predicate : x is even if and only if x is divisible by 2. In the above examples, definitions are formulated without any formal '·'index distin­ guishing them from theorems. Such an index is not necessary provided that its '�function is performed by context. Otherwise we usually write "df" either as the subscript or as the superscript of the definitional functor, i.e., the central functor of identity or of equiv­ alence; this can also be placed either before (often with the number of the definition) or after the definition. In a natural '�language we have at our disposal various forms of definition, for example : ( 1 ) a number is said to be even if it is divisible by two; (2 ) "even" denotes a number divisible by two; (3) "even" means the same as "divisible by

De Iorio

177

two." The first is understood to be an equivalence (though by itself "if" indicates a con­ ditional) since the context "is said to be . . . if" expresses the intention of providing a definition in normal form. W.M.

De Iorio, Andrea (1 769- 1 8 5 1 ) La Mimica degli Antichi Investigata nel Gestire Napoletano (The Mimic Art of the Ancients Investigated in Neopolitan Gesture), written in 1 832 by Canon Andrea De Iorio, is per­

haps the most complex and systematic treatise of kinesics to have appeared before Darwin (1 872). The work aims to present a vast �·phenomenology of gesture through an investigation not only of the internal correlation in the •�sign between a gesture (an expressive configuration) and an idea (the content) but of the complex process of inter­ pretation to which the correlation belongs in its totality, in so far as it is a trail one may follow by way of inferences, hypotheses, and reconstructions to the gestural language of the Ancients. At question is a process in the sign in which one thing stands for another (aliquid stat pro aliquo) according to the modes of an inference (p :::> q) whereby p is a class of perceptible events on the basis of which one infers the absent, the imperceptible. Hence "modern mimic art must consider itself . . . an extremely exact guide to the com­ prehension of the ancient one, for the passage from the known to the unknown" (De Iorio 1 8 32 :4). De Iorio notes the "double" aspect of gesture, stating that "The gesture, or the atti­ tude of any bodily limb, may be considered in two aspects : as the mode in which one physically executes a movement and as the idea bound to it. These two extremely diverse aspects must be considered in such a way that the exact configuration of the first be abso­ lutely necessary to the comprehension of the second" (De Iorio 1 8 32 :3 ). While the word always signals the same denotative content despite variations in pronunciation, the slightest variation in a gesture produces a different •:•meaning. The questions foremost in the author's mind concern the nature of gestural language. He discovers a profound analogy with "alphabetical languages," stating that "Gestures may be deemed the Alphabet of mimic language, and every grammar of any idiom first presents its elementary signs and, then, following the path of their progressive composi­ tion, forges a path to its own comprehension" ( 1 8 32 : 1 ) . The parallel with natural lan­ guages prompts De Iorio to adopt analytical criteria borrowed from a linguistic model. In fact, once the author has executed a division into '�distinctive features, he delves into *syntactic composition of the gesture, its �·articulatory relationship with other gestures within complex narrative sequences, its occurrence in a context of enunciation, and, finally, the rhetorical devices which render the gesture a discourse in the proper sense of the word. The work sets itself forth as an agile instrument for consultation, as a simple guide within the labyrinth of the great "edifice of gestural language. " Structured like a *dic­ tionary, it is articulated in three indexes which reflect the semiotic vision the author has of gestural language. In addition to the general index, the second index in fact orders the gestures alphabetically, as pure expressive configurations, while the third orders meanings, the ideas to which the corresponding gestures refer. The richest, most com­ plex index is the third, in which the author initially proceeds by attempting a •:•definition

178

De Iorio

of the entry and of the various meanings it may carry when joined with other entries and in Neopolitan usage. He then passes to a careful description of the different modes of the gestural expression of each of these signifieds. Every expressive configuration is ana­ lyzed in relation to a context of other gestures and to precise conversational circum­ stances and is studied in its various gradations, which range from the diminutive to the height of expressive violence and indicate, for instance, the greater or lesser intensity of an emotion. Lastly, each gesture's specific ''modality of sign production is analyzed. In this matter, the point of departure is generally the signified : content, in fact, seems to motivate �·expression, by means of ratio di/ficilis. The author in fact rejects an arbitrary relationship between expression and content. This relationship might be called analogical in today's terminology, i.e., a full-fledged iconic representation. De Iorio affirms that "gesture is but the counterfeit of or simile for the internal sentiment expressed by the external" ( 1 8 32:272). Nevertheless, a process of strong, conventional stylization actually selects various properties which are reproduced in the kinesic repre­ sentation, for instance in ''metonymical procedures and metaphorical transpositions. The analogy with verbal language does not apply only to the morphological division of gestural language into prepositions, comparatives, verbal predicates, nominal or modal, nor does it stop at the syntactic articulation. It may also be applied to a rich rep­ ertory of tropes intrinsic to the elocutio of verbal discourse, in other words of all the rhe­ torical devices which render Neopolitan mimic language a persuasive discourse. De Iorio maintains that "The gift of speech would indeed be a poor thing were it limited to the mere manifestation of ideas. And mimic discourse would be equally lacking if it were to fail to move the will of others and to incite the affects" (1832 :204). The device of amplification serves as an example: "When faced with the expression of passions, the Neopolitan is never content. In executing a gesture he includes each of its corollaries, he magnifies it in various tasty ways; to give the gesture greater force he repeats it, he adopts similar ones, he mixes and grafts the first with the second, he interrupts the latter and takes up the former again, polishing all with a thousand insets" (De Iorio 1832 :48). A second example is the sustained use of Irony, which like "a Simulation or Fiction, saying one thing when one intends another, has the effect of changing a habitual meaning into an ironic one" (De Iorio 1832:179). However, certain figures such as metonomy and metaphor go beyond other rhetorical artifices such as the periphrasis or the use of simile and constitute operations in the production of sings which come to our aid when the poverty of language requires an invention, an act of instituting a *code. An example is offered by the representation of abstract nominal predicates "as moral beings do not have gestures naturally destined to express them, it shall always be necessary to fall back upon a metonomy, denoting the category either by effect or cause" (De Iorio 1832:193). Thus, either the sword which provokes death or the shriveled face which is its effect may denote Death. The same shift occurs in transpositions of literal and figural meanings, linked by a reciprocal analogy of a symbolic character: "Numerous gestures could not be explained if not in a metaphorical sense" (De Iorio 1832 :192). One has a metaphor when "gestures express a transposed sense. This transposition of the meaning is not limited to a single idea. The metaphor so takes hold of the imagination of the Neapolitan that he continually adds new figures to the previous ones, . . . he weaves a discourse fulfilled in the metaphorical sense itself" (De Iorio 1832 :39). In so far as it is a "continuous metaphor, " the unified conceptual unfolding of a series of metaphors, Neopolitan mimic discourse takes on the

Della Volpe

179

configuration of a grandiose allegorical fresco. Thus the identity of the allegorical and the figural turns a narrative ''structure capable of establishing the overall order of a representation. This continual play of transfer features suggests that the metaphorical process is born out of an already spoken texture of �·culture : a code is in fact presupposed and forms the basis of *oppositions and similarities which emerge and allow the advance of metaphorical �·transformations. Hence De Iorio's semiotics resembles a form of cultural �·anthropology which recognizes the categorical indices upon which metaphors play and which investigates the historical conditions, the birth, and varieties of the indices. The book is a veritable encyclopedic representation which draws not only on direct observa­ tion but from literary sources, philosophy, previous kinesic theorizations, art history, and, above all, from the traditions of the Ancients, both literary and figurative. Rather than a dictionary work presents itself as a veritable encyclopedia, in which each entry is explained in relationship to all of the other entries through a play of continuous, internal referrals. It is within this theoretical framework, which is also a method, that the hypothesis that one can investigate the mimic tradition of the Ancients through the gestural lan­ guage of contemporaries acquires sense. And through the author's discourse begins with the modern, viewed as ''the first step in the ascent to the piano nobile of the grand edifice of the mimic art" (De Iorio 1 832 : XXIII) in fact what occurs in this treatise of ancient and modern mimic language is that "the one lends a hand to the other so that each may in turn aid the other, both in order to open a new path to the explanation of the ancient and in the interest of adjusting to the modern" (De Iorio 1 832 : XIII). In this way, one mimic language makes itself the *interpretant of the other in a process of continous ''semiosis which can circulate only within an encyclopedic representation of this kind. P. M.

Della Volpe, Galvano (1 895-1968) The work of Galvano Della Volpe is held by many authorities to occupy an important place in philosophical thought. The problematic and productive character of Della Volpe's work is most evident in his contribution to �·aesthetics and, secondly, to linguis­ tics and semiotics. Della Volpe's Critica de! Gusto (1960) provided contact with a number of studies that were little known .or even alien to contemporary Italian culture (from the ''New Criticism to American neo-Aristotelian criticism, from the new *stylistics to the linguist theories of *Saussure and �-Hjelmslev). The Critica de!gusto starts from one basic princi­ ple : the *semantic nature of thought. In the key chapter of the work, entitled "The Semantic Key to Poetry," the semantic nature of thought is specified in terms of the interdependency between thought and �·language, insofar as language is intended as a system of sings, a social-historical institution. From this point of view, poetic discourse, like any other kind of discourse ( ordinary, historical, scientific), is an intellectual-discur­ sive, logicalsemantic ''structure. In other words, it is one form of �·communication among many others, analyzable by reference to the linguistic ''code. Poetic discourse nevertheless has a peculiar feature of its own : in terms of its techni­ cal-linguistic organization, it appears to be characterized by a specific semantic

1 80

Della Volpe

organicity, that is to say by the mutual contextual solidarity between the ,:•symbols that comprise it. These contextual iterations are said to produce the particular meaningful­ ness and communicative richness of poetry, and the *ambiguity and polysemic quality of the organic context taken as a whole. The poetic symbol sets poetry far apart from the equivocal �·meanings of ordinary discourse and the univocal meanings of scientific-his­ torical discourse. It does-so by setting up that semantic dialectia which the "critical paraphrase" has the task of revealing by holding up the poetic ''text against a series of equivalent rewordings expressed in ordinary discourse (following an open-ended and inexhaustible process of gradually making the sense explicit). But despite Della Volpe's insistence that the poetic text is a formal-semantic module, he does not appear to take sufficient account of the organic nature of the poetry. This is not only a question of the organic nature of thought - that question one also finds in nonpoetic works - but of the thought imprisoned in words.) He not only undervalues the semantic weight of the phonic, metrical-rhythmical, and intonational aspects of verbal language, but holds that poetry (or at least great poetry) is wholly translatable, regardless of the features of the linguistic signs of which it is composed. The assumption that poetry is fully translatable is certainly a reflection of his cultural motivation. But his theoretical reasons are questionable. Della Volpe maintains that the arbitrariness, the emptiness, and the biplanarity of the linguistic sign are the concrete features of the *message itself, and, for this very reason, "language wants to be over­ looked" even in poetry. By confusing the sign as a theoretical entity with the sign as an empirical entity (i.e., maintaining that the purely distinctive, differential, and "incorpo­ real" character of the sign is an observable feature of verbal language and hence of poetic discourse), Della Volpe not only undervalues the formal and epistemological claims of Saussure's and Hjelmslev's linguistic theories, but appears to be a victim of an empiricist illusion, as has been pointed out elsewhere (Modica 1978 : 82-85). The final part of Critica de/ gusto, entitled ''Laocoonte '60," is even more problemat­ ical. Here Della Volpe endeavors to move from a theory of poetry toward a theory of the arts, holding that it is possible to identify certain semiotic functions that cannot be translated into verbal language inasmuch as each of them appears to be endowed with a peculiar *cognitive value. What he is attempting to do here is to construct a system (not a hierarchy) of the fine arts, applying the criterion of semantic organicity to nonverbal "artistic languages, " extending the character of emptiness - which is proper to the lin­ guistic sign - to what he calls "the other primary signs of thought, " in other words, the cinematographic, the figurative, and the musical. There are considerable difficulties in attempting this construction, and they imme­ diately become manifest. Della Volpe was not a professional linguist or semiotician. Indeed, quite a few of his proposals are ambiguous, and some do not stand up to close investigation. This does not mean that Della Volpe's contribution to linguistics and sem­ iotics is purely a matter of historical interest, or that it can be reduced merely to his preemptive pioneering work. Compared with his "Laocoonte '60," Della Volpe's attitude in his last work (1968) changes considerably. Not only does he recognize that what he called semantic dialectics is a model serving to interpret verbal material, but he contrasts the rigorous semiotic ana­ lyzability of poetry and music with the nonrigorous analyzability of the *cinema and the figurative arts ; in other words, cinema and ''painting do not constitute a message to the same degree as do poetry and *music, which can be related to an underlying code.

Denotation I Connotation

181

It is true that today we would tend to be much more cautious when tackling the problem of poetry and music. Music, which Della Volpe likened to the category of semiotic systems, following Hjelmslev's hypothesis (Hjelmslev 1943 ), is still a problem. It would also seem that poetry, as some students have shown (Garroni 1977: 1 28-3 3), borrows important features from nonlinguistic operations, to constitute what we define as its "specificity." But on other matters, and apart from a little polemical shortsighted­ ness, some of Della Volpe's ideas can still find acceptance. We are referring not only to his notion of "critical paraphrase," which should be carefully developed further (even beyond Della Volpe's original intentions), but to his slightly self-critical stand against the immediate extention of the object of linguistics to become the object of semiotics in other words, against the application of linguistic categories to objects that do not permit of simple analogous approaches. To a certain extent, perhaps we can say that Della Volpe's objection to the tendency to change semiotics into a "total" discipline anticipated, albeit in embryonic form, the need to thoroughly axamine the status of semiotics as a science. M. Mo.

Denotation/Connotation

These two terms have been used with different meanings throughout the course of semiotic and philosophical tradition. The differences are so great that Geach (1962:55) suggested that the term denotation should be "withdrawn from philosophical currency" since it has produced "a sad tale confusion." Denotation/ connotation are alternatively considered as properties or functions of (1) single terms, (2) predicative sentences, (3) descriptive noun-phrases or definite descriptions. Denotation is used either in an exten­ sional framework or in an intensional one. Depending on the above choice, connotation is frequently used also as affective or emotive *meaning, as opposed to the cognitive one (mother denotes "female parent" and connotes "love" or "protection"). Since the his­ tory of this terminological couple should be identified with the whole history of semiotic thought, one can try to outline only a clear-cut opposition between two principal streams in the speculation about meaning and reference. In *philosophy of language, the denotation of a �·term is usually the set of objects to which it refers, and the denotation of a predicative sentence is the corresponding state of the world. In this sense denotation can be roughly identified with reference (the denotatum of an expression being its referent). It remains to be decided whether the denotatum of a term is a single existing object or a class of objects and whether the deno­ tation of a sentence is what is actually the case or the state of affairs in a •�possible world. In any case, many authors follow the basic distinction outlined by John Stuart *Mill (1 843 ), 1;2,5) : "the word white denotes all white things, as snow, paper, the foam of the sea, and so forth, and implies, or as it was termed by the Schoolman, connotes the attribute whiteness." In this sense an expression denotes the class of individuals of which it is the •�name, while it connotes the property or the properties by virtue of which indi­ viduals are recognized as members of the class in question. If the couple denotation/ con­ notation is taken as synonymous with the couple extension/ intension, then connotation determines the possibility of the denotative or referential uses of an expression: exten­ sion is a function of intension (*Carnap 1955). The following pairs can also be consid-

1 82

Denotation I Connotation

ered as more or less synonymous : denotation/ meaning (�·Russell 1905 ), referent/ refer­ ence ('fOgden and ,;Richards 1923), extension/comprehension (*Port-Royal's logic), breadth/ depth (*Peirce 193 1-66, 2.4 1 8), although certain authors use denotation for the reference to individuals and extension for the reference to classes. It is also commonly suggested that the denotation/connotation opposition corresponds to *Frege's (1 892) distinction between Sinn 'sense' and Bedeutung 'reference,' but the second term has been sometimes translated as 'meaning.' Lyons (1977, 1 :199) remarks that in current German terminological usage what Frege called Bedeutung is replaced by Bezeichnung 'designation' (reserving Bedeutung for 'meaning'). Designation seems to be a less ambig­ uous term for the relation of a ,;sign with what is the case. In the medieval tradition, denotatio, as well as designatio, have been frequently used in order to express an intensional relationship: thus �-Abelard says that voces 'terms' denote or designate their ''definitions, while they significant the corresponding intellectus (as mental concept) ; the reference to things is instead called nominatio or appellatio. For other authors, the intensional relationship is generally called signi/icatio; for *Ockham, significatio acquires both an intensional sense (terms signify concepts) and an extensional one (concepts signify things), but the specific way in which terms within a �·proposition stand for their referents is called suppositio. It was Ockham however to use connotatio to define these terms that, beyond their direct signification, signify also something in a second sense ( Summa Totius Logicae 1, 10). This Scholastic usage has probably influenced linguistic ,:•structuralism. In a structu­ ralistic framework problems of reference are irrelevant since linguists are mainly inter­ ested in the correlation between words and their socially coded meaning: thus the Saus­ surian distinction between ,;signifier and signified has nothing to do with extensional problems. *Hjelmslev (1943) has further developed the Saussurian dichotomy into the notion of *sign-function posited between the two entities ,:•expression and ,;content (both being the elements of two structured systems). He has distinguished between a denotative semiotic "none of whose planes is a semiotic," a connotative semiotic "whose expression plane is a semiotic," and metasemiotic or ''metalanguage whose content plane is a semiotic ( 194 3: 1 1 4), a distinction that can be represented as:

E

denotative semiotic connotative semiotic

E E

C C C E

metasemiotic

E

C C

While in a denotative semiotic an expression plane denotes a content plane, in metalan­ guages the content plane is given by both the expression and content planes of the object­ language. A connotative semiotic is a language where both expression and content of a denotative semiotic become in turn the expression plane of a new content plane. In this

Descartes

18 3

sense connotations correspond to a sort of intensional superstratum which cannot be reduced to idiosyncratic emotive reactions since they belong to a structured content-sys­ tem. Moreover, they are conveyed only by a previous denotative relationship. Hjelmslev's theory proposes a clear formal definition of connotation, but his formal model can be interpreted at least in two different (even though not necessarily contrast­ ing) ways. Hjelmslev seems to have intended by connotators the phenomena depending on stylistic, vernacular, tonemic differences, in the sense in which a given pronunciation or a given usage of certain archaisms can denote membership in a social class, or group. In this line of thought, connotators can depend either on the expression or the content of the basic denotative language. However, *Barthes ( 1964a, 1967b) and in general structuralistic semiotics have con­ sidered as connotations all those second (and more or less peripheral) meanings or con­ tent-units that a given code or subcode assign to the denotative one. Thus the word dog denotes "canine" (or other zoological properties) and connotes, according to given cul­ tural conventions, "fidelity." A skull on the label of a bottle denotes "poison" and con­ notes "danger" or "do not drink it" (see Eco 1976c :2.3). A semiotic representation of the various possible connotations of a given expression should take into account also the so-called world-knowledge of a cultural group, therefore it can be implemented only by assuming a '�semantic (intensional) model in the format of an encyclopedia. Recently Lyons (1977, 1 :208 ) has proposed to use denotation in a neutral way as between extension and intension : "We will normally say, for example, that dog denotes the class of dogs (or perhaps some typical member, or exemplar, of the class), but that canine denotes the property, if there is such a property, the possession of which is a con­ dition for the correct application of the expression." The same flexibility is also advisable as far as connotation is concerned : a given pronunciation can connote "high class," a political expression can connote a certain ideological attitude, the word lion can connote "courage" or "ferociousness." A more precise terminology for each specific extensional or intensional phenomenon will be welcomed, but at the present state of the art denotation and connotation should be considered as polysemic U. E. terms.

Descartes, Rene (1596-1650) Descartes did not deal anywhere at any great length with ,:•language and ,:•signs, but there are several passages scattered throughout his works in which one can find elements that make up a coherent semiotic theory. Though language was not his main concern, his methodology and philosophy of the mind played an important role in the development of seventeenth century grammatical analysis. For instance, it is usually admitted that Claude Lancelot and Antoine Arnauld, the ,:-Port-Royal authors of the Grammaire Generale et Raisonnee (1660; English translation. A General and Rational Grammar, 1753), and Arnauld and Pierre Nicole of La Logique, ou /'Art de Penser (1662) found in Descartes' writings some of the philosophical bases for their approach to language. A theory of the sign is outlined in Le Monde ( The World), a work of Descartes' earlier period published posthumously in 1664. In opposition to most of his contempo­ raries, Descartes considers the word, seen as the basic semiotic unit, as a sign of institu-

18 4

Descartes

tion. For him there is no original natural affinity between words and the things they sig­ nify, yet signwords-sequences of sound-are a means of representing things and natural phenomena (also La Dioptrique 1637, 4 : 68 4; Les Principes de la Philosophie 1 644, 4, art. 197). The three-term relationship sound/ letter-*meaning-thing meant sug­ gests Ogden and Richards' semiotic "triangle" ( The Meaning of Meaning, 1923 ), but it is not historically new since it was already expressed in the Scholastic vox signi/icat medi­ antibus conceptibus ('the voice signifies through mediating concepts'), or even earlier by the Latin grammarian Priscian (ca. 500 A. D.). Descartes extends this analysis to nonlin­ guistic signs such as light, laughter, tears or the pressure of a leather strap, sign being consistently taken in the etymological sense of signum. Words, then, are dualistic by nature insofar as they are sounds (or letters) whose function is to direct the attention to concepts. In this respect language sounds are signs: in medieval parlance, aliquid stat pro aliquo 'something stands for something (else).' Descartes' conception of language - like �-Saussure's in the twentieth century - is to a large extent rooted in Scholastic and even Ancient theories. In this particular case, the Stoics had already distinguished in the word the "•:•signifier" (to semainon) and the "signified" (to semainomenon), that is outer •:•form and inner meaning. Descartes' origi­ nality lies in the analysis of the relationship between what the Port Royal Grammar (vol. 2,) calls the "material" part of the word (sound) and its "spiritual" part (thought). This of course reflects the Cartesian dichotomy between *body and mind. For Descartes, as indeed for the Port Royalists and later Cordemoy (Discours Physique de la Parole, 1 668) or Lamy (La Rhetorique, ou /'Art de Parler, 1676) language, as a system of signs, has a purely instrumental and communicative �·function which is to express, "explain" or "un­ fold" man's thoughts ( Grammaire Generale et Raisonnee, vol. 2.1 ). In fact, language presupposes a thought which is fully structured and in which the three basic operations of the mind - perception, *judgment, reasoning - have already come into play. From one idiom to another the surface structure of language (sounds) may vary, but in deep structure, the mental operations which they express are stable, just like wax may be molded into different shapes without changing in nature ( Second Medi­ tation, 164 1 : 425 ff.). Sounds are variable and arbitrary, not ideas. So each language expresses in a different way identical mental operations whose conjunction constitutes reason. Consequently, the structure of thought is common to all men and reason is uni­ versal. This will form the epistemological basis for the establishment of a universal gram­ mar. The separation of thought and language and the primacy of the former over the latter is exemplified by the fact that we may perfectly well remember what was said to us without remembering in which language it was actually said (Le Monde: I). Another example is offered by the case of the deaf and dumb who, being deprived of the organs of phonation, are nevertheless capable of inventing signs to express their thoughts and communicate with others (Le Discours sur la Methode 1637, 5 : 630). Descartes postulates the existence of mental discourse independently of verbal discourse. It naturally follows that the material part of language plays an ancillary role in Car­ tesian semiotics and that, being variable and arbitrary, it must be submitted to doubt. As signs, words express thoughts, but as material objects, they can conceal or distort the underlying operations of the mind, and this mainly through habit, wrong usage, �·ambi­ guity and a close adherence of the user to the word itself. But this does not entail a funda­ mentally skeptical attitude towards the value and use of words, like ''Hobbes' in the

Descartes

185

fourth of this Third Objections to the Meditations ( 1 64 1 :608). In his answer, Descartes clearly rejects Hobbes' '�nominalism. Neither is he a "realist." As it was already sug­ gested, he does not consider the sign as a direct indication of reality. Language is not the "picture" of the universe, it is primarily a representation of concepts in the mind and only secondarily of external phenomena. In other words, as Leibniz was to put it later, language is le miroirde l'entendement, the mirror of human understanding. This mentalistic view is developed in Descartes' best known and oft-cited passage in the fifth part of the Discourse on the Method (628-632) where he opposes man to the machine and to the animal. His main contention is that there is a radical difference between the man and the animal, the latter being at best a sophisticated machine, like a clock, capable of uttering sounds, but totally unable to adapt itself to new situations through the use of signs. Man alone can "think what he says" and arrange different words together to express his thoughts. He possesses a species-specific capacity which manifests itself in the "creative aspect of ordinary language use" (Chomsky 1 966a :4-5 ) . This is due to the fact that the animal has no soul (mens) and consequently, the soul being the thinking substance, that it cannot think (Letter to Moms 5 Feb. 1 649 : 884-85). The comparison between man and beast raises another problem concerning the nature of signs and their etiology. If the animal cannot think, hence express itself through sign-words, it can at least produce and react to sounds or other signs that do not express thoughts but feelings. And so can man to the extent that, having a body, he is an animal. Descartes distinguishes signs of reason which affect the soul (human lan­ guage) and natural signs which affect the body and excite feelings regarded as "con­ fused thoughts" (Discourse: 630). Yet some natural signs excite thoughts as well as feel­ ings (Letter to Chanut 1 Feb. 1 647 :712) and conversely words excite passions as well as thoughts (Principles, fourth part, art. 197). Thus, towards the end of his life, the clear-cut distinction body vs. soul, feeling vs. thought, natural sign vs. rational sign seems to become somewhat blurred (e.g., Letter to Morus 5 Feb. 1649 :8 84-87). Descartes' attitude towards universal language is interesting in relation to his semiotic theory. It is well known that language planning was generally considered in seventeenth century scientific circles as a real possibility for advancing human knowl­ edge. In a letter to Mersenne of 20 November 1 629 (227-32), Descartes comments on a scheme that had been recently published and he proposes a symbolical system that would represent every idea which the mind can conceive. The purpose of this ideal uni­ versal language is to rearrange thoughts in an orderly way and thus to help man's reasoning and to preclude all errors. It should be remembered that in later years (La Geometrie 1637) Descartes was to elaborate a system of mathematical symbolism; (see also Rules for the Direction ofthe Mind, written in 1628, 1 4 : 1 85-190). Descartes' approach to language has never been investigated thoroughly and it has given rise to conflicting interpretations. According to most commentators, Chomsky's essay (1 966a) offers a distorted view of Cartesian linguistics (e.g., Salmon 1 969; Aarsleff 1 970, 197 1 ; Joly 1977). Descartes' contribution to the development of semiotic and linguistic analysis has to be reevaluated. A. Jo.

1 86

Description, definite

Description, definite A definite description, also called an individual description, is, roughly speaking, a singular ,:"term (i.e., a term denoting exactly one object} that is not a proper ,:•name, given that proper names are construed as mere identifying labels attached to individuals without any reference to their properties. A definite description has a descriptive content con­ cerning that set of properties of an individual that is characteristic of it, that is to say, not shared by it with any other thing. That uniqueness of object is indicated in En­ glish by the definite article, e.g., the father of *Plato, the capital of France, the highest mountain. Apart from definite descriptions, logic also deals with indefinite descriptions, which may be roughly described as those phrases that in some natural �-language are formed by means of the indefinite article. When speaking of a description without any adjective one usually means a definite description, which is the convention followed here. The object referred to by a description is called its descriptum. The technical term definite description was not used until •:•Russell ( 1 905 ) introduced it, but the problems of dealing with singular terms go back to antiquity (cf. Barth 1 974). The first modern formulation of the idea of what was later called "description" is attributed to *Frege (1 893). A description is an expression of the form (tx}( .. x. . ), where the dots around x represent a sentential context of the variable bound by the operator "(t} . " The expres­ sion "(tx}( .. x .. )" is read : the unique object x such that so-and-so. The symbol " ( 1 ) " is called the iota-operator, sometimes written as inverted iota " ( 1 ). " The following are examples : ( 1 ) the description (tx} ( l < x< 3 ) denotes the number 2 , provided that x ranges over integers, (2) the descriptum of the phrase (tx}(x is the father of Plato) is that (unique) male who begot Plato; (3) the descriptum of the phrase (tx}(x is the largest city in England) is London. To define the notion of description the following procedure is adopted. A schematic expression representing description, e.g., (tx}(Ax} , is taken in which A is a variable ranging over sentential formulae. The expression (tx}(Ax) is placed in a context in which something is said, say B, about the individual (tx}(Ax); this yields : B(tx}(Ax). The whole context is defined by declaring it equivalent to another *sentence. The definition is as follows : Df. 1 B(tx}(Ax) for (Ex)(Ax & (y)(Ay :::, (y= x)) & Bx). For instance, when taking x as ranging over integers and taking for A the expression "is a number greater than 1 and less than 3" and for B the expression "is a prime," we read the definiens as follows : there is a unique integer that is greater than 1 and less than 3 , and this integer is prime. The second member of the definiens, (y)(Ay :::, (y= x) }, expresses the unique­ ness condition by saying that whatever satisfies A is identical with x, hence there is only one x such that Ax. In an equivalent formulation, given by Russell, the definition is : Df. 2 B(tx}(Ax) for (Ey)((x)(Ax= (x= y)) & By). W. M. Design The design of *objects has to do with �·semiotics because objects of use can be shown to possess *meaning and to become, under certain conditions, ''signs in a ''communica-

Design

1 87

tion process. There are two basic types of semiotic preoccupation with design - one with designed objects and one with the design process from which they result. Semiotic theories of design objects deal with the �-structure of objects created by the designer, semiotic theories of the design process with the structure of data, design decisions and processes necessary to arrive at a designed object. Semiotic theories of design object have become relevant because they help us to understand the current trend of desemantization in the modern environment (e.g., Menna 1 967-71 ). This loss of meaning is said to be a result of the uniformity and stand­ ardization of object design due to industrial production. The more uniform the appear­ ance of objects serving the same function, the less variety will there be in the world of objects. But uniformity is now even creeping into the design of objects with different functions. For instance in *urban design "modern" buildings with different functions are being constructed according to an uniform "international" �·style. The result is that buildings all over the world are beginning to resemble each other. According to Menna, design has to restore diversity and richness of meaning to the human environment. Greimas ( 197 4) has suggested that desemantization is a general semiotic phenomenon based on the substitution of meaningful human behavior by automated programms (for example, the substitution of the open hearth fire by thermostat-regulated central heating). According to Greimas this process of desemanti­ zation may be compensated for, however, by a process of resemantization (for example, the introduction of fireplaces into otherwise centrally heated apartments). Another design-oriented semiotic object theory is that of Bense (197 1 ; cf. Walther 1 974). He divides all object into four categories : natural objects, technical objects, objects of art, and design objects, by assigning to them different degrees of functional determination. Design objects, like all other "artificial" objects, are constructed and, like all technical objects, are planned, but are not as fully determined in their functions as the latter. Objects in the category of art are the least functionally determined. In the development of his semiotic classification of design objects, Bense ( 1 97 1 ) adopts ''Peirce's (193 1-66) triadic conception of semiotics defining each •�sign in rela­ tion to its material dimension, in relation to the object it stands for, and in relation to its interpretation and use. According to Bense the design object is a special type of sign, in that it realizes a combination of particular characteristics from all three dimensions. It has a material aspect, which he calls hyletic. As an object it possesses a •:•semantic dimen­ sion, which is called morphetic. Finally it exhibits a functional aspect, which is called syn­ thetic. In less complex terminology a design object has the aspects of material, form, and function. Its �·pragmatic dimension (its use), is present in hyletics as well as in morphetics and synthetics : The application of the functional aspects (synthetics) to the material aspects (hyletics) creates its form (morphetics), which is then available for con­ sumption and use. According to the theory, three phases in the creative design process result. In the first phase, the planning phase, the functional dimension is adapted to the material. In the production phase, the object is realized in form. Finally, in the phase of utilization, the object satisfies the human needs which gave rise to it (Walther 1 974 ). Whereas this subdivision of the design process is not specifically semiotic, semiotic theories of the design process should profit from the notion of encoding and decoding. From the point of view of the designer, the design process is a process of encoding properties into an object, from the point of view of the user a process of decoding these properties. If design objects are only in part functionally determined, the question is

188

Design

what makes them different from objects which are completely determined by functional considerations (such as a generator in an automatic pumping station). The difference must have to do with the amount of direct contact, by the senses such as touch and sight, between the human user and the object. The task of the designer would then be to encode into the object not only functional properties in a narrow sense, but also properties which arouse certain sensory impres­ sions in its user. In the design process, therefore a double correlation must be established - that between functions (in the engineering sense) and the object and that between means of expression and the impressions engendered by them. In that sense a design object is "doubly coded." In addition to the functions incorporated into the object, the user is confronted with the means of expression selected by the designer. These means of expression will create certain impressions. The designer, knowing the impressions he wants to arouse in the user must therefore try to present the means of expression which best correlate with these impressions. One of the central problems of design could thus be defined as the "management of impressions by expressions." In this respect design dialectically unites the mere ful­ fillment of functions typical of engineering and the mere expression of subjective feel­ ings which is the domain of the artist. Taking a city as an example, the task of urban design is not only to make sure that the buildings are statically sound but that the functions of buildings are recognized by their features. Users of a city see the expressive shapes selected by the architect-designer. If these shapes are selected properly, the users recognize the functions of the buildings. Therefore, the architect-designer must first conceptualize the different functions of buildings in order to be able to select the appropriate expressions for each of these func­ tions. This task may be defined as differentiation of functions by means of *function •fin­ dices. Within the scope of one given function, further diversification and innovation are possible by impression management. Users may be faced with different forms of build­ ings which, although they serve the same function, engender different impressions. The architect-designer is thus faced with the task of first deciding on appropriate impres­ sions for buildings of a given function and subsequently finding the formal means of expression to engender these impressions in users. This task may be described as diffe­ rentiation of *styles by means of stylistic indices. Turning from architectural design to the design of three-dimensional objects as in product design or industrial design the same principle holds : object design aims not only at the mere functioning of the object but also at the establishment of the correct correla­ tion between means of expression and the impressions engendered by them. It has been found in empirical research, for example, that simple three-dimensional forms like cubes, pyramids or spheres have different and specific impressions associated with them. The knowledge of which impressions are correlated with which simple three-dimensional forms is important to the designer, because more complex three-di­ mensional objects may be thought of as being put together by simpler three-dimensional forms. The impressions associated with complex three-dimensional objects should con­ sequently be related to the impressions associated with these basic components although certainly not in a simple fashion. For instance, two car bodies might awaken different impressions in the beholder because they comprise different component forms.

Design

189

Again, research in the field of typographic design has shown that the basic difference in the impression conveyed by different type-faces is based on the presence or absence of an ''opposition between thick stems and hairlines combined with serifs. The presence of this opposition including the serifs produces a narrow range of impressions which could be summarized by such adjectives as "classical," "traditional," etc. The absence of these formal characteristics , i.e. , stems of equal thickness and the absence of serifs, produces a wide range of impressions which range from "modern" through "precise" to "efficient" and "functional," in short, impressions which are compatible with the content of a very wide range of printed *messages. As has been shown so far, in the design process a correspondence is established between a set of functions (selected from a universe of functions) and a set of functional properties of an object (selected from a universe of functional object properties ) , on the one hand, and between a set of impressions (selected from an universe of impressions) and a correlated set of means of expression (selected from an universe of means of expression) on the other. In a very general sense, two-dimensional or graphic design is the projection of three-dimensional means of expression onto a surface. It seems reasonable to suppose that the impressions associated with a three-dimensional form (for example a cube or a pyramid) are similar to the impressions associated with their two-dimensional projec­ tions (for example a square or a triangle). Therefore, the problem of graphic design is not only to transmit *information at an optimal rate (in the sense of information theory) but to find a correct correlation between the graphic means of expression and the impressions engendered by them. Again it is important to know which forms of expression are correlated with which impressions. Empirical research has shown , for instance, that the impressions associated with the outside geometric form-frame influence the impressions associated with the forms which the *frame surrounds. Under certain circumstances, the frame of a visual message may even become more important than the message it contains. The graphic designer must therefore be aware that the selection of a triangle , a square or a circle as the outer form of, say, a trademark, promotes in each case different impressions along with those produced by other forms or messages within that frame. The correspondence between the set of intended impression and the means of expression must be realized for three different levels. One decisive level is the situation in which the design object is going to be used. The correlation to be established here is between a set of circumstances and a set of intended impressions the design object is meant to arouse. The designer must predict which situations will favor the arousal of impressions intended by him, and select them from the universe of all situations in which the design object could be potentially used. In doing so he will, on the one hand, be able to save in coding. Taking into account, for example, the circumstances of day and night in the design of a watch, instead of using a 24-hour dial, a 1 2-hour one may be suffi­ cient. On the other hand, he takes on the whole responsibility for the failure of object use and impression management, if the situation of use is incorrectly or incompletely predicted, since the user cannot be blamed for the circumstances in which he wants to use the design object. Another decisive level of design is that of choosing the means of expression from the correct design �·code. A design code, again, consists of two correlated universes - one of means of expression and one of the impressions engendered by them. The particular

1 90

Developmental Semiotics

co-ordination of a set in one of these universes and a set in the other constitutes what is called an index of a style. A design code is thus a whole system of indices, each of them being made up of a means of expression and a impression correlated with it, the entire system representing a design style. For the designer, then, it is important to chose the correct design code in order to allow the user to effectively reduce the area in which the impressions the designer intends to convey are located. The choice of the wrong code (e.g., a style of typography unknown to the public instead of a known one) leads to a failure of impression management. The third and fundamental level of design involves making clear the intention to convey impressions by the design object which go beyond the mere functioning of it. The level seems so obvious that it tends to be underrated or overlooked. As was pointed out, the circumstances of use may help to pinpoint the particular impression correlated to a means of expression presented by a design object. The design code stakes out the stylistic realm in which the user must look for indices. But only the clear presentation of the *intention to convey impressions lays the foundations for the user's attention to the means of expression employed by the designer. This fundamental level of design has been variously called marking or foregrounding in the semiotic literature (e.g., �Jakobson and Waugh 1 979; Garvin 1 978). M. K.

Developmental Semiotics Definition Developmental semiotics, in analogy to developmental psychology, is a study of the ontogenetic evolution of the capacity to use different *sign systems in children. This has become an important problem in semiotics connected with questions of the phylogenesis of sign ,;use and of the specificity of human sign use and its (pre-)history.

Developmental Semiotics of ,;Language Linguistic Approaches Developmental semiotics has long been established as a question of how children acquire language. Work in this area has been carried out by the Sterns ( 1 92 8 ), �-Buhler ( 1 930), ''Piaget ( 1 923a) and ,;Vygotskij ( 1 962). But the most important contribution, from a semiotic point of view, has been made by ,;Jakobson (l 971f; 3 1 7ff.) in his early drafts of •:•structural analysis concerning the development of language in children and in his classic "Kindersprache, Aphasie und allgemeine Lautgesetze" ( 1 942). It is especially on the level of the development of �-distinctive features underlying phonemes that the basic tenets of this early work are still valid today, although Jakobson has repeatedly pointed out that they were not meant to be understood mechanically or as dogma, but as working hypotheses open to corrections Qakobson and Waugh 1 979). Jakobson's hypothesis is that the development of distinctive features in children universally follows the same relative chronology and order despite individual oscilla­ tions. In a first babbling stage, many sounds are produced which later drop out of the child's repertory whereas the sounds of his own language are retained. •:•Intonational

Developmental Semiotics

191

contours expressing appeals to *action (imperative •·utterances) and to answer (inter­ rogative utterances) are a first linguistic system used by children Qakobson and Waugh 1979). The system of distinctive features is acquired according to the following sequence of phonemic '�oppositions: -

+Vowel ~ + consonant (e.g., a~ m or b) oral ~ nasal consonants (e.g., p or b~ m) labial ~ dental consonants (e.g., p t, m~ n) open ~ close vowels ( e.g., a~i) palatal ~ velar close vowels (e.g., a~i~u) or "linear vocalism" (e.g., a~ e~i)

There are thus, according to Jakobson, two universal systems of minimal differentiation, one for consonants and one for vowels. Although beyond these earlier stages, child language development will differ according to the distinctive features of particular languages, it will nevertheless follow certain general *rules. The first words appear at the age of approximately one year. By the age of two, the child has mastered about 200 words which by now have aquired the function of denotative signs (cf. Buhler 1930). Between 18 and 24 months of age, single word utter­ ances are replaced by two- or three-word utterances. The source of this development presupposes a social (i.e., parental) speech input (parole) processed by an inborn capacity for detecting its underlying linguistic system (/angue). Jakobson's hypothesis that all kinds of language development (language change, language loss as well as - in reverse direction - language [re-]acquisition), are subject to the same set of universal principles has not remained unchallenged. While the evidence brought forth in its favor is predominantly from phonology, the sequence of morphemes in language loss, e.g., seems to differ in an as yet unpredictable way from a mere reversal of the acquisition sequence. Nevertheless, similar universal hierarchies such as the above for phonology have in the meantime been postulated for various aspects of other levels of description (such as morphology, •·syntax, ,:•semantics). Furthermore, the range of this kind of approach has been considerably extended. Jakobson himself has pointed out that historical linguistics can profit from language typology. J. H. Greenberg has been able to establish sets of implicational •·universals for the interlinguistic distribution of a number of linguistic phenomena. Pidgin and creole languages, notably the functional and structural reduc­ tion in pidginization and the corresponding elaboration in creolization, have been shown to provide further important fields of evidence. In addition to child language, creolistics can be regarded as an another laboratory of linguistic development, where development in progress as well as the creation of linguistic structures can be studied. In recent years, partly as a generalization from pidgin and creole studies, the scope of this kind of approach has been further widened to include foreign language acquisition, and even foreign language learning (by formal instruction). The search for causal explanations for the restrictions governing language and lan­ guage development has prompted an increasing interest in bio-neurolinguistics, (experi­ mental) phonetics, and sociolinguistics. Developmental linguistics has used converging evidence from all the above-men­ tioned fields to underpin a unitary explanatory and predictive theory of language in terms of naturalness - a central notion in most of the approaches in which the concept

192

Developmental Semiotics

is admitted. What is connatural is a basic concept of developmental linguistics, for which it is claimed that it determines the range of linguistic processes and relations free of interference from other systems, i.e., developments which are not abnatural, viz. those resulting from socially conditioned contact, hypercorrection, borrowing and other developments. Connaturalness as characterized builds upon two insights of Greenberg:

( 1) the dynamidprocessual principle:

the more-marked tends to change to the less-marked (unless overruled by predictable higher-level processes such as assimilations, sound shifts, etc.)

(2) the typological/relational principle:

the presence of the more-marked will, as a result, imply that of the less-marked.

It is important to note that in development linguistics these principles are not to be understood in a statistical manner, but have an axiomatic status which involves a strong empirical claim: Exceptions to the principles above must be due either to influences from outside the language system or to higher level processes. Time is thus a fundamental dimension of linguistic theory and description in the developmental framework. All kinds of language variation (not only historical, but also regional, social, *stylistic, acquisitional, etc.) have been shown to fit into (gradient) rela­ tional (namely implicational - cf. principle 2 above) patterns, which can be causally related to (natural) processes (cf. principle 1) in a timeful (dynamic) framework, as opposed to the static frameworks with thing-entities in earlier approaches, where pro­ cesses are viewed as replacements of one thing by another. In other words, linguistic variants are the patterned output generated by (natural) •�rules from one underlying system of entities. For example, some varieties of English may have [o1 ] for the vowel in wife, but [ae ] for the vowel in wives; in unmonitored styles, the former may have [ae] too; similar alternations of [o1 ] and [i] can be observed. Obviously, then all of these outputs can be derived from an underlying vowel /III/, the implicational patterning being such that a given stage of the change before voiceless fric­ atives implies the same stage before voiced ones, but not vice versa. Acquisition of this kind of grammar involves the application of the comparative method (to the different stylistic as well as regional variants); this is why developmentalists have likened the child in the process of language acquisition to a historical linguist. The developmental framework as outlined above presupposes a philosophical out­ look different from Platonism (relationalism) as well as positivism, namely a mediating conceptualist position: it denies neither the reality of the flux of phenomena (as does Platonism), nor that of abstract relations. A more detailed account can be found in various publications by C.-J. Bailey. The child in the phase of language acquisition has often been likened to a linguist in the process of trying to write the grammar of a foreign tribe's language on the basis of speech samples. He starts off with very basic hypotheses about the rules of the game, and keeps restructuring his grammatical system until he ends up with an adequate grammar (which) - although generating an adequate set of acceptable outputs - may be different from that of his parents. These simple hypotheses are maintained by special­ purpose rules formulated ad hoc in the face of new evidence until these special-purpose rules become too numerous and are replaced by a new and more economic hypothesis about the grammatical structure of the speech input.

Developmental Semiotics

193

Theories of linguistic development such as that of McNeill (1970) maintained that the child's first attempt, between one and two years, is to construct a holophrastic *dic­ tionary containing words with *sentence interpretation. The more problems have to be semantically differentiated, the more precision is needed and the more ambiguous and less efficient holophrastic expression becomes. Thus the dictionary project changes in the direction of a word �·lexicon and syntactic hypotheses become necessary. Between 18 and 24 months, the first hypotheses about the ordering of two- or three-word sentences are advanced. These grammatical hypotheses yield as their result the produc­ tion of what has been called "telegraphic speech." The underlying grammar is very basic and follows the same principles, whether the child is exposed to English, Russian (Slobin 1966) or Japanese (McNeill 1966). This basic grammatical hypothesis is main­ tained by special-purpose rules up to the point where it is more convenient for the child to advance hypotheses about a new grammatical tool, i.e., (abstract) *transformation rules, enabling him to account for many variations in the surface structure of the (parental) speech input. M. K. Development Semiotics of Other Sign Types Development of Gestures

Whereas the developmental semiotics of linguistic signs is fairly advanced today, the semiotically oriented study of the development of �·iconic signs in children has apparently been neglected. There are, to be sure, some reflections on the ontogenesis of gestures and their role in language development. For instance, Vygotskij ( 1962) describes the development of the pointing gesture as follows: At first, the child notices an object but his attempt to grasp it fails for lack of coordination. The mother interprets the child's unsuccessful action of reaching for the object as a gesture of pointing at it and hands it to the child. Only on the basis of the mother's social interpretation of the failing action does the child finally come to use it for pointing at objects which are out of reach. Well known are also, in this connection, the considerations of Mead (1934) on the origin of self by internalization of the gestures of others. But a coherent theory on the develop­ ment of iconic signs in children is still lacking. This is all the more surprising as a semioti­ cally oriented developmental psychology of different types of signs already exists. It is the merit of Jean Piaget to have provided this connection between developmental psy­ chology and semiotics. Piaget's Theory of Semiotic Development

Piaget's interest in the concept of �·symbol dates back to his early work in �·psychoanalysis (Piaget 1923b). The fact that, in the symbolic thinking of children and tribal religion alike, the borderline between reality and sign is "magically" blurred requires special consideration of the connection between signans and signatum. In his thinking on this problem Piaget was directly influenced by the "Geneva School" of lin­ guistics based on the famous courses of ''Saussure at the University of Geneva. The con­ cept of sign as a unity of ''signifier (signi/iant) and signified (signi/ie) is already present in La representation du monde chez !'en/ant (Piaget 1926) and La causalite physique chez !'en/ant (Piaget 1927). In Le jugement moral chez !'en/ant (Piaget 1932), one finds the fol­ lowing footnote: "We use the term 'symbol' in the sense of the Saussurean School of Jin-

194

Developmental Semiotics

guistics as the opposite of the 'sign.' A sign is arbitrary, a symbol is motivated. lnciden­ taly, *Freud also talks about symbolic thought in the same sense." An even broader space is given to the definition of the term symbol in the Saussurian sense in La formation du symbol chez !'en/ant (Piaget 1946) and still in Le structuralisme (Piaget 1968) Piaget defends the Saussurean acceptance of the term symbol against the use of the equivalent Peircean term icon (cf. M. D. Smith 1977). Apart from the terms symbol and sign, Piaget also uses the term �·index. Whereas symbol and sign represent objects while they are not present in perception either in a motivated or in an arbitrary (i.e., verbal) fashion, the index is a "concrete signifier" (Piaget 1936) directly connected to the presence of objects in perception. The development of Piaget's triad of index, symbol and sign (paralleling index, icon and symbol in *Peirce) becomes clear in relationship to his general developmental ''schema of intelligence in the child and the relationship of this schema to the construc­ tion of the child's world first by perception, then through represention by means of sym­ bols and words. According to this, intelligence develops in three phases: The first phase of this development, from birth to the age of three to four years, brings about what is called sensorimotor intelligence. During this phase, perception develops from static, object-centered registration of incoherent, isolated aspects to active search by increasing decentration. Semiotically speaking, these two kinds of perception serve the production of *meaning by indices derived from present objects and processes. The second phase of intelligence development serves the preparation and organiza­ tion of concrete mental operations. It takes place between the age of four and twelve years. Now the child begins to "conjure up," by different methods of symbolic represen­ tation, objects which are not present in perception. Thus, semiotically speaking, the pro­ duction of meaning by symbols (i.e., icons in the Peircean sense) develops during this phase. During the third phase of the development of the child's intelligence, from the age of twelve to fourteen years, logical operations are mastered. In this phase, concept for­ mation permits operations on entire ''classes of the symbols which the child was able to form in the previous stage. In this new stage, intelligence deals not only with the repre­ sentation of single objects but with representation of entire classes of objects by words and numbers and with operations applied to these classes. Semiotically speaking, meaning is now produced by abstract (verbal, or mathematical) signs (symbols in the Peircean sense). While, towards the conclusion of the development of intelligence in the child, sym­ bols are relegated to a function which is inferior to that of verbal or mathematical signs, this subordination is nevertheless limited to problem solving by classification only. In the field of geometry, or in the design of plans and other whole objects, symbolic representa­ tion remains an important intellectual tool throughout adult life. Neither, of course, is perception obliterated by the capacity of representing absent objects through symbols or signs. Another Theory Regarding the Development of Different Sign Types The theory of Piaget on the development of the three sign types creates the impression that it is basically a biological process which evolves independently of social or cultural factors. If this were true, children from different cultures should develop according to

Developmental Semiotics

195

the same phases mapped out by Piaget. Also, children from rural and city environments would reach the different phases at the same age. There have been doubts as to whether this is the case. Bruner (1967), for one, claims that the intellectual development of children also depends to a large degree on cultural input. Although Bruner's sequence of intellectual development - the enactive, iconic and symbolic phases of representing reality - is very similar to Piaget's stages of perceptual activity, imitative and verbal representation, their assumptions regarding the "motor" behind development are diffe­ rent. For Piaget, a change in the way of modeling the world is brought about by a dis­ equilibrium between accommodation and assimilation. Thus, imitative imagination is replaced by verbal representation if the former becomes insufficient to accommodate more complex experiences. Bruner (1959) has criticized this concept of Piaget as being circular and suggested instead that the motor of development is a lack of concordance between the enactive, iconic and symbolic modes of representation. If *action is contradictory to what one sees and its verbal �·description, the child may change his problem solving behavior drastically. Thus, the visual surface of an object is certainly not identical with its whole reality and iconic representation may have to be sacrificed in favor of verbal description. How important the concordance between action, image and word is for a child depends, however, according to Bruner (1967), directly on the action patterns, images and concepts furnished by a given culture. Even the value placed on caring for such a concordance might be different from one culture to the next. For Bruner, the highly verbal development of Western culture is a direct consequence of its complexity, origi­ nating in the division of labor. AWestern child cannot personally experience all the tech­ niques necessary for survival in his culture, as is still possible in a less complex society. These techniques must be taught to him verbally in a school. Whereas his peer in the less complex society assists the activities of the grownups in his daily (visual) experience, the Western child remains, for a long period of his life, separated as a learner from the adults as the "professionals." The so-called "generation gap" thus turns out to be an effect ofWestern cultural circumstances. A Developmental Classification of Iconic Signs For a developmental semiotics of iconic signs, Piaget's classification of symbols (i.e., icons in Peircean terminology) is of special interest, because it also follows a develop­ mental sequence. This classification is proposed in La psychologie de !'en/ant (Piaget and lnhelder 1966) in a chapter explicitly carrying the title "The Semiotic or Symbolic Func­ tion." There are five classes of semiotic functions developing one after the other. The first four concern motivated symbols, the last arbitrary signs: (1) Delayed imitation (e.g., a child imitates the temper tantrum of a playmate two hours after it has happened); (2) Symbolic play (e.g., a child mimics his own behavior of when he is tired (closing his eyes, sucking his thumb, etc.); immediately afterward, he puts his teddy bear in a toy bed making the bear a symbol of himself); (3) Drawing (in children's drawings, acquired graphic symbols, like circles for a head, a ball or a wheel, are used at first in the same way as toys in symbolic play to assimilate, later to imitate "reality");

196

Diagram

(4) Imagination (an object or a process is imitated "internally" without the appear­ ance of outside gestures or traces; the child accomodates to reality rather than assimilating it); (5) Mentioning (e.g., a child mews without a cat being present; the absent cat is verbally represented by "signs" in the Saussurean sense).

From this classification it can be seen that children's drawings are considered a specific type of symbol which occurs in the transition between symbolic play and imagination.

Developmental semiotics of children's drawings. Although the psychological study of children's drawings can be traced back to the time around the turn of the century (Cooke 1885; Ricci 1887; Barnes 1893; Clark 1897; Kerschensteiner 1905) the study of this material in terms of developmental semiotics was only begun with Piaget's above classi­ fication of symbols (icons in Peircean terminology). Following Luquet ( 1927), Piaget and Inhelder ( 1948) adopt the following three phases in the development of children's drawings:

( 1) The stage of "synthetic incapacity": After a phase of mere scribbling, indicative of sensorimotor intelligence, at the age of about three, first round and closed as well as open and linear forms are produced by the slowing down and segmentation of arm movements. Some topological relationships such as inclusion, exclusion and intersection can soon be rendered by these simple initial forms. This development calls to mind the babbling stage in language. With the same repertory of forms, children try to put together their first drawings of little men, houses, animals, etc. (2) The stage of "intellectual realism": In this phase, beginning approximately at the age of four or five, the repertory of forms is completed and topological relations are rendered systematically. The term "intellectual realism" refers to the fact that children draw not only what is visible of an object, but "everything which they know is there" (e.g., what is happening inside a house). This yields sometimes so-­ called "transparent" drawings of objects. First, nonsystematic attempts to draw perspectives are made (e.g., different perspectives may occur in one drawing). (3) The stage of "visual realism": Towards the age of eight or nine, children begin to draw what is visible from a certain point of view only. They begin to apply perspective systematically and to coordinate it with the proportions of the objects they draw. M. K.

Diagram From its etymological meaning of graphic representations (6taypaµµa·ta) and its appli­ cation, in classical times, to a table on which all the sounds of a given musical system were serially represented, several meanings of the term diagram have been derived over time. It is possible now to distinguish two main acceptations, a modern, technical one and a traditional, ambiguous one. In the former, diagram is a synonym of function (in the mathematical sense, i.e., a diagram is nothing but the representation of the variation of an entity y with respect to the variation of an entity x), for example: statistical histo­ grams, evolution curves in economy, star distribution diagrams in astrophysics, or

Diagram

197

temperature diagrams in medicine. The latter sense is philosophically more interesting inasmuch as it concerns the relationship of geometrical intuition with *writing; it refers to sketches, maps, graphs, outline, profile, schematic representation of an organized entity, and, more generally, all schematic representation of the relations holding between the components of a system. This latter sense is of particular interest for the structural disciplines, notably for semiolinguistics. These disciplines' aim is to succeed in conceiving first how a given '�structure is articulated, i.e., to make explicit the connections through which its parts are related and then how this structure is regulated (for instance a stable and autoregu­ lated organization such as a biological organism, or a ,:-Gestalt, or a semic taxinomy). Diagrammatic representations of such structures are usually laid out once their systemic '�articulations and regulations are discovered. *Cybernetic diagrams are used for the rep­ resentation of structural regulations. As for the representation of articulations, two examples can be given:

( 1) The syntagmatic trees of structural syntax. These graphic representations of gram­ matical connections are derived from Tesniere's Stemmas: "for the sake of clarity the connections existing between words will be graphically represented by lines that we will call connection lines" (Tesniere 1969: sec. 1. 13). For him, a stemma is ''the visual representation of an abstract notion which is precisely the structural schema of the sentence" (sec. 3.10). It "clearly shows the hierarchy of the connections, schematically reveals the various nodes which bind them in bundles, thus visually materializing the *sentence's structure" (sec. 3.9). (2) The semiotic square. This square develops a binary semic category and is, according to Greimas, "a canonical form" that is immanent (Greimas 1970a: 161), a "primary taxinomic instance" (p. 163) which organizes signification prior to its manifestation. It is defined as "the elementary structure of signification used, as form, in order to articulaJe the semantic substance of a micro-universe" (p. 161). Its formal expression is, like Tesniere's stemmas, a graphic representation that visu­ alizes a structural organization.

Structural articulation, either in a paradigmatic form (e.g., semiotic square) or in a syntagmatic form (e.g., Tesniere's stemmas) is conceivable only through its diagram­ matic schematization because it is formal, and not phenomenal (in the usual sense of the term). This raises a critical theoretical problem because, by the same token, the objective reality of structural phenomena is thus reduced to mostly arbitrary representations which cannot claim to have any ontological value. However, as formal grammar theories have shown, it is possible to deal with these representations as formal objects and to develop a mathematic from them; but this does not solve the crucial problem, i.e., going from diagrams to schemas endowed with objective value. However, the situation has improved since Thom's catastrophe theory has made it possible to develop mathemati­ cally founded structural models. Let us illustrate this with the case of phonology as an example of paradigmatic artic­ ulation. Since the *Prague School's and more particularly Jakobson's work, phonemes are described as positional units, relational entities, formed by bundles of distinctive fea­ tures, a concept which was applied to semantic analysis by '�Hjelmslev, Levi-Strauss and Greimas. For a long time phonology, which was dealing with the form of phonetic expres­ sion, and audio-acoustics, which was dealing with its substance, remained two separate

198

Dialogue

endeavours. But fundamental experiments on categorical perception phenomena which have been made for the last fifteen years, have made it possible to achieve a synthesis of these two viewpoints. For instance, in order to describe the systems of stop consonants, a bidimensional space W of relevant acoustic indices (voice and place of articulation) is considered. As phonemes, stop consonants correspond to a categorization of W by a system K of discontinuities, interfaces, and borderlines. In other words, categorical per­ ception phenomena are perception cases of critical phenomena analogous to other critical phenomena of a physical nature such as phenomena of phase transition, and the morphology K categorizing W into phonemic regions is analogous to a phase diagram. Catastrophic models are the first general mathematical models of critical phenomena. They make it possible to mathematize paradigmatic articulations and, in particular, to transform semiotic square diagrams into topological schemata endowed with objective value (Petitot 1977, 1982c). For syntagmatic articulations, the problem consists in finding a general principle allowing the ,:-transformation of stemma representations into real syntactic morpholo­ gies, or actantial graphs that are not arbitrary but deduced from a principle. According to Thom, this general principle consists in reducing the "':•actants" to their localization in an (abstract) space underlying the stemma and in deducting actantial graphs as inter­ action diagrams between various possible localizations. Note that these interactions are not arbitrary because they are geometrically constrained. They can be classified, and thus yield a list of elementary stemmas that Thom has called "archetypal morphologies" ( 1972, 198 1). J.Pe.

Dialogue In different disciplines the term dialogue is used in various technical senses which are to be precizations (in the terminology of Naess 1966) or ,;explications (Hanna 1968) of a corresponding everyday conception. The everyday use of the term not being syste­ matic, no attempt will be made here to give a complete characterization of it. Only the main aspects of the concept which have been treated in philosophical, formal, or descrip­ tive studies are included. The importance of dialogues for semiotics lies in their being paragon cases of ,;semiosis and �·communication (Kevelson 1977b and some further hints in Sebeok 1974).

Characterization At least four senses of the term dialogue can be distinguished:

( 1) A dialogue is an individual event occurring during a certain span of time at a certain place in which at least two persons, the dialogue partners, treat one or more ,;topics, interacting by using *signs, among them, of necessity, ,;language signs. (2) A dialogue is a class (or property) of individual events which are dialogues in sense (1 ). (3) A dialogue is a *text (or discourse) which consists of a sequence of phrases, ,:•sentences, or texts of a language, reported or imagined to have been uttered with speaker/hearer alternation by two different speakers.

Dialogue

199

(4) A dialogue is an abstract object, or a complex of signs representing such an object, which is structurally similar, according to some theory of dialogues, to a dia­ logue in sense (1 ).

Dialogues in sense ( 1) are clearly basic, as dialogues in senses (3) and (4) may be thought of as representations (or models) of dialogues in sense (1) (Gunther 1977). Therefore this discussion is devoted only to dialogues in sense ( 1). Some immediate consequences to be drawn from this characterization are that ( 1 ) dialogues are semiotic processes; (2) dialogues are interactions; (3) dialogues are (parts of) conversations; and (4) dialogues are governed by �·rules.

Analysis The semiotic analysis of dialogues starts from the fundamental observation that dia­ logues are not just alternating sequences of the dialogue partners' *utterances but highly structured complexes of ;'actions (Reichman 1978). When engaged in a dialogue, the participants are performing essentially two types of actions, those that establish, maintain, modify, or end social contact between the dia­ logue partners (Goffman 197 1), and those involved in treating one or more ,;topics (van Dijk 1977c). Both types of actions must necessarily be performed in a dialogue. A topic cannot be treated in a dialogue without social contact between the participants, and, on the other hand, social contact between persons without the treatment of a topic is not a dialogue at all. The social contact basic to a dialogue is typically, although not exclusively or even necessarily, regulated by the use of nonverbal signs, while the treat­ ment of a topic is typically, although not exclusively or even necessarily, performed by the use of verbal signs (Yngve 1970; Beattie 1978b), for example by performing and interpreting ,;speech acts. The basic constituents of a dialogue are ( 1 ) the utterances of language signs made; (2) the goals and (other) propositional attitudes (especially, but not exclusively, ,;beliefs) of the dialogue partners; (3) the systems of signs used; (4) the rules and conversational maxims according to which the dialogue partners act and interact; and (5) the topics (or themes) of the dialogue. Higher-order constituents of dialogues may tentatively be reduced to (and eventu­ ally be defined by reference to) these basic constituents; (for the pertinent concept of reduction, see Montague 1969): (6) the speech acts performed, reducible to utterances and propositional attitudes; (7) the dialogical moves, reducible to utterances, goals, beliefs, and rules; (8) the strategies (or tactics or plans) the dialogue partners use in their interaction, reducible to goals and beliefs; and (9) the phases of a dialogue, reducible to utterances, goals, and topics.

Classification There is as yet no generally agreed upon classification of dialogues. Within different approaches to the analysis of dialogues, some classes of dialogue are regularly distin­ guished; e.g., everyday dialogues vs. institutional dialogues in conversational analysis, natural language dialogues vs. formal language dialogues in man-machine studies, etc.

200

Dictionary

Lines of Research

Work on dialogues relevant for semiotics can conveniently be classified as formal or descriptive. Formal work on dialogues - i.e., work done mainly in logic, linguistics, �·artificial intelligence research, and, generally, in *cogn itive science - has centered on the topic­ treating aspects of dialogues, including: studies of information-seeking dialogues (Hin­ tikka 198 1a, 1982; studies of logical dialogues as games between a proponent and an opponent (Lorenzen and Lorenz 1978; Drieschner 1966); construction of models or attempts to construct a mathematical theory of dialogues (Hamblin 197 1; Nowakowska 1976; Gunther 1977; Heidrich 1977; Bottner 197 9); studies of natural language dia­ logues for the sake of computer simulation (Schank 1 977a; Levin and Moore 197 8 ). Descriptive studies - i. e., work done in linguistics, poetics, *sociology, and conversa­ tional analysis - deal with, among others, *literary dialogues (Deimer 1975; Muka­ fovsky 1967), everyday dialogues (Dressler 197 1; Henne and Rehbock 197 9), micro-di­ alogues (Posner 197 2c); and conversations (Sudnow 197 2; Sacks, Schegloff, and Jefferson 1974). A. Gu.

Dictionary Semiotically speaking, a �-language dictionary is both a textual, ''metalinguistic *struc­ ture and a cultural, communicational product - a book. Dictionaries and encyclopedias are similar from the second viewpoint, but are structurally and textually different. A (language) dictionary is a �·description of part of the *lexical component of a natural language in cluding a partial description of the other components and using either the same language (in monolingual dictionaries) or one or more other languages. It is divided into many entries, often classified in a formal, perhaps alphabetical, array. Each entry contains the same types of ''information. The first level of organization of this information is its "macrostructure," the internal structure of the entries being called "microstructure" (Rey-Debove 197 1). Every entry is "mentioned," not "in use": its status is one of autonymy. Among the types of information given by the dictionary there are •:• definitions and equivalences. The definition is a paraphrastic *synonym, while equivalences are lexical and idiomatic synonyms in another language, the first one being metalinguistic, the second interlinguistic. Every definition or equivalence is supposed to give the �-reader a (better) knowledge of a subclass of *uses for a lexical item. Monosemic items need onl y one definition; others are analyzed according to a linear or to a hierarchical ordering, which is methodologically typical. In order to describe the functioning of each lexical unit in linguistic use, the dictionary needs examples, which can be phrases, ''sentences or parts of sentences, including the item. Some dictionaries are derived from a corpus of texts, written or not, and the examples are extracted from the corpus; in other dictionaries, examples are generated with a didactic purpose. The obj ect of a dictionary is to provide an "image" (a) of the entire lexicon (Rey-Debove 1977), as a functional set (general dictionaries), or (b) of part of it: spe­ cial dictionaries may deal with a formal, *semantic, or thematic subset of the lexicon.

Dictionary vs. Encyclopedia

201

Semiotic Types of Dictionaries

While dictionaries, when linguistic, describe uses and *functions of lexical units, termi­ nological dictionaries and vocabularies deal with concepts and sets of features. When monolingual, both use definitions, when multilingual equivalences and/or definitions. Encyclopedias may analyze concepts, but describe classes of *referents: proper �·names may appear among their entries. The object of encyclopedias is not the set of lexical units (*symbols) but the definitional or descriptive *interpretant of names. The present dictionary is terminological and encyclopedic. Some dictionaries and most encyclopedias include iconic devices. Iconism is better suited to logically singular contents (e.g., proper names) and descriptions, than to the general, abstract contents of the language dictionaries (classes of linguistic uses). The art of compiling dictionaries and the methodical study of dictionaries is called lexicography; the latter, sometimes, metalexicography. A Communicational View

Both dictionaries and encyclopedias give an image of a cultural set of information and knowledge. Only dictionaries are metalinguistic, but both are metasemiotic and can be considered as didactic texts, whose '�meaning is interpretative of a set of semiotic - cul­ tural and ideological - systems. These text are *indices of a society as a source of *judg­ ments and opinions. Last but not least, dictionaries and encyclopedias are, among reference books, strong communicative media: millions of copies are sold and used each year in developed countries. A. R.

Dictionary vs. Encyclopedia In contemporary semiotic discussions, dictionary and encyclopedia are theoretical '�models accounting for a possible interpretation of the *semantic component of a *grammar or for given mental or '�cultural �·competence. As such, they do not corre­ spond to the "flesh-and-blood" dictionaries and encyclopedias as practical repertories sold under the form of one or more books, even though, also in this case, a dictionary is supposed to provide mere "linguistic" �·information while an encyclopedia provides pieces of world-knowledge. For an overview on this opposition see Wilson 1967; Katz 1972, 1979a; Leech 1974; Lyons 1977; Eco 1979c. The analysis of "flesh-and-blood" dictionaries (see, for instance, Weinreich 1980, Rey-Debove 1971) shows that they do not distinguish clearly between linguistic information and world-knowledge. If one con­ sults, for instance, the Merriam Webster Dictionary of 1974 one finds that bull is defined as an 'adult male bovine animal,' and (as we shall see) this can be taken (even from the theoretical point of view) as a dictionary-like *definition; on the other hand, tiger is defined as a large tawny black striped Asiatic flesh-eating mammal related to cats and this looks like an encyclopedic definition relying upon elements of world-knowledge.

202

Dictionary vs. Encyclopedia

Theoretical Dictionaries The main problem of a theoretical dictionary is that it is controversial which criteria one can select to distinguish linguistic information from world-knowledge. According to Katz (1972 :6), a dictionary should predict and explain: ( 1) •:•synonymy, that is, sameness of �·meaning or paraphrases; (2) semantic similarity or difference (aunt and cow have a semantic component in common, while are both opposed to shadow and reflection) : (3) antonymy; (4) superordination and subordination, or hyperonomy and hyponymy; (5) meaningfulness and semantic anomaly ("a smelly soap" is meaningful while "a smelly itch" is not); (6) semantic *ambiguity (ball can be either a missile for playing or a large formal dance); (7) semantic •:•redundancy ("my female aunt" contains superfluous information); �·analytic *truth ("uncles are males" is true by virtue of the definition of uncle) ; (8) contradiction ("uncles are woman" should be false by virtue of the definition of uncle) ; (9) syntheticity ( "uncle are verbose" is neither true nor false on the basis of dictionary meaning alone); (10) inconsistency ("John is alive" and "John is dead" cannot be true or false together when they refer to the same individual); (11) entailment ( "this car is red" entails "this car is colored"); (12) self answered �·questions (the interrogative "is this spinster a female?" expresses a question which is answered in the asking); ( 13) �·presupposition (the interrogative "where is the key?" presupposes the truth of the declarative "the key is someplace"); ( 14) possible answers (the *sent­ ence "John arrived at noon" is a possible answer to "when did John arrive?"). All the cases listed above (except perhaps 13 and 14) seem to be reducible at the double feature of analyticity and entailment: the definitions of a dictionary must be ana­ lytic, that is, they belong to the defined item by virtue of the definition, irrespective of any evaluation in terms of truth values; the system of features is a hierarchical one, so that the inferior features entail the superior ones. Thus it seems that a theoretical dictionary requires that the number of the semantic features must be finite (a closed system of relationships of entailment can provide such a guarantee) and that the semantic features must be theoretical constructs and cannot be further analyzed or interpreted by other features. This second requirement can be met either by assuming that the semantic feature belongs to a finite set of semantic *uni­ versals or by establishing for a finite set of *lexical entries a closed system of meaning postulates. Many objections have been raised against the notion of a theoretical dictionary. First of all, there is no criterion for establishing what an analytic property is. Second, all the criteria for establishing a set of universals of *language are ad hoc. Third, it is impossible to describe through purely dictionary-like properties, the real competence of a speaker; a dictionary holds only for an artificial language or for a dummy natural language. On the grounds of �·Peirce's notion of �·interpretant a more radical stricture sounds as fol­ lows : A "flesh-and-blood" dictionary explains words through words; the defining words must be interpreted in their turn, and so on ad infinitum, in a process of unlimited �·semiosis. To avoid such a condition, a theoretical dictionary should use abstract *sym­ bols or, in any case, theoretical constructs (represented by other words only by virtue of a convention), but these constructs cannot be interpreted, since of they are interpreted the set of the semantic features will be open-ended. That they cannot be interpreted to say, for instance, that an uncle is a "male human relative, brother of the father or of the mother of x" does not explain how this lexical entry can be understood, since it is impos-

Dictionary vs. Encyclopedia

203

sible to tell what relative or brother represent. The only solution would be to assume that the words representing the ultimate semantic features are object-words defined by a mere acquaintance with extralinguistic reality. If so, a dictionary, devised in order to de­ scribe linguistic knowledge as independent of factual matters, will depend entirely upon factual knowledge. Moreover, the problem of object-words is closely linked to that of logical proper names and with a possible criterion for their ,:-rigid designation (Kripke 1972). Thus a dictionary would account only for a restricted set of dictionary words de­ finable through object words which are not described by the dictionary itself. Encyclopedias

In order to cope with these problems, many semantic theories look for a semantic representation in the format of an encyclopedia, taking into account not only so called world-knowledge but also ,;pragmatic phenomena. Since the most common objection to an encyclopedic representation is that it should be potentially infinite and that it relies upon factual rather than linguistic knowledge, the encyclopedic project more or less explicitly assumes that: (1) the encyclopedia is a semiotic postulate, a regulative hypothesis that encourages partial and local exploration of the encyclopedic universe; (2) there is no difference between linguistic knowledge and factual knowledge because the knowledge represented by an encyclopedia is a "cultural" knowledge, as reflected by interpretants. It is not an indiosyncratic competence but a cultural competence, recorded by an intertextual body. Though the concept of encyclopedia risks being a mere philosophical hypothesis, it must be used to provide local semantic description because even dictionary-like representations are "polluted" by an encyclopedic element. The so-called "universals" or "primitives" that work as semantic markers in a dictionary are mere linguistic labels that cover more "synthetic" properties. Usually the local representation of the encyclo­ pedic knowledge assumes the form of a set of instructions for the proper textual inser­ tion of the ,;terms of a language into a series of contexts (as ,:-classes of co-texts) and for the correct disambiguation of the same terms when met within a given co-text. An ency­ clopedic version of componential semantics should then appear as a Instruktions­ semantik which is text-theoretically oriented (see, for instance, Schmidt 1976). Eco proposes ( 1976) a model of componential analysis in an encyclopedic format, where the spectrum of the sememe (corresponding to the content of a given expression) is analyzed in terms of contextual and circumstantial selections. This kind of representa­ tion should not only hold for so-called categorematic terms but also for the syncategore­ matic ones. Eco ( 1979c) insists on the fact that a sememe is a virtual or potential ,;text and a text is the expansion of one or more sememes. The ecyclopedic representation of the sememe has been reinforced with the reference to *frames, scripts and other instruc­ tions concerning coded circumstantial and contextual occurrences. Nevertheless, there are cases in which such notions as contextual selections are not enough to establish the possible textual insertion of given lexemes. Take, for example, expre_ssion such as "John was sleeping when he was suddenly awakened. Somebody was tearing up the pillow." A computer fed with dictionary-like information would be able to understand what to sleep and pillow mean but would be unable to establish what the relation is between John and the pillow (and which pillow?). Current research in •:•artificial intelligence has elaborated the notion of frame (Minsky 197 4; Winston 1977;

204

Dictionary vs. Encyclopedia

Schank 1975a, 198 1 ; van Dijk 1977). The *addressee (be it a computer or a human being) is endowed with an enlarged encyclopedic competence which encompasses also a set of frames or scripts, among which - for instance - are the frames "sleeping" and "bedroom." By resorting to this store of competence, the addressee knows that human beings usually sleep in bedrooms and that bedrooms are furnished with beds, beds with pillows, and so on and so forth. By amalgamation of two or more frames the addressee realizes that the pillow just mentioned can only be the one John was resting his head on. To what extent can one assume that frames too are elements of an encyclopedia-like componential analysis (an extremely rich one ideed)? Probably the very notion of en­ cyclopedia has to be revised and re-organized according to different rates of social accessibility or of textual necessity. For instance, when reading "John was sleeping and dreaming of . . . ," one does not need to have recourse to the frame "bedroom" (even though one shares this piece of knowledge), and only the co-text leads one to blow up and to narcotize given sememes or frames (see Eco 1979c). When a text theory aims to establish a "frame for frames" (Petofi 1976a), it is attempting both to discover textual '�rules and to set up a more organized and compre­ hensive notion of code as encyclopedic knowledge. Schank and Abelson (1977) not only attempt to represent the lexical meaning of a given expression but also all the connected forms of world-knowledge that allows the interpreter to draw (from the �-utterance of a term or of a sentence made up with the analyzed terms) co-textual inferences through the use of certain primitives representing fundamental human *operations (such as ATRANS, EJECT, INGEST, MOVE, and so on). Schank represents in the following way (see figure 1) the verb to eat, as an item sus­ ceptible of being inserted in contexts such as "John ate a frog :"

t

mouth1 mouth

= 'ART OOHN,,_,__

hand

V

Figure 1.

i�

V

mouth

The project for an encyclopedic semantics is still in progress. Up to now we are wit­ nessing a series of alternative or complementary proposals. Partially encyclopedia-ori­ ented is the *structural semantics of Greimas (1966a, 1979) with its notion of �·actant and classemes or contextual semes, as well as with the idea of "narrative programs." The early *case grammar of Fillmore (1966), along with the more recent research of Eco and proposals by Bierwisch ( 1970, 1971) are encyclopedia-oriented. A new semantics in terms of encyclopedia and world-knowledge (and phenomeno­ logical experience) is evident in the latest research of Lakoff (1980a) and Lakoff and Johnson (1980b). Undoubtedly any encyclopedia-like semantics must blur the distinctions between analytic and synthetic properties and in this sense nothing has been said that is better

Dictionary vs. Encyclopedia

205

than Quine's "Two dogmas of empiricism" ( 1951b). Naturally, nothing prevents us from using analytical markers, provided one knows that they are shorthand devices used to include other analytic properties they entail and to summarize all the synthetic markers they name. In this sense it the proposal of Putnam ( 1975a) is more than acceptable: The normal form description of the meaning of a word should be a finite sequence, or "vector," whose components should certainly include the following (it might be derisable to have other types of components as well): ( 1) the *syntactic markers that apply to the word, e.g., "noun;" (2) the semantic markers that apply to the word, e.g., "animal," "period of time;" (3) a description of the additional feature of the stereotype, if any; ( 4) a description of the extension. The following convention is a part of this proposal : the components of the vector all represent a hypothesis about the individual speaker's competence, except the extension. Thus the normal form description for water might be, in part:

Extension

Syntactic markers

Semantic markers

Stereotype

mass, noun, concrete

natural kind liquid

colorless transparent H20 (give or take tasteless thirstquenching impurities) etc.

The idea of the stereotype represents probably the most fruitful suggestion coming from Putnam's theory of language. If it had to be accepted, semantic markers should also be considered pseudoanalytical shorthand labels, since there must be also a stereo­ type for liquid and perhaps for natural. As for the extension, Putnam's proposal depends on his sharing of Kripke' s ( 1 972) theory of rigid designation. Many of Putnam's suggestions (for instance, the distinction between stereotyped knowledge and expert knowledge) are accepted by Petofi theory of encyclopedic representation. Let us consider, for instance, the tentative representation of chlorine pro­ posed by Nebauer and Petofi ( 198 1: 367): A. Sector of Commonsense Knowledge

generic term color

smell

element

greenish disagreable, bad

B. Sector of Expert Knowledge

(1) Chemical knowledge

element category

family

valence

chemical symbol natural occurrence chlorine compounds (2) Physical knowledge

nature state or matter other states weight

atomic number atomic weight

non-metallic

haloge univalent polyvalent Cl in chlorides NaCl, HCI etc. gas

liquid chlorine

2 times as heavy as air

17 33.453

206

Dictionary vs. Encyclopedia

(3) Biological knowledge effect on living po!SOn organisms

(4) Geological knowledge amount in earth's crust 0, 15%

(5) Historical information discovery Scheele 1774, Davy 1810 further research production of liquid chlorine in 1823 (6) Etymological Information origin Greek chloros

(7) Industrial knowledge electrolysis from common production salt bleaching in paper and uses textile industry; disinfectant (germicide and pesticide); chemical warla.re cool, dry conditions, in storage iron etc. container

All the information listed above are part of a possible linguistic competence, irre­ spective of the difference between dictionary and encyclopedia. Probably the difference between commonsense knowledge and expert knowledge should be traced each time according to the specific co-text. The advantage of Petofi's model over Putnam's is that the former gives up the dis­ tinction between *intension and extension. Any item of expert knowledge can be intended as a meaning component that serves to establish the extension of the term under certain circumstances. There are many other models for an encyclopedic representation and at the present state of the art it would be embarrassing to decide which one is the more suitable. Rey-Debove (197 1 ) speaks of bricolage a propos the work of lexicographers and their "natural" definitions. She also remarks. that, looking at existing dictionaries, it seems that it is easier to define infrequent expressions like infarct than frequent ones likes to do. A semiotic encyclopedia, even though only designed under the form of local exam­ ples, is subjected to the same restrictions. However, the choice of the encyclopedia as against the dictionary is not a free one: dictionaries cannot exist, if not as theoretical fig­ ments. The universe of natural languages (and not only of verbal ones) is the universe of semiosis. The regulative idea of encyclopedia is the only way to outline a possible format of such a universe and to try tentative devices for describing part of it. U. E.

Diderot

207

Diderot, Denis ( 1713-1784) Denis Diderot, Langrois by birth and Parisian by stubborn and passionate choice, was not un philosophe, but le philosophe of eighteenth-century France. Diderot devoted his entire life to language, from the massive �·Encyclopedie to his creative works. In each piece he provided a systematic concept of ,:•signs, which he anchored firmly in humanity. For Diderot, signs were visual, auditory, ,;cultural, and potentially *tactile ("Lettre sur !es aveugles," 1749). He felt that one could only capture the constantly evolving uni­ verse by *paintings or words (see his article "Encyclopedie," 1 75 1-80). Since words can only approximate reality, he saw them as metasigns: the tools for subjective decoding of human perception in a dialogical society ("Paradoxe sur le Comedien," 1773). Diderot, as the destinateur, wrote for the exceptional, intelligent, educated �·reader, as his destinataire. Thus, he considered the habitual use of conventional signs as intellec­ tually marginal. The creative artist had to forge beyond the marginality of the norm, which was precisely what made him an homme de genie ("Refutation d'Helvetius," 1774). Diderot categorized signs according to ,;semantics, ,:•grammar, ''syntax, diachrony, and synchrony. His definition of man (like the actor) as a neutral sign, whose plus and minus features are determined by experience and sensations, endowed man with a shifting paradigm of possible ,;actions appropriate for the ever-changing universe ("Re­ futation d'Hemsterhuis," 1 773; and Paradoxe sur le comedien). Diderot embodied a combination of ,;theory and praxis. (See the explanatory dia­ gram of Diderot's tree of knowledge for the Encyclopedie, which reduced the universe to a system - Figure 1., p. 208) His fascination with paradox demanded system while shunning it. He subdivided life into art, ,;music, �·theater, and *mathematics. After decoding each into its semiotic components, he then interrelated them to form a single unit. Diderot recognized the value of simplification through taxonomic decoding ("Memoires sur differents sujets de mathematiques," 1 748; "Lettre sur !es aveugles," Encyclopedie 175 1; etc). As a theoretician of language, Diderot began from the premise that the vocabulary of any natural language is a faithful table of its people, and that languages dictate ideas to the speakers, pre-Humboldtian concepts of language and culture. Semiological variety in perception stems from the fact that the signified does not change; rather, the ,;signifier mirrors the prismatic interpretations of the viewer, guided by sensations and governed by memory ("Refutation d'Hemsterhuis"). Given these limitations in contrast to potential linguistic freedom, Diderot explained that one's mastery of *style provides one with the tools to decode matter and thereby control nature artificially (commentary on Beccaria's "Recherches sur la nature du style," 1 77 1). In his chats and harangues with his readers, and as their faire trans/ormateur, Diderot outlined all of modern linguistics, both implicitly and explicitly. To show his total in­ volvement in semiotics and linguistics would mean listing the index of all the many volumes of his (Euvres Completes. J. P. K.

208

Diderot

Illustrated System of Human Knowledge M'"-olre (Histoire, etc.) Science de la nature etc.

sl ruct\Jre Ki nes.cs para language gral)hemes

L-surface structure __J ENCYCLOPEDIE textual level

Content (Articles)

Expression (Phemes, phcnemes)

l

Surface level (Oi.SCUl"$i'Ye enunciation) senuotic grammar

j

Deep level (Narrative enunciation) cul1ural level

Figure 1.

faire cr0tre (faire persuaslf)

nConviction (example article) H

,sem:,,,,.,.

X

(dissuasklf!

stz .

persuasion

+classemes>

I

I Sys1em

[dist>et leij

sem.,llC square

Trulh semes

�.

_s_ - • •2

;; +-----+ s

L.,.,rpt,o1ogyJ

I

Process conviction

'--syntax -.J I parole (lhe article "Persuasion" wllhall lls�

PRAXIS

�.

�

Diegesis

20 9

Diegesis As a semiotic concept, the English term diegesis comes from French '�narratology. It is the equivalent of two expressions in French: (1) diegesis (611']y11cr�), as opposed to mimesis (µtµT)cr�); and (2) diegese, referring to the spatiotemporal universe designated by the narrative �·text. It is important, therefore, that caution be exercised when employing the English translation. Diegesis in its first sense is derived from Tbe Republic (392c-398 b). •:• Plato, in distin­ guishing between logos and Lexis, maintained that the latter can be either in the pure nar­ rative mode (as in the ancient dithyramb) or in the mimetic mode (as in the theater), the mixed form appearing in the epic. •:•Aristotle (Poetics, 144 8 a) subordinated both the nar­ rative and the dramatic to mimesis as two forms of poetic imitation. In both cases, how­ ever, the two forms remain opposed to each other. During the 1960s, the French structuralists adopted the diegesis vs. mimesis dichotomy as the equivalent in narratology of the telling vs. showing concept of Anglo-American criticism (cf. Lubbock 1921; Friedman 1955b). This principle, man i­ fested particularly in narrative summary and scene, has appeared in narrative theory under various names: e.g., panoramic vs. scenic presentation (Lubbock 1921: 67); eigent­ liche vs. szenische Darstellung (Ludwig 1891 :202ff.); berichtende Erziihlung vs. szenische Darstellung (Stanzel 1955: 22, 1964:11-16); dire vs. montrer (Pouillon 1946: 136). How­ ever, the structuralist position is distinguished from other approaches in that it avoids visual *metaphors. It has been pointed out, for instance, that in *natural language narra­ tive, *non verbal events can be mediated only through verbal events and that mimesis remains subordinated to diegesis (Genette 1966: 152-56). Similar positions, related to �·Ben veniste's linguistic criteria for disco urse and story (1966: 237-50), appear in Todor­ ov's "narration" vs. " representation" (1966: 144) and in Weinrich's "erziihlte" vs. "be­ sprochene Welt" (197 1). One observer has suggested that Genette's use of the diegesis vs. mimesis principle corresponds to Stanzel's modal opposition between narrator *func­ tions vs. reflector functions (Cohn 198 1: 160, 170-71; cf. Stanzel 1979: 189ff.; 198 1). Genette's narrative model, where this principle accounts for the relations between the narrator's and the characters' discourses, incorporates summary and scene as varia­ bles which are dependent upon the speed of narration and the degree of presence of the narrator's voice (197 2: 184-203; cf. Debray-Genette 1971). Accordingly, the term dis­ tance is used to co ver the two aspects of this narrative category: (1) narrative of events (cf. diegesis) and (2) narrative of words (cf. mimesis). The latter aspect consists of the various types of characters' discourse: (1) narrated, (2) transposed (in indirect style), and (3) reported (in direct style). (This use of the term is therefore not to be confused with distance in Booth's sense [1961a:68-73; 1961b: 155-59], or with the temporal dis­ tance separating narrated event from narration.) Distance, together with perspective, constitutes mode: the "regulation of narrative •:• information" (cf. To dorov 1966:14 1-46; Lintvelt 198 1:51-52; for an author who rej ects the diegesis vs. mimesis principle, see Bal 1977 :26-28). In its second sense, diegesis is the spatiotemporal universe in which the story is situated. Theoreticians (even those writing in French) frequently employ diegesis as a synonym for story - a practice which partly originates in the misleading identification o f diegesis with Todoro v's narrative as story (vs. narrative as discourse) (Genette 197 2:72, n.1; cf. Todoro v 1966). Actually, the expression diegese comes from •:• film theory, where

2 10

Diegesis

it designates the spatiotemporal relations within the represented world, as opposed to those to which the signifying �·medium (or film) is subject (Souriau 1948; 1953b; cf. Metz 1968, 1972. For the use of this term in the semiotics of *theater, see Souriau 1956: 1 1 and Issacharoff 198 1). Diegesis in this sense, while a necessary ingredient of story, clearly does not cover '�actions and their '�actants, and it should therefore not be used interchangeably with s to ry. Once this aspect of diegesis appears in narratology, it is subject to two sets of deter­ minations (Genette 1969:201-03; 1972:238-59). The first - person - concerns the absence vs. the presence of the narrator in the narrated world: heterodiegetic vs. homo­ diegetic (autodiegetic, when the narrator is protagonist). The traditional grammatical criteria (third person vs. first person) is thus dropped, as all narration is seen as "virtual­ ly" in the first person. (See the commentary in Cohn 1978 [272-73, n.35]; for the use of this terminology in the study of autobiography, see Lejeune 1975.) The second determination - narrative level - is based on the contextualization of nar­ rative discourse: ( 1) the extradiegetic level is external to the events of the primary narra­ tive; (2) the (intra)diegetic level is that of the events situated within the primary narra­ tive; (3) the metadiegetic level is that of the events situated within the secondary narra­ tive. Given the dynamic relations which link these levels one to another, the narrative instance of the primary narrative is extradiegetic, just as that of the secondary narrative is diegetic. However, various forms of infractions of these levels are possible (e.g., when an extradiegetic narrator intervenes in the diegesis), and they are grouped under the term metalepse. Finally, the intersection of person and the first two narrative levels results in four categories which constitute the status of the narrator. The prefix ,:•meta-, as used in this context, has provoked some controversy. Genette stipulates in his study that metadiegesis (and metanarrative) are to be taken in the oppo­ site sense of the logico-linguistic *metalanguage (cf. *Jakobson 1960:356): rather than a level about the primary level, metadiegesis is a level embedded within it ( 1972: 139, n.1). One critic has argued, however, that in order to avoid terminological confusion, a hierarchy should be established in which hypo- replaces meta-: hyp odiegesis (and hypo­ narrative) would then correspond to a level which is inferior to diegesis, whereas meta­ diegesis would correspond to a level which is superior to diegesis - a level at which would appear a narrative on narrative (Bal 1977:24, 35, 198 1). But it seems that, had this critic faithfully adhered to this visual logic, she would have named the superior level not meta­ diegesis, but hyperdiegesis. (cf. Genette 1982.) The central factor remains (as Bal is well aware) that a distinction must be made between primary narrative and secondary (i.e., embedded) narrative and their respective levels (cf. Rousset 1973 :69-70; Lintvelt 198 1:209-14). This is not to say, however, that this distinction does not prove problem­ atical in the analysis of some kinds of texts (cf. Rimmon 1976a:359-60, 1976b; for its use in the analysis of mise en abyme, see Dallenbach 1977:70-72). Another commentator considers that the relation between the extradiegetic and the diegetic levels is subject to a metalinguistic function (Bronzwaer 1978 : 1-2). This scheme (similarly to the previous one) reverses that proposed by Genette, since it seeks to identify the extradiegetic level with metadiscourse. One can, of course, speak of a meta­ linguistic function in *literature, insofar as literary texts contain ,:•indices which point to their own paraphrase and/ or interpretation (cf. Hamon 1977; Heuvel 1977; Prince 1977; Rossum-Guyon 1980-8 1). As for Genette, he does include among the five func­ tions of the narrator a metalinguistic (or metanarrative) one - /onction de regie 'control

Distinctive Feature

21 1

function' - which serves to organize the text internally (1972 :261-65). It is on the basis of these and other criteria that we must conclude that extradiegesis - unlike metalan­ guage - is not "about" or "on" diegesis, but rather that it is a narrative level which is external to that of the world represented in the primary narrative. J. Pi.

Distinctive Feature Distinctives features are those minimal properties of sound which, combined in bundles to form phonemes, are used to differentiate one word from another. Their function is the discrimination of words differing in �·meaning (sense-discrimination - see '�Jakobson and Waugh 1979). Thus, they are �·signs the �·signified of which is "mere otherness" (Jakobson, Fant, and Halle 1952; Jakobson and Halle 1956b) and which serve as the •:•signifiers in other signs, e.g., words. They thus evidence double •:•articulation and mediacy (an indirect relation with meaning), although as a secondary and opposite tend­ ency they may evidence a drive toward an immediate, direct relation with meaning (Ja­ kobson and Waugh 1979). The distinctive features are differentiated from the redundant features, which serve as their perceptual enhancement and may substitute for them in elliptical modes of speech (see Jakobson 1968 ; Jakobson and Waugh 1979), as well as from the configurative features (which signal the integrity and/or boundaries of words), the expressive (stylistic) features (which signal the speaker's attitude), and the physiognomic features (which serve to identify the speaker in various ways). Given the use of distinctive features only for sense-discrimination, if in a given context the substi­ tution of one distinctive feature for another is not possible, their sense-discriminative capacity is suspended or neutralized and an incomplete or neutralized phoneme results. There is a *universal, hierarchical set of features both inherent (segmental) and prosodic (suprasegmental), divided into three types: sonority (force), tonality (tone), protensity (quantity). The features are perceptual (acoustico-auditory) in nature, with articulatory/motor coordinates. They are also binary (except perhaps for the feature compact/diffuse when associated with vowel height - see Jakobson and Halle 1956b; Fischer-Jorgensen 197 5 : 1 40; Jakobson and Waugh 1979) and based on �-opposition and hence markedness. Thus, the two poles of any feature are defined in relation to each other and may exhibit considerable variation in different simultaneous or sequential contexts or in different *languages. The two poles are generally related either as the presence vs. absence of a given property (e.g., [ ± nasal] where [ - nasal] = lack of nasality) or as opposite ends of a continuum (e.g., [ ± grave] where [ + grave] = low tonality and [- grave] = high tonality), although with markedness either there is the presence of a mark (e.g., [ + nasal] = a mark) or the absence of a mark (e.g., [- nasal] = a nonmark). There is systematic �·ambiguity in the use of the term feature in the litera­ ture, to mean either the opposition as a whole (e.g., the nasal feature = ± nasal) or one pole of the opposition (e.g., the nasal feature = + nasal). In the speech chain, divisions between distinctive features are easier to make than divisions between phonemes, for the latter tend to overlap and not to show the same discreteness as the distinctive fea­ tures. In generative phonology, distinctive features are based on sense-discrimination combined with morpho-phonological alternations and are articulatory in nature. They

2 12

Divination

are binary and may be somewhat abstract at the systematic phonemic level and n-ary and more concrete at the systematic phonetic level (see Chomsky 1964a, 1966c; Chomsky and Halle 1968b). The list of features is currently in flux (see Anderson 1974), as is their function in generative phonology and in particular the question of their abstractness (see Dinnsen 1979). The notion of features or components or properties of sound was perhaps first discussed in the Stoic theory of dynameis, and was present in the work of Baudouin de Courtenay ( 1910); the concept of "correlation" introduced by Jakobson (1928b, 1929b) implies the concept of distinctive feature (see also Trubetzkoy 1929). The first definition of the phoneme as a bundle of distinctive features is by Jakobson (1932a; see also 1939, 1949; Jakobson and Lotz 1949a; Trubetzkoy 1939; Bloomfield 1933). The concept of distinctive features has been fairly widely accepted, although their status as signs and their binary nature have not (see, e.g., Martinet 1957-58, and Wilson 1966; cf. Halle 1957). In some work on semantics, semantic components are termed "semantic distinctive features" (see Schooneveld 1978). L. R. W. Divination Divination, the prognosis or diagnosis of events from a ritually prescribed object (e.g., an animal bone) or occurrence (e.g., a dream), satisfies the basic criteria of semiotic pro­ cess. It entails interpretation on the basis of a more or less elastic encoding of sign-event correspondences; it can be retrospectively adjudged true or untrue, in accordance with either its own constitutive �-rules or extrinsic evidential criteria; and its primary goal is to render an absence present and conceptually accessible. This last criterion is fundamental. As the etymology of divinatio implies, the act of divination is an attempt to mediate the boundary between immediate human experience, on the one hand, and hypothetical domains of causation - including the supernatural - on the other. Divinatory procedures are conventionally represented as rule-governed and investigative. These characteristics have led to their representation as a form of nor­ mative science (e.g., Leibovici 1968 : xix; cf. also Vernant 1974:24, for a more semiotic formulation). Such perspectives lead logically to a critical scrutiny of the ethnocentric '�ideology that underlies the very distinction between divinatory and experimental practi­ ces. Even here, however, there may be room for a more radical critique: it is by no means a foregone conclusion that prediction must always be the primary aim of divina­ tion, rather than - for example - the symbolic exploration of generic social problems and ambiguities. For some audiences, fallible divination may express the ultimate opacity of all social experience, making particular failures as instructive as successful predictions. Evans-Pritchard's (1937) lucid study of Zande divination illustrates the central diffi­ culty. He noted that failures reflected badly on individual divernes' skills, rather than on the theory, practice, and rationale of divination as such. For him, therefore, Zande div­ ination represented a conceptually closed system; its results constituted steps in a pro­ cess of self-fulfilling prophecy. According to this and related perspectives, the looseness of the mantic '�rhetoric obscures the internal contradictions between predictions and

Divination

2 13

actual outcomes. Thus, in a somewhat similar view of Korean geomancy, we find a mea­ sure of definitional fluidity attributed to the boundary between events predicted and actions taken subsequently in the light of the prediction (Yoon 1976: 194-97). Such treatments of divination take definitional flexibility to be a mechanism for maintaining the general principle of oracular wisdom. That principle is then viewed as a part of the overall moral structure, and its maintenance is treated metonymically as a means of rein­ forcing the encompassing values of the society in question (especially Turner 1975a: 3 1, 242). There may be some justification for treating divination as a circular evidential system in particular ethnographic cases. Nevertheless, any assumption that this is a general characteristic of all divination can be rejected for two major reasons:

(1) The circularity may not be located in the oracular process itself, but in the *form and transmission of the narratives about it; the historiography of the Delphic Oracle illustrates how the narratives about past prophecies were selected and mod­ ified to suit expectations (Fontenrose 1978 ; cf. Herzfeld 1982). (2) The notion of divination as a closed or circular system falsely assumes the impossibility of an internally generated critique (Horton 1967), or of the develop­ ment of a "revolutionary" model created in reaction to a socially unsatisfactory paradigm (Bauer and Hinnant 1980). Here, moreover, ethnographers may have been tempted to attribute to divination a tendency to self-fulfilment analogous to what JulesRosette ( 1978) has suggested may be a pervasive feature of their own theoretical assumptions.

As an interactional and political phenomenon, divination may certainly be expected to resemble ethnographer-informant relations in certain highly revealing ways. Fernan­ dez' (1967 :28) insight into the ways in which a client may enable the diviner "to define the situation as one in which the pertinent superstructure is being effectively revealed" anticipated the shift away from a simplistic view of divination as mere self-delusion and towards a more comprehensive understanding of its socially disclosive qualities. Semiotically, moreover, divination can itself be seen as standing for the political status of its performers. In Grodzynski's (1974) account of the progressive and eventually exclusive arrogation of divinatory authority to itself by the Roman imperium, it is possible to find a complementary explanation of how the continuation of folk divina­ tion may have reproduced an ideological opposition to imperial and, later, ecclesiastical legislation. The continuing use of scapulomancy among politically marginal pastoralists in modern Greece is certainly suggestive in this context; in the nineteenth century it was especially popular among the brigand-guerrillas of the mountain areas (Megas 1926: 1 1). In general, it is becoming increasingly clear that divination cannot simply be de­ scribed in terms of *belief systems, or as a rigidly rule-bound activity protected by arcane or circular reasoning. The performative context and rhetoric of divination may be expected to furnish more effective insights into the indigenous semiotic principles involved, and to dispel the once widespread scholarly view of divination as a literalistic and naYve mode of prophecy. M. He.

214

Doctrine

Doctrine

In the natural language environment of the seventeenth century, the term doctrine was a direct derivative of the Latin philosophical term doctrina, which expressed with great clarity and richness the notion of a level or type of knowledge which, we can see in hind­ sight, is critically distinct from scientific knowledge or science (scientia) in the modern sense that term has - like theory - acquired since �·Locke's day. Attention to this differ­ ence can be of considerable help in clarifying the dimensions of semiotic development in our own day. A "doctrine," in Latin or English, refers first of all to a "teaching" in the most gen­ eral sense, i.e., to what some person or persons, e ven unkown, assert to be the case about something. Thus, doctrina may concern natural reality or social reality, or any web and admixture of the two. In the systematic context of Latin scholasticism, with a remarkable constancy spanning the later centuries of that epoch - as illustrated, for example, early in *Aquinas (c. 1266 ( 1952] :I q. 1) or late in *Poinsot (1637 (1931] : disp. 2 - doctrina referred to a body of thought sensitive to its own implications and striving for consistency throughout, while achieving explanations (however provisional) at a level beyond what can be empirically circumscribed in unambiguous ways. Thus, the notion of doctrina is one of the avenues expressing the differing ways in which the sensory core of �·cognition is relied upon in dominant moods of thought which are typi­ cally "scientific" as contrasted with those that tend more to typically "philosophic" analysis. J. N. D.

E Economics

In speaking of the semiotics of economics we may give the impression of wanting to make of semiotics a science so all-embracing as to include within itself every other disci­ pline which in some way is tied to *communication. In reality, however, we could inter­ pret this expression as meaning that it is semiotics that belongs to economics, where "be­ longs" indicates adherence to a specific point of view, so that semiotics inserts itself within the perspective of economics and takes up economic categories. It is precisely this interpretation that we find confirmed in the history of the relations between the two dis­ ciplines. In fact, when semiotics was but a vague hypothesis, *Saussure, who was one of the first to envisage semiotics as a science, used economics as a model for linguistics, which he viewed as a branch of semiotics, thus conferring scientific status on it. Such a stance produces far more than a marginal effect on the development of semiotics, given that, despite Saussure's specific interests in verbal �·language, he theorizes a general model of the *sign which still today continues to influence semiotic theory. It is precisely in Saussure ( 1 9 1 6) that we have the encounter between the study of the sign and the study of economic exchange. He uses, as his methodological model, con­ cepts taken from "recent works" of political economy "which tend to be scientific." The very expression ''which tend to be scientific" contains a general reference to the so­ called "marginalistic revolution" headed by Menger, Jevons, Walras, and Pareto, which, in contrast to classical political economy and the German historical school headed by Schmoller, aims at acquiring scientific status. It is above all the school of Lau­ sanne (Walras and Pareto), with its application of the "pointe de vue statique," which exerts the deepest influence on Saussure's theory of the value of the sign. In Saussure, the concept of value is of primary importance and consequently the relationship between linguistics and economics plays a central role, since both are value sciences. The distinction between diachrony and synchrony is also the result of the relationship between the two disciplines. As early as 1883, Menger, in his Untersuchungen uber die Methode, distinguishes between the theory and the history of economics, and the same distinction is confirmed by the School of Lausanne. It is along these lines that Pareto dis­ tinguishes between a static and a dynamic perspective in economics. The former is con­ cerned with a particular state of economic stability while the latter examines subsequent balances and developments of economic phenomena. A procedure analogous to the theory of economic balance of the School of Lausanne is to be found in Saussure's

Cours.

Using the theory of value of the School of Lausanne as a model, Saussure reduces linguistic value to exchange value. The value of a sign is given by its position within a *sign system just as the value of a commodity is determined in "pure economics" (Walras) by its relationship to the other commodities on the market. Saussure says nothing about linguistic production just as economic production is not mentioned by the School of Lausanne, which studies the market in its ''purity," thus emptying it of his­ torical perspective. As Rossi-Landi says, "Saussure . . . does not appear to possess a theory of linguistic work, the most viable foundation for any theory of linguistic value"

216

Economics

(1915a: 139). The reason for this is to be found in the fact that Saussure overlooks the theory of labor value elaborated by the classical school of economics (Smith, Ricardo) and by 'fMarx; consequently the langue as a social phenomenon is reduced to the sum, the average of individual acts, the parole. The concept of the social as the product and as "external constriction" reveals a clear ideological tendency of individualistic flavor. Traces of marginalistic 'fepistemology are present also in Levi-Strauss's structural 'fanthropology, where categories used in the study of verbal language are applied to the study of 'fnonverbal communication and in particular to exogamic exchange. This is due both to the indirect presence of Saussure's structural anthropology and to the explicit relationship established by Levi-Strauss between structural anthropology and the eco­ nomic sciences modeled on the marginalistic theories of Neumann and Morgenstern. We do not here intend to analyze Saussure's theories, as this would of necessity entail a comparative study of the official Saussure and the Saussure who has recently emerged from the direct analysis of the course notes taken by his students (see Godel 1957). What we do want to insist upon, however, is that the Saussurean model of sign, which has had and continues to have a far-reaching effect on semiotic theory, reveals the influence of the marginalistic theory of the School of Lausanne. The comparison with Marxian criticism of political economy therefore appears to be pertinent and in some ways inevitable in the discussion of the concept of sign and sign value (see Rossi­ Landi 1968 , 1975a; Ponzio 197 8 , 1980 ). It is not merely a question of "transferring" Marxian theories of labor value from the field of political economy to semiotics. Such a proposal would simply be naive, for it presents Marxian theory as a definitive system, ignoring current discussion of the Marxian theory of value together with the subsequent interpretations which have accumulated around the works of Marx. In his criticism of political economy, Marx aims at deciphering the "language of commodities" (Das Kapital, I). He considers commodities as �-messages and looks for the explanation of the whole process of their functioning. By analyzing commodities in the context of communicative and social �· structures, we are able to overcome fetishistic visions which conceive of commodities as simply given and natural and which interpret the relation between commodities as a relation between things and not between men inside specific social structures. Marxian demystification of bourgeois economy and analysis of commodities in p ar­ ticular appear as semiotic analyses and are inseparable from the consideration of com­ modities as messages: the commodity is studied not only at the level of exchange but also at the l evel of production and consumption (see Rossi-Landi 1968 : 116, 1975a: 133-37). A commodity is a commodity rather than a mere product because it functions as a mes­ sage. If we deprive commodities of their character as messages and subtract the character of sign communication from the market, we will have neither commodities nor market. The production and consumption of a product is one thing, the production and consumption of a sign is another. A commodity is a commodity not when a product is produced and consumed in its use- value but when it is produced and consumed as an exchange-value, that is, as a message. All this makes economics a sector of semiotics. In this perspective, the semiotics of economics has its starting point in M arxian anal­ ysis, while on the other hand Marxian criticism of political economy can lead semiotics into a field of research yet to be explored. The result is that, despite the inclusion of eco­ nomics in semiotics, semiotics is not to be seen as an imperialistic and all- inclusive

Ecriture

217

science, as it accepts indications and suggestions from political economy. In the semiotic analysis of every kind of social sign, as in the Marxian analysis of commodities, it is a question of passing from the level of sign exchange and sign market to the underlying levels of the social work of communication and signification. In this way, sign structures appear as structures of human relations. In other words, we must achieve in semiotics what Marx achieves in his analysis of commodities and capital: social relations must be made to emerge in the place of mere relations between things and individuals reduced to things. A. P./S. P. Ecriture Ecriture is the common French noun for ";writing,' 'script.' As the term for the graphic

form of *language, it plays a preponderant part in Derrida's theory, where it is first opposed to parole, the spoken form of language, and then, as archi-ecriture, extended to refer to the fundamental differential system of language itself. Derrida's new concep­ tion of ecriture, combined with an equally influential revaluation of the same term by ''Barthes, has endowed it with a central importance in French *literary semiotics. Among progressive theorists it has come to replace the term literature in discussions of creative writing. This substitution may be held to have started with Barthes's Le Degre Zero de l'Ecriture ( 1953), although in that early book ecriture means no more than a "so­ ciolect," or *style of writing common to a generation of writers in particular sociohistor­ ical circumstances. In Barthes' subsequent work the term evolves to acquire the richer, more polemical sense of "writing" as a productive process whose raw material is words. The new notion of ecriture is part of a wider ambition to "demystify" bourgeois institutions; literature is one such institution and ecriture seeks to demystify it by exposing the secrets of its manufacture. Ecriture was presented - originally, at least - as a material ist theory of writing, neo-Marxist in inspiration and opposed to the idealism embodied in the traditional theory of literature. In their wish to fuse •:• aesthetic with po­ litical arguments, Barthes and his radical followers of the time (mostly associated with the Parisian review Tel Q ue/) assimilated the writer to a language-worker and the •�sig­ nifiers of the language he works with to the materialist infrastructure of which the super­ structure of signifieds or *meanings is derivative. Where the "bourgeois" �·ideology of literature gave the superstructure precedence, believing that content dictates expres­ sion, ecriture assumes the opposite, that expression dictates content. Along with the sacrosanct term literature, ecriture also rejects the two basic categories of literary thought: those of "book, " or "work of literature," and of "�·author." The work itself now becomes a "*Text," a term preferred because it does not imply "clo­ sure" into a self-contained and complete entity and because its etymology reminds us that a text is a weaving-together of a multiplicity of strands. The text is a tissue, not of original thoughts exclusive to a unique author, but of elements drawn in profusion from the writer's •:• culture: "a multi-dimensional space in which a variety of writings, none of them original, blend and clash" (Barthes 1968 : 146). Ecriture is thus collective and plural where literature was singular. Ecriture has no author; it is essentially anonymous, the product of the encounter between writer and language. This writer is sometimes allowed a modest degree of iden­ tity as a "scriptor". But his ecriture is never expressive of his self or inner state, nor is it

2 18

Ejxenbaum

representative of some preexisting state of things in the external world. The scriptor is not the master of language, as the author of old was presumed to be, but its servant or plaything. Ecriture is the activity in which he creates himself. It is both a theoretical and a practical activity: it teaches us a lesson in the actual creation of meanings. What we get with ecriture is a demonstration of the "practice of signification." As such, it does not lead us on teleologically, to some final, Platonic meaning, but scatters meanings indi­ scriminately along the way; it is a true process of dissemination, an exploration of verbal and hence '•semantic potentialities. As he works, the scriptor will yield to the superior force of words themselves and do their bidding, just as, in the Freudian *schema, the dream-work is performed without the conscious intervention of the dreamer. Ecriture is Freudian where literature was Cartesian, soliciting the cooperation of the unconscious mind and harking back to the belief of the French surrealists in the virtues of "automatic writing." The scriptor invites language to subvert whatever preconceptions he brings to his work. In so doing he works always toward meanings, never away from them, and recognizes the validity of Derrida's thesis that the signifieds of a language are also signifiers, forming as they do an inescap­ ably differential system of cross-reference. Ecriture lacks the implicit unity of literature: it is a chorale, where that term implies a solo performance, and generates innumerable and possibly contradictory meanings. It can be focused only in the mind of a '•reader, who is given a demanding and responsible role to play as the person who "rewrites" the text in the act of reading or "making sense" of it. Despite the brashly antithetical manner in which the notion of ecriture was first launched and then upheld in France - and its logical but also optional convergence there for a time with extreme left-wing political views - it can be valued as a theory of writing closer to the facts of the case than some versions of the theory it aspires to replace, ver­ sions which quite certainly accord too little influence to the autonomy of language and too much to the mastery and singularity of the author. J. S.

Ejxenbaum, Boris Mixajlovic ( 1886-1959) Leading theorist of *Russian formalism and its historian, one of the most productive of Soviet *literary, scholars and author of over 300 works, Ejxenbaum graduated from St. Petersburg University in 1 912 and taught at Leningrad State University from 1918 to 1949, and at the Leningrad Institute of History of the Arts from 1920 to 1931. Member of Opojaz later, during the 1920s, together with '•Tynjanov, he established Formalism in academic circles in Leningrad. He elaborated several crucial Formalist concepts including the notion of the dominant as the leading constructive principle of the literary '•text ( 1922a). In the theory of narrative he was responsible for the notion of skaz (narration organized by discourse rather than event). The idea for the analysis of narrative as discourse was first approached in his study of Gogol's story " The Over­ coat" ( 19 19), and later developed in "Leskov and Contemporary Prose" (published in Ejxenbaum 1927a). Ejxenbaum was also the author of a pioneering study of the *structural significance of '•intonation in verse as formed by the interaction of rhythm and '•syntax ( 1 922a); he

Eisenstein

2 19

tackled brilliantly the problem of the literary biography ( 1922b) and the problem of the interaction of literature and milieu ("Literary life," published in Ejxenbaum 1929a). Ejxenbaum is the author of a concise history of the formalist movement ("The Theory of the 'Formal Method'," published in Ejxenbaum 1927a), and he took on the role of the group's "official" spokesman. In Ejxenbaum's works problems of *history and *theory are closely interwoven (see especially his [ 1924] study of Lermontov). His main interest in the field of literary history were the writings of Tolstoj and Lermontov, although he wrote on most of the Russian classics. Ejxenbaum was also the editor and a contributor to the Formalist classic on theory of the *cinema, The Poetics of Cinema (1927c). All his chief literary theoretical works were republished in the Soviet Union in the late 1960s. A. S.

Eisenstein [Ejzenstejn ], Sergej Mixailovic ( 1898- 1948) Renowned Soviet director, theoretician of ,:•cinema, and aesthetician, perhaps more than any other major creative personality of his time Eisenstein combined the practical explo­ ration of a rich new *medium with a wide-ranging theoretical investigation into its place in the system of the arts. For Eisenstein cinema represented the pinnacle of the arts ; its synthetic and syncretic nature not only opened up new expressive possibilities but made it possible to discover the basic aesthetic laws governing art in general. Eisenstein's aesthetic theory emphasized the *pragmatics of art as a communicative process, i.e., the relation of ''sender and receiver to the �·text. Art models the complex inner state of the artist in both its "thematic-logical" and "image-sensual" dimensions and re-creates an analogous state in the receiver of the text. Hence, on the one hand, Eisenstein's theory of the cinema of "shocks" or "provocations", and, on the other, his abiding interest in the organic and synthetic nature of primitive *cultural forms, especially magic and ritual. · Eisenstein consistently insisted on the *sign character of the filmic image. As opposed to the more recent theoretical tendency to see the essence of the cinema in its indexical aspect, its ability directly to fix reality (cf. Bazin 1959-62 ; Henderson 198Gl), Eisenstein stressed the conventional (symbolic) nature of the object as captured on film, its status as sign rather than mere denotatum. While fully aware of what he called "the shot's tendency toward complete factual immutability" (1934, 1949:5), Eisenstein battled for the cinema of signification ; as he wrote of the close-up, "its principle *func­ tion . . . is - not only and not so much to show or to present, as to mean, to signify, to sym­ bolize" (1940, 1949 :238). It is indicative that major sources of inspiration for Eisenstein were highly conventionalized media like the Japanese Kabuki or the Chinese *theater, where "every situation, every object is invariably abstracted in its nature and purely sym­ bolic" (1935 :764 ). Furthermore, while recognizing and indeed intently studying cultu­ rally coded �·meanings (for example, in his writings on color, 1940), Eisenstein promoted an essentially textual semiotics : the signified was seen as determined primarily by the text-system rather than by pre-existing codes. "In art it is not the absolute rela­ tionships that are decisive, but those arbitrary relationships within a system of images dic­ tated by the particular work of art" (Eisenstein 1940, 1942 : 1 50).

220

Emblematology

Eisenstein wanted to deprive already given signs of their static character and elevate them into "symbols in formation" through juxtaposition with other signs and creation of new paradigmatic associations, or, to use his own favorite term, through montage. In this respect Eisenstein was fascinated by Japanese ideograms, both for their historical evolution from concreteness (*denotation) to *abstraction (connotation) and by their very '�structure, which involves the combination of discrete elements giving rise to a new whole expressing a concept greater than the sum of the parts comprising it ( 1929; Bystr­ zycka 1970). On the further evolution of Eisenstein's montage theory, especially after the introduction of sound, and the concept of vertical montage, see Eisenstein 1941 and Wollen 1972. The essentially semiotic nature of Eisenstein's theoretical and creative work is summed up by Eagle as an "awareness of the sign functions of cinema on diverse levels, with the existence of mutually overlapping signifiers which operate with respect to vastly different codes" (Eagle 1978 : 183). As Ivanov has shown ( 1976a), Eisenstein's synchronic and structural approach to film form was always counterbalanced by an evolutionary (diachronic) approach that was historical and psychological in orientation. Art was seen as limited to a small number of basic situations and symbols, the latter as forms whose appearance could be explained by tracing the early phases of the evolution of man and art, with the correlate attempt at unearthing archaic (prelogical) forms in contemporary sign systems. Although at times outmoded methodologically, Eisenstein's diachronic investigations are rich in observations of value for the '�semiotics of culture. Eisenstein's crowning works, Method and Grundproblem (still unpublished in full but extensively cited and analyzed in Ivanov 1976a), reveal the tremendous range of semiotic problems Eisenstein tackled. S. R.

Emblematology The word emblem (from Greek eµ�aUoo 'to insert') generally denotes a pictorial *sign composed of one or more elements which are taken to have symbolic value. Man's earliest emblems may well have been those cave drawings in which the image of the animal is marked by an arrow or a cross, i.e., in which two signs not naturalistically linked have been conjoined to produce more '�meaning than the sum of the signs taken individually. A more familiar historic form of the emblem is the coat-of-arms - to which is often joined a brief verbal motto, - first devised by medieval knights for purposes of identification and now used by individuals and families as evidence of their ancient and noble lineage. Because emblems are practically by definition smaller than pictures, their content must be relatively uncluttered and correspondingly highly symbolic. In antiquity the reverse of coins often bore an emblem (the dolphin intertwined about an anchor, with the motto "Festina lente," is one famous example), and in the modern world the prac­ tice has continued and been extended to postage stamps and even paper money. The reverse of the American one-dollar bill offers us a classic emblem: within a cir­ cle, a decapitated pyramid is set on a flat empty landscape; in the place of the missing apex an eye within a triangle radiates light; three inscriptions are linked to the pictorial elements: ''MDCCLXXVI," ' 'ANNUIT COEPTIS," and ''NOVUS ORDO SECLO­ RUM." The inscriptions are enigmatic both by their terseness and by being couched in

L 'Encyclopedie

221

an unfamiliar tongue, while the pictorial elements are strange by their form and by their lack of relationship to each other or to anything evidently American. In short, some sort of interpretation seems required. The Egy ptian reference of this emblem derives from a tradition current among Christian neo-Platonists and revived in the *Renaissance. It was believed that pre-lapsarian knowledge had been transmitted to the ancient Egyptians and had been conserved in their hieroglyphics, which were hence thought to be unmediated signs of that knowledge. In 1419 scholars discovered the so-called Hieroglyphica of "Horapol­ lo," reputed to be a pre-Christian book of hermetic wisdom. Though not printed until 150 5, it circulated widely in manuscript form and inspired an enormous interest in the possibilities of pictography. The most important descendants of the Hieroglyphica are probably Francesco Colonna' s Hypnerotomachia (Venice 1499) and Andrea Alciati's Emblemata (Augsburg 1531); it was this latter work which made the word emblem cur­ rent in the modern languages. From the sixteenth century to the beginnings of the Romantic period emblematic literature flourished and even spawned its own metalitera­ ture. (For the history of the phenomenon and its vast bibliography, see Klein 1957; Praz 1964 ; Gombrich 1972d; Henkel and Schone 1976; Daly 1979, 1980). It may be possible to see in this fascination for emblems - and for their relatives, the device and the impresa - the antithesis of the prevailing trend toward rationalism. Com­ menting on the Hieroglyphica, Ficino argued that whereas the human mind can grasp only sequentially the various ,;propositions of a symbolic image, the divine Mind can encompass their totality simultaneously (1576: 1768 ). Thus the more meanings one might instantly and intuitively perceive in an emblem, the higher one raised one's mind toward participation in the divine Mens. The concision and incisiveness which are conse­ quently implicit in the concept of the pictorial emblem related it closely to the poetic epi­ gram. Indeed, most emblem books were collections of pictures accompanied by short poems (e.g., Sceve 1544), the images and the words supposedly complementing each other. Francis �·Bacon was the first to equate gestures with hieroglyphs and emblems (1605, 1623). His implication is that gestures are immediate signs of the mind's conceptions, hence a direct, even pictographic form of �·communication. By the same token they stand in the same need of interpretation as do emblems. Bacon's use of emblem to signify a type of gesture was revived by Efron (1941), and subsequently by Ekman and Friesen (1969b, 1972b), and by Morris (1979). Whether pictorial, verbal, or gestural, the idea of the emblem corresponds to an apparently fundamental semiotic longing, that the mind may devise a sign so polysemous and multivalent, yet so evident, that it will transcend our normal epistemo­ logical processes. J. M.

L'Encyclopedie L' Encyclopedie (1751-1772) is perhaps the greatest testimony to, and at the same time the greatest intellectual catalyst of, the bourgeois drive for a transition from a textual­ ized to a grammaticalized cultural universe (to use the terminology of Lotman) or (to use a terminology a la •:• Peirce) from a *culture that exalts the *index to a culture that

222

L 'Encyclopedie

exalts the *symbol. This undertaking passed through a passion for totalizing semiotic discipline, often identified by critics (from Horkheimer and Adorno from 1947 on) with a program of dominion over nature both internal and external to man. But Diderot, Editor of the Encyclopedie, although perfectly conscious of the symbolic power of the *rules and habits of behavior ("establishing manners by law means making a cult of virtue, " he writes in the article on ' 'Maniere, " almost anticipating the Jacobin spirit), is keenly alive to the specificity of cultures and of *semantic fields and to their relativity to the different histories of peoples and languages, as well as being aware of the provisional nature of all *codes and hypotheses and of the daunting difficulty of drawing up ethical codes (see especially Le Neveu de Rameau) . The accusation of authoritarianism and standardization is also traditionally leveled against the theoretical linguistics developed in articles of the Encyclopedie (by Du Mar­ sais, up to volume VII, and then by his pupils Douchet and Beauzee). Despite the reas­ sessment by Chomsky (1966a), who, however, tends to reduce the linguistics of the Enlightenment to the '�Port-Royal model (as argued by Rosiello 1967 :140-48), the accu­ sation comes up again, for example, in Gusdorf 1973(31 2), �here the importance at­ tached to a general grammar as an instrument for analyzing particular grammars of lan­ guages that actually exist is seen as the outcome of an aprioristic philosophical ,:•anthro­ pology and serves to make way for value judgments and prescriptive ideologies in which "rational imperialism" leads in the end to a sort of "grammatical Gallicanism." This interpretation seems, however, to be contradicted by the repeated statements of principle made by the grammarians of the Encyclopedie regarding the absolute pri­ macy of usage: grammar must not impose rules on usage but must elicit and describe the reasons underlying usage. The grammarians' program is implemented by a method that is not so much empiricism, as argued by Rosiello (1967), as a kind of analytical and ,:•pragmatic rationalism. This leads to the identification (with particular clarity in the articles on "Grammaire" and "Langue," both by Beauzee and Douchet) of three levels or analytically detectable aspects of the constitution of the signification of �·speech in any language. The deepest is the *structure of intellectual *judgments revealed by logical analysis; this structure is common to all languages and is the subject of general grammar. Next we have the level of *syntax, of the various constructs whereby the deep structures are expressed in the various languages according to the rules dictated by usage. Finally, we have the phonological and surface semantic level, in which ideas, *uni­ versal or peculiar to a given culture, are enunciated in various ways in the various partic­ ular languages. The reference to a logical order common to all languages certainly also implies an attempt at philosophical justification that assumes, on the one hand, the immutability of human nature and, on the other, the uniformity of certain fundamental relationships in external nature. But this is secondary in the eyes of the grammarians of the Encyclope­ die: the logical a priori is a pragmatically necessary postulate to explain the fact and the analysis of the universally observable "*communication between men of different epochs, between peoples of different regions of the Earth, between any two individu­ als" (article on " Langue"). The article on "Sensations," perhaps written by Diderot himself or in any case inspired by him, is the one that expresses most clearly the philosophical-gnoseological background of the semiotic concepts of the Encyclopedie, precisely delineating the course from preconscious and prelogical perceptions to perceptual conceptions or

Endosemiotics

223

�·ideas. Having shed �·Locke's leveling approach, which reduced the materials of the life of the conscious mind to a single kind of brick, "simple ideas," the Encyclopedist makes a penetrating distinction, drawing on a module, later to be taken up in various ways, between (a) preconscious perceptions or "small perceptions," which are imprinted singly but overlap one another and crowd in on our nervous system, present to but not distinct for and not directly identified by the conscious mind; (b) unitary sensory percep­ tions or actual sensations, rich and vivid but "obscure," because they constitute "con­ fused perceptions" by the conscious mind of immediate small perceptions, affording an organized, though somewhat simplified, picture of them and one not devoid of ambiguity and uncertainty; (c) ideas or conceptual perceptions, voluntary intellectual representations, that the process of logical structuring, analysis, and definition by the reason renders determinate and hence "simple," "clear," and "distinct"; and (d) per­ ceptions of feeling, or passions, which, unlike ideas, refer not to external objects but to "the mind that produces them." Sensations, unlike ideas, present themselves to us as "in­ separable from bodily objects"; sensations, not ideas, pick up the profound message of preconscious "small perceptions" and testify obscurely but unequivocally to the condi­ tioning of the conscious mind by something outside it. Apart from the Encyclopedie articles already cited, the following ones on linguistic themes are particularly relevant : "Abstraction, " ''Adjecti/," "Alphabet," " Conjugaison, " " Consonne, " '' Construction," "Declinaison, " ' ' Etymologie," ''Figure," '' Langage," "Mot," " Sens," " Voyelle. " M.A. B.

Endosemiotics Sebeok ( 1 967 :3) differentiates the total scope of semiotics into anthroposemiotics, •fzoosemiotics and "a third domain, endosemiotics - which studies �·cybernetic systems within the *body." He stresses that "in this field the genetic *code plays a role comparable to that of the verbal code in anthroposemiotic affairs . . . and that it is still broadly true that the coding and transmission of �·information of differences outside the body is very different from the coding and transmission inside." There is a growing knowledge about sign processes and *sign systems inside the body and their importance for the maintainance of life. But there is a wide gap between our knowledge in different fields, as for instance between molecular biology, immunol­ ogy, endocrinology, neurology, metabolics, and other diciplines, on one side, and efforts towards a comprehensive *semiotics of biology, on the other. This gap is also responsible for the ironic fact that more and more biological disciplines adopt a termi­ nology of signs and signprocesses but rarely give themselves an account of the signifi­ cance and the epistemological background of the reasons which force them to do so. There is still a deep insecurity about the question of whether speaking of signs and sign­ systems in connection with biological phenomena says something about the "nature" of these phenomena or is only a *metaphor or •fanalogy of doubtful value, whereas the "real" events can only be "detected" by an analysis of the causal connection of physical and chemical facts. This shows the pressing need for a semiotic analysis of the relation between natural scientists and their objects. This analysis will reveal that molecules which we isolate by painstaking biochemical procedures can by their physical or

224

English Theories of the Sign in the Seventeenth Century

chemical qualities alone never be signs. They can at best become signs if they are appro­ priate to enter the exchange of information as carriers of *meaning in a living system of cells, organs etc. The '�structure of living systems and the interrelation of their elements can be revealed only by an analysis of their relations of meaning (J. v. ,:-uexktill 1940a; T. v. Uexktill 1950, 1979b). So it turns out that the use of a semiotic terminology in the disci­ plines which examine living systems within the body is far from accidental or pure conve­ mence. T. v. U.

English Theories of the Sign in the Seventeenth Century The most important aspect of the theories of *sign in English linguistics in the seven­ teenth century was the rise and development of the theory of the arbitrariness of the lin­ guistic sign. The origin of this idea must be connected with the general tendency toward what we could call a critique of �-language. This tendency was connected with contem­ porary considerations of scientific method and with the establishment of rigorous empir­ icism in *epistemology. Even from this point of view the influence of Francis '�Bacon was essential. A recur­ rent theme in his work (1957-59,1 : 1 35-39, 151-54, 161, 6 1 4 ff. and 3 :606-08) is the contraposition of dialectics, which are inadequate in the face of the complexity of nature, to induction, which, instead, stimulates the senses, and in a certain sense recon­ structs and repeats the course of nature. The objectiveness of reasoning, as it is expressed in dialectics, never coincides with the objectiveness of science. Sensory evi­ dence is always richer than rational evidence; true scientific method is not dialectical but experimental. This sort of skepticism was also encouraged by Bacon's theory of idols, or false images of things, that vitiate scientific operations. Among the different classes of idols, the most dangerous are those concerning '�names, inadequate signs that reflect the preju­ dices of the speaker more than the essence of the things designated (1951-59,1 :164, 170-72, 645-46; 3:396-97, 599). Bacon considered it possible that in Adam's language there was a perfect correspon­ dence between signs and things. But in human, conventional language, we cannot aspire to more than a partial congruity obtained through induction and the interpretation of nature, with the help of a philosophical study of grammar and of the "genius" of the var­ ious languages (1957-59, 1 : 1 32, 434, 465-66, 476, 654). Since the best guarantee of con­ gruity between sign and thing is sensory evidence, signs are the more adequate to their function the nearer they are to the images of the things represented. The highest degree of congruity will therefore be found in classes of signs such as gestures, emblems, *sym­ bols, and hieroglyphics, which have an immediate sensory correspondence to the thing designated. This linguistic skepticism was widespread in English linguistics in the seventeenth century. We also find it, together with elements that had their origin in the *Renaissance, in the mysticism of the logos, inspired by the works of Cornelius Agrippa and Jacob Boehme (translated and diffused in England in several editions between 1623 and 1661) and by the magic platonism of Robert Fludd. According to Webster (1654), for example

English Theories of the Sign in the Seventeenth Century

225

there is a "mystical Idiome" innate in man seen as microcosm and in the universe seen as macrocosm. But in man this language has been disfigured and cancelled by the super­ position of institutional languages, in which the congruity of sign and thing has been lost. Just as science is the inexhaustible reconstruction of a knowledge lost by man, so language is the reconstruction of a repertory of signs that, however, can never repro­ duce the image of natu re in its whole authenticity. If Webster used the "pars destruens" of Bacon's theory of sign, others used its "pars construens." Thus, for example, Ward, answering Webster with his Vindiciae Academiarum (1654), said that a language made of adequate signs must not be looked for in an innate language, which has never existed, not even in the earthly paradise, but in the advancement of science. Science will make possible the construction of an artifi­ cial language, based on a rigorous analysis of ideas and therefore able to reflect the struc­ ture of reality. This idea of an artificial language was very widespread during the middle of the seventeenth century. The two most famou s examples are the Ars Signorum (1 661) by Dalgarno and the Essay Towards a Real Character and a Philosophical Language (1668 ) by Wilkins. These projects are based on the idea that, by analyzing mental content, clas­ sifying all simple and complex ideas, and giving to each a symbol, we can obtain a system of absolutely univocal signs, free from the �·ambiguity and imprecision that char­ acterize natural languages. The theoretical premise for these projects was the linguistic coventionalism of the Aristotelian tradition. According to *Aristotle's De lnterpretatione, names are the con­ ventional signs of ideas, and ideas are the natural signs of things. What guarantees the relationship between name and thing is in fact the mediation of the idea. The theoreti­ cians of a universal language came to the conclusion that a rigorou s classification of ideas should therefore assure a univocal, congruous, and adequate �·meaning to the arti­ ficial signs. But the greatest epistemologists of the time - from �·Hobbes to *Locke - destroyed just this conception of the relationship between the artificial sign (name) and the natural sign (idea), and therefore the conception of language as simply the '�translation of a system of natural signs. The sign was seen by Hobbes as an element of a calculation; this calculation consti­ tutes human thought, and it is possible only because there are notae or marks artificially assumed to correspond to notions (1839, 1 : 12; 183 9-45, 1 : 14-15, 5: 1 9-20). The u niver­ sality of most of the signs that form language makes us think that the things themselves are u niversal, whereas this is a prerogative only of the signs that are said to be indefinite, that is, those signs whose area of significance is determined in the linguistic act (1839-45, 3 : 21-22). The relationship between language, which is formed mainly of general or indefinite '�terms, and experience, which, instead, is always concerned with particu lar objects, was one of the critical points of epistemology in the seventeenth century. Digby (1644) and Burthogge (1678 ) discussed it, partly anticipating the solution that Locke gave to it in the third book of his Essay on Human Understanding (1690b). Locke's semiotics is a radical formulation of the theory of the arbitrariness of the sign. The meaning of a sign is the idea that this represents. But ideas are not, according to Locke, as they are for Aristotle, natural signs of things. On the contrary, the ideas themselves are the produ ct of a process of *abstraction effected by the intellect, which

226

Enonce I Enonciation

can never use a real pattern. This is true in the case of the so-called mixed modes, whose abstract ideas are collections of simple ideas that have no natural pattern: they are therefore arbitrary to the highest degree. But this is true even in the case of ideas of sub­ stance; not even for them can we ever have any knowledge of their real, essential nature. It is just by the use of abstraction that we construct nominal essences, while the collec­ tion of the simple ideas that form them depends on the choice of the speakers, which in turn is widely determined by the specific necessities of �·commu nication, the conditions of knowledge involved, linguistic habits and conventions, in a word by different ambient causes. In other terms, the sign never expresses the real essence of the thing it indicates, but always and only a more or less wide and arbitrary collection of attributes. There is never a correspondence between the real classes of obj ects and the classes of obj ects that are the *referents of signs. Classification is never a specular representation of reality. The doctrine of the arbitrariness of signs is an essential aspect of what the adversaries of Locke immediately denounced as his "ideism, " or the "way of ideas." It was also the starting point of the most interesting developments in semiotics in the seventeenth cen­ tury, from •�Berkeley to *Hume, up to •�Condillac. It certainly marked the crisis of Aris­ totelianism in linguistics, and constituted therefore, even though still vaguely, the pre­ mise for the development of linguistics in the nineteenth and twentieth centuries. L. Fo.

Enonce/Enonciation When �·Saussure (1916) defined the obj ect of lingu istics as the study of •�language considered as a social product and a collection of conventions, he excluded speech as being too heterogeneous because it straddled several areas simultaneously - physical, physiological and psychological - and so belonged to both the individual and to society. Until recently enonciation, or the act of linguistic production, was assimilated to speech, and opposed to enonce or *utterance, the result of linguistic production. Moreover, since for Saussure speech was not a collective instrument and its manifestations were individual and momentary, its study, and subsequently that of enunciation, were thought to be impossible and were excluded from the domain of linguistics stricto sensu. The term utterance designates a segment, undetermined in length, of words emitted by one or several speakers, and a set of utterances constitutes the empirical data or corpus of linguistic analysis. For example, in distributional analysis, an utterance is con­ sidered a segment of variable l ength of the spoken chain delimited by formal marks, whereas for a Bloomfieldian structuralist, an utterance is made up of different levels, each level is made up of constitutes which can be analyzed into immediate constituents, and each constituent, except for the morpheme or ultimate constituent, can in turn be considered as a constitute. However, as Bally (1932) pointed out, "enunciation" is not a clearly defined notion. It is an individual act of the speaking subj ect made up of the totality of factors and acts which produce it, and herein lies the ambiguity of the notion. In an attempt to clarify this problematical area Greimas and Courtes (1 979: 126) proposed two different ways of considering enunciation: first, as a non-linguistic structure underlying lingu istic

£nonce I Enonciation

227

communication, or second, as a linguistic instance logically presupposed by the existence of an utterance which bears its marks. In the first case, terms such as communication situation or psychosociological context should designate the production of utterances actualized in such situations. In the second case, an utterance is considered as the result attained by enunciation and the latter appears as the domain of mediation ensuring the process by which the virtualities of language become the utterances of discourse. This theoretical position was handed down from �·Benveniste (1966a) who gave the first formulation of enunciation as a process by which natural language (Saussure's langue) is turned into discourse. Even though Benveniste (1966a:257 ) introduced the subj ect in the process of enunciation and distinguished language as a system of signs and �·rules from language as activity manifested through discourse, he nonetheless constituted the enunciative subject as only an achronological, positional and nononto­ logical being. Benveniste founded his analysis mainly on the presence of deictics (personal pronouns, certain adverbs) as well as *syntactic �·functions: interrogation, inti­ mation, assertion, but the phenomenon of enunciation remained for him essentially an act of appropriation of language by a speaker which permits him to situate himself only in relationship to the world and no more. Dubois (1969 : 10 8 ) stressed the necessity of substituting the continuous properties of enunciation for the discrete character of utterance and of constructing a theory which, unlike Benveniste's, gives the privileged marks of subjectivity to a greater number of elements. To this end he proposed four "concepts" which define enunciation in dis­ course: distance (the relationship between speaker and utterance), adhesion (marked by the presence of modals indicating the speaker' s degree of adhesion to his utterance), transparency (the preceding phenomena considered from the point of view of the hear­ er), and tension (the relationship between speaker and hearer). By reformulating the dichotomies of language/speech, ,;competence/performance, Dubois remained faithful to the linguistic tradition (for a critique of this tradition see Todorov 1 970b). *Austin (1962a) approached the problem from a more philosophically oriented perspective. He distinguished two different classes of utterances: "constatives" and "performatives." Of the two �·sentences "I am saying the whole truth" and "I swear to say the whole truth, " the first utterance is constative because it only describes an �·ac­ tion by its speaker, whereas the second is performative because to utter such a sentence is not to state that I am doing it, but to do it. The advantage of Austin's position is that, as Todorov (1970b :5) argues, �·speech act theory straddles linguistics and *anthropology since speech acts as " acts" are anthropological, and as "speech" they are linguistic. Austin's attempt to establish grammatical categories enabled him to define the intrinsic nature of the performative verbs and laid the foundations for speech act theory. His work, like that of Searle (1969), provided criteria for distinguishing in utterances three simultaneous acts: the locutionary act, the illocutionary act, and the perlocutionary act, thus reintroducing the notion of situation not only at the level of speech but also at the level of language, from a *pragmatic point of view. An utterance is a locutionary act insofar as it articulates and combines sounds and syntactically links the notions repre­ sented by words. It is an illocutionary act insofar as its enunciation constitutes an act. And it is a perlocutionary act insofar as its enunciation can have consequences or effects on the actions, thought, or *belief of the hearer. The context of enunciation (';intention, discursive situation, reference, *presupposition) becomes an operative notion, since for

228

Episteme

example, the analysis of presupposition enables one to consider enunciation from a lin­ guistic point of view and at the same time to open up investigation into extralinguistic fields (see Ducrot and Todorov 1972a, where the notion of presupposition raises the possibility of linguistically analyzing conventions and laws, or the institutional frame­ work which regulates intersubjective discourse). The analysis of context not only eluci­ dates the coded elements of enunciation but also goes far beyond its grammatical and deictic aspects by raising the possibility of studying the conditions of discursive production in concrete social situations. P. P.

Episteme Episteme refers to a group of interrelated sociocultural concepts defining the orientation, development, and conditions of birth of all forms of knowledge of a determined period. Foucault ( 1966) defines episteme as an epistemological field in which specific knowledge (scientific, philosophical, artistic, etc.) is viewed not in relation to any criterion of rational or formal value but in connection with its very conditions of being. As such, the study is more an archeology of thought than a *history of ideas. For instance, Foucault sees two important discontinuities in the episteme of Western '�culture, the first being that of the classical age and the second commencing in the early nineteenth century and continuing up to the modern age. Despite an apparent continuity in the theories of representation from the *Renaissance to today, analysis reveals that in the classical period, language and representation were indivisibly linked, in contrast with the modern period, in which theories of representation are not seen to be the basis for ordering nature and have no bearing on language. Thus, according to Foucault, defining the episteme of a specific age is essential for understanding how thought was generated and evolved during that period. Greimas takes a more rigid approach, considering an episteme as "the structure that defines the hierarchy of systems present" (1970a). This leads him to perceive a stratifica­ tion of semiotic systems potentially capable of generating all sociocultural manifesta­ tions within a given group. This all-encompassing point of view requires a great deal of structuralization but allows the transfer of hypotheses and results from one 'field of inquiry to another. Clarifying the attitude of a sociocultural community in relationship to the ,:•signs it generates, an episteme has to be seen as a connotative metasemiotic phe­ nomenon (Greimas and Courtes 1979) projecting a unifying view of the thinking and creations of a period. The notion of an episteme, though extremely useful, can easily be abused if no rigorous scientific methodology (especially for the transfer of results) is applied. A. G.

Epistemology Introduction

In another encyclopedia, epistemology, or the theory of knowledge, is defined as "that branch of philosophy which is concerned with the nature and scope of knowledge, its

Epistemology

229

*presuppositions and basis, and the general reliability of claims to knowledge". Hamlyn, to whom this definition is due (1967b:8-9), has given a fairly comprehensive survey of the history of this philosophical discipline up to the middle of this century. We aim to supplement his essay by pointing to some more recent developments. Nature of Knowledge The gist of �-Plato's investigations into the nature of knowledge is usually summarized by the so-called "classical" definition of knowledge as justified true *belief. Gettier' s famous note (1963) purports to refute this conception by the following argument: Sup­ pose someone, a, believes and is j ustified in believing that something, p, is the case; unfortunately, p isn't really the case, so a does not know that p. Suppose further that something else, q, is the case while a neither believes nor would be j ustified in believing that q; thus a doesn't know that q either. Now consider the *proposition that p or q: (pvq) ; believing p evidently entails believing (pvq) ; also, since a is justified in believing p, he must be j ustified in believing the logical consequence (pvq) ; finally, (pvq) is true since q is true. Hence, according to the "classical" definition, a must be taken to know that (pvq). B ut our intuitive opinion denies this: p not being the case, we do not want to admit that a knows (pvq) unless he knows q. Numerous amendations of the "classical" definition have been suggested to cope with this difficulty. Most of them consisted in strengthening the j ustification requirement or in adding some fourth condition of knowledge (e.g., "defeasibility;" cf. Lehrer 1974 ). It has been shown, however (Lenzen 1978 : sec. 2.3; Kutschera 198 1: sec.1.3), that each purported improvement along the "classical" lines was bound to fail; provided that such an analysis of knowledge takes the j ustification requirement, however strengthened, to be independent of the truth- condition, it can necessarily be refuted by a modification of the Gettier argument. Scope of Knowledge The problem of skepticism has occupied epistemologists for more than two millennia, and it is going on to occupy contemporary philosophers as is witnessed by the recent appearance of full-fledged books for and against skepticism (Unger 1975; Rescher 1980). We cannot enter this discussion here; suffice it to say that, in our opinion, skepti­ cism is readily refuted by a careful distinction between epistemic possibility and logical possibility (Lenzen 198 3; cf. also Kutschera 198 1: sec.1.7). Assuming, then, that skepticism is untenable and that, accord ingly, the scope of knowledge is not j ust null, we may ask whether our knowledge of the world is restricted to singular facts. This, of course, is the problem of induction. *Hume clearly saw that the method of inductive generalization cannot be justified rationally (in the philosophical sense of the word); inferences leading from the observation of some cause C having in the past always been followed by some effect E to the conclusion that also in the future C will always be followed by E are not logically valid. Yet Hume believed that they were basically sound inferences resting upon "a kind of pre- established harmony between the course of nature and the succession of our ideas" (18 8 2:40 ). In his famous book of 1965, Goodman set out to show that such an optimism con­ cerning the reliability of inductive generalization is without foundation:

230

Epistemology

Suppose that all emeralds examined before a certain time t are green. At time t, then, our observations support the hypothesis that all emeralds are green; . . . Now let me introduce another predicate less familiar than "green." It is the predicate "grue" and it applies to all things examined before t just in case they are green but to other things just in case they are blue. Then at time t we have, for each evidence statement asserting that a given emerald is green, a parallel evidence statement asserting that that emerald is grue . . . . Thus . . . the prediction that all emeralds subsequently examined will be green and the prediction that all will be grue are alike confirmed by evidence statements describing the same observations (pp. 73-4 ).

The upshot of this "new riddle of induction" is that inductive generalization - taken as a universal method - leads to inconsistencies. Thus the problem arises of finding criteria for distinguishing between admissible ("green-wise") and inadmissible ("grue­ wise") inductive inferences. None of the diverse attempts to solve this problem has gone without objection (cf. Lenzen 1974: ch.5). To be sure, the problem is only a theoretical or epistemological, but not a practical one. Humans simply do not generalize in the "grue-wise" fashion; nor is there any reason why they should do so. Our inductive practice has, until now, proven successful and it is thus pragmatically justified. What Goodman's riddle shows, however, is that alternative ways of inductive generalization would equally have proven successful; other inductive practices would also have been in harmony with nature's course. Therefore our choice of one particular inductive method rather than another lacks any theoretical or epistemological foundation. Presuppositions and the Basis of Knowledge

Speaking about presuppositions of knowledge one will almost automatically be reminded of ,;Kant's "Bedingungen der Moglichkeit aller Erkenntnis." As a matter of fact, certain trends in contemporary philosophy display a kind of renaissance of at least some of Kant's ideas. According to the so-called evolutionary epistemology not only man as a species but also man's •:•cognitive apparatus has to be viewed as a product of evolution. By the process of adaptation to the environment, this apparatus has become fit for yielding as much "knowledge" about the world as is necessary for the species to survive. But we are not entitled to assume that the picture of the world as produced by that apparatus is a true and objective mirror-image of the world "an sich." Visual per­ ception, for instance, must not be thought of as a neutral (re)presentation of something immediately given; rather it must be viewed as the organized output of our cognitive apparatus the input of which is a chaotic mass of sensory stimuli. This very rough sketch of an evolutionary theory of knowledge raises a lot of questions that cannot be an­ swered here. In particular, we cannot enter the discussion of idealism and realism. We have to confine ourselves to refer to some recent publications in this field: Campbell ( 1974) has argued in favor of a ''hypothetical realism;" a modern case for idealism may be found in Foster ( 1982); and a very good survey of the whole debate is given in Kut­ schera ( 198 1). Claims to Knowledge

It is an ideal of human ,;communication that everyone makes an assertion only if he knows that what he asserts is true. This ideal involves a (still ideal) sincerity requirement according to which one is justified in asserting that p only if one believes to know that p. However, since none of us is infallible, sincerity alone cannot guarantee the former

Ethnomethodology

231

ideal; even our most sincere assertions may fail to be reliable in the sense of warranting the �·truth of what is asserted. In this respect, claims to knowledge do not differ from other assertions at all. Moreover, someone's claim to know that p doesn't seem to be in any way stronger than the unmodalized assertion of p simpliciter although, of course, the statement that a knows that p is stronger than (i.e., entails but is not necessarily entailed by) the state­ ment that p. The above sincerity requirement which is but one of Grice's conversational postulates (cf. 1975b. 46) in conjunction with some simple epistemological principles imply that asserting p and asserting to know that p are pragmatically equivalent. For a is justified in asserting that p if and only if a believes to know that p. Now, epistemic logic (Lenzen 1980) tells us that believing to know (or being absolutely sure) that p is equiva­ lent to being absolutely sure that one knows that p. Hence a is justified in asserting that p if and only if a is justified in asserting to know that p. W. L.

Ethnomethodology The sociological approach to the analysis of the orderedness of everyday social activities (including verbal activities) as the ongoing achievement of situated organizational phenomena. Ethnomethodology is indebted to Schtitz's (1970-73) thinking in that social reality is considered to be created by a culture's members' everyday acting and everyday experience, by the methods members use to attribute sense to and disambiguate their social activities. Garfinkel, whose name is primarily associated with ethnometho­ dology and who also introduced the term, meant the word ethnomethodology to refer to these methods. The ethnomethodologist studies how members do methodology (cf. Gar­ finkel 1974:16), that is how they bring about that what they are doing is available to them, is "observable and reportable," i.e., "accountable." "Accounting practices" are always "reflexive," i.e., "they are carried on under the auspices of, and are made to happen as events in the same ordinary affairs that in organ­ izing they describe" (Garfinkel 1967:1). The reflexive quality of accounting practices constitutes the fact that their functioning is taken for granted by members; '�competence in their accomplishment of recognizability with respect to the quality of the ongoing activity is unproblematic for members. This is one of the decisive differences between members' methodological activities in everyday interaction and the ethnomethodolo­ gist's activities in doing ethnomethodology: "In unknown ways that accomplishment is for members a commonplace phenomenon. And in the unknown ways that the accom­ plishment is commonplace it is for our interests, an awesome phenomenon . . . " (Garfinkel 1967: 10). If it became an "awesome phenomenon" for members, the phe­ nomenon to be studied, the normality of everyday affairs would be destroyed. So it is not the methods used to make contextually situated activity intelligible that distinguish professional ethnomethodologists from the other members they study, but the fact that they are concerned with problems that are never viewed as problems by those who, in their unproblematic acting, create these problems. The reflexivity of accounting practices is also the basis for the indexicality of speak­ ing. Speaking is viewed as something like the prototype of social activities; competence in speech is seen as the competence of members that furnishes the normality to be de-

23 2

Ethnomethodology

scribed. Starting from observations made in logic and linguistics (cf., e.g., *Bar-Hillel 1954) that there are expressions in •:• language (like here, there, I, you tomorrow) that cannot be interpreted without making use of the contextual setting of the •�utterance, Garfinkel argues that indexicality is the destiny of all speaking, be it a lay activity or the scientifically ambitious endeavor to be obj ective which is characteristic of the professional. Since members normally try to repair indexicality, the insight into the ever­ lasting situational boundedness of all *communication is another quality of the profes­ sional ethnomethodologist in contrast to the members whose practices are studied. Due to the indexicality of speaking, the accomplishment of common understanding always has to rely on the specific situational setting of the activity in question, and the context is - as a consequence of the reflexivity- nothing but a part of the activity in question. There are two striking connections with semiotics involved in ethnomethodological thinking. One is the rej ection of a certain ''sign theoretical '' presupposition which is implied by the ethnomethodological concepts of reflexivity and indexicality. The other is the '�analogy with (generative) �·grammar which is often (cf., e.g., Gumperz and Hymes 1972:304; Weingarten and Sack 1976:15) used to characteri7 dtmomethodolo­ gy as the ''explication and *description of unknown regularities which unconsciously underlie speaker members' linguistic-social behavior: The analyt ic procedure of an ethnomethodologist does not distinguish between "what was said" and "what was talked about," assigning the semiotic conception of a correspondence between sign and '�referent to the relation between the utterances and what was actually meant (cf. Garfinkel 1972:3 17). The operations of accomplishing common understanding consist in the methods of speaking, "to recognize what is said means to recognize how a person is speaking" (Garfinkel 1972:320 ). The idea of the reflexive and indexical quality of verbal interaction rej ects the conception that *mean­ ings are expressed by signs, that a referential relation links two corresponding but inde­ pendent entities. Speaking itself is saying something; understanding, i.e., the accomplish­ ment of sense, of meaning, is recognizing that the manner of speaking is in accordance with a ''rule. The possible analogy between generative grammar and ethnomethodology is mainly associated with the name of Cicourel, although he is the one who in a very detailed way shows the differences between generative grammar and ethnomethodology, the short­ comings of the grammarian's rule explication in view of the ethnomethodologist's recon­ struction of the constitution of meaning and understanding. Cicourel (1973 ) argues that grammatical analysis of the form-meaning correspondence ignores the situation boundedness of speaking and the situation- specific quality of actual meaning which can never be grasped by the general "laws" and norms which are used in linguistic descrip­ tion. He gives credit to the fact that there are nonlinguistic ways of communicating meaning, and he uses experimental results of �·cognitive psychology concerning the pro­ cesses of perception and memory to argue that the verbalization of rules of understanding can never wholly account for the actual process of understanding. Conversational analysis, as a somewhat specialized approach within the ethnometho­ dological framework, seemingly shows an especially close connection to linguistics in that it deals with the organizing methods of verbal exchanges in conversations and thus seemingly shares with linguistics an interest in language. This superficial similiarity may be one of the reasons why ideas, observations and mainly terms that stem from the " ana­ lytic mentality" ( cf., e.g., Schenkein 1978 b) of conversational analysis have been largely

Ethnoscience

233

adopted by linguists, especially by those who are interested in the analysis of ,;texts and discourse. The superficial quality of the similarity, however, has been misleading in some cases (for critical reviews see Bergmann 198 1; Quasthoff 198 1) in that the ethnomethodo­ logical foundations of conversational analysis have been ignored, together with the fact that the interest of conversational analysis is not in language itself (Schegloff and Sacks 1974:233). The ways in which turn-taking (Sacks, Schegloff, and Jefferson 1978), the accomplishment of reference to persons (Sacks and Schegloff 1979), and storytelling (Sacks 1970-71; Jefferson 1978; Quasthoff 1980) - to name only a few of the aspects of conversation studied by conversational analysts - are organized in conversations have to be seen as the members' methods of acting sensibly and not as evidence for some con­ text-free *structure as part of a linguistic system which can be described objectively. If the mateials (records of natural conversation) were orderly, they were so because they have been methodically produced by members of the society for one another, and it was a feature of the conversations that we treated as data that they were produced so as to allow the display by the co-participants to each other of their orderliness, and to allow the participants to display to each other their analysis, appreciation and use of that orderliness. (Schegloff and Sacks 1 974: 234 ).

To produce sense, to "display orderliness," is always context-free and context-sen­ sitive, acting sensibly means to show how this acting in this situation conforms to this socially organized way of acting. U. M. Q.

Ethnoscience Ethnoscience is that subdivision of �·cognitive *anthropology devoted primarily to the study of folk systems of classifiying the external world. The term has sometimes been synecdochically used to denote the whole field of cognitive anthropology, as have its close synonyms ethnographic * semantics and ethnosemantics. In its widest meaning, eth­ noscience denotes the study of all systems of folk knowledge, and in this sense is some­ times known as ethnoepistemology, but its narrowest, and most common meaning is the study of native classifications of the natural environment. Its aims are twofold : to pro­ vide *descriptions of native systems of classification, and to discover *universals of human thought. Much of this program overlaps with other fields of inquiry, especially symbolic anthropology, *structural anthropology, sociolinguistics, semantics, and ,;artificial intel­ ligence. Ethnoscience differs from these fields in its presuppositions, methods, and focus. Having developed largely out of earlier *language and ,;culture studies, ethno­ science naturally presupposes that language is the principal representation of thought and the best means of studying it. The methods of ethnoscience are often formal, inspired by direct linguistic, mathematical, and logical analogies, and by an urge for quantification. Ethnosciencie is also rather narrowly empirical and oriented to visual perception, focusing on "*names for things" - on perceptually based systems of catego­ rizing the external world rather than on systems of concepts or symbols. Consequently, referential semantic ,;functions dominate its concerns and methods, with relatively little

234

Ethnoscience

attention being given either to sense or to figurative ,:•meanings or to questions of non­ cognitive meaning. Ethnoscience is structural rather than processual, its notion of structure based on vocabulary rather than discourse. A typical structure is a list of words arranged in some pattern of ,:•implication. Some of the more common structures are taxonomies, para­ digms, typologies, trees, scales, series, complexes, and focal/peripheral. Taxonomies, for example, are ordered by class inclusion, paradigms by class intersection, and com­ plexes by associative chains. Of these, taxonomies have been the favored subject of both descriptive and theoretical research. There are now a great many studies of native taxon­ omies and a substantial body of information on the taxonomy as a universal form of clas­ sification. Apart from universality, the major theoretical issues more or less specific to eth­ noscience revolve around the concepts of semantic domain, taxonomic rank, covert category, and variation. These have to do with how a culture's taxonomies are separated from one another, how a taxonomy's internal levels are differentiated, the role of unnamed categories, and the fact that different informants often use varying systems of classification. Prominent in earlier discussions, but less relevant now, are issues involving the con­ cepts of lexeme, marking, feature, contrast, and component. These concepts pertain little to questions of how words interact to constitute larger structures, but focus instead on the word and its constituents. Thus lexeme addresses the problem of what linguistic utterances count as labeling units, and markedness, feature, contrast, and component all deal with the composition of labels and their modes of differentiating themselves from one another. Less specific to ethnoscience are such theoretical issues as psychological reality, *synonymy, and polysemy. Does a particular taxonomy, for example, represent the native's way of seeing things, or is it just the ethnographer's arbitrary ordering? In eth­ noscience problems of synonymy and polysemy are narrowly interpreted within a refer­ ential context. The word-thing relation is either many words for the same thing (synonymy) or one word for many things (polysemy), and all instances of meaning transference implicated by these relations are construed metonymically as direct substi­ tutions that entail changes only in levels of referential inclusion or ,:•abstraction. All other tropic *transformations are excluded or abitrarily reduced to metonymic transference. In recent years various forms of focal/peripheral ordering have been favored topics, somewhat replacing the earlier interest in taxonomies and paradigms. Focal/peripheral forms of ordering are characterized by: (1) one or more lexemes of a domain functioning as foci; (2) all other lexemes relating to these foci as an overlapping but more or less graduated set of similarities. The focal lexemes are either prototypes, exem­ plars, paragons, or ideal types, which in ethnoscience are taken to function indexically or synecdochically, as when one lexeme denoting a particular category comes to signify not just its own category but the whole domain or some large part of it. Focal categoriza­ tion exemplifies the fact that all members of the same class of things are not equally good representatives of the class. This phenomenon received early attention in several areas of ethnoscience, but was developed primarily in studies of color terminology and later extended to a wider range of '�object classifications. The concepts basic object refer­ ence and category salience developed from these studies. Both of these signify the fact that some members of a domain are better representatives of that domain because they:

Ethnosemiotics

235

( 1 ) are learned first; (2) have more stable physical properties and are easier to learn; (3) are better represented, and more available to be learned; (4) are functionally more signifi­ cant and overlearned; or (5) are harmonies of the mechanics of perception and the nature of nature and are unavoidably learned. These observations address a major preoccupation in ethnoscience - the quest for universal principles of classification, or "laws of thought." The notion, only indirectly supported by evidence, is that modes of categorizing the world and the world's catego­ rizable objects are universal. This is a controversial position and one that is unacceptable to some who would categorize themselves as ethnoscientists. Basic works in ethnoscience can be found in several collections, most notably those by Romney and D'Andrade ( 1964), Hammel (1965), Tyler ( 1969), Spradley (1972), Casson (198 1 ) in the American Ethnologist, Special Issue on Folk Biology ( 1 976), and a useful introductory text is Burling 1970. Outstanding monographic examples of ethnoscientific research are Agar 1973, Berlin, Breedlove, and Raven 1 974, and Conklin 1957. Apart from the collections noted above, summaries of ethnoscience methods, aims, and research can be found in Evens et al. 1980, Frake 1980, Goodenough 1971, Kay 1 970, and Werner 1972. In addition to bibliographies in the works cited above, Conklin's extensive bibliography ( 1972) is a standard source. S. A. T.

Ethnosemiotics

Ethnosemiotics is a special domain of social and/or cultural semiotics which deals with �·sign systems and their *functions from the viewpoint of ethnographic approaches. The term was coined simultaneously and independently by four scholars at the beginning of the 1970s. As the Russian linguist Stepanov points out (197 1 : ch.2 p.2) ethnosemiotics studies the "hidden" level of human •:•culture. While, for example, folklore investigates a wedding song as a such, semiotics tries to detect its function and existence as one of the sign systems within a given society. It is well known, for example, that postural habits have a specific world distribution (Hewes 1955, quoted in Stepanov 1971) which indi­ cates physical and mental behavior as well as cultural fixation. Following the ethnolin­ guistic tradition of B. L. �-whorf and E. Sapir, Stepanov points out, cultural relativism is the very basis of different sign systems, and this is the reason why ethnosemiotics has been deeply intertwined with culture ,;history and linguistics. As Levi-Strauss explains the origin of verbal *semantics, *language does not absorb one by one the different �·meanings of words, but rather, by an epistemological leap from non-semantic animal speech, achieves fully semantic human language (Levi-Strauss 1968b). Stepanov, on the contrary, argues for a step-by-step development of human lan­ guage and its sign systems, since he does not consider the global *signifier and the global signified as central phenomena. Greimas ( 1973a), at the first congress of European ethnology (Paris, August 1971), envisages the possibilities for "ethnosemiotic" studies in three fields: poetics, *music and gestures. In general he is of the opinion that, in ancient societies, ethnosemiotic sign sys­ tems predominate, i.e., the global ,;myths are manifested at the same time in poetic, musical and gestural �·codes. In modern societies, where instead of "ethno-semiotic"

236

Ethnosemiotics

sign systems "socio-semiotic" sign systems exist, signs are produced and used by individ­ uals. Folklore, myths, and traditional lore in general are capable of expressing the sign system of a given society. Its efficiency attracts not only the scholars of social history, but also scholars of social semiotics. The Hungarian ethnographer Hoppa! (1971), in his outline of "ethnographic semiotics," focuses on *communication theory, stressing the fact that social �·anthro­ pology determines the addresser and the �-addressee in communication, while social psychology determines the coding and decoding, whereas folklore tradition is manifest mostly in the *message. This code-oriented approach may be based upon a paradigm by *Morris (1 938 :79) in which "man is the dominant sign-using animal." Usage plays a central role in Hoppal's ethnosemiotics, the task of which is to determine the elements of a given code, the connections between the elements, a semantic system of the ele­ ments, and the �-pragmatics of the sign system, which becomes manifest in the behavior of the individuals in every society. Dream analysis, folk '�beliefs, and cu stoms definitely show that in folk culture signs regularly occur both overt and covert. A more general outline of ethnosemiotics was given by Voigt (1971), who divided the scope of ethnosemiotics into three maj or chapters: pragmatic, ,:• syntactic, and semantic ethnosemiotics. It is a clear resemblance to the system of semiotics inaugurated by Morris (1938 ). Pragmatic semiotics deals with the function and use of specific sign systems in folk culture, as has been exemplified already by *Bogatyrev (1937 in S lovakian, 1971b in Russian, English translation 1971a), who studied folk costumes, folk songs, the Christmas tree, some folk legends and customs according to their mani­ festations in Slovakian folk culture. All ethnographic phenomena can be analyzed by pragmatic rules, as a scope belonging to semiotics, becau se in traditional folk culture usage is not determined by written and conceptual regularities, as is current in modern cultures, but by obsolete rules, which might be seem symbolic or cryptic even to their users. As for syntactic ethnosemiotics, the forerunners of this branch of studies are Mele­ tinsky and his collaborators (1974) who, following the morphological tale analysis by '�Propp (1928 in Russian, 1958 in English), have stated that morphological and structural regularities of folk narratives are su perposed by the system of functions of dra­ matics personae, which might be called a function circle, i.e., a constant scheme for all heroes within the folktales. S yntagmatic units thus change themselves into a more general system, the given positions of what can be filled by semantic rules. The most difficult, yet the most common domain of ethnosemiotics is ethnosemantics. The fact that various forms of folk culture are of a definite sign character (*writing, property marks, tattoos, masks, grave crosses and other symbols, symbolic colors, forms of folk costume, etc.) enables us to decipher the signifying character of a given culture. In his later summary Voigt (1975b) stresses the common but distinctive features of "ethnographic semiotics" and "semiotic ethnography." The former is an aspect of gen­ eral semiotics, in which signs are analyzed according to their different cultural and ethnic backgrounds. American trends in social sciences, like cultural anthropology and ethnolinguistics describe the differences in behavior, manners, customs and way of life. If we study the manifestations of signs within a given culture, where the culture is archaic enough and can be called a traditional one, in fact we can have "ethnic semi­ otics." On the other hand, semiotic ethnography is a transitional field between eth­ nography and semiotics. There exist possibilities for such a study in three main direc-

Ethology

237

tions : ( 1 ) narrative structure *universals are more than larger structural units ; (2) perfor­ mance in itself is a way of communication within a specific society; and (3) �·genre theory offers good prospects for semiotic research in general. The latest folklore or ethnographic reviews (see, e.g., Dorson 1973; Bauman and Sherzer 1 974; Ben-Amos and Goldstein 1975 ; Ben-Amos 1 976 ; Jason and Segal 1 977b) have clearly pointed out that in folklore research post-structural and post-communica­ tional trends regularily use semiotic terminology, and sometimes their results also are relevant for general semiotics. In ethnography proper, specifically, in the study of material culture, there exists a firm tradition of ethnosemiotic methods only as far as folk costumes are concerned, fol­ lowing the pioneering studies of Bogatyrev and �·Barthes ( 1 967). Lonnqvist ( 1 972) has analyzed the changes in folk costume of Swedish Finland between 1 870 and 1 920. Prop­ erty marks offer a good possibility for semiotic research ; the most proper method of analysis seems to be the mathematical one (Grafik 1975). Contacts of folklore with lin­ guistics and ethnography from a semiotic point of view are sketched by Kosova ( 1 973 ). In social anthropology symbols are very often understood by nearly semiotic means. As Firth made it clear ( 1 973 ), symbols are always interconnected with myth and ritual. The London school of symbolic anthropology (summarizing evaluation by Newall [ 197 6] from a traditional point of view) has offered good solutions for ethnic food and dress. Kinship system studies properly fall under semantics and not under semiotics ; eth­ noscience and * ethnomethodology are special forms of * cognitive systems. They should be studied by ethnosemiotic methods, but until now the main domain under which eth­ nosemiotics can be posed is �·semiotics of culture, which is a different field but has many similarities with ethnosemiotics, as Umiker-Sebeok has clearly pointed out ( 1 977a). In folklore research every meaningful *text can be analyzed by semiotic procedure : tales, proverbs, legends, songs, pictures, rituals and beliefs by that type of analysis are transformed into "texts " (see Voigt 1 976 ). In general, ethnosemiotics gave an impetus to various other fields of sign studies : myth and genre studies, summarized as "symbolic production" by Maranda ( 1 977); past and present forms of signs which have been studied by Hungarian scholars (see, e.g., Voigt, Szepe, and Szerdahelyi 1 975a; A tdrsadalom jelei, 1977; Grafik and Voigt 1 980), etc. Soviet folklore and myth research is indispensible for the semiotics of the Moscow­ Tartu School (see Ivanov 1 976a), in which both reconstruction of lost sign systems and general semiotics of culture are closely interconnected with proper ethnosemiotical investigations (Lotman and Uspenskij 1 976k). V. V. Ethology

The field of ethology developed as biologists became interested in the evolutionary origins and adaptiveness of the patterns of behavior that characteristically differ among species. Each species of bird, for instance, has a distinctive song or set of songs and other signaling patterns, species-specific habitat preferences and ways of foraging, and partic­ ular, seasonally adjusted social arrangements. These behavioral characteristics are most clearly evident in natural circumstances. Indeed, in captivity they often disappear or become grossly distorted. As a result, much ethological work involves careful observa-

238

Ethology

tion of free-living animals, and ethological experiments address aspects of behavior that are significant in the social and ecological environments within which the subjects evolved. The term ethology has often been identified in recent years with the procedure of naturalistic observation and with some of the typical subject matter of traditional ethology, suchs as acts of signaling, dominating, and territoriality. Since neither the procedure nor such subjects are exclusively within the province of biology, the term is now sometimes applied to research in other fields, e.g., to much of what is known as

human ethology.

The interests of biological ethologists span all behavior, from the sleep and foraging of individuals to social activities such as pairing, fighting, and living in groups. Among the first kinds of behavior to attract ethologists was social signaling, especially that involving conspicuously elaborated signals of individuals (termed displays whether vis­ ible, audible, *tactile, or specialized for other sensory modalities) or cooperative signaling performances (recently termed formalized interactions). Indeed, communica­ tive behavior was even the subject for a book, The Expression of the Emotions in Man and Animals, by Darwin ( 1872), whose interest in the evolution of behavior antedated the field. Darwin's eventual successors, for instance Julian Huxley, Oscar Heinroth, Konrad Lorenz, Karl von Frisch, and Niko Tinbergen, all did considerable work with signaling behavior. However, the field of semiotics did not begin to enrich ethology uni! the early 1960s, when several individuals, especially Marler (1961), Sebeok (1962), and Smith ( 1963), independently applied some of its concepts to the signaling behavior of nonhuman species. It is as yet unclear to what extent semiotic concepts devised to deal with human forms of behavior will influence the development of ethological theory and research. This is partly because we are still at a very early stage in our understanding of both the com­ municative and the cognitive procedures of nonhuman species. However, at a very gen­ eral level the basic semiotic division of the study of communication into three levels of abstraction - *syntactics, •:•semantics, and *pragmatics - is having an important influ­ ence now. For instance, ethologists have tended to confound interpretations of the '}in­ formation made available by signals by incorporating interpretation of the functional significance of signaling, not considering that •:•functions follow from responses (both pragmatic subjects) and that responses are elicited by information (a semantic issue) from many sources in addition to that from any particular signal. Full recognition of the •:•context-dependency of communication developed slowly in ethology, in fact, at least partly because of a failure to conceptualize semantic issues as distinctive. Most of the material that follows is developed more fully and with numerous exam­ ples in Smith 1977, and where no references are cited that source may be assumed. It is a synthesis of ethological research on communication, and employs syntactic, semantic, and pragmatic distinctions within an explicitly interactional perspective to provide the basis for organizing concepts and evidence. Other major reviews of the field, combining diverse theoretical positions and extensive reviews of the biological literature, have been edited by Sebeok (1968a, 1977a).

Syntactics Perhaps the first feature that strikes an ethologist comparing signaling specializations among species is their enormous diversity. Even nonspecialists tend to be aware on warm

Ethology

239

spring days that a great variety of "songs" can be heard from birds, and pet-owners know the growls , whines, barks, purrs, tail wags and twitches, ear-flattening, and hair­ raising of their dogs and cats, but this is only the slightest indication of the real diversity. Naturalistic study of any species of bird reveals numerous vocalizations, postures, and movements, most different from thos e of other bird species, and the same kinds of results are obtained, adding other sensory modalities, with mammals, fishes, insects, s piders, squids - the whole host of animal life. Some species even produce their own light, synthesize and deposit chemicals s pecialized as odors, emit strong pulses of elec­ tricity, or generate wave patterns on the surface of water. We will never know the full extent of the diversity, as there are simply too many species for us to be able to catalogue all the signals that have evolved. Even on the basis of a limited and sel ective survey, though, there does appear to be a syntactic characteristic that is very widely shared among animals and may be a funda­ mental feature. This is a common and rather low upper limit to the number of signals in each s pecies' repertory. Even highly social birds and mammals rarely have more than about 40 to 50 displays, for instance, counting those specialized for all sensory modali­ ties. The actual number recognized varies with the criteria of different ethologists, and some workers might tally twice as many units; there is also some difficulty in ass essing the numbers of units that are represented in the complexly graded signal complexes that comprise parts of the repertories of some species. By the most generous criteria, however, signal repertories are small. Why most species have only small repertories of signals is not known, although per­ ceptual limitations seem to be implicated. At least three syntactic devices appear to pro­ vide some extension of animals' abilities to signal, however. First, most displays vary at least somewhat in form, and the repertories of many species include sets or families of clos ely similar variable, often intergrading, displays . To the extent that variants remain distinguishable in spite of the problems recipients must have in categorizing them, e.g., because of environmental *noise, they should add to the effective size of signal reperto­ nes. Second, displays are often performed simultaneously: the human face and '�body provide perhaps the richest example of simultaneous arrays of visible signals, but other species commonly perform simultaneous combinations, for instance of sounds and pos­ tures. Potentially, these could have special significance qua combinations, although whether simultaneous combinations are ever informative other than as some sum of their parts remains to be tested. (While the question of how they inform is a semantic issue, it has syntactic implications : if the combinations are unspecialized concatenations of more basic recombinable units, then only the latter should be considered as s pecial­ ized s ignals. Nonetheless, performance of simultaneous combinations gives an animal control over some of the sources of information that occur contextually to each s ignal, and is thus a means of extending the capacity to signal. Third, displays are performed in sequential combinations, and in many species some of thes e strings are *rule-boun d. The rules govern such variables as the durations of intervals between successive units, and which units can be performed in sequence with one another. Ty pically, adherence to such rules of performance does add information not made available by the individual s ignals themselves, and thus sequencing rules are devices that can significantly extend the signaling capacities of s pecies with limited numbers of displays. In fact, the most conspicuous examples of ex panded signal reperto-

240

Ethology

ries are found in the sequentially organized singing behavior of some species of birds, raising the possibility that such organization permits significant additions to basic reper­ tories. Semantics

The initial assumptions in ethology were that displays and other specialized signals com­ municated emotional and motivational states. These assumptions guided semantic inter­ pretations from Darwin through Tinbergen, and still influence many studies. They lead to oversimplified semantic models, however, and have the problem that they must remain very imprecise until we have much clearer understanding of the physiological nature of motivational and emotional states. Further, they tend to focus attention on the internal condition of signaling animals rather than to ask what information the signals make public. The procedures ethologists devised to make motivational and emotional interpre­ tations involve study of the circumstances in which signals are performed, the other activities with which signal performances correlate, and to some extent the form of the signals (many of those resembling biting or striking, for instance, reflect aggressive motivation). The first procedure yields criteria that can be difficult to apply to the study of *referents, since most displays of at least birds and mammals are performed in diverse circumstances. The second procedure has been more useful thus far: the performance of most, perhaps all animal signals does appear to correlate reliably with particular kinds of behavior. A signal thus provides information about that behavior, and (viewed prag­ matically) could be used by a recipient to predict what acts the signaler is selecting (or may be about to select) from its behavioral repertory; such information is relevant as animals deal with one another, negotiating status orders or territorial boundaries, reaf­ firming and modifying relationships, keeping their foraging groups coherent, and so forth. Surely the information also reflects motivational states; any display correlated with a restricted range of activities must share whatever motivational causes those activi­ ties have in common. Many displays, however, are not consistently related to any partic­ ular emotional states. Comparative research on the behavioral referents of animal displays is still based on a limited number of species. Nonetheless, it appears to be revealing a widespread semantic pattern among birds, mammals, and at least some other vertebrate animals. Behavioral referents of the displays of different species are remarkably similar, and seem to involve few classes of behavior: by conservative criteria, only about a dozen are recognized at present. Even species of very different evolutionary lineages, ecological habits, and social lives share these classes, although exceptional behavioral referents may be most frequent for displays of species with either very brief and simple or unusually continuous and complex social arrangements. In addition, many referent classes of behavior are delimited quite broadly. For example, displays that inform about interactional behavior occur in most or all reperto­ ries, and in the case of many such displays no particular kind of interacting can be pre­ dicted. Such displays are performed in correlation with the whole gamut of interactional behavior, including activities as diverse as attack, copulation, and simply "associating" (i.e., staying in the presence of another individual or permitting it to remain in the signaler's presence). Other displays, with narrower interactional corre-

Ethology

24 1

!ates, may have attack, copulation, or association behavior as their referents or, in rela­ tively few species, other classes of behavior, such as mutual grooming or preening, but many kinds of interactional behavior are not specifically predicted by displays. Other classes of behavioral referents include the behavior patterns performed when the signaler is vacillating indecisively, when the signaler is seeking opportunities for some kind of activity, when the signaler is locomoting, and for a class of diverse activities dis­ tinguished only by their incompatibility with some class such as interaction or escape. A display with the class of incompatible acts as referents must have as additional referents both the alternative(s) and the class of indecisive behavior : the display thus predicts choice from among these possibilities. Among the more narrowly delimited classes of referents is escape behavior. Still, dis­ plays making information available about this class may, in different events, correlate with anything from slight withdrawal or turning away to precipitous fleeing, and with other tactics such as freezing motionless. There are species, especially in very open habi­ tats in which neighboring individuals are unusually dependent upon one another for information about dangers from various kinds of predators, in which different displays subdivide the class of escape behavior into different tactics. Although these species seem at the moment to be exceptional, the evolutionary conditions under which any of the rather grossly defined classes of behavioral '�messages is subdivided among different dis­ plays within a repertory, and the nature of the resulting subdivisions, need further atten­ tion. These are important issues, and much zoosemantic research can be expected on them in the immediate futre. Displays do not make reference simply to different kinds of activities. They also contain information about the probability or relative probability that each of the two or more kinds of activities among a display's referents will be performed. Within the reper­ tory of a single species, two or more displays may provide information about the same kinds behavior, say the possibilities of associating with a companion, or fleeing, and indecisive activities in which the signaler may engage while not yet committed to either associating or fleeing. For each such display the probability of indecisive acts will pre­ dominate, although to different degrees, and for one display the probability of fleeing may be next highest, for another associating. The same displays will also make available information about how the predicted activities will be performed, e.g., how forcefully or completely, and with what orientation. Viewed pragmatically, such an array of signals must provide useful clues and options as individuals negotiate their interactions and move to appease or confront one another. Information about the probability that behavior patterns will be performed is crucial to signaling, yet it is difficult for observers to measure. This is because the probabilities with which different actions will be selected at the instant of signaling do not become static. Circumstances change, sometimes in no small part due to responses recipients make to a signaling act, but also for various other reasons, many of which cannot be fully anticipated. These changes can increase, eliminate, or otherwise alter the usefulness of performing any particular behavior pattern. A signaler is not bound to perform as its signals indicate; the probabilities they express are conditional. But the signals are nonetheless real for being conditional, and can be assessed at least by relative measures. Thus, for any display, the average relative probabilities of alternative behavioral pat­ terns can be determined from numerous observations of the patterns actually performed

242

Ethology

in a full range of circumstances, and similar observations permit comparisons of infor­ mation about probability among different displays. Displays make available information about more than the signaler's behavior: they also have nonbehavioral referents. For instance, they identify signalers to species and often even to individual levels, and sometimes indicate their extent of maturity, their sex, or their breeding condition, and even their social affiliation with certain groups. Most audible signals contain information that gives away the location of the signaler. And ethologists have long been interested in the additional possibility that some displays have as referents things or situations apart from the signaler itself and its own activities. In many circumstances the recipients of a display respond as if information had been provided about something such as a predator, food, a nest site, or a territorial boundary. It does not follow, however, that the display makes available all the information on which the recipients base their response - indeed, that would be exceedingly unlikely. Much information comes from other sources, e.g., from evidence of the location of the event, from the direction in which the signaler is seen to attend or act, from the behavior (e.g., food seeking or spatial patrolling) underway when the display is performed, and from the responses of other recipients. Even conditions that deny important information, such as the recipient's being in a position from which neighboring individuals cannot be seen, are informative and affect response selection - in that example often by biasing animals to act as if assuming the worst or causing them to give priority to seeking further information. Other information is brought to bear from the experience of the responding individuals. That displays themselves might have referents apart from the signaler has been sup­ ported until recently only by inconclusive evidence. New studies, especially that of Owings and Leger ( 1980) on California ground squirrels, have demonstrated such cases. Distinguishable variants of a vocalization of this rodent correlate both with different activities such as fleeing, monitoring, or approaching (or different assortments of such activities, or different relative scalings of probabilities) and also with different eliciting situations. The latter are variously defined : some by particular classes of preda­ tors (e.g., coyotes vs. badgers vs. snakes) and others by intraspecific social events such as chases. Each vocal form that correlates with a particular class of predator has as one of its several referents an entity other than the signaler. Similarly, observations and experi­ ments by Seyfarth, Cheney, and Marler (1980) showed that different calls of vervet mon­ keys correlate with different classes of predators. Evidence was not offered to show whether these calls also enable prediction of signaler behavior, although some results suggest that a predator alone is not sufficient to elicit vocalizing by a vervet who does not respond in other ways as well. The vocalization uttered by ground squirrels and vervet monkeys in response to pre­ dators belong to complex sets of displays with variable and intergrading forms. Many species of birds and some mammals have such complexes in their repertories, and addi­ tional cases may be shown to involve referents apart from the signaler. Even if such refer­ ents are uncommon, however, the variable complexes of signals suggest considerable diversity and perhaps subtlety of referent classes and are among the richest subjects for current research in animal communication.

Ethology

243

Pragmatics Why signals have their forms and their referents, what responses they elicit in different events, and how signaling and responding function for the individuals involved are prag­ matic questions. They are understood to very different extents, and differ greatly in the difficu lties they pose for ethological research. Some aspects of syntactic form and diversity are readily studied. Ethologists have found, for instance, that the babel of many species is a local source of confusion and signal masking. Some differences among the signals of species that live within the same region are adaptations to this background, as members of each species must be able to distinguish one another's signals from those of other species. Other aspects of form and differences among signals correlate with other environmental features: the density and kind of vegetation, for instance, affects the distances to which sounds of different fre­ quencies travel and the colors of dermal or epidermal ornaments that can contribute to making signal movements conspicuous. On the other hand, still other differences must resu lt from the divergence of evolutionary lineages responding to selection pressures not directly related to signaling. Thus while the form of each signal may be adaptive in many respects, not all of the differences among signals need be functional per se. Why the display repertories of most species of animals may deal in the same small set of behavioral referents is another pragmatic issue. The most promising clue is surely syntactic: the limited number of displays available in each species' repertory. With small repertories, evolution must usually favor displays providing information that can be usefu l in many different circumstances. There may be relatively few classes of behavior significant enough to be selected as referents, leading to little semantic divergence among species. Evolutionary pressures must also favor expending at least some displays on information that may not be functional in the broadest ranges of circumstances because it can be crucial when a display must elicit an appropriate response with minimal delay. Events in which attacking and fleeing are of central importance are common examples. Again, rather few classes of behavior appear to be significant enough to have influenced the evolution of signaling in most species. It is also possible that the syntactic clue here is sometimes misleading. The apparent grossness of most classes of behavioral referents must in some cases result from research that considers together as one unit a range of display forms. This procedure is appropriate if animals must often have to do similar lumping becau se in noisy situations recipients cannot determine which variant of the display has been performed. However, the greater the range of the variation the more likely it becomes that recipients do make distinctions (more or less finely under different circumstances), and hence do recognize subdivisions and other modifications or derivatives of the gross referent classes. As already indicated, the extent to which this occurs, and its implications for semantics and pragmatics, are important issues in current ethological research. Studying responses to animal signals is difficult because recipients respond to infor­ mation from many sources in addition to the signal in any event. Experimental control of contextual sources is difficult without grossly altering circumstances and violating the expectations of subj ects. "Natural experiments" do occur: careful observers can make good use of events that differ from one another only minutely, or primarily in one very important variable (such as the sex or age of a recipient), and in which recipients of the same kind of signal behave differently. Limited experimental intervention can also

244

Ethology

be done in the field, for example by presenting a tame predator (to obtain examples of circumstances that would otherwise be observed only rarely) or by artificially placing scent markings. But limitations are quickly encountered. For instance, when recorded songs are played back to territorial birds, only the initial responses are readily interpreted; situations rapidly deteriorate as the test individuals become disturbed by the presence of signals in the absence of signalers with whom they can interact. Laboratory experiments greatly restrict the options that more natural environments present animals, for instance in removing the opportunity for each participant to with­ draw from the other. Yet the laboratory offers much better control over available sources of information. It can be used to address questions that have been found signifi­ cant in the field and are understood well enough to be phrased with some precision and appreciation of the effects of impoverished circumstances. To be effective, all possible techniques have to be brought to bear on the study of responses : field observation to reveal naturally occurring patterns and suggest apparent responses, the seeking of situa­ tions that will produce repeated natural experiments, limited experimental intervention in nature, and carefully controlled laboratory experiments. The responses elicited by signals have consequences for both the signalers and the responders. Some consequences are disadvantageous (a singing bird will attract hawks; a male firefly approaching the flashes of a female of a predatory species, who mimics flashing patterns of his own species, will be eaten). Others are advantageous (the male bird or firefly may also meet an appropriate female, and breed). The advantageous con­ sequences of responses generate their functions for both parties - in evolutionary biol­ ogy, the concept of function involves adaptiveness. Studies of the costs and benefits of signaling and responding to signals are the bases of functional analyses. Costs and benefits are difficult to measure, however, even apart from the need to show the involvement of a signal in generating a response as a prior condition to investigating function. To an increasing extent, research on functions is being guided by theoretical models of evolutionary processes, particularly those arising in what is popularly known as sociobiology (Wilson 197 5). At present, however, most representations of communication in this modeling are highly simplified. A current theme in evolutionary theory stresses the advantages of "selfish" behavior, and has generated interest in misleading signaling. Certainly, the extent to which signaling is reliable is a pragmatic issue of fundamental importance. However, extending the selfish theme to its limit has led to the suggestion that reliable communica­ tion "does not accord with the theory of natural selection" (Krebs and Davies 1978 : 1 57). This extreme view overlooks at least two issues. First, in many kinds of cir­ cumstances the participants are so closely interdependent that each has nothing to gain by misleading the others, and only reliable informing provides any advantages. For instance, the vocal signaling while a flock of chickadees (Parus sp.) or a troop of baboons (Papio sp.) forages appears to help the groups to remain coherent even though their members often cannot see one another through the vegetation. Each member needs the group and can both keep with it by attending to the vocalizations and help it to keep with him by uttering them. Second, there is the problem that misleading signaling is mimicry, trading upon the pervasiveness of reliable communication. Neither this nor any other mimicry can succeed without the prior and continued existence of its reliable models.

Etic I Ernie

245

Although much signaling must therefore be reliable, we still must ask to what extent misinforming evolves. It is well known in interspecific predator-prey interactions where various, often remarkably detailed kinds of mimicry have been developed. But preda­ tor-prey interactions are extreme and, as social events, minimally complex. lntraspecific misinforming could be much more intricate. All misinforming, however, leads to evolu­ tionary instability because the pressures of natural selection acting on signalers and recipients are in opposition. If signalers gain by misinforming, their misled recipients must almost always lose by responding. There is inevitably evolutionary selection for skepticism, and evidence is beginning to appear showing both that recipients have ways of actively checking and testing signalers and that unreliable signals or signalers can be devalued or recalibrated. Skepticism itself creates pressures on signalers to be able to cer­ tify their own reliability, for instance by accepting handicaps (Zahavi 1975, 1977). Mis­ leading signaling must occur primarily where it is hardest to detect (e.g., if done rarely or irregularly, or in relationships too ephemeral to permit much checking) and where the gains to the signaler far outweigh both the costs to the misled recipients and the costs of being discovered. Fortunately, because biologists are currently so interested in this topic, ethology should soon have a clearer understanding of the extent to which misinforming limits the evolution of reliable communication. W.J. S.

Etic/Emic Two terms first coined by Pike and used in the first edition of his monumental work on language and culture (1954) (and derived from the linguistic therms phonetic and pho­ nemic). From the beginning, Pike intended these terms to apply in the analysis and de­ scription of both verbal and �·nonverbal behavior - i.e., to the analysis of languages and •=·cultures. The terms are crucial to the definition of the central unit of Pike's linguistic theory (tagmemics) - i.e., the •=·tagmeme. Pike states that the etic approach classifies systematically all comparable data of all systems, provides a set of criteria for such a classification, organizes the elements into �·types, and studies and identifies any new found data in reference to the system which the analyst has created. On the other hand, the emic approach is valid for only one lan­ guage or one culture at a time and is an attempt to discover and describe the patterns of that particular language or culture in reference to the way in which the various elements within that one system are related to each other (Pike 1967 :37). The etic view may be pre-analytic or extra-systematic. That is, etic units may be uni­ versally defined (e.g., with universal �-descriptions such as "voiceless bilabial stop") or they may be components within a system which are non-significant to that system (such as allophones, allomorphs, etc.). That is, etic units may be non-constrastive, predictive components. Ernie units, however, are always systematic, and within only one coherent system. They are significant and contrastive within that system. An analyst endeavoring to describe a new system will first note units which are sim­ ilar (from some point of view) to those in his own system or in other systems with which he is already acquainted. At this point, to the analyst, all units are etic - and only after he has discovered systematic relationships between units can he define them emically, as to their contrastiveness and significance in their own system.

246

Explication

Some anthropologists have found this viewpoint useful in describing behavioral and symbolic differences across cultures. A thorough discussion of anthropological use of these terms is found in Fisher and Werner 1978. For other applications of these terms to both verbal and nonverbal behavior, see various articles by Pike in Brend 19726 and the bibliography therein. Thus, in semiotic studies, signs may be viewed etically or emically: a typology of the signs in a number of systems would be etic; a description of the *function and significance of the signs within one particular system would be emic. R. M. B.

Explication Giving an explication for (or explicating) an unclear concept (one that is vague, ambig­ uous or simply lacking in �·definition) consists in replacing it by a new, exact, and well­ defined concept that is close enough to the former one to take over its role in scientific (philosophical etc.) discourse. The earlier concept is called the explicandum, the new one an explicatum of the former. The condition of adequacy of an explication (that makes clear the sense of the expression "to be close enough" as used above) goes as fol­ lows: The concept C' is an adequate explicatum of the concept C if and only if no impor­ tant *proposition containing C changes its supposed ,;truth-value (the property of being true or false) when C is replaced in it by C'. Thus, the adequacy of an explication depends upon the set of contexts of the explicandum that are taken to be important. Classical examples of explication are: ,:•Frege's definition of number, ,;Russell's theory of �·descriptions, Tarski's definition of truth, and ,:•Carnap's definition of necessary truth. The above defined, technical sense of the term explication as used in contemporary philosophy is due to Carnap (see 1947:7 f.). Earlier philosophers used the term analysis in a similar, but non-identical sense : the difference consists in their claim that analysis should result in the concept which is �·synonymous with the earlier one, or, in other words, that analysis should uncover the hidden ,;meaning of the concept being analyzed. This assumption underlies the so-called paradox of analysis (see Carnap 1947:63, Quine 19606:258 f.). Explicatum is not supposed to be synonymous with explicandum: instead they are required to satisfy the weaker condition of being cognitively equivalent in the sense explained above. B. S.

Expression vs. Content "Le signe linguistique unit non une chose et un nom, mais un concept et une image acoustique" e saussure 1967:98). With this statement Saussure gave twentieth-century linguistics a new direction by demanding that *language should be studied as a system of immanent entities forming a *structure (langue) and not as the individually produced linguistic *text, or speech (parole). Saussure calls the acoustic image �·signifier (signifiant) and the concept signified (signifie'), and their combination constitutes the ,;sign (signe).

Expression vs. Content

247

Although their combination is arbitrary, in the sense that there is no necessary or moti­ vated connection between signifier and signified, it is this combination that creates the division into differentially articulated form elements both within sound and within thought. These form elements, called values (valeurs), constitute two systems, one on each language plane, and it is only because of the internal differences within each system that it is possible to recognize the elements of a given sign and thereby differen­ tiate it from other signs. Saussure does not use the concepts expression and content in any systematic way, but �-Hjelmslev uses them, identifying expression with signifier and content with signi­ fied. However, whereas Saussure's point of departure is the sign, and signifier and signi­ fied are defined as the two parts of it, Hjelmslev's starting point is the text and the system behind it, and he often talks of the expressionplane and of the content plane of lan­ guage; the partition of the text into these two planes is the first to be undertaken in an analysis and is presupposed in every other partition of the text. He introduces the con­ cepts in this way : We have here introduced expression and content as designations of the *functions of the func­ tives that contract the function. This is a purely operative definition and a formal one in the sense that, in this context, no other meaning shall be attached to the terms expression and con­ tent. (Hjelmslev 1953: sec. 1 3.)

This very strict operative definition, however, is followed by a subdivision of both the expression plane and the content plane into form and substance, and even if the parallelism, or isomorphism, of the two planes is maintained, the expression form and the content form are defined in relation to the substances they articulate. Hjelmslev gives several examples of this and the following formulation is typical of his point of view : A paradigm in one language and a corresponding paradigm in another language can be said to cover one and the same zone of purport, which, abstracted from those languages, is an unanalyzed amorphous continuum, on which boundaries are laid by the formative action of the language. (Hjelmslev 1953 : sec. 3)

This is followed by an example of how the content systems of the color designations divide a part of the color continuum differently within English and Welsh :

gwyrdhd green blue gray brown

glas ilwyd

While the content system, or rather the content systems, constitute specific divisions made within a given language with regard to its designative or referential function, the expression form of a language is the specific way in which, for instance, the phonetic zones of purport (i.e., expression substance) are divided into different •}figura (pho­ nemes).

248

Expression vs. Content

The parallelism between expression and content is dearly seen in the way in which it is possible to analyze them below the sign level. This can be illustrated as follows:

A

:

Expression

._ I __ :_ ___L_

B

A

Content

B

d_ _

__J

Both expression and content are analyzed in such a way that each element (phoneme and lexeme) is defined relationally, differentially, and negatively in relation to the other elements in the paradigm. The paradigms can be further divided into constitutive com­ ponents, as in this example the content can be analyzed into four elements ('�semes, or in Hjelmslev's terminology, pleremes) : A: male; B: female; a: adulthood; and b: child­ hood. The expression paradigm, can be analyzed into A: bilabial; B: labiodental; a: unvoiced; and b: voiced. Hjelmslev's claim that the expression plane and the content plane are isomorphic and can be analyzed in the same way has been very influential. Greimas has developed this point of view in his studies in structural *semantics, and in his paper "Pour une theorie du discours poetique" ( 1972: 14), where this isomorphism is represented in this way:

expression plane =-----------------_ j

q) . q ) :::> p. From a logical point of view Aristotle's second type of metaphor seems more acceptable, since it represents a correct case of modus pon­ ens: ((p :::> q) • p) :::> q. Unfortunately a material implication can sound very unconvincing from the point of view of natural language. Thus the metaphor of the second type is a really unsatisfactory one. The example provided by Aristotle is "Indeed ten thousand noble things Odysseus did," (where ten thousand stand for 'many,' a genus of which it is allegedly a species). The clumsiness of Aristotle's example is self-evident. In fact, ten thousand is necessarily 'many' only in a Porphyrian tree that is based on a cer­ tain scale of quantity. One can well imagine another scale oriented toward astronomic sizes, in which ten thousand, even a hundred thousand, is a rather scarce quantity. In other words, while it may seems more or less necessary for a man to be mortal, it is not as necessary than ten thousand be a lot. This notwithstanding, ten thousand suggests 'many' intuitively and with an undeniably hyperbolic tone, while men for 'animal' is not intuitively perceived as an interesting figure of speech. Both examples depend, however, on the same logical scheme. Probably, according to the code of the Greek language in the fourth century B.C., the expression ten thousand was already over­ coded (as a ready-made syntagm) and was used to designate a great quantity. In other words, Aristotle explains the modes of interpretation of this synecdoche taking as already disambiguated the synecdoche itself - a new example of confusion between the structure of language, or of the *lexicon, and the structure of the world. The surprising conclusion is that metaphors of the second type are logically correct but rhetorically insipid, while metaphors of the first type are rhetorically acceptable, but logically injus­ tifiable. As for the third type of metaphor, the Aristotelian example is two-fold : Then he drew of/ his life with the bronze and Then with the bronze cup he cut the water. Another transla­ tion would have a bronze sword, in the second case, cutting the flow of blood or life. These are, in any case, two examples of a passage from species to species : Drawing of/ and cutting are two cases of the more general 'taking away.' This third type genuinely seems to be a metaphor : it could be said right away that there is something "similar" between drawing off and cutting. This third type so truly seems to be a metaphor that many of the later theories work out of preference on examples of this type. It accounts for such expressions as the tooth of the mountain (peak and tooth partake of the genus 'sharpened form'), or she was a birch (girl and birch partake of the genus 'flexible body'). Contemporary theories say birch acquires a "human" property or that girl takes on a "vegetal" property, and that, at any rate, the units in question lose some of their

Metaphor

537

own properties (cf., for example, the theory of trans/er features in Weinreich 1972 ; see also Levin 1977). But at this point, two problems arise. First, to define which properties survive and which must drop away, we must by rights construct an ad hoc Porphyrian tree and this operation must be oriented by a universe of discourse or frame of reference. Second, in this operation of sememic intersection, a phenomenon arises that is new with respect to synecdoches or metaphors of the first two types. Consider the two-fold movement that is at the basis of both the production and interpretation of tooth of the mountain. In a synecdoche, in which a peak was named as something sharp, peak would lose some of its characteristic properties (that of being mineral, for example) and instead share with the genus to which it is reduced some morphologic properties (in particular, that of being sharp). In a metaphor of the third type, peak loses some properties in becoming a sharp thing and regains others in becoming a tooth. While peak and tooth, however, have in common the property of sharpness, their mutual comparison also focuses on those very properties that stand in opposition. This is so much the case that the phenomenon is referred to as a transfer of properties, as we mentioned above (peak becomes more human and organic, tooth acquires the property of being mineral) . What makes theories of transfer features questionable is always the fact that we cannot tell who gains what and who loses instead something else. More than of a transfer we could speak of a back-and- forth of properties. This phenomenon was called "condensation" by *Freud, and also characterizes the fourth type of metaphor. At any rate, what likens the metaphor of the third type with the fourth type is that mere identifications or absorp­ tions (from species to genus) are no longer the rule of the game: now it is both similarity and opposition, or identity and difference, that are in question. The metaphor by analogy or by proportion is a metaphor with four terms, which are no longer A/B = C/B (e.g., peak is to the genus of sharp things in the same way as tooth) but A/B = CID. The cup is to Dionysus as the shield is to Ares, suggests Aristotle. In this way, the shield can be defined as the cup of Ares or the cup as the shield of Dionysus. And again: Old age is to life as sunset is to day, and thus old age can be defined as the sunset of life and the evening as day's old age. Aristotle' s definition has always seemed superb for its concision and clarity. In fact it is, and undoubtedly the idea of finding a sort of propositional function, in which infi­ nite concrete instances can be inserted, represented a stroke of geniu s. Even more so given that this proportional formula permits the representation of even those cases of strict catachresis where the vehicle stands for a tenor that, lexically speaking, does not exist: A/B = x/C. Aristotle provides his own, linguistically complex example, but we can also turn to two familiar catachreses, the leg of the table and the neck of the bottle. It becomes clear right away that the way leg is related to body is not the same way in which neck is to body. The leg of a table resembles a human leg provided we have a frame of reference that puts into relief the property of "support," while the neck of a bottle is not exactly the support of a cork nor, on the other hand, of the entire container. It seems that the analogy on leg plays on functional properties at the expense of morphologic similarities (themselves reduced to very abstract equivalences, and quantity having been put aside as nonpertinent), while the analogy on neck drops the functionally pertinent features and insists on those that are morphological. Which is to say that, yet again, different criteria for constructing a Porphyrian tree are in question - if it were even still possible, though, to speak of Porphyrian tree tout court.

538

Metaphor

Consider the typical situation of a metaphor of the fourth type, like cup/Dionysus shield/Ares. How can we accomodate it into any Porphyrian tree? To begin with, the relation cup/Dionysus, according to the criteria of the later theories of rhetoric, is of a metonymic type. Cup and Dionysus are commonly associated by continguity, through the relation subject/instrument, through a cultural habit (without which cup could stand for many other obje cts). This relation is not at all amenable to being expressed by a Porp hyrian tree, unless we want to draw broadly inclusive equivalencies (of the type : cup belongs to the class of things characterizing Dionysus, or alternatively, Dionysus belongs to the class of all beings that use cups). And the same goes for the rela­ tion shield/Ares. In other words, it is very difficult to recognize in this relation a case of embedding of genus within species. The case of man/ animal presents us with •�analytic properties, while that of cup/Di­ onysus presents us with synthetic properties. Man is animal in virtue of its definition, while cup does not necessarily refer to Dionysus except in a very restricted, •�co-textual situation in which the variou s pagan gods are listed iconographically along with their characteristic attributes. While it is not possible to think of a man who is not an animal, it is always possible to think of Dionysus without thinking of cup. Even if one should grant that it is possible to group together the relations cup/Dionysus and man/animal, a new problem surfaces: why should Dionysus be placed in a relation with Ares and not, for example, with Ceres, Athena or Vulcan? Aristotle himself would find it difficult to name the spear of Athena as the cup of Athena and the wheaten sheaves of Ceres as the shield of Ceres (even if baroque contexts where that is possible are not excluded). Intuition says that shield and cup can entertain a relation because both are round and concave (round and concave in different ways, yes, but therein is the metaphor's cleverness, in making us see a certain resemblance between different things). But what matches Dionysus and Are s? In the p antheon of pagan gods it is their diversity that unites them (strange oxymoron). Dionysus, the god of joy and of the peaceful rites, and Ares, the god of death and war: a play of similari­ ties, then, mingling with dissimilarities. Cup and shield become similar because of their roundness, dissimilar because of their functions; Ares and Dionysus are similar because both are gods, dissimilar because of their respective domains of action. The metaphor with four terms does not set into play verbal substances alone. No sooner has the proportion been established than it is impossible to see, as something incongruous, Dionysus actually drinking out of a shield or Ares defending himself with a cup. In the first two types of metaphor, the metap horizing term absorbs (or confuses itself with) the metaphorized term, much as a �·figure enters a multitude - or leaves it - without our cognitive habits coming into question. At best, the result is something impoverished, both conceptually and perceptually. In the third type of metaphor, instead, a superimposition of plant and girl is created that is almost visual, as in the fourth type. Aristotle realizes, albeit confusedly, that by naming one thing with something else's name one denies the first thing those qualities proper to it. Ares' shield could also be called cup without wine (Poetics 1 457b32). Henry (197 1 ) notes that this is no longer a metaphor, but instead a "secondary phenomenon," as a consequence of the preliminary metaphor. That is true, but it means that, as the metaphor starts to be understood, the shield becomes a cup, even as this cup, while remaining round and concave (though in a different way from the shield), loses the property of being full of wine. Or, in reverse,

Metaphor

539

one forms an image wherein Ares possesses a shield that acquires the property of brimming with wine. In other words, two images are conflated, two things become dif­ ferent from themselves, and yet remain recognizable, and there is born a visual (as well conceptual) hybrid. One can maintain that the whole of further studies on metaphor have played upon Aristotle's proposals, at least through three main lines of thought. In the first line of thought, the metaphor was seen as a device producing an exchange of properties between conceptual entities, such an exchange producing in some way an increasing of knowledge or a different understanding of the concepts involved in the metaphorical "short circuit." In this perspective metaphors have been seen as equivalent to oneiric images, quite like what Freud (1899) called "condensation," where noncoincident traits can be dropped while those in common are reinforced. The process is typical not only of dreams but also of jokes ( Witzen): that is, of those psychra 'puns' or compound words (Rhetoric 1406bl ), or even better, or those asteta 'witticisms' (Rhetoric 140b6), which seem so similar to some of the categories of Witze, Kalauer, and Klangwitze ana­ lyzed by Freud (1 905). If the Freudian typology can be compared to a typology of rhe­ toric, there can be no doubt that, at least, the final result of the Aristotelian proportion is a process very much like Freudian condensation, and that this condensation, as will be better demonstrated later, can be described as far its semiotic mechanism is concerned in terms of the acquiring and losing of properties or semes, whatever we should want to call them. Clearly the properties set into motion by the third and fourth types of metaphor do not have the same logical status as those set in motion by the first two types. To obtain the condensation cup/shield, it is necessary to activate properties or semes such as "round" and "concave," "war" and "peace," "life" and "death." It is clear here that a difference is being outlined between a semantic description in the form of a �-dictionary and a description in the form of an �·encyclopedia, or even, with inconsistent variations, between :E properties and n properties (Groupe µ 1 970), between semantic properties and semiological properties (Greimas 1966a), or between dictionary markers and world knowledge. In the second line of thought, the cognitive aspect of the metaphor has been stressed. Undoubtedly Aristotle assigned a cognitive function to the metaphor, not only when we associates metaphor with enigma - an extended sequence of metaphors - but also when he says that creating metaphors "is a sign of a natural disposition of the mind," because knowing how to find good metaphors means perceiving or grasping the similarity of things between each other (Poetics, 1 459a6-8). But if the proportion between cup and shield and between Ares and Dionysus where already overcoded, that metaphor would not say anything other than what is already known. If it says something new, it means that either (1) the proportion was not so commonly accepted, or (2) if it was accepted, it was then soon forgotten. And thus the metaphor posits {'posits' in a phil­ osophical sense, but also in a physical sense, as in "putting before the eyes," as Aristotle beautifully said), a proportion that, wherever it may have been deposited, was not before the eyes; or it was before the eyes and the eyes did not see it, as with Poe's pur­ loined letter. To point out, or teach how to see, then. To see what? The likeness between things, or the subtle network of proportions between �- cultural units? To this question Aristotle gives no answer, as was only appropriate for one who had identified the modes of being of being {the categories) with the modes of being of language.

540

Metaphor

According to the first line of thought metaphors have been seen as devices providing a transfer of cultural properties : such a project will be sufficiently illustrated by the position of Tesauro and are still influencing the structural approaches to metaphors. According to the second line metaphors have been considered as models or analogical images of something otherwise unspeakable : such a project will be illustrated by symbolical theories of the Middle Ages. According to a third line of thought the pair metaphor/ metonymy characterizes the whole of human thought and of the semiotic activity. The position of Freud, quoted above, ranks in some way with this line. During the last decades the most impressive example of such a position has been undoubtedly the one of *Jakobson, who was equated the pair metaphor/metonymy with the pair selection/combination (or para­ dygm/syntagm) and has applied this model to language (from aphasic disturbances to *literature), to magic, *cinema, and the *visual arts. Under the influence of Jakobson, *Lacan has applied the same model to unconscious phenomena.

The Medieval Encyclopedia and analogia entis We have seen how Aristotle's limitation consists in his identifying the categories of lan­ guage with the categories of being. This identification is not questioned by post-Aristote­ lian rhetoric - from the Rhetorica ad Herennium, through �-Cicero and �·Quintilian, and all the way to the medieval grammarians and rhetoricians - which in the meantime works out the traditional classification of figures. However, a panmetaphorical attitude is established in the Middle Ages, which deserves a brief discussion, since it helps resolve (even if in a negative manner) the question with which we are concerned. Saint Paul had already affirmed that "we see through a glass, darkly" (1 Corinthians 1 3 : 1 2). Medieval neoplatonism gives a metaphysical frame to his �·hermeneutic tendency. In a universe that is nothing more than an emanative outpouring from the unknowable and unnameable One down to the furthest ramifications of matter, every being functions as a synecdoche or metonymy of the One. When Hugh of Saint Victor affirms that the "entire sensible world is, so to speak, a book written by the hand of God, " and that "all visible things, visibly presented to us by a symbolic instruction, that is, figured, are pro­ posed for the declaring and signifying of things invisible" (Didascalicon, PL, CLXXVI, col. 8 1 4), he gives us to understand that there exists a sort of code that, assigning to things emergent properties, allows them to become metaphors for supernatural things, in accordance with the traditional theory of the four levels of exegesis (the literal, alle­ gorical, moral, and analogical). This is the project taken up in the bestiaries, the lapida­ ria, the imagines mundi, all formed on the Hellenistic model of the Physiologus: certain properties are predicated of every animal, plant, part of the world, or event in nature, and a correlation is established on the basis of an identity between one of these proper­ ties and one of the properties of the supernatural being that is to be metaphorized. A net­ work of cultural information exists, which functions as a cosmological code. The code is nevertheless ambiguous, since of all the properties there are to choose from it chooses only a few, and those are contradictory. The lion erases his tracks with his tail to throw the hunters off his track, and is thus a figure of Christ canceling the traces of sin ; but in Psalm 21 the terrible maw of the beast - " Salva me de ore leonis" - becomes a metaphor of Hell, and ' 'per leonem antichristum intelligitur'' definitively.

Metaphor

54 1

Even though medieval neoplatonism was not aware of it, the universe, which seems to be a maze-like network of real properties, is in effect a maze-like network of cultural properties, and those properties are attributed both to the earthly beings and to the heavenly beings in order that metaphorical substitutions may be possible. What medieval neoplatonists knew was that, in order to decide whether the lion must be seen as a figura Christi or as a figure of the Antichrist, a co-text is necessary. The tradi­ tion provides a typology of possible co-texts; so that the best interpretation is always the one recorded by some (intertextual) auctoritas. Tesauro An interesting return to the Stagyrite' s model is found in the Cannocchiale aristotelico ( The Aristotelian Telescope) of Emanuele Tesauro, wirtten in 1655 at the very height of the Baroque period. From Aristotle, Tesauro inherits the tendency to call every trope and figure a metaphor. We will not speak here of the detail and enthusiasm with which the author of the treatise studies puns both in single words and in actual microtexts, and of how he extends the metaphorical mechanism to visual puns, *painting, sculpture, *ac­ tions, inscriptions, proverbs, truncated phrases, laconic messages, mysterious charac­ ters, hierograms, logogriphs, cryptograms, gestures, medallions, columns, ships, gar­ ters, chimeric bodies. Tesauro knows that metaphors are not created out of a pure joy of invention, but that they impose on one a labor, which takes practice to master. The first exercise is the reading of catalogues, anthologies, hieroglyphic collections, medallions (and their reverse sides), emblems - in a way, a pure invitation to intertextuality, to the imitation of the "already said." But the second stage of the exercise presupposes learning a combi­ natory mechanism. Tesauro invites his reader to draft a categorical index, that is, a model of an organized semantic universe. Such a model begins with Aristotle's categories (Substance, Quantity, Quality, Relation, Place, Time, Position, Possession, Activity and Passivity; cf. Catego­ ries l b25-2a8 ) and then organizes under each of these categories the various members that are inclusive of everything susceptible to such categorization. Suppose we have to make a metaphor on a dwarf. We leaf through the categorical index until we find the entry Quantity; we then identify the concept "Little things," and all such microscopical things as may be found under that rubric may be divided still further according to contex­ tual selections (as we would say now): astronomy, human organism, animal, plant, etc. But an index organized by substances would have to be integrated with a second index in which each substance were analyzed according to the particles that define how the object in question manifests itself (e.g. , under the category of Quantity we would then have to find what measurements it has, what weight it has, and how many parts it has, and under Quality there would be the specifications whether it is visible, whether it is hot, and so on). As can be seen, this is essentially a system of content organized as an encyclopedia. At this point, we will find that the smallest measure is the "Geometric Thumb," and we will say that in order to measure a dwarf's body, a geometric thumb would be too gross a measure. While he is a careless structuralist, Tesauro knows nevertheless that it is no longer ontological relations but the structure itself of language that guarantees the metaphorical transfers. Look, for example at the Aristotelian metaphor of old age as the sunset of life

542

Metaphor

(or of youth as spring). Tesauro still proceeds by analogy, but the relation is one of con­ tiguity in the index. This is how the transfer is structured: Analogous genus . . . . . . . . Subordinate genus ... . Analogous species . . . .

Duration of time Human age Season of the year + + Spring Youth

The higher nodes become classemes or contextual selections of the lower nodes. One can see that the analogy Aristotle perceived between draw off and cut functions when the act of drawing off is considered under the category of Passivity; but if it is considered under the category of Possession, drawing off becomes analogous to other p rocesses of acquisition and not to processes of deprivation (take away). Henceforth, there is the pos­ sibility of searching through the categorical index ad infinitum, and of uncovering a reserve of untapped metaphors, and of metaphorical *p ropositions and *arguments. Tesauro's model still represents the framework of medieval neoplatonism, which it has deliberately resolved, though, into a network of units of purely cultural content. It is the model of an unlimited *semiosis, a (still too) hierarchical system of semes, a web of �·interpretants. Vico An overwiew of the history of metaphorology and of its epistemic breaks, however brief, must not omit Vico, if only because of the fact that La Scienza Nuova (and its Chapter "Della logica poetica") seems to put in question the existence of a cultural network, of semantic fields and universes, and of a pre-established process of semiosis, which should p roceed (on the basis of the foregoing observations) the p roduction and interpretation of metaphors. Vico discusses the "first tropes," and the phenomenon of speaking by m eans of ani­ mate substances, whereby natural objects and phenomena are named by reference to parts of the body (1967 : 1 29), for example, the eyes of needles, the lips of a vase, etc. Now, much too much has been said about this "auroral" moment of language; in the view of some interpreters, Vico argues that the creating of metaphors is an inborn ability in beings who are at the dawn or awakening of their own intelligence; metaphorical speech, furthermore, would be iconic, insofar as it instituted a kind of native onomatopoeic relation between words and things. But the fact is that Vico knows and says that, outside of the utopia of an Adamic language (an idea already in Dante, and later elaborated in seventeenth century England, as well as in Vico's own time), what is indisputable is the diversity of languages. As the people have certainly by diversity of climates acquired different natures, from which have sprung as many different customs, so from their different natures and customs as many different languages have arisen. For by virtue of the aforesaid diversity of their natures they have regarded the same utilities or necessities of human life from different points of view, and there have thus arisen so many national customs, for the most part differing from one another and at time contrary to one another; so and not otherwise there have arisen as many different languages as there are nations. (Vico 1 967 : 1 4 8 )

Vico makes the following fundamental observations: languages, like customs, are born as the response of group s of human beings to the material environment in which

Metaphor

543

they live; even if the disposition toward language functions and develops in all human societies according to the same logic, and even if the utilities and necessities of life are the same for everybody, nevertheless human societies have looked at these material uni­ versals from different viewpoints, which is to say they have made pertinent different aspects of their universe. Catachreses are created out of transpositions of natural objects "according to their natural properties or sensible effects" (Vico 1967 : 1 47). In this sense the labor of meta­ phor is always motivated. What must be asked here is if those effects and properties given that metaphors are the result of a selection of pertinent aspects - are not already cultural constructions. If metaphors require an underlying cultural framework, then the hieroglyphic language of the gods cannot be a merely primitive stage of human con­ sciousness: it needs the presence of both the symbolic language of heroes and the episto­ lary language of men as its starting point. Thus Vico is not speaking of a linear develop­ ment from a metaphorical language to a more conventional language, but of a continual, cyclical activity. The language of the gods is a heap of unrelated synecdoches and metonymies: thirty thousand gods as identified by Varro, as many as the Greeks counted, including stones, fountains, reefs, brooks, minute objects, signifiers of forces, causes, connections. The language of heroes already creates metaphors (which thus are not so primeval), but the metaphor or catachresis invents a new term using at least two terms that are already known (and expressed) and presupposing at least another one that is unexpressed. Could that symbolic language establish itself without the support of an epistolary language, the only language recognizedly conventional? On this point Vico is very explicit: To enter now upon the extremely difficult [question of the] way in which these three kinds of languages and letters where formed, we must establish this principle : that as gods, heroes, and men began at the same time (for they were, after all, men who imagined the gods and believed their own heroic nature to be a mixture of the divine and human natures), so these three lan­ guages began at the same time, each having its letters, which developed along with it. ( 1 96 7 : 1 49)

In light of these considerations, Vico' s semiotic resembles, more than an aesthetics of ineffable creativity, a cultural �-anthropology that recognizes the categorical �-indices on which metaphors are based, indices whose historical conditions, birth, and variety it researches even as it explores the variety of brave deeds, of medallions, and of fables. U.E.

Contemporary Philosophy of Language The present interest in metaphor denies the observation Murry (1972) made in 1931 in his essay on the subject: "Discussions on metaphor - there are not many of them - often strike us as superficial." The ever-increasing number of works on the subject, which by now run in the thousands, reflects the fact that the problem of metaphor has been recog­ nized as basic in the philosophy of language. The outstanding book on the subject to emerge in recent years is the monumental study by Ricoeur, The Rule of Metaphor. It gives a comprehensive and elaborate account of linguistic and philosophical aspects of metaphor as they appear both in historical perspective and in contemporary authors.

544

Metaphor

The importance of metaphor for the *philosophy of language lies i n the fact that it brings into focus basic questions concerning the nature of language and the fundamental problems of epistemology. The study of metaphor involves problems of the nature of meaning and semantic deviation, standards of exactness of literal versus metaphorical discourse, and conventionality and creativity in language. Theories of metaphor may be roughly divided in two groups: comparison theories and interaction theories. The former follow the tradition of classical rhetoric and regard metaphor as transfer of name based on resemblance of objects; one thing is referred to by a name which properly belongs to another thing for the purpose of decorativeness or economy of discourse. Interaction theories do not deny the role of resemblance but explain the metaphorical meaning as interaction between two terms that metaphor brings together; the metaphorical meaning does not originate on the level of words but within the sentence as a whole. Some theories go still further: the mechanism of meta­ phor involves something more than the semantic clash between interacting terms, and an adequate theory of the metaphorical process is supposed to account for the role of imagination or explain it in terms of the •:-speech act theory as an incongruence between the meaning of a sentence and of a speaker's *utterance. Philosophical Theories of Metaphor The basic idea of the comparison theory can be traced back to Aristotle, who said that " . . . a good metaphor implies the intuitive perception of the similarity in dissimilars" (Poetics, 1 459a5-8) and regarded metaphor as an elaborate simile. The radical contem­ porary version of the comparison theory identifies metaphors and similes. It denies they differ in any essential property and defines the metaphor as a compressed simile (Murry 1972), from which a term of comparison has been deleted. The iconic theory (Henle 1958) bases metaphor on similarity but differentiates between metaphors and similes. The metaphorical statement does not directly compare two objects but it mediates figur­ atively their qualitative or structural resemblance. The figurative sense is explained in terms of *Peirce's concept of icon. The metaphorical statement does not actually con­ tain an i con but it makes an indirect reference to something by describing or symbolizing its icon. There must be an initial similarity to make both metaphor and simile possible. It is peculiar of the latter, however, that it involves two semantic levels: some terms in metaphor refer to the icon and the others to what is iconized, as i n Keats' s metaphor " . . . hateful thoughts enwrap my soul in gloom, " where gloom is mediated by the description of something capable of surrounding and enwrapping. The iconic theory comes close to interaction theories in that it introduces the differentiation of semantic levels within the metaphorical statement. It explains the aptness of metaphor in terms of the capability of extending the parallel structure, as in Kant' s metaphor of a des­ potic state as a mere machine (like a hand- mill) where there is no similarity in proper­ ties. Various obj ections to the comparison theory have been raised. For one thing, the statement of similarity is reversible, that is, "A is like B" implies "B is like A, " as in the proportional metaphor, stating that A is to B like C is to D, where, e.g., evening is described as the old age of the day and old age as the evening of life; however, in many metaphors the inversion does not work, e.g., in Baudelaire' s metaphor of the flowers of evil - although we know something of evil when we think of flowers, we will never find

Metaphor

545

anything about flowers when we think of evil. Further, the very notion of similarity is vague, if not empty; given a suitable aspect, any two things appear similar. The propo­ nents of the theory often stress, however, that the metaphorical statement formulates a striking, previously unnoticed resemblance. Granted even that, another objection is still valid, namely that, under the comparison interpretation, metaphors appear, in most cases, either trivially true or false. The statement of similarity "Juliet is like the sun," cor­ responding to Shakespeare' s metaphor "Juliet is the sun," is either trivially true (since, e.g., she shares some chemical characteristics with the sun) or plainly false. Therefore the explanation in terms of similarity is at best circular; it offers a stylistic variant of the original statement which, if not trivially true or false, must again be interpreted figura­ tively. The interaction theory radically opposes the comparison theory in that it denies that metaphor can be dissected into two separate elements which are linked by resemblance. On the contrary, the metaphor makes an integral whole constituted by the cooperation or interaction of its elements. The point is stressed by Richards (1936) , who maintains that the figurative meaning of the metaphorical statement results from the copresence of its two parts: the vehicle, or the underlying idea, and the tenor, or the idea in terms of which the first idea is apprehended. He locates, then, the metaphorical meaning at the level of the sentence, not of words. The vehicle is not a mere embellishment of a tenor but it is the cooperation of the tenor/vehicle pair that gives rise to a new meaning. The interaction may be based on various relations, not necessarily on resemblance. An influential form of the interaction theory was developed by Black in his now classic paper, supplemented by a later study (1962, 1979). In his view, the metaphorical statement has two distinct subjects and it works by applying to the principal subject, or the literal �·frame, the system of "associated implications," usually commonplaces, char­ acteristic of the subsidiary subject, or the metaphorical statement' s focus. The interac­ tion of the subjects consists in creating implications about the principal subject which normally apply to the subsidiary one. Black sets forth the creativity thesis: the interac­ tion metaphor may offer cognitive insight, it is not a dispensable stylistic ornament. There are metaphors, then, that generate new knowledge. However, it has been observed (Ricoeur 1977) that the explication of interaction in terms of the system of commonplaces, that is, *connotations already established, is confined to trivial meta­ phors only. Black admits that the system of connotations on which the interaction ope­ rates may also include deviant implications, established ad hoc by the author of a meta­ phor. The way those are established is left unexplained. According to the verbal-opposition theory (Beardsley 1958, 198 1 ) the interaction within the metaphorical statement is explained in terms of the opposition created by the two levels of meaning in the modifier applied to the subject (the modifier/subject pair is analogous to vehicle/tenor and focus/frame) . In the metaphorical attribution the modifier acquires, besides the connotation determined by its ordinary extension, also a new *intension, sometimes peculiar to the particular context. It is the opposition or ten­ sion at the levels of attribution that creates the "metaphorical twist" of meaning. The type crossing theory explains the metaphorical incongruence of meaning in terms of a category mistake of some kind. A category mistake consists in "the presenta­ tion of facts of one category in the idioms appropriate to another" (Ryle 1953 ) . According to Orange (1966) , metaphor is a *type crossing sentence which is grammatical but conceptually meaningless, that is, it expresses an unthinkable proposi-

546

Metaphor

tion which combines concepts in the way we cannot understand them. The metaphorical sentence is meaningless if interpreted literally; it is precisely the impossibility of taking it literally that forces the reader into reinterpreting it. The theory does not make clear what sort of interpretation is required to make a metaphorical sentence intelligible. On the one hand, a type crossing, being grammatically correct, may be turned into a mean­ ingful sentence by a suitable substitution (e.g., "The theory is blue" into "The sky is blue"). The result, however, is deprived of the metaphorical character. On the other hand, a reinterpretation limited to the class of type crossings leaves metaphor conceptu­ ally meaningless. According to a theory formulated by Turbayne ( 1 962), the sort crossing does not necessarily lead to a mistake ; it may be intended, as in make-believe, and the intended sort crossing is characteristic of metaphor. The use of metaphor involves pretense that something is the case when it is not. The metaphorical statement, then, involves, first, the awareness of sort crossing and, second, re-presentation of facts of one sort in the idiom appropriate to another. In other words, the use of metaphor involves both the duality of sense and the pretense that the two senses are the same. The mistake is not to use a metaphor but to be used by it, that is, to misinterpret its dual nature. Another view of metaphor as a calculated category transfer was formulated by Goodman ( 1 968). In his view metaphor is applying an old label in a new way. The meta­ phorical application of a label (e.g., "grey") involves a change of not only its range (gray things) but usually of the whole realm of its application (colored things). This results in a transfer, to some extent contraindicated, of the schema in which the label is functioning, into a new realm ("gray" as applied to thoughts). The transfer of schemata is not peculiar to metaphor but is characteristic of all discourse ; it is specific of meta­ phors, however, that they lead to a notable organization of a new realm. Speech act theory opposes both the comparison and the interaction theories (Searle 1 980). Their failures, speaking in •:•Fregean terms, consist in that the former explains the metaphorical meaning as a relation between references, the latter as a relation between senses associated with references. They both fail to distinguish between the sentence meaning, which is never metaphorical, and the utterance meaning, which may be meta­ phorical. Searle's view is formulated within the framework of speech act theory. In case of a literal utterance, the speaker means what he says; the basic principle on which meta­ phor works is that the literal utterance may invoke another meaning. The core of the theory is the account of the procedures required in recognizing and comprehending met­ aphorical meaning. First, on the basis of the linguistic and factual knowledge a speaker and a hearer share, the utterance " S is P" appears defective if taken literally; hence it is taken metaphorically to mean "S is R." Second, there are shared principles of associating P with possible interpretations R. Third, there are shared strategies to re­ strict possible interpretations to the actual meaning. The imagination theory elaborates a structural analogy between the three comple­ mentary components of metaphor: the cognitive, the imaginative, and the emotional (Ri­ coeur 1 9 8 1 ). As in the interaction theory, the metaphorical meaning is located at the level of the sentence as a whole. Against that theory, however, metaphor involves some­ thing more than the semantic clash between the interacting terms. The role of imagina­ tion in metaphor is established by abolishing the dichotomy between sense and represen­ tation. Its role goes beyond a mere psychological factor and is explained not in terms of �-Hume's concept of image as a faint impression but *Kant's concept of imagination as schematizing. The productive imagination works in metaphor in three steps. First, it

Metaphor

547

schematizes, that is it gives an insight into the "like" through the "unlike," which results in a reconstruction of semantic fields and possibilities of a new congruence. Second, it depicts, that is it provides an image on which a new congruence can be read. Imagining does not consist in having a mental picture of something but in displaying relations in a depicting mode. Third, it leads to a "split of reference" which consists in suspending the reference of ordinary discourse and projecting a new, secondorder refer­ ence. The role of feeling is elaborated in terms of parallel stages. The new congruence is felt as one's own in the way proper to the illocutionary force of the metaphorical act. Moreover, feelings also involve the suspension and transformation of the literal emo­ tions of everyday life. Therefore they also display a split structure which completes the split structure pertaining to the cognitive and the imaginative components of metaphor. The emotive theory is based on the distinction between the cognitive and the emotive functions of language (*Ogden and �·Richards 1923) and the empiricist criterion of meaningfulness. It denies metaphor any cognitive import and must be clearly distin­ guished from the imagination theory which ascribes the imaginative and the emotional components of metaphor a role analogous to that of the cognitive component. According to the emotive approach, the cognitive function, typical of the language of science, consists in making statements of fact, the emotive function in expressing and evoking feelings and attitudes. Hence the metaphorical statement is regarded as a mean­ ingless pseudostatement. The emotive theory shares the shortcomings of the oversimpli­ fied positivistic approach to language. Under this interpretation it is impossible to distin­ guish metaphors from emotive utterances of other kinds, e.g., exclamations. It also e ncounters the difficulty inherent in the verification theory of meaning: an affirmative statement ' 'Man is a wolf" is regarded as a pseudostatement and a mere expression of feelings, while its negation appears to be a cognitive statement, despite the fact that the negative statement is just as metaphorical as the affirmative one. Philosophical Aspects of Metaphor The most significant philosophical aspects of metaphor are involved in the problem of its cognitive content. There is a wide variety of possible solutions based on different assumptions concerning the nature of meaning and knowledge. On the one extreme, metaphor is denied any sense at all, while on the other it is regarded as an essential condi­ tion of all discourse and thought. The position that the metaphorical statement is de void of sense assumes an unproblematic standard of meaning by which metaphor is ruled out. The approach is characteristic of logical positivism and analytic philosophy, where the analysis of language is based on the semantics of word-thing relations. In *Carnap (1937) meaningfulness is granted to expressions formulated according to the rules of denotation and transformation which, in particular, determine the range of application of predicates, thus ruling out other applications as meaningless. The cognitive, then, is identified with the literal and descriptive and the metaphorical stateme nt may have, at most, some emotive import. That position leads to a grave philosophical consequence, namely that language cannot reflect on itself and its relation to the world (Apel 1967a). The procedure of constructing formalized metalanguages is infinite and the metalan­ guage actually used in the se mantical interpretation does not satisfy the require ments of the procedure. The philosophical investigation must by necessity be metaphorical as it goes beyond a mere description of the world. The point is made forcibly by *Wittgen-

548

Metaphor

stein, who admits that his own reflection on the nature of language and its relation to reality is meaningless : "Whereof one cannot speak, thereof one must be silent" ( 1 9226 :7). According to a less radical approach the metaphorical statement is meaningful but excludable without a loss of cognitive content in favor of a literal paraphrase. Metaphor, then, does not tell us anything new. Its role is purely ornamental and consists in saying in a more suggestive or pleasing way what is expressible in a literal language. The meta­ phorical/literal distinction is valid neither semantically nor epistemologically. There is no figurative meaning not reducible to a literal description of the world, usually to a statement of resemblance, and the metaphorical statement expresses no special cogni­ tive insight. Accordingly, the study of metaphor is confined to rhetoric and poetics. The metaphorical/literal opposition assumes a duality both in language and in cog­ nition. The metaphorical statement conveys a meaning which cannot be reduced to a literal language. A literal paraphrase may approach, more or less closely, the metaphorical statement, but an adequate translation is not possible. Moreover, at least some metaphors may generate new knowledge ; they are cognitive instruments indispen­ sable for grasping certain aspects of reality which they themselves help to create. The problem of the cognitive content of metaphor is especially interesting in science. Although metaphorical expressions are, no doubt, widely employed in science, it is pre­ cisely the metaphorical use which often helps to contrast the scientific discourse with ambiguous and vague linguistic usages which fail to meet the requirements of precision and explicitness of scientific statements. Accordingly, the role of metaphor in science, if any, is merely a subsidiary one. On the one hand, metaphor is regarded as characteristic of pretheoretical stages of development of a discipline and, in consequence of a subse­ quent theoretical refinement, it is supposed to give way to a literal articulation (Quine 1 978). On the other hand, its function is reduced to an auxiliary role in heuristic tech­ niques. It is argued, however, that the above account is connected with the empiricist criteria of linguistic precision and procedures of introducing theoretical concepts (by means of operational definitions, reduction sentences, and the like). In the philosophy of science, theories of nondefinitional procedures of introducing general terms have been developed which throw a new light on the role of metaphor in sience. Boyd ( 1 979) distinguishes the class of "theory-constitutive" metaphors which are employed in nondefinitional procedures of reference fixing of theoretical terms in science. Their role consists in introducing new terminology. Boyd argues, taking examples from metaphors in cogni­ tive psychology (thought as information processing, brain as computer), that theory­ constitutive metaphors provide theoretical insights. They refer to theoretically impor­ tant phenomena which cannot be otherwise formulated. The theory-constitutive meta­ phor does not presuppose similarities but calls forth the similarity upon which its func­ tion in determining reference depends ; it expresses theoretical truth. There is no a priori reason, however, why the cognitive content of the theory-constitutive metaphors cannot be eventually explicated in terms of literal paraphrases. The creative role they play in scientific investigation is not diminished by their (at least temporal) inexplicit­ ness and has nothing to do with whether a literal paraphrase is available or not. The metaphorical/literal distinction is rejected on the ground that metaphor is the omnipresent principle of discourse. The view that it is something special in the meta­ phorical use of language, a deviation from its normal mode of working, results from the

Metaphor

549

"proper meaning superstition" (Richards 1936). All language consists in "remaking reality" and difficulties in explaining how metaphorical and literal applications work are basically the same (Goodman 1968). The only difference is that of degree, inasmuch as the metaphorical application is novel, therefore less sharp, while the literal one is more stable. Cassirer (1946a, 1953-57) opposes the "translation" and "radical" concepts of metaphor. The former consists in the transposition of one thought content for another, somehow similar to it, which presupposes that both ideas are already given; the latter is essential to the creative articulation of reality. The radical metaphor is the basic principle of creative power of language, of which *myth and discursive logic are but two different modes. Both mythic and scientific thinking are governed by the same meta­ phorical principle which operates in concept formation. Metaphor is not reducible to a certain development of speech but is one of the essential conditions of symbolic knowledge. U.N.

Rhetoric of Metaphor Metaphor as a basic concept implies a reduction, be it explicit or implicit, of any trope, ore even any figure (synecdoche, metonymy, oxymoron, etc.) to a common structure. In this process the latter becomes, of course, all the more vague and ill-defined. Most often, the authors supporting this theory (e.g., Breton) consider metaphor and image to be two equivalent concepts. There was thus a crying need for clearer definition, and the Groupe µ ( 1 970; refined in 1 977b) proposed a general pattern in which metaphor (as well as metonymy) is con­ sidered as a compound figure made of two combined synecdoches. In this figure, the two synecdoches have opposite directions, but follow the same (referential or semantic) decomposition pattern. In this general pattern one may distinguish two different types. The first one, in a semantic mode, combines a generalizing synecdoche ("slenderness" for "birch") with a particularizing synecdoche ("girl" for "slenderness") resulting in the metaphor "girl" = "birch." The second one, in a referential mode, combines a par­ ticularizing synecdoche (in French voile 'sail' for voilier 'sailship') with a generalizing one (veuve 'widow' for voile 'veil'). The resulting figure is the metaphor "widow" = "sailship." It seems that the first type of pattern is by far the most frequently used, but the reasons therefor are not yet clear. Metaphor may be in praesentia (when both the source term and the target term are expressed), or in absentia (when only one of them is in evidence). Grammatically, the two terms are located in coupled positions (Levin 1964) and linked by a copula expressing all possible degrees of comparison, ranging from likeness to total identity. Hence by means of metaphor, the reunited semes of the two concepts acquire a property of identity that actually is only valid for their intersections. Like metonymy, metaphor does not modify the level of generality of the utterance to any significant extent. In overlooking distinctions, metaphor implies a regression toward indistinction, which has been used to explain the euphoria-promoting ethos of this trope (Edeline 1965, 1979 ; Arnheim 1966; Groupe µ 1977b). In combining two units that may be extremely distant in the semantic universe, metaphor also generates what has been called a certain exoticism (Ricardou 1964) or even, given its connection with vertigo and nonsense, "the constant possibility of a delirium" (Mannoni 1962). F.E.

550

Metonymy and Synecdoche

Metonymy and Synecdoche Metonymy is a figure of speech formerly defined as taking a part for a whole, while syn­ ecdoche is a figure of speech which rhetoric defined by giving an enumeration of con­ crete cases : more to less, the material for the object made of it, species to genus, part to whole, etc. (Lausberg 1 960 ; Morier 1 96 1 ). In his famous paper on *aphasia, Roman *Jakobson merges synecdoche within the concept of metonymy as opposed to *meta­ phor ( 1 956a). Even more radical than this dichotomy, doubt has been cast up on whether synecdoche is in itself a figure or not (Sato 1 979; controversy between Ruwet [ 1 976] and the Groupe µ [ 1 977a, 1 977b, 198 1 ] ). The Groupe µ ( 1 970, 1 977b), has dis­ tinguished four types of synecdoches, according to the following table :

type of analysis Operations

addition

suppression

referential mode (parts)

semantic mode (semes)

generalizing synecdoche (type 3) "cake" for "piece of cake"

particularizing synecdoche (type 2) "sampan" for "boat"

particularizing synecdoche (type 4) "sail" for "sailing ship"

generalizing synecdoche (type 1) "mortal" for "man"

Just like metaphor, metonymy is a metasememe. It is an �·operation acting on the sememe of a word without altering the level of generality of the '�utterance. When the effect is taken for the cause, the producer for the product, the material for the object made of it, the number of elements deleted or added has no influence on the poetic ethos. On the contrary, in the synecdochical process, the modified term implies a change in the degree of generality of the utterance. In a scholastic language, it might be said that there are four types of synecdoches : genus to species, species to genus, whole to part and part to whole. Nevertheless, a careful examination leads us to confer a special status to type 1 syn­ ecdoches as opposed to the other three. The statement "I saw a pachyderm" may pro­ duce a deviation in a context where the expected term would rather be elephant. Yet this statement is true since an elephant is a pachyderm. On the contrary, the other three syn­ ecdochical modes always produce false assertions. Saying about pachyderms something that is only true for elephants would be erroneous ; reduction only could make the meaning acceptable. As for the confusion between synecdoche and metonymy, it can be explained by the fact that the referential decomposition of a sememe is much more empirical, by defini­ tion, than the semantic analysis itself. "Smoke" is an effect of the cause "fire " but to

Mill

551

some extent w e can imagine that there is a whole phenomenon of "combustion" including at the same time the combustible, the flame, the smoke, the heat, etc. What is relevant from the semiotic point of view is the expected level of encyclopedical compe­ tence (Klinkenberg 1 983): in the universe of representations "there is no smoke without fire" (Minguet 1980). Theoretically distinct, synecdoche and metonymy may actually be equivalent. Take, for example, "No smoke without fire: " in a common sense interpretation, the phe­ nomenon "fire" is followed by the phenomenon "smoke" and hence one may go back from the sign, "smoke," to the cause, that is the "fire. " What is generally considered as a typical case of metonymy (cause to effect, i.e., fire to smoke) may thus be analyzed as a synecdoche (part to whole, "smoke" being the residue of the deletion of all other determiners of "fire"). Even if we consider that there may exist a more elevated degree of competence (of high school level) then, the phenomenon "fire" might be analyzed in such a way that the process allowing for the transition from "fire" to "smoke" might belong to the metonymic type: in this case, "fire" and "smoke" would be considered as two entities belonging to a encompassing whole that would be called "phenomenon of combustion. " In other words, the further encyclopedical competence deviates from the established perceptual patterns, the more refined a trope appears to be, at a strictly semantic level or, using a one-author expression, on a sigma plane. As Jakobson already said in his famous lecture on poetics (1960), a pars pro toto syn­ ecdoche may be at the borderline between metonymy and metaphor: "horse" is a part of the whole "horse + horseman" (synecdoche), but at the same time, "horse" is a pos­ session of the "horseman" (metonymy). As a matter of fact, the relationship between early and modern rhetorics seems ambiguous and difficult to establish (V alesio 1 980). P.Mi.

Mill, John Stuart ( 1 806-1 873) English philosopher, economist, and spokesman for a liberal humanistic morality, Mill was the main representative of British empiricism in the nin eteenth century and in many respects a follower of �·Hume; his originality lies in the investigation into inductive methods of experimental inquiry and in the radical empiricist interpretation of *mathe­ matics as a generalization from experience. Mill' s major and influential work is System of Logic ( 1 843), devoted mainly to the methodology of empirical science. His account of inductive methods is an improvement and completion of *Bacon' s tables of investigation. Mill distinguishes ( 1 843, Bk. 3) the "eliminative methods of induction" which include the methods of agreement, differ­ ence, the joint method of agreement and difference, the method of residues, and that of concomitant variation. For example, if we observe that every positive instance of a phenomenon under investigation has only one circumstance in common, the method of agreement allows us to infer that the circumstance is the cause, or a part of the cause, of the phenomenon. The methods of eliminative induction make it possible to discover causal relationships, which Mill defines, following Hume, in terms of invariable sequence. The methods operate on two assumptions: that every event has a cause (the principle of universal causation) and that similar causes produce similar effects (the prin-

552

Mimicry

ciple of uniformity of nature) . According to Mill these principles are generalizations from experience. Mill's connotation theory of �·meaning is especially interesting from a semiotic point of view. Mill maintains (1 843, Book I) that a *name denotes the things to which it is prop­ erly applied and connotes attributes of those things. The connotation of a name is given in its *definition and is identified with the meaning of a name. It is important to note that Mill's theory of meaning as connotation does not apply to proper names, which are said to have *denotation but no connotation, hence no meaning; proper names do not apply to things but refer to them. The meaning of a *proposition is a function of meanings of names, hence it is determined by their connotations. The connotation/denotation dis­ tinction is not altogether clear. An interpretation of the concept of connotation hinges on that of the concept of attribute. The concept of denotation is ambiguous since Mill maintains that a general name, e.g., man denotes not only the individuals it is correctly predicable of (e. g. , Plato, Socrates, etc.), but also the class of such individuals. Compare the distinction between comprehension and �·extension in the *Port-Royal logic, depth and breadth in *Peirce, and sense and reference in *Frege. U.N.

Mimicry Biological mimicry provides a signal to show that an organism displays adaptations of a special type in order to protect it from its enemies. These adaptations involve two alternative conditions which are distinct from one another yet capable of intergrading. The situation is best illustrated by the Lepidoptera (butterflies and moths). Some of these insects belong to groups that secure considerable immunity from insectivorous animals, chiefly birds, because of an unpleasant taste or poisonous qualities. It is evident that such species should be easily recognized so that they can be avoided as potential prey. This is generally achieved by a striking and characteristic appearance, made the more evident by a slow flight and flaunting behavior. Yet other species that are edible protect themselves deceptively by taking as a "model" and copying those genuinely unfit for food. This is Batesian mimicry (named after its discoverer, H. W. Bates, 1 825-1 892), and its results have been established by experiment. Birds that have been brought up in cap­ tivity willingly feed upon such mimics until supplied with their models. On sampling these they reject them with signs of disgust and often nausea, after which their mimics, previously accepted, are avoided also, sometimes for months, owing to the long memory of birds. It is evident that this situation, tho ugh an advantage to the mimic, constitutes a threat to the model when a given signal, resulting from appearance and behavior, comes on balance to be associated with something edible rather than inedible. That difficulty is, however, met in two ways. The inedible species have developed a tough, leathery cuticle. Thus they can survive pecking or crushing to an extent that reveals their unpleasant qualities; those that belong to groups not so endowed would be killed by such treatment because it would crack their body wall. Furthermore, Batesian mimics are often polymorphic, that is, they include distinct forms capable of mimicking quite dissimilar models. They may also possess a nonmimetic

Modality

553

phase. Whether they do so or not, in these circumstances they adjust numerically until all receive equal protection. A single edible species may copy several distinct inedible ones. Such mimics, though so different, are each of them very constant in appearance so as to be easily confused with their respective models. Yet when such polymorphism is not mimetic, its various phases are themselves variable. In all circumstances, the condition is controlled geneti­ cally. The Batesian mimicry of most but not all butterflies affects one sex only, the female. The female is in the greater need of protection since it tends to be exposed during egg-laying and it is the less adroit on the wing due to its relatively heavy body. Moreover, a differential elimination of males may not be harmful, as a single individual can often fertilize several females. Furthermore, it may be important to keep the males uniform, since their color and pattern frequently act as a sexual stimulus. Mimicry can also be of the Mullerian type (named after F. Muller, 1831-1897, who discovered it). Here two or more species protected by unpalatable qualities come to resemble each other. In this we find an adaptation to reduce the destruction due to birds, which have to learn by experience what forms can and cannot be eaten; Mullerian mim­ icry ensures that each sampling is of wide application. The condition does not involve polymorphism, since that is an adaptation to produce diversity whereas the Mullerian situation is one that promotes similarity. Both types of mimicry can result in a resemblance between species that belong to widely distinct groups : yet the resemblance may be produced by different means, since it is only necessary that the forms should resemble each other superficially. Thus, their similar color can be due to chemically different pigments, or a characteristic mark on the abdomen may be effectively mimetic when reproduced on the wings if these overlap the body. Mimicry can be remarkably accurate, the species involved departing to an extreme degree from the appearance of those related to them. In addition, it may affect habits and behavior, and even, perhaps, scent. It may be noted, however, that size does not usu­ ally enter into mimicry, as it is not easily assessed on the wing. The adaptations of mimicry are not confined to insects, nor indeed to animals. Cer­ tain weeds have come to resemble the crops among which they grow. Thus fibre-flax is impoverished by a plant ( Camelina linicola) very similar in appearance. The selective agent here has been man, eliminating those undesirable seeds that are the most easily recognized. E.B.F.

Modality Let the term modal functor be a means of defining the concept of modality. A functor is said to be modal, in the broadest sense of the term, if it is a nonextensional (nontruth­ functional) functor forming a *sentence out of a sentence. Thus, F is a modal functor (modal operator) operating on a sentence A if and only if the *truth-value of F(A) is not determined (at least, not entirely determined) by the truth-value of A. In a narrower, more traditional sense a modal functor is identified with an alethic modal functor (see below).

554

Modality

When a sentence A is made subject to a qualification by means of a modal functor (e. g., "it is possible that somebody has no friends"), this qualification is said to represent a modality to which A is subjected, in the above example it is possibility. The above definition is concerned solely with modality (Latin modalitas) de dicto (i.e., about a sentence), which was named thus in contradistinction to modality de re (about a thing). The distinction was introduced by the schoolmen *Abelard, Peter of Spain, Thomas *Aquinas, etc. Modality de dicto attaches to the whole statement (e.g., "It is possible that Sherlock Holmes lived in London"); modality de re attaches to the verb (e.g., "Sherlock Holmes possibly lived in London"). The term de re indicates that the statement in question asserts something about an actual thing that possibly did such-and- such (cf. Prior 1962a; Kneale 1962; Snyder 197 1 ). There is an obvious analogy with the two uses of negation. Negation de dicto (if we adopt the same term) (e. g., "It is not the case that Sherlock Holmes lived in Paris") affects the whole statement, and yields a sentence that is true both if Holmes never lived and if he lived somewhere outside Paris. Negation de re (e.g., "Sherlock Holmes did not live in Paris") allows us to infer that Holmes existed and that he lived outside Paris. In generative grammar, a functor de re is part of a verb phrase. In ancient and medieval logic, modality was construed in a narrower sense, being equivalent to the contemporary concept of a/ethic modality, that is, modality relating to truth (Greek: aletheia). It is necessarily true (or: false) that A. It is actually true (or: false) that A. It is possibly true (or: false) that A. The phrase "necessarily true" may be abbreviated as "necessary, " and "possibly true" as "possible." In a symbolic formulation the above formulas may be rendered as N(A), A, and P(A), respectively. There are relations between them that are analogous to the known relations between a universal sentence, a nonquantified sentence, and an existential sen­ tence. These are as follows: (1 ) N(A) ::> P(A), (2) N(A) ::> A ( a necesse esse ad esse valet consequentia), (3) A ::> P(A) ( ab esse ad posse valet consequentia), (4) P( ~ A) = ~ N(A), (5 ) N( ~ A) = ~ P(A) . Von Wright (195 1 ) noticed that parallel relations hold within other triples of expressions, viz., those relating to knowledge (Greek episteme), which he called epistemic modalities, and those relating to duties (Greek deon 'that which is obligatory'), which he called deontic modalities. The logic dealing with epistemic notions is called epis­ temic logic. The logic dealing with deontic notions is called deontic logic. There are two areas related to deontic logic, namely, logic of norms and logic of commands or impera­ tives. An important contribution to logic of commands is found in Rescher 1966. W.M.

Modality

555

Greimasian Modality Starting from the division of elementary enunciates into *actants and *predicates, Greimas defines modality, in accordance with the definition given by B. Pottier, as " that which modifies the predicate of an enunciate." Modalization is then defined as "the pro­ duction of an enunciate called modal, overdetermining an enunciate called descriptive" (Greimas and Courtes 1979 : 230). This analysis of predicates is central in Greimas' s theory for at least two reasons: (1 ) considered as a *syntactic class, a predicate is constitutive of the relation between actants (i.e., subject/ object); and (2) divided into "static" and "dynamic" predicates, it permits a m inimal typology of elementary enunciates: enunciates of state and enunciates of doing. On the basis of this typology it is possible to distinguish among "modal" and "descrip­ tive" enunciates, according to their reciprocal position (thus avoiding an inductive inventory of verbs or modal locutions which in any event vary from one natural lan­ guage to another). Enunciates of doing and enunciates of state (also called "of being") are, then, capable of occurring either in a syntactic situation (descriptive enunciates) or in the hypertactic one (modal enunciates) . Through the combinatory of modal or descriptive enunciates of doing and enun­ ciates of being we can thus obtain: (1 ) doing modalizing being (*pragmatic performance or act: faire-etre 'to make someone or something to be or to become' ) ; (2) being modal­ izing doing (pragmatic *competence presupposed by the act) ; (3) being modalizing being (veridictory modalities, which can be interpreted as *cognitive com petence); (4) doing modalizing doing (factitive modalities, which can be interpreted as cognitive per­ formance) . Being modalizing being and doing modalizing doing are also called "transla­ tive modalizations" (Greimas 1 976a:93) in order to distinguish two different subjects, one for the modal predicate, one for the descriptive predicate. Competence can receive modal overdetermination from an inventory which (as Greimas emphasizes) is provisional and which is formulated in a hypothetical-deductive way. The inventory contains four modalities: wanting, having to, being able and knowing which can modalize being and doing. On the other hand, a reflection on modes of semiotic existence - virtual, actual and realized - permits the reorganization of the modalities in the following table (Greimas and Courtes 1979 : 231 ) . Modality

virtualizing

actualizing

realizing

esotactic

having to

being able

doing

endotactic

wanting

knowing

being

Here the term esotactic refers to modalities "capable of entering into translative relations," while endotactic refers to simple modalities whose subjects (of the modal predicate and of the descriptive predicate) are identical or in syncretism. As it is not possible to give a formulation of the four modalities considered as "sup­ porting terms" inde pendently of their "supported terms, " we will speak syntagmatically of modal structures (i.e., wanting to be, wanting to do, having to be,

556

Modal Logic

having to do) while from a paradigmatic point of view they can be considered as modal categories and represented in a semiotic square. Every modal category represented in a semiotic square is capable of receiving an "appropriate and arbitrary" denomination. This procedure, which Greimas calls "denomination," is an arbitrary passage by con­ densation from two predicates to a single modal value. See, for example, the diagram of the ' 'deontic" modalities along with their denominations :

having to do

having not to do

not having not to do

not having to do

(prescription>
. . . , Bk) and B' = f'(B t > . . . , Bk), then not in general F(B) = F(B'). (34) guarantees the dependence of F(B) on the interpretation of its parts, (35) provides for the contribution of the structure of B to its interpretation, such that different combi­ nations of the same constituent parts may have different interpretations. The dual code of natural numbers discussed above clearly meets (34) and (35), if f is construed as the ordering of digits and g is the operation defined in (11b). The following simplified example shows how the relevant notions apply to the more complex structure of natural languages. Let (36) be the structure of "John melted the ice" relevant for its semantic interpretation. In other words, (36) is to be construed as f (John, Past, melt, the, ice). Then (37) indicates the interpretation of its ultimate constit­ uents and (38) their integration into the interpretation of (36). (36)

John

I NP

Past

melt

the

Infl

V

Det

I

I

s

\

�VP

I

ice

I

�NP

�/

(37) (a) FQohn) = JOHN (b) F(Past) = PAST (c) F(melt) = [CAUSE X [GO Y [FROM SOLID)[TO LIQUID ) ] ] ( = (24) (a)) (d) F(the) = DEFINITE (e) F(ice) = ICE (38) F((36)) = g((37)(a), (37)(6), (37)(c), (37)(d), (37)(e) = [PAST [CAUSE JOHN [GO [DEFINITE ICE)[FROM SOLID)[TO LIQUID] ] ] ]

Semantics

875

In effect, g consists in four operations: (a) substituting JOHN for X, (b) applying DEFINITE to ICE, (c) substituting the result for Y, and (d) applying PAST to the out­ come of these operations. There are various ways to formalize this procedure in a more general way, such that g derives in a systematic way from f for arbitrary syntactic struc­ tures, but we need not go into those details here. Notice, incidentally, that the structure of (38) derives in part from the internal structure inherited from the constituents and in part from the combinatorial operations determined by the syntactic structure of the sen­ tence. This is due to the fact mentioned earlier, that syntactically primitive elements might have a semantically complex interpretation. The next point to be noted is that within the range defined by the two conditions (34) and (35) various types of compositionality can be distinguished according to the way in which g depends on f. I will briefly characterize and illustrate two variants, which might be called strong and weak compositionality. We will say that F is strongly compositional if the following additional condition is met : (39) C = g(C t > . . . , Ck ) properly contains Ci for 1 .:5 i .:5 k Intuitively speaking, strong compositionality does not change or distort the interpreta­ tion of the constituent parts. Thus, in the above example, F(the ice) = [DEFINITE ICE] would be strongly compositional, as the constituent meanings are not changed by their integration into the meaning of the whole. Things are less obvious with respect to F(melt), as here the two variables are substituted by the interpretation of the subject and the object, respectively. Hence the effect of (39) depends in part on the interpretation of "properly contains." Thus, we might in fact distinguish two variants of strong composi­ tionality, one that allows the substitution or specification of free variables - the filling of empty slots, so to speak - and one that does not. If we call the latter "strict composi­ tionality," then (38) would be strongly, but not strictly compositional. Although the dis­ tinction is rather subtle and in part a purely technical issue, there seem to be certain inter­ esting phenomena turning on precisely this distinction. Consider, for example, cases like (40) and (41): (40) (a) (b) (41) (a) (b)

We promised John to melt the ice. John promised us to melt the ice. We promised John to help each other. John promised us to help each other.

Here the interpretation of the infinitives to melt the ice and to help each other - more precisely: the determination of the element that occupies the subject position - depends on the matrix sentence in which they are embedded: In the (40a) and (4 1a), we specifies the subject variable, in (406) and (41 b) John, leading to the ungrammaticality of (4 1 b ). Hence, sentences like these are not strictly compositional, illustrating non-trivial conse­ quences of this property. Turning to the far less subtle distinction between strong and weak compositionality, we notice first that weak compositionality is something like a compromise between strong compositionality and the fixed, non-compositional inter­ pretation of syntactically complex expressions. To illustrate the latter case, consider an idiom like kick the bucket. Although syntactically a verb-object-construction, it cannot be interpreted compositionally. Its meaning is simply that of die, and even if we assume that F(die) = F(kick the bucket) is semantically complex, namely something like (42),

876

Semantics

there is no way in which this structure derives from the interpretation of kick and the

bucket.

(42) [GO X [TO [NOT ALIVE] ] ] In short, kick the bucket must b e treated a s a syntactically complex ,;dictionary entry. This kind of non-compositionality is generally considered as a defining property of idi­ oms. There is, however, a large class of idioms that are weakly compositional in that they meet conditions (34) and (35), violating only (39). To illustrate the point, consider cases like pull strings or spill the beans. The fact that they are compositional in the sense of (34) and (35) can be seen as follows. Condition (35), which requires the meaning of the whole to be different from the collection of the meaning of its parts, is trivially met. Hence the crucial point is (34), which requires the interpretation of the whole to depend on the meaning of the parts. That this is the case is shown by the possibility of composi­ tional modification as in ( 43): (43) (a) He pulled strings you wouldn't think of. (b) John spilled every single bean. The interpretation of (43a) obviously requires that F (pull strings) is built up from semantic constituents that are available for combinatorial operations in a way that does not hold for strictly non-compositional expressions. Thus, weak compositionality as exhibited by (compositional) idioms combines and simultaneously changes the interpre­ tation of the constituent parts. In other words, in these cases g depends on the Ci it integ­ rates, but it does not preserve them as proper parts of the resulting C, thereby violating condition (39). It is an open question how this peculiarity is to be captured in a systematic way. Its solution is closely connected to the analysis of figurative meaning, which differs from idioms only by its lack of conventional fixation. Without going through further differentiations of compositionality, we may note that sign systems can be classified with respect to the type of compositionality. This does not mean that a sign system is necessarily based on only one type of compositionality. It seems, in fact, to be a characteristic property of natural languages, that various types of compositionality are used in their semantic interpretation. To complete the discussion of compositionality, it must be noted that any composi­ tional interpretation needs something to begin with, i.e., an interpretation of the basic constituents. In symbolic sign systems, this initial interpretation must always be speci­ fied by a list. For natural languages this list constitutes the dictionary. It would go beyond the present limits to discuss the vast complexity of problems connected with the structure of the dictionary, which is, in fact, anything but a mere list, but rather a highly structured system on its own, organized in part by the internal structure both of the syn­ tactic and the semantic structure of the individual entries. I will merely point out that the assumption that compositionality is always rooted in elementary signs the meaning of which are specified in a list must be restricted in two different respects. First, there is an interesting class of structural properties that enter the form B of a sign and contribute to its interpretation C but do not have the status of an independent sign. The most obvious phenomena pertaining to this class are stress and ''·intonation in natural languages. Obviously, (44a) and (44b) differ in meaning in a systematic, compositionally determined way, although stress cannot occur as an independent min­ imal sign.

Semantics

877

( 44) (a) John ate only two apples. (b) John ate only tw6 apples. Secondly, the nature of compositionality varies in essential respects with the different types of sign systems discussed above. Thus, for iconic signs there need not be a list of basic signs from which compositionality might draw. Two-dimensional representations of three-dimensional objects, maps, blueprints, and other visual sign systems illustrate the point in question. Although some of the characteristics of compositionality will also be preserved in those systems, its role in the organization of super-signs is essentially different and must accordingly be analyzed in a rather different way. Further Perspectives The previous discussion concentrates almost exclusively on systems of conventional or symbolic signs, and more specifically on languages - which are a rather special type of semiotic systems - mainly because more detailed notions of semantics have been developed primarily with respect to this domain. It must be added, though, that the treat­ ment is rather incomplete even within this restricted area : some central notions, such as ''truth, ''ambiguity, �-synonymy, have not even been mentioned. Instead of filling these gaps, we will return to the problem of general semantics and ask to what extent the semantics of linguistic systems is representative in principle for other semiotic systems. At least two types of restrictions connected to the choice of symbolic systems should be distinguished: First those that separate properties of symbolic systems from other, mainly iconic systems, and second those that separate human sign systems from those of other species. The first restriction concerns the conceptual aspect, the second the bio­ logical foundation of semiotics. With respect to the conceptual aspect, we might ask whether there are interesting generalizations that hold for the structure and content of the semantics of arbitrary sign systems. A basic condition for any semantic analysis is, of course, the development of a general framework within which more specific question can be pursued. It should be obvious from the previous discussion, however, that these general guidelines are revealing only to the extent that they are filled in with principles and specifications that capture generalizations determining more specific properties of classes of sign systems. We may ask for instance whether the general principles determining the structure of meaning can be characterized in a uniform way, irrespective of the particular class of sign systems. It should readily be seen that, although there may be interesting generaliza­ tions for the class of natural languages, or even of languages in general, it is not very likely that many of these generalizations hold also for systems of pictures or *music. It seems, in fact, more promising to look for specifications that distinguish principles of the structure of meaning in different classes of semiotic systems. (See, e.g., Bierwisch 1979 and Jackendoff and Lehrdal 1980 for contrastive approaches to natural language and music.) A similar point may be made with respect to the content of meaning: rather different kinds of entities are expressed - or projected, for that matter - by means of a novel and a film even if they are based on the same story. Most obviously, the kind of compositionality, and more generally the properties of the mapping from the form onto the meaning of signs, is radically different for different types of sign systems. To men­ tion just one arbitrarily chosen point : real time plays a crucial role (though in different

878

Semantics

ways) for both film and music, while it is irrelevant for the denotative aspect of lan­ guage. The meaning of a sentence does not change with the rate of speech. Turning next to the biological foundation of semiotics, it is obvious that different species have access to extremely different sign systems, although these differences do not correspond in any direct way to the classification of sign systems according to the conceptual aspect. But even the different sign systems acquired by human beings are not very likely to correlate particular inherited capacities to sign systems classified according to purely conceptual criteria of general semantics. It rather seems that certain mental dispositions are exploited simultaneously by quite different sign systems, and that one and the same sign system draws on different modules of mental organization. In other words, empirically interesting generalizations concerning the organization of human sign systems may cut across defining characteristics provided by a purely conceptual classification of these systems. To the extent that semantics is ultimately to be construed as an empirical discipline in the sense discussed earlier, this observation is not accidental but belongs to the very core of semantics. Seen in this pespective it is not an incident that revealing semantic activities usually cluster around rather specific classes of sign systems, with general semantics in the strict sense providing merely a framework of general and fairly unspecific concepts. M. B.

Action-Theoretic Semantics Thanks mainly to �·Peirce, ,:-Morris, and Wittgenstein, the basic idea of realistic or correlational semantics (meanings are entities) has been opposed by the idea that mean­ ings of signs consist in nothing other than their use. Semantics is to be based on pragmat­ ics. Action-theoretic semantics tries to give this idea - left too often too vague in discus­ sion - its proper basis in general action theory. Realization of this program proceeds, roughly, in two steps. ( 1 ) explication of a concept of communicative act general enough not to presuppose concepts of regular/ conventional/linguistic meaning; and (2) explication of the latter concepts by means of the former one. The central thesis is that as far as our understanding of the meaning of signs is concerned, their communicative function is to be given (explicative) primacy which is not to deny that they may have other functions as well. On step ( 1 ) a particular action is said to be understood by us if we know the intention with which it was done. This holds with communicative acts, too, but their distinguishing feature is that, from the utterer's point of view, they will be successful (i.e., will reach their respective primary aims, e.g., audience's doing or believing something) if and only if they are understood by the addressee. Thus, communicative acts necessarily have to be explicated with reference to reflexive intentions. This idea was first examined by Grice ( 1 957), but his conditions were not sufficient by not implying that the audience is intended to recognize the utterer's act as being a communicative one. For this to fol­ low, recognition of all intentions necessary for communicative acts (and not only of the primary one) must be intended; this solution was hinted at by Strawson ( 1 964 ), exploited by Schiffer ( 1 972), and discussed and formally reconstructed by Meggie ( 1 98 1 ) . The generality of the resulting concept of communication is reached via abstrac­ tion from the different reasons the speaker may have for his expectation of really being

Seme

879

understood by his audience - and so from the meanings the signs uttered may have, in particular. Step (2) concerns the meaning of signs as types of acts c:·speech act forms) or as the products thereof (expressions). For whole utterance types the meanings of both may be equated. Regular meaning (of sign S relative to the population P and situations of type T) can be identified with the communicative intention, with which the respective utter­ ances (of S by P-members in T-situations) are regularly connected (Alston 1 964; ch.2; Grice 1 968). For this connection to be reliable, the relevant regularity must be (in P) common knowledge - which a fortiori must be the case, when regular meanings are to become conventional. Conventions in general and conventional signifying systems in particular have found the long-awaited clarification in Lewis 1 969, where, using the frame of game theory as a part of rational decision theory, they are treated as regularly chosen solutions of coordination problems (see the critical remarks on D. Lewis's game­ theoretic notions in Gilbert 1 98 1 and Lewis' application of these notions to signifying systems modified - in a way accepted by him in Lewis 1 975 - in Bennett 1 973 ). Unfortu­ nately, when coming to languages and linguistic meanings of structured expressions, Lewis opted out of the whole program by putting his proposals on the shoulders of realistic semantics. How this predicament may be avoided in relation to a simple lan­ guage (of first order '�predicate logic) is shown in Kutschera 1 976. See the broad discus­ sion of the program in Bennett 1 976 and Meggie 1983, the former trying to couch it in more behavioristic terms, the latter in more formal terms. In its main premise, i.e., in starting to talk about sign meanings by talking about actions, there are analogies with endeavors of the German "Erlanger Schute" (Kamiah and Lorenzen 1 967; Lorenz 1 970), but differences in method. The program will turn out to have wide applications, its most important one being to constitute the only approach by which conversational �·implicatures (Grice 1 9756) can be treated more systematically. G. Me./M. U.

Seme Seme is a term used to indicate a unit in ,:•semantic analysis. There is little agreement on the place the seme occupies in the semantic hierarchy or the question to what unit, if any, it corresponds in phonological or morphological analysis. In spite of a great variety of definitions, it is possible to distinguish two major groups, both European. The first group is small compared to the second, and for that reason is much more homogeneous. ( 1 ) Seme is a term introduced by Buyssens ( 1 9436: 1 2) to indicate "any ideal proce­ dure whose concrete realization makes communication possible" ('tout procede ideal dont la realisation concrete permet la communication'). The concrete act producing �·communication is called a semic act and will be defined as the concrete realization of a seme. Following Buyssens and relating the term to ,:•Saussure's linguistic sign, Luis Prieto ( 1 968: 1 25) uses the term seme in the same sense as Saussure used linguistic sign. By choosing seme for the unit consisting of a signified (signifie) and a *signifier (signifiant), Prieto wants to avoid the ambiguity arising from the fact that sign is often used as an equivalent of signifier.

880

Seme

In Prieto's theory, the seme or sign is, as it was for Buyssens, at the level of the •·ut­ terance (enonce') , and not at the word level. In his book, published in 1 964 but written after the 1 968 publication, Prieto posits a unit of content, the noeme ( noeme), which con­ sists of the maximum number of features that are in what he calls an "ab relationship." In such a relationship, all features are expressed formally and cannot be left out or replaced without changing the form. For that reason, spoken French does not have the same noemes as written French, which expresses features that are not expressed in spoken language. Consider, for example, "elle regarde," which in its written form has the following noeme: (subject) determined - (subject) feminine - (subject) third person - (subject) singular ( 1 04 ). As "elle regarde" and "elles regardent" are pronounced iden­ tically, the element "(subject) singular" is left out of the spoken form. There are gram­ matical and lexical noemes, the latter equally depending on formal expression. The Buyssens-Prieto seme concept is not very well known outside French-speaking structuralist and functionalist circles, yet the conflict between this view and the seme as defined by the second group is not imaginary. Therefore, Claude Germain ( 1981: 1 34) prefers to avoid the term altogether, introducing ,:• meaning features (trait de sens) to replace seme as interpreted by the second group. (2) Seme is a term to indicate a small unit of meaning, a feature of *content. The Saussurean signified may contain several semes. Heger (1969 :55) uses the same termi­ nology as Prieto, but with an entirely different meaning. He speaks of a minimal unit of content, arrived at by the strictly immanent analysis of a given language. This seme (seme) is opposed to minimal units that are defined independently, without referring to a specific 'flanguage, and can thus be considered to be semantic 'funiversals. For this second category Heger proposes the term noeme (noeme), pointing out that a unit may have double status when independent and immanent procedures yield identical results. The combination of semes and noemes occuring together forms a sememe (sememe) . The analysis of language ( 1 ) 'tongue' and langue (2) 'language' may illustrate Heger's ideas: langue ( 1 ) has the elements 'organ' + 'fleshy' + 'placed in the mouth,' etc. ; langue (2) is decomposed into 'system' + 'of means' + 'of communication,' etc. Heger does not indicate any formal criteria, nor does he tell what analytical procedures he uses. Thus, his analysis makes the impression of being wholly intuitive and the addition of "etc." at the end of the enumeration confers inevitably an open-ended character to the series of semes of the lexical items involved. In order to circumvent these problems, many definitions and descriptions of the seme include procedural requirements. The most common procedure goes back to •·Hjelmslev ( 1 943, English translation 1 953, French translation 1968). On the level of content Hjelmslev works in the same way as on the level of •·expression, resolving the minimal ,:•sign's content through a further analysis on the basis of functions into smaller compo­ nents. Take, e.g., the units 'ram,' 'ewe,' 'sheep,' 'he,' and 'she.' The elements 'ram' and 'ewe' can be split up into 'he- sheep' and 'she-sheep'. We can oppose 'ram -+ he - sheep' to 'stallion --+ he-horse'. The meaning components are called /igurae of content. In a discus­ sion of Hjelmslev's work, Martinet ( 1 946:39) criticizes the fact that the figurae are in themselves independent content for expression, i.e., words. If one replaces 'she' by 'femi­ ninity', thus suggesting a more abstract character for the component in question, one obtains a picture of Greimas' views ( 1966a:22ff.). He speaks of differential elements which are at the same time constituent elements: s (is a quality of) A, where s 1s femininity and A girl. The quality is identified by way of the opposition girl-boy.

Seme

881

Minimal pairs of binary ''oppositions form the cornerstones of most analyses. This kind of analysis yields, however, for any given language an unmanageably large number of semes, so it is not surprising that efforts have been made to create some order by sub­ dividing semes into different categories. Important representatives of this trend are Coseriu (1973) and Pottier (1974 ). The former tries to distinguish between classemes, i.e., features that make it possible to structure the ''lexicon, and semantic features that do not have that property. Very general classemes encountered in many languages would be male/female, animate/inanimate, and human/non-human (see Coseriu 1973 : ch. 4). Classemes have an important role in the structuring of semantic fields, where they often represent the common element that holds the items in the field together. Pottier introduces a number of subcategories of semes, defined in the glossary of his book (1974:3 30-31). Besides specific semes, generic semes, virtual semes, applicative semes, and descriptive semes, Pottier uses classeme, semanteme, virtueme, and sememe in his ana­ lyses. He applies his method to full lexical terms as well as to function words, for which he uses the label grammeme. A few examples (1974 :70) will help to clarify this terminol­ ogy : (1) gueule

applicative descriptive / orifice / / to swallow / semanteme / material / / animal / classeme , connotes "popular. if applied to hwnan virtueme

(2) armoire, with a dictionary defintion "meuble, generalement en bois, ferme par une ou deux pones, garni interieurement de planches ou de tiroirs, et servant a ranger !es objets domestiques" (piece of furniture, generally made of wood, closed by one or two doors, with shelves or drawers inside and used for arranging household objects), yields the following result :

semanteme

classeme

virtueme

descriptives

applicatives

/closed by doors/ /provided with shelves or drawers/

/to range household objects/

/material/ Idiscontinuous! (derived from the coverword meuble) /of wood!

(3) jusqu 'd

Semanteme: classeme: virtueme:

/movement, limited, the end term is stressed/ /spatial, or temporal, or notional domain/ enters into fixed sequences such as jusqu 'd la Saint Glin-Glin 'until the time when the moon is made of green cheese.'

Pottier defines seme as ''distinctive feature of the substance of a sign's signified (at the morpheme level), and relative to a given set of signs. For the applicative seme, the feature should evoke the function of the element under consideration or what it is intended for.

882

Seme

The descriptive seme gives the nature of the element under consideration. As is illustrated by the examples, the subcategories cut across the following three subsections: the generic seme, which is an element of the classeme, enabling us to bring together two neighboring sememes by reference to a more general class; the specific seme, constituting an element of the semanteme and allowing us to oppose two sememes that are very close thanks to an individual characteristic; and the virtual seme, which is an element of the virtueme, representing unstable and often individualized associations, but within the �·competence of the speaker. Whereas the sememe is the total set of semes of a sign at the morpheme level and forms the substance of the signified, the semanteme includes only the specific semes of a sememe and the virtueme only the virtual ones. It should be noted that the distinctive feature seme is not accepted without criticism as a linguistic unit by everyone. Perrot (1958:692) rejects it on the grounds that there are no formal criteria on which to base the analysis. Dubois, though less categorical in his criticism (1 966:47), draws the attention to the danger of analyzing social structures instead of linguistic ones. When he studied the semantic field of domestic animals, Mounin ( 1 965) acknowledged the fact that the criteria for inclusion in the field were heterogeneous (linguistic, biological, social), and the choice of criteria a subjective one. This statement brings into question the status of the classeme and of the pertinent semantic feature. Kassai (1 969: 343) warns that it is extremely difficult, when trying to define connotation and grasp the nature of meaning, to stay within the framework of linguistics. The difficulty of pinning down the exact difference between two synonyms, for which one would need specific semes, is discussed by Schogt (1976: ch. 2) with refer­ ence to Saussure's synonyms craindre, redouter, and avoir peur. Fran�ois (1 979: 249) stresses the open-ended character of the componential analysis into semes, especially the connotative ones (or as Pottier calls them: virtual ones). The whole area of virtuality appears to be fuzzy, the just-quoted analysis of jusqu'd revealing a virtueme simply because St. Glin - Glin is virtual. One could disagree here with Pottier and posit tempo­ rality within the classeme exactly as for jusqu'd la Saint Nicolas. This is a secondary point, however. The main objection is, as we have seen, the lack of formal criteria, and, even if there are criteria, it is difficult to apply them beyond the mostly well-chosen examples that are given. Zawadowski (1975: 91) deplores - not without reason - that little research has been done on the seme, be it in field studies or in the works on semes as immediate morpheme-meaning constituents. There are, however, some recent studies where min­ imal semantic features are examined much more thoroughly. Ebeling operates with mi­ crosemes, which he defines as follows: "A minimum set of (one or more) semantic particles that correlates with something in the formal plane. If the formal correlate contains one or more units, the (micro)seme is a sign­ meaning. If the formal correlate is an *intonation, then the (micro )seme is an intention . . . . If the formal correlate does not contain any unit and is not an intonation, then it consists of the arrangement of units correlating with other semes; in that event the (micro )seme is a supplemen­ tary seme . . . . ( 1 978 :26)

Furthermore, Ebeling develops a hierarchy to take into account the grouping together of microsemes (occurring in �-texts) : Two hierarchies in a language concern alternations. There are a formal and a semantic hie­ rarchy. The units in these hierarchies are, in descending order:

Seme

883

sememe morpheme macroseme macromorph (micro )morph (micro )seme A macromorph is a set of formally identical (micro )morphs with a significantly similar meaning. A morpheme is a set of macromorphs with a significantly similar meaning. A macroseme is a set of semantically identical (micro )semes with a significantly similar form. A sememe is a set of macrosemes correlating with the same morpheme. ( 1 97 8 :32)

Ebeling sets up a discovery procedure that will have to furnish the criteria for the estab­ lishment of the required similarity. His unusually comprehensive approach and the very complex discovery procedure - there are more than two hundred rules closely interrelated and noted in abstract formulas - make his important theoretical contribu­ tion not easily accessible, and, as Mignot (1980) states in his review, the practical applica­ bility of the theory has yet to be tested. The influence of North American linguistic theories on Ebeling is undeniable, and the same is true for Mel'cuk, who belonged to the Soviet school of semantics, other important members of which are Aleksandr Zolkovsky and its founder, Jurij Apresjan. Mel'cuk moved to Montreal, where he started a gigantic enterprise, the Dictionnaire Explicati/ et Combinatoire du Franrais Contemporain. He and his collaborators aim at giving an exhaustive description of the French lexicon. An entry has three major sec­ tions, one of which is purely semantic and provides a �-definition, while the other two deal with �·syntactic combination possibilities and lexical compatibilities, respectively. All occurrences, including those in idiomatic expressions, are examined, and ultimately the unit called a vocable is decomposed into lexemes that are in turn decomposed into semantic elements. If the decomposition is pushed to its limits, primitive semantic ele­ ments will be discovered. The lexemes represent the different meanings of one word, e.g., the vocable to respect has different lexemes in to respect the great statesman and to

respect the neighbor 's afternoon nap. The work, based on meaning-text theory (see Mel'cuk 1974), has only just started,

and it will take decades before a sizable part of the contemporary French lexicon will have been covered. The primitive semantic elements remind one of semes but point also toward North American work in which to kill is analyzed as cause to cease to live (see Lyons 1977: sec. 12.5). The term seme is not very common in North American semantics. A major exception is Lamb (1964, see also 1966 ), in whose stratificational grammar there is a sememic level or stratum. In Lamb's theory there are four strata and each of these strata has an elemen­ tary unit as well as a unit relating to the next higher stratum (1964: 60): Principal Structural Units

Stratum Sernernic Lexernic Morphemic Phonemic

Elementary unit

Unit relating to next higher stratum

Semon Lexon Morphon Phonon

Sememe Lexeme Morpheme Phoneme

The units in the column at the right correspond with those in the middle one at the next higher stratum. Thus, apart from what Lamb calls special representational relations, morpheme corresponds with lexon and lexeme with semon.

884

Semiosic Competence, Development of

Lamb's terminology has little in common with either of the European terminologies discussed above, and Lamb's theory has remained outside the mainstream of North American semantics. In general, one can say that interest in analysis into meaningful ele­ ments manifested itself later in America than in Europe. Theoretical approaches and ter­ minologies are not the same, and the seme, as a term for a unit of semantic analysis, what­ ever its place in the hierarchy of units may be, is absent. However, some of the results and problems that remain unresolved on both sides of the Atlantic show a striking resemblance. Lyons ( 1 977: 327-29) discusses the similarities between classemes and semes, on the one hand, and markers and distinguishers, introduced by Katz and Fodor ( 1 963 ), on the other. It is true that, contrary to the seme, the distinguisher is the one feature that lends its specific identity to the lexical unit opposing it to all other units, but that does not make it any easier to see a clearcut distinction between markers and distinguishers. As a com­ parison with European seme discussions, it is interesting to see how the distinction was eventually dropped, and to read reactions to the Katz-Fodor theory by, among others, Weinreich (1972) and Lehrer (1974: 47ff.). H.G.S.

Semiosic Competence, Development of Given a semiosic *competence produced by phylogeny and genetically inscribed, the problem arises as to how this competence is acquired by an individual. This problem can also be called "ontogeny of *codes," stressing that the term refers to the development of the general code-using competence and not the learning of a single code, for instance a foreign language. Remarkable attention has been paid to the psychological and linguistic aspects of this matter, but very scarce are the data on its biological background. It should be born in mind that: (1) A code is a system of intermediation between the individual and reality, (2) numerous codes exist for the various aspects by which reality is interpretable. Codes, at the human level, are reciprocally related and combine to form a •�culture. It follows that there are as many single competences as types of codes. However, the capacity of developing single competences probably depends on a general semiosic func­ tion (a general competence) naturally constructed to decipher what is meaningful for a given organism of a given species. (3) Though semiosic competence is a general feature of all organisms (allowing them to select and use what specifically corresponds to their biological categories and is useful for their survival), only for man is it possible to construct complex codes, historically determined through cumulative transmission and learning and basically represented by means of language. The development of semiosic competence in man is, therefore, based on the use of language, and other language-correlated functions, such as logic and memory. ( 4) A code, in this sense, is historically developed by a community and is therefore largely impersonal with respect to the individual: it is in itself a reality, just as the reality

Semiosic Competence, Development of

885

it serves to interpret. The individual, who is part of the objective world, can interpret the environment by means of objective reality of the codes. The development of semiosic competence involves three aspects: First, a suitably adapted genetic structure in the individual (i.e., a potential for developing the actual competence) bearing information for the neural structures. This can be called "genetic competence." Second, the existence of actual codes in the individual's environment. This situation as a whole can be called the "intersubjective system condition." These two factors are determined historically and through evolution (the former by natural history, the latter by cultural history), but can be taken as invariables for the brief period during which the individual's genetic competence encounters the system condi­ tion in the process of ontogenetic differentiation. Third, an interaction between an emergent individual and a code, that is, between a potential competence and a pre-established system condition. The individual thus models his genetic potentialities on the existing systems, and so implements his compe­ tence. These three aspects must be examined separately.

Genetic Competence This is characteristic of the species, and within the species it is subject to the fluctuations typical of all genetic features. These fluctuations are never such as to prevent individuals using the same code in common. This condition is not necessarily limited to man: many animals are behaviorally inserted in a code system (�·zoosemiotics). The problem of ontogeny is thus relevant to all species capable of using codes. In the case of man, the genetic competence to use a code is a manifestation of the more general competence to perform logico-formal operations that makes it possible to carry out reality-stimulated hypotheses. Thus, as regards the human language function, some kind of genetic innateness of the competence must be postulated (Chomsky 1966a; Fodor 1968), in analogy with the innateness (different from species to species) that must be admitted for code systems in zoosemiotics c:•uexkiill 1940b; Sebeok 1972, 1976a) but a much more relevant importance assumes the impact with the historically determined codes which exist in the environment. In terms valid for all species, this genetic competence can be seen as a "':•structure of categories" forming the code-using competence or "species categoricity" whereby each species sees reality through its own codes. This competence develops as a differentiated function in the course of phylogeny. In the human domain, the genetic code-using competence is investigated by psy­ chogenetics, neurolinguistics, neuropathology, and neurobiology, which deal with two main aspects: (1) the normal development of the child's language competence in connec­ tion with the gradual development of logical operations and affectivity (*Piaget 1968), and (2) dysfunctions in the language competence produced by various kinds of lesions ('�Luria 1959a; Lenneberg 1 967). It must be noted, however, that semiosic competence is essentially unknown at the molecular-neurological level, just as is the molecular mechanism of logical operations.

886

Semiosic Competence, Development of

The Condition of the Intersubjective System The general features of the intersubjective system are : ( 1 ) Codes are outside the individual and have a high degree of inertia. Changes that can be made to the code by the individual alone are negligible. Thus, codes can be consi­ dered invariable, at a first approximation, from the standpoint of the individual who enters into contact with them during his life. (2) Although individual changes are negligible, numerous changes by many individ­ uals can cause language to evolve. In fact, no change is random, but occurs by means of the language and in its direction. Every change, in this sense, is "selected" by the lan­ guage. Thus, the code evolves in an ordered way, preserving its character of "system." (3) As a system of intermediation between the individual and the surrounding world, a code varies in order to correspond with new relationships within the code-using indi­ vidual toward reality. These new relationships are represented by new knowledge and new experiences. The code, in so far as it is a system, is continuously being adapted to reality. Its evolution depends on the necessity of being adequate as a system of intermedi­ ation. This is why codes accumulate human experiences as transmutable languages. His­ torical evolution affects both the 'rsemantic and the ,:•syntactic aspects of codes which may, of course, also be found in ,;nonverbal codes (for instance, kinesics and proxem­ ics ) .

The Interaction Between Individual Competence and Code The interaction between individual competence and codes is the ontogeny of compe­ tence in the strict sense. The semiosic competence of man, for all these reasons, is not a mechanical translation of behavior in a fixed ,:•schema but represents the chance of using a flexible code, which serves to construct hypotheses concerning the world. The individual must be confronted with a concrete code (or set of codes) to develop his genetically established competence. In this process he enters into �-history, from the ahistorical status of his genetic potentialities, and molds himself on the codes he finds. Ontogeny of codes has thus both an aspect of freedom and of constraint, since, on the one hand, the use of codes permits ,:•communication (which, in the case of human codes, also means choice, opportunity, alternatives, etc.) while, on the other, it �arks the constitution of the individual with the imprint of the code or codes that have actually guided his ontogeny. The physiology of the ontogenetic development of codes is based on a chronology of the various stages correlated with phases in the development of impersonal relations that language gradually renders possible. Each phase of development both presupposes phases reached earlier and links up with a network of situations reached synchronically. In all probability this also depends on the maturation of certain cerebral structures through reciprocal conditioning and conditioning by use : it has been ascertained, in fact, in the case of other neurological functions, that the synaptic connections, though genet­ ically established, can be "more precisely fixed" only by use during the first phases of ontogeny. Psycholinguistics in particular studies the definitive inability to develop lan­ guage due to lack of stimulation during the appropriate ontogenetic period and lan­ guage acquisition in the deaf-and-dumb, the blind, etc. Some abnormalities are due less to genetic defects than to the desynchronization of the stages in ontogeny, each of

Semiosis

887

which is linked to given external stimulations. It must also be remembered that the ontogenetic development of the semiosic, code-using competence means not only the evolution of a function, but the discovery of the objective world, and hence an affective, psychological evolution. Ontogeny of codes forms part of the overall development of the personality. Although ontogeny has been taken here as an individual, historical interplay between two structures (the individual genetic competence owed to the species, and the code owed to the culture), it is necessary to make it clear that these structures them­ selves are the historical product of interactions. Two hypotheses can be considered. The first is that semiosic competence has de­ veloped, like all biological functions, through selection. But, unlike other functions, its selection has taken place not against natural situations, but against code-situations - i.e., language. Out of given, primitive forms of communication more and more efficient communicative functions have developed by selection. Not only do these bring with them the advantages of communication, but they are also capable of giving rise to even more complex code-situations capable in turn of acting as selective conditions for even more effective communicative abilities. Thus, it has come about that linguistic compe­ tence has selected itself, and codes have shaped man's genetic and neurological struc­ ture and not just his competence, such that humanity is intimately conditioned and char­ acterized by its genetic semiosic competence. The second hypothesis is that particular codes are the product of cultural history, which is not an abstract process but is influenced by the use of codes by individuals who thereby necessarily keep the code continually adapted to their constitution. If the genetic structure is determined by codes, on the one hand, on the other codes are originated and caused to evolve by being constantly employed in concrete situations. This is why they continually adapt to the natural and cultural realities with which the community comes into contact in the course of time. G.P.

Semiosis The Term "Semiosis " The occurrence of the term semiosis in the literature is for the most part consistent with the broad sense of signification process or '',ign process. Thus, the term signifies for �·Peirce the triadic nature of the operation of a sign (Peirce 1931-66 : 5.484; Greenlee 1973 :105) and ''·Morris uses it to refer to "the process in which something functions as a sign" (1938 :3). For Eco, the term is synonymous with "the process by which empirical subjects communicate, ''·communication processes being made possible by the organiza­ tion of signification systems" (1976c :316 ). Wells equates semiosis with signification ( 1971:99), and Deely equates it with the action of signs ( 1978 :6). The etymology of the term semiosis is clearly outlined by Max Fisch in a discussion of Peirce's theory of signs (Fisch 1978 :32, 41). Unlike most of Peirce's technical sign vocabulary, which was derived from Latin, semiosis may be traced back to a Greek term - crt]µEtrocn� 'semeiosis.' The plural form adopted for the term by Peirce, although seldom used, is semioses (Peirce 1931-66 : 5.489).

8 88

Semiosis

Peirce claims that semetosis, referring to practically any variety of sign action, occurred as a vocabulary item in the "Greek of the Roman period, as early as �-Cicero's time" ( 1 931-66 : 5.484 ; Sebeok 1 976a:48; Greenlee 1 97 3 : 1 9). Greenlee suggests that �·Locke's writing, rather than the Greek literature of Cicero's era, was the actual prece­ dent of Peirce's use of semiosis as well as related lexical items (Greenlee 1 973: 1 9 ; Locke 1 96 1 , Bk. 4: ch. 21, sec. 2-4, pp.309-1 0). The Greek term syntactics and 'pragmatics (Morris). This is the main use of semantics among logicians (since '>Carnap and Tarski), but the frequent adoption of pragmatics by the linguists might give a chance to the concept in linguistics as well. (*Benveniste's opposition between "le semantique" and "le semiotique" uses semantique in an entirely different meaning.) "Semasiology" Prior to Brea!, the German Latinist Karl Christian Reisig had named the historical study of lexical meanings with a term derived from the Greek sema. In 1825, then 1839 ( Vorle­ sungen uber lateinische Sprachwissenschaft, Leipzig), Reisig divided "grammar" into three parts: etymology, semasiology (Semasiologie), and syntax. The term was used during the nineteenth century by F. Haase, Russel Martineau, Heerdegen ( Un­ tersuchungen uber lateinische Semasiologie, 1875), and Lazar Saineanu (lncercare asupra semasiologiei limbii romane, 1887, which appeared the same year as Breal's Semantique). But, besides the fact that Gustav Stern, in his influencial Meaning and Change of Mean­ ing, preferred semasiology to semantics, the term was restricted, in German linguistic from which it was borrowed in several languages - to the opposition between onomasiology and semasiology (the study of naming concepts and classes of things versus the study of meaning of linguistic units such as the word, or the '>name). A variant of semasiology, sematology, was used by Smart, commenting on Locke ( Outline of Sematology, 1831 ), where Sematology, distinguished from Physicology and Practicology, equals Semiotics, on a philosophical level. "Signifies" The word signifies, form Latin signi/icare, significatio, was coined by Lady Victoria Welby but did not prevail over semantics or semiotics. Only the Greek root sema seems to be accepted in modern usage.

900

Semiotics

"Semiology" The term semiology, and its variant semeiology, have been used in several languages since the seventeenth century in order to designate the science of medical signs and symptoms. Semeiotics and semiotic(s) have also sometimes been used in that sense. As is well known, the word semiologie was adopted in French by Saussure, in his Cours de Linguistique Generale, to name the science studying "the life of signs within society." Discussed by Naville (Nouvelle Classification des Sciences, 1901), the term was scarcely in use, even in French, until Buyssens, and then Prieto, gave it a more restricted definition. Later, still in French, *Barthes took up the term in quite a different way (Elements de Semiologie, 1965), defining it as "the part of linguistics (dealing with) the great signi­ fying units of discourse." The view is derived from Hjelmslev's theories, and entirely opposed to Saussure's. The use of semiologie by some French-language philosophers and linguists (such as Ricoeur, Derrida, Benveniste) is not always clear, somewhere between Saussure's and Barthes' positions. The attempts to define semiologie, either by itself (G. Granger) or in contrast to semiotics (semiotique) (A.J. Greimas) did not resolve the terminological problem raised by the existence of two traditions, derived from the Ecole de Geneve (semiologie), and from Peirce and Morris. '

1

Semiotics" ( "semiotic")

The term semiotics and the concept came from Locke's use of the Greek word Semeiotike in his Essay (1690). Semeiotike is opposed to Physics and to Practice (ethics), and include the theory of ideas and the �·philosophy of language with logic. After Locke, Lambert, Smart (see sematology above), and the Polish scholar Hoene Wronski used Locke's conceptualization, where semiotics (or sematology) practically stands for "philosophy of knowledge." Conceptually, then, a scholar like *Cassirer is a ' 'semiotician" in this sense. But the thinker who gave semiotic (firstly semeiotic) its modern content, as well as its terminological success, is, of course, Peirce. It will suffice to recall here that Peirce divides semiotics (including logic proper) into "pure grammar" (the study of the "ground"), "logic" (the study of the "object") and "rhetoric" (the study of the "*in­ terpretant"). Charles Morris must be credited with having subdivided semiotics as a science into syntactics, semantics (see above) and pragmatics. The trichotomy was terminologically useful, especially to improve the ambiguous status of the word semantics. Then, within the framework of the "unified sciences," Morris gave semiotics its full interest for lin­ guists, logicians, social scientists and specialists of the humanities. The next and last step toward the triumph of semiotics - the form with an -s is intended to align it with linguistics, mathematics, etc. - was the 1964 meeting at Indiana University (with Margaret Mead, T.A. Sebeok, A.S. Hayes, M.C. Bateson, etc.), after which semiotics came to be used all over the world (e.g., Italian semiotica, French semio­ tique). It was already known in the German philosophical tradition (from Lambert), and more recently, in Russia. Its few competitors (German Zeichenlehre, French semiolo­ gie) do not have the same interlinguistic advantages. Of course, such success entails many ambiguities, and it is one of the intentions of a dictionary of semiotics to describe and reduce them as much as possible.

Semiotics

90 1

Besides the different meanings of the term semiotics as a name for a science - or a type of knowledge - whose topic is signs, sign systems and functions, it has a different meaning, e.g., in Hjelmslev's works, where it refers to a sign system. In that kind of ter­ minology, the term metasemiotic is equivalent, either to semiotics, or to Saussure's use of the term semiologie (semiology being, in Hjelmslev's terms, a metasemiotic whose object is a nonscientific semiotic). "�·Semanalysis" ("semanalyse") See this word. Semiotic Domains

The names for semiotic domains are either compound words (anthroposemiotics, endosemiotics, ,;zoosemiotics - Sebeok) or phrases (pure, descriptive, applied semiotics Morris; semiotics o/. . . ) or specific words, such as kinesics, proxemics, etc. In the areas where semiology is used the term is included in similar phrases. A.R.

Methodological Nature of Semiotics

Five Notions of Semiotics One must realize that the word semiotics is polysemic. What, then, are the notions of semiotics? In other words, what meanings does the word semiotics have ? It appears that five meanings can be distinguished: thus, we would have five notions of semiotics, a fact we are generally unaware of. First, semiotics can amount to syntactic, semantic, or pragmatic properties of the sign, or to properties or features of one of the above kind, or of any two kinds, or of all three kinds. Moreover, either selected properties may be considered in their entirety. For example, instead of asking about the semiotic properties of a given sentence, we may ask about the semiotics of the sentence. Similarly, we could ask what is its syntax, or syntactics, what is its semantics or its pragmatics. Since in all these cases we are inquiring about properties, we propose to introduce the terms semioticsp , syntacticsp, semanticsp and

pragmaticsp ,

Second, semiotics can mean a theory or discipline studying the above-mentioned semiotic properties, that is, a theory or science the object of which is semioticsp and hence also syntacticsp, semanticsp, and pragmaticsp , Such a theory or discipline des­ cribes and defines these properties and their mutual relations, explains their conditioning and the consequences they bring, it classifies and introduces ordering sys tematization and typology of semioticsp elements. As all these tasks are usually per­ formed by science, we suggest for the semiotic theories or the scientific discipline of semiotics as defined above the terms: semiotics5 , syntactics5 , semantics5 and pragmatics5 . Third, semiotics can be a theory or science which examines semiotics5 and its separate fields from a higher level. This higher level is called metatheoretical or meta­ scientific. The Greek prefix •:• meta-, which means after, applies here because a theory must first exist before it can be analyzed on the level of metatheory; we must first have a science before we can elaborate a science of it, that is, its methodology, or metascience. Historically speaking, the terminological practice of attaching the prefix meta to names of theories about theories or sciences of sciences dates from antiquity. The name meta-

902

Semiotics

physica, or more accurately i-a µha i-a or of the semiotic method, it is usually this latter that one has in mind. In its non- or pre-scientific form it was since the oldest times the tool of augurs, dream readers, astrologists and priests explaining the meaning of sym­ bols. Its transformation into a reliable explanatory and heuristic method was slow - the first small advance was made by dowsers and those who forecast weather or the harvest, then, by much larger steps, progress was made by the first doctor-philosophers who began diagnosing illness on the basis of symptoms. Patterns for the method of interpreta­ tion were supplied by another semiotic method, the language text analysis flourishing since antiquity within philology, and particularly in two of its procedures: *hermeneu­ tics, which strove to determine the correct text, and exegesis, which furnished the text with philological commentary. Over the course of time, the scope of these procedures was widened and they were transferred from language to other, extralinguistic, fields. Even the word hermeneutics itself slowly began losing its original meaning and, for about a hundred years now, the meaning we attach to the phrase "method of interpreta­ tion" has been gaining acceptance. The importance of this method increased with the development of sciences such as *history, psychology, and *sociology, in which explana­ tion consists in, among others, understanding human intentions and motives which are expressed indirectly, by means of signs, in both the actions and behavior of man, as well as in his creations (particularly in the sphere of spiritual ,;culture).

91 0

Semiotics

SemioticsA (Applied Semiotics) The applications of the semiotic methods discussed last, namely the method of interpre­ tation, sometimes take the form of distinct scientific disciplines or branches of knowl­ edge. This happens when the method of interpretation is applied not to a single thing, event, or phenomenon but to a certain complex thereof, a set forming a whole, particu­ larly a systematized whole. The obtained result or results constitute what we call the semiotics of the given discipline, semioticsA, or applied semiotics, for example, the semi­ otics of culture or the semiotics of art. There are regions in which the semiotic approach has been natural for a long time. One such sphere is �·literature. Being an art of words, an art using linguistic signs for material, it imposed on researchers, literary historians, and critics the methods of lan­ guage analysis and interpretation. The fact that poetry and prose pieces are capable of being understood, which makes them complex signs, was never disputed. Hence the science of literature is in part its semiotics ; in part, because elements like biographic data on the authors and bibliographic data on their works are not a part of semiotics of litera­ ture. On the other hand, what certainly does belong to it are speculations around the issue of how the experiences of the •=·author have been reflected in his work, how his times are mirrored in it, what is the relation of the artistic representation to the matters referred to in the text - the problem of mimesis, or imitation of reality - next a whole group of issues related to influences and dependencies in literature, the questions of truth and fiction, and also of the so-called literary convention, of realism and fantasy, and of a host of other issues. It is understandable that, given the obviousness of choosing the semiotic method of interpretation for studying literature resulting from the fact that it uses linguistic mate­ rial, the same method was naturally applied to other disciplines based on linguistic texts: to *theater, film, television, religion, or �·myth. This resulted in the appearance of the relevant branches of semioticsA" The next step led beyond the direct relation to verbal signs, for what one artist expresses with words or with words and images, or with words and �·music, another expresses with image only, music only, or sculpture only, without the assistance of words. And so appeared the semiotics of *painting, music, sculpture or architecture, successive applications of the method of interpretation to various arts, those intended by the artists to have a sign character (for example, representational painting or sculpture), as well as those without such intent (music, architecture). How­ ever, all works of art, whether representational or not, have one important feature in common - they are an expression of the artist's experiences. But, one might ask, are not un-artistic objects, our behavior and ways of life, also an expression of our experiences ? And so the semiotic approach, the use of the method of interpretation, finds its way into constantly new areas: the fashion and style of house decoration, customs, and traditional meals, games, and ceremonies; finally, to the entire spiritual and material culture. Not only culture, however, is bestowed with sign character. It is also a feature, or at least it can also be a feature - thanks to our interpretation endeavors - of nature. Semiotic applications reach there too, evidence of which is the semiotics of animals (zoosemiotics) or the semiotics of living organisms (biosemiotics). Characteristic of all kinds of applied semiotics is that they strive to uncover the �·meaning of the investigated reality. They encounter two serious obstacles in this task.

Semiotics

911

The first is our frequent unawareness of the difference between the meaning found by the wielder of the semiotic method of interpretation and the meaning bestowed by him in the course of simply using the interpretation procedure. An easy example will illustrate the difference. The sentence "The sun shines" has a given meaning, and the interpreter encounters the meaning and reveals it; a cloud in the sky has no meaning in itself but we may lend it meaning in the very fact of perception connected with interpretation. This is proved by certain experiments performed by the adherents of the so-called Gestaltpsychologie. The second obstacle in the quest for meaning is the recognized polysemy of the word meaning, in particular the fact that it is being understood also unsemiotically as an aim, result, role, function, positive value, importance, cause, motive, and origin. Those dealing with applied semiotics often overlook these differences in the meaning of the word meaning, and no wonder, since many of their number are experts in the fields to which they would like to apply the method of interpre­ tation rather than being experts in theoretical semiotics. And so they include in the cate­ gory of "text" a variety of objects, events and phenomena either because they are con­ vinced of their importance or value, or because they have discovered their cause or source, or finally because they see that the alleged "text" has an aim or plays some role - all these decisions arising from the different meanings of the word meaning. Hence, semioticsA is a field of thriving weeds of equivocation, and perhaps for this reason attrac­ tive for all those ignoring the clarity of thought and precision of formulation. In both semioticsA and, on the lowest level of theoretical semiotics, semiotics5 , we describe and interpret semiotic properties of the separate signs, words, phrases, senten­ ces, or texts. Thus, our remarks on the Latin sentence "veni, vidi, vici" analyzed earlier may equally well belong to semiotics5 as to semioticsA. There can be nothing in the for­ mulations that would signal, for example, that we are dealing with semioticsA and not with semiotics5 , or vice versa. In descriptions and analyses of semiotic properties of signs, both semiotics5 and semioticsA use the three semiotic metalanguages mentioned above. However, the role of a given description of the semioticsr of concrete signs within semiotics5 is different than within semioticsA because the research tasks posed by the two are different. The former occupies itself with semiotic properties of concrete signs in order to gain information about the semiotic properties; the latter aims at gaining infor­ mation about those concrete signs. Semiotics5 treats the semiotic analysis of separate signs as a basis for generalizations concerning the semiotic properties of the sign in gen­ eral. SemioticsA treats the signs it analyzes from the point of view of semiotic properties as the only or the main object of its interest, as an aim and not as a means towards an aim, because it wants to find out what these particular signs mean, and not what meaning is in general. In view of the above, within semiotics5 the concrete signs we analyze from the point of view of semiotic properties do not constitute a uniform body. On the contrary, we try to gather material for analysis from various fields, and we study all kinds of signs, natural and also artificial languages, and, among the former, inflectional, isolating and agglutinative languages. Semiotic analyses of a picture are found side by side with specu­ lations on semiotic properties of a piece of music. Such a differentiated empirical mate­ rial is to protect semiotics5 from one-sided and premature conclusions about the nature of the separate semiotic properties, and to provide it with a sufficiently wide basis for generalizations.

912

Semiotics and Linguistics

In semioticsA, on the other hand, which also describes and analyzes concrete signs from the point of view of semiotics, these signs usually form some uniform whole such as : a literary work or a literary trend, Easter celebrations, the fashion of a certain period, place, and social circle, an architectural style, a romantic park. It is about this whole, about the Christmas Eve supper, that we want to know what is its "syntax" and what is the "meaning" of its elements and of combinations thereof. However, in spite of the fact that semioticsA aims at understanding the studied object (considered as its proper and adequate object of cognition), and not generalizations about the nature of meaning, the concept of truth, or the essence of intensionality, and despite the fact that within semioticsA we use intuitive notions of meaning, denotation and text, the description and analyses performed by applied semiotics, similarly to the descriptions and analyses of semiotic properties of the sign provided by semiotics5 , con­ tribute to the formation or modification of the concepts of these semiotics properties. Ideally, a theory should precede applications, and in the applications one should use terminological tools prepared earlier by theory. In practice, however, it is not always so. As we see in history, semioticsA often precedes semioticsT or appears simultaneously with it. This is both good and bad : bad because many misunderstandings and errors have arisen in semioticsA out of faultily conceived notions; good because there exists and is constantly expanded, an empirical basis of semiotic theory. As we see, there are various notions of semiotics. The word, it seems to us, is used in five senses, although we are not always fully aware of the fact. We have distinguished semioticsr and semiotics5 from semioticsMs> semioticsM> and semioticsA . Our use of letter symbols instead of full names is intentional, for it is not the names that matter but the distinction lying behind them. We realize that one often attaches greater importance to the name of something than to the nature of that something. Thus, should someone prefer calling semiotics5 descriptive, semioticsT general, and semioticsA special, one is free to follow one's preferences and to change the letter symbols suggested here (See :

Philosophy of Language.).

J.P.

Semiotics and Linguistics The intimacy of the relationship between semiotics and linguistics is accountable on the one hand to the role assigned to language within semiotics, and on the other hand to the role assigned to linguistics within language. Specifically, the arrangement of the argu­ ments of these "within" relations prefigures the controversial hierarchical entangle­ ments between and among semiotics, language, and linguistics.

Linguistics Recent linguistic history on semiotics has its roots in the semiology of Ferdinand de Saussure, whose widely read Course in General Linguistics (1959) dominates linguistic semiotics in the twentieth century. While Saussure's semiology would elevate linguistics to the position of "the master-pattern for all branches of semiology" (68 ) , it is obvious in the final quarter of the century that the Saussurean observation "Linguistics is only a part of the general science of semiology" (16) is proving to be more prophetic.

Semiotics and Linguistics

913

The Glory and the Burden In Peirce's taxonomy of the sciences (1931 :1.§ 183, 186, 189, 191, 200) linguistics is as­ signed to classificatory psychics, or ethology branching from the psychical sciences, which, in turn, branch from idioscopy. In contrast, Peirce derives semiotics from logic, a normative science branching from philosophy. Nonetheless, Peirce foreshadows the glory days of linguistics in his appraisal of "the vast and splendidly developed science of linguistics [among the] studies of mental performances and products" (1931 :1.§ 271 ). This value judgment on linguistics is echoed throughout the century by linguist and non-linguist alike. Leonard Bloomfield (1939 � :25 ) 1 states confidently that "Linguis­ tics is the chief contributor to semiotic," and with equal conviction Jean Piaget (1966:25) writes "La linguistique est sans doute la plus avancee des sciences sociales." The movement toward recognition that the "problem of interrelations between the sciences of man appears centered upon linguistics" (Jakobson 1970 � :26) culminates in Roland Barthes' ( 1 967 :11) inversion of Saussure's declaration [Saussure 1959:16 "Lin­ guistics is only a part of the general science of semiology"]: "linguistics is not a part of the general science of *signs, even a privileged part, it is semiology which is a part of lin­ guistics." In the early seventies Thomas A. Sebeok's Volume Twelve of the Current Trends in Linguistics series addresses "Linguistics and Adjacent Arts and Sciences." In this volume semiotics is a self-contained topic somewhat on the same level with such disciplines as philosophy; the verbal arts, subsuming poetics, folkloristics, rhetoric, stylistics, literary genres, and metrics ; artificial languages; translation; psychology; anthropology/sociol­ ogy; economics; education; bio-medicine; and computer science. This assemblage of lin­ guistics-related arts and sciences rests upon a widespread conviction that linguistics functions for them as a "pilot science" (Marcus 1974b:2871), "the primary science which serves as the unifying focus of the language sciences as a whole" (Garvin 1974 � :2890). With the maturation of the discipline of semiotics, however, the singular glory of linguistics begins to assume the characteristics of a burden, since the linguistic model, which is understood as a structuralist model (cf. below), is expected to account for "any study concerned with the production and perception of 'meaning'" (Pettit 1975 � : Preface) . By the late seventies then, that is, within half a decade, the linguistic model, the " 'archetype' . . . for semiological inquiry" (Pettit � 1 09) experiences a stark shift in its hierarchical position among related sciences. Sebeok's sobering characterization of linguistics as being "alias semiotcs" (1979b:66) signals the turning point.

Language Linguistics is singled out as the essential paradigm for semiological research because it claims language as its primary research object. Accordingly, the literature is rife with the indiscriminate synonymizing of linguistics with language. There is no doubt that with or without linguistics, language is a most powerful organ. At the same time, it remains a fact that linguistics is an extremely successful structural model.

References followed by this symbol ( t,. ) are listed in the Supplement to the Bibliography, pp. 449 ff.

914

Semiotics and Linguistics

The *Sign System par excellence Tzvetan Todorov (1975 � :97), echoing others such as Edward Sapir and Uriel Wein­ reich, declares language to be "the *sign-system par excellence." Certainly the amena­ bility of language, specifically the linguistic sign, to the scrutiny of structural principles supports the esteem for language as a sign system. In fact, it leads to an array of studies treating non-linguistic signs via language-analogical tools (cf. below). But language, for the moment now without reference to any one discipline, traditionally holds a superior position in writings on the sciences of man. Witness, for example, George Steiner's view (1978 � :208) : "The creative well-being of an organic system depends on intricate bal­ ance between stimulus and repose, between use and recuperation. This balance, in turn, derives from adjustments between inner and outer environment. Language constitutes both in the most immediate and dynamic sense. It is the pulse and skin of conscious be­ ing." In this context, Bloomfield's view (1939 � :55) that "In the cosmos, language pro­ duces human society, a structure more complex than the individual, related to him some­ what as the many-celled organism is related to the single cell," appears rather as an anomaly, since Bloomfield is customarily interpreted as antimentalistic. Finally, we see in Edmund Leach's statement (1976 � :88): "From some points of view spoken lan­ guage is not only a part of culture, but prototypical of all culture," the rapid accelera­ tion toward deification of a particular phenomenon, this time not the linguistic model, but language itself. Once more Sebeok intervenes in a seeming reversal of the trend with his comment "Genetic copying is the semiotic process par excellence " (l 979b :120). The reversal, however, is a mirage. Sebeok's biologically based semiotics is not entirely novel (cf. Baer 1 979 � :354), nor, of course, does he claim it to be. The Bloomfield quote (immediately above) can be said to prefigure it serendipitously. The Steiner � quote (also above) accords well with concepts embodied in von Uexktill's "Funktionskreise" (1 940 :9) which, thanks to Sebeok's efforts, represent the most current stream of semiotic research. The main cur­ rent in this stream is the catastrophe theory of Rene Thom. Most importantly for the present discussion, Thom's explanation of the rationale for the existence of language (1975:309-11): " l . For a personal evolutive constraint, aiming to realize the perma­ nence of the ego in a state of wakefulness; 2. For a social constraint, expressing the main regulating mechanisms of the social group," shows Sebeok's epitomical genetic copying on one plan to underlie another "paramount" (Sebeok 1979b :63) phenomenon, which is language, on another plane. The *Language Inlay The reputed role of language within semiotics is understandably richer than that of lin­ guistics within language (cf. above). This is eminently reasonable since "One can raise an enormous variety of questions about language. Language is used for a great diversity of purposes, it affects many different aspects of personal and interpersonal activities, it has its physical qualities in terms of sound patterns, it evokes reactions from people, etc. It is senseless to expect that there should be one unitary science [linguistics] that would answer all these questions." (Moravcsik 1975 � : l l ). Beginning with Saussure (as a point in recent history) we read that (1959 : 1 7) "the characteristic that distinguishes se­ miological systems from all other institutions shows up clearly only in language." Emile Benveniste (quoted in Todorov 1970 � :125) formulates a simple axiom, stating "The

Semiotics and Linguistics

915

configuration of language determines all semiotic systems," while Jakobson avidly ex­ ploits this axiom in his work; he is convinced that "any human communication of non­ verbal messages presupposes a circuit of verbal messages without a reverse implication" (1970 .t, :32). Again we witness a steam-roller effect when Claude Levi-Strauss (1963a:83) postulates that "language comes into play" in the exchange of messages, of goods, and of mates-the linchpins of his interpretation of society within the science of communication. One of the most current and explicit reiterations of the Benveniste axi­ om is found in the thinking of Sebastian Shaumyan, who attributes the ability of natural language to translate a message from any '�sign system into itself to the "cognitive flexi­ bility" of natural language (1976 .t, :60). And finally, we may note that it does not ap­ pear accidental that on occasion Francis Whitfield, in translating Hjelmslev's Prolegome­ na to a Theory of Language, employs English 'semiotic' as a gloss for Danish sprog. Hjelmslev, after all, forms another link in that chain of thought which holds that "a lan­ guage is a semiotic into which all other semiotics may be translated-both all other lan­ guages and all other conceivable semiotic structures" (1963a:109). It seems crucial that, in the maze of literature dealing with language as underlying semiotics, at the very least a clear distinction ought to be drawn between ,:•metalanguage, sometimes called metasemiotic, and �·language infrastructure or inlay. Sebeok, quite rightly, bursts yet another inflated concept when he questions the "asymmetrical con­ vertibility of codes" (1979b :256 ), that is, the translatability of any semiotic message into language but not necessarily the reverse. Certainly language as a representative or surro­ gate for another ,:•semiotic system ('�metalanguage) cannot modify the nature of that sys­ tem. Accordingly, the translatability problem is trivial. The case is otherwise, however, if features or characteristics of language are integral to another �·semiotic system (lan­ guage inlay). Here, we are dealing with the Soviet semiotists' concept of secondary ,:•mo­ deling systems, that is, systems based on the so-called primary system, which is lan­ guage. In fact the Soviet School (Lotman et al. 1975d :76) defines culture as a system of secondary ,:•modeling systems and Lotman considers it to be more complex than natural language, or, for that matter, biological systems (Rewar 1979:282, 276). As always in discussions of crossdisciplinary systems, a healthy skepticism for the value of, indeed the justification for, positing degrees of complexity among the systems is in order. There is, however, no disputing the fact that the problem of the language inlay in semiotic modali­ ties resides ultimately in the nature of language. Defining Language "If we are to discover the true nature of language we must learn what it has in common with all other semiological systems." This dictum of Saussure (1959:17) represents the Sisyphean task confronting linguist as well as semiotist. Among the most common pa­ rameters characterizing language are the features verbal versus nonverbal and human versus nonhuman. Adam Schaff defines verbal as follows (1970 l> :114): "By verbal signs I mean the proper parts of the sound language used by men for the purpose of com­ munication (written language being treated merely as a peculiar transformation of sound language)." While Schaff's definition is incisive, it is debatable since "language al­ so contains non-verbal components, such as the prosodic features of intonation and stress, which serve to distinguish meaning and should thus be counted as 'verbal sig­ nals"' (Schmitter 1 979 .t, : 1 46). Prosodic features are frequently relegated to ,:•paralan-

916

Semiotics and Linguistics

guage, the delimitation of which is obviously as uncertain as that of language itself (Rauch 1980 /:::,. ). The expanding bounds of language are implied in Erving Coffman's ob­ servation (1979 :6) that "the terms 'speaker' and 'hearer' imply that sound alone is at issue, when, in fact, it is obvious that sight is organizationally very significant too, some­ times even touch." These bounds become virtually unlimited in the provocative state­ ment of Madeleine Mathiot (1 979 /:::,. ) that "Discourse is not a part of language." Schaff's definition indicates that language is a human property. Although there is an increasing tendency to expand the application of the term language, the tendency is not necessarily accompanied by an expansion of the traditional concept of language very recently reiterated by Ronald Wardhaugh (1977 /:::,. :3): "Linguists are in broad agree­ ment about some of the important characteristics of human language, and one definition of language widely associated with linguistics may illustrate areas of agreement. This par­ ticular definition states that language is a system of arbitrary vocal symbols used for human communication." Thus Noam Chomsky (1979 :32), for example, is willing to extend the application of the term language beyond the human species, "If by language we mean symbolic system, or system of •:•communication." Nevertheless, Chomsky holds firmly that "human language . . . is outside of the capacities of other species, in its most rudimentary properties. The differences appear to be qualitative; not a matter of 'more or less' " (43). Contrariwise, William Orr Dingwall writes (1980:78) "the determination whether one communicates in a human manner cannot be made in absolute terms; rather some gra­ dient measure must be employed." Even though it is obvious that linguists are, in fact, actively divided on a definition of language, the question of what in the nature of language is semiotic, namely, Saus­ sure's dictum above, is rarely pursued directly. Jakobson's insight that (1971b /:::,. :703) "The cardinal functions of language-referential, emotive, conative, phatic, poetic, me­ talingual-and their different hierarchy in the diverse types of messages have been out­ lined and repeatedly discussed. This pragmatic approach to language must lead mutatis mutandis to an analogous study of the other semiotic systems" serves as a lonely beacon and at that a weak one, since the mutatis mutandis remains unspecified. More fruitful is his conviction that (197 1 b t::,. :708) "only natural or formalized languages are able to generate . . . judgments, general and especially equational propositions," for it can be challenged as being human-language specific. Frequently intention is entailed in the pro­ duction of a proposition. John Lyons (1977:32-3), who considers language as but one in­ strument of communication, requires of all communication that it be intentional. Even this requirement can be circumvented for nonhuman communication since, according to Jonathan Bennett (1976:205) "intention is significantly like biological function." In­ deed, Peirce determined a syllogistic effect in the pinching of the hind legs of a decapitat­ ed frog, namely "the way a decapitated frog reasons when you pinch his hind legs" (Fann 1970 /:::,. :28). Accordingly, "the illative relation is the primary and paramount '�se­ miotic relation" (Peirce 1932 :2.§ 441n.1). If language is posited as an inlay in all modali­ ties of communication, whether human or nonhuman, verbal or *nonverbal, it is due to acceptance of the exceptionally broad �·definition that "Language is signifying through an illative-type process" (Rauch 1979 /:::,. :7 4).

Semiotics and Linguistics

917

Linguistic Semiotics •:•Structuralism, whether in linguistics, biology, genetics, physics, chemistry, geology, or sociology, holds that being resides in relationship, that a structure or system consists of elements having certain mutual relations rather than of a mere accumulation of inde­ pendent items. Before Saussure, Jakob Grimm systematized the evidence for the First or Germanic Sound Shift through the use of the structural principle. Pai:i.ini (fourth cen­ tury B.C.) was a structuralist and Chomsky is one as well. It is, however, to Saussure and the Prague School of linguistics that credit is given for formulating the consistent theory of *structuralism which dominates modern linguistic method. During the first third of this century, the Prague School, with its strong concentration on phonology, was able to apply structural strategies comparatively easily, in particular feature analysis. In the second third of the century, the emulation of the successful application of structural principles to nonphonological data was, however, not as facile, since semantic data, which become primary, are not automatically amenable to the analytical tools of pho­ nology. Linguistics' concentration on semantics has led into another member of the semi­ otic triad, *pragmatics. Pragmalinguistics, as well as attempts to extend the linguistic pa­ radigm to *nonverbal data, encourage the refinement of the *structural model through *action theories in the final third of the twentieth century. Feature Analysis Saussure ( 1 959: 1 20) considers a linguistic system to be "a series of differences of sound combined with a series of differences of ideas." The differences are perceived by Saus­ sure and the Prague Phonologists as diadic oppositions. In fact, Saussure's linguistic model rests largely on binary conceptual characteristics such as signifier : signified, langue : parole, paradigm : syntagm, matter : form, sound : meaning, diachrony : syn­ chrony. This pervasive habit of the linguist of bifurcating his cosmos continues to per­ meate semiotic method. Consider, for example, Sebeok's characterization of the zoo­ semiotic features of human communication in terms of binary components ( 1977b ), or Lotman's discussion of a dynamic model of a semiotic system by means of the antinomic features systematic : extrasystematic, monosemic : ambivalent, nucleus : periphery, de­ scription : nondescription, necessary : superfluous ( 1 977b). It is quite another challenge, however, to define and isolate feature which are min­ imal and distinctive internal to a •:•semiotic system. The linguistic method employed in ferreting out features is *commutation (Hjelmslev 1 963a:73), whereby the substitution of one feature by another in a given context may effect a difference in meaning. In anal­ ogy to phonemic structure, semiotic method seeks to identify the mytheme, filmeme, danseme, vesteme (Greimas 1 97 1 ), proxeme (Hall 1 963), kineme (Birdwhistell 1 970), among other emes, but such data from semiotic modalities other than language tend to defy their definitive reduction to discrete elements consisting of distinctive features (cf. Jakobson 1971a ..6.. : 105-6 ; Lange-Seidl 1 977: 1 5- 1 9). The �·Language Act As diadic as feature analysis in the linguistic framework may seem to be when used in an act of signification or semiosis, a feature necessarily participates in a triadic act. The •:•language act itself consists of an organism (a human being) producing an expression

918

Semiotics and Linguistics

(a sound, orthographic sequence, gesture) in order to refer by it to something (an ob­ ject) (Carnap 1 961 : 8-9). Thus, the �· language act, comprising a speaker, an expression, and a designatum correlates directly to Charles Morris' (1 955b :217-19) tripartite div­ ision of semiotics, namely, �·pragmatics, �·syntactics, and �·semantics, respectively. In turn, the members of these two sets correlate further to the �-interpretant, �-sign, and *ob­ ject integral to Peirce's definition of a ,:•sign (1931 : 1.§ 339) : "A ''sign stands for some­ thing to the idea which it produces or modifies." Of the three divisions of semiotics, �·semantics "occupies a central place" (Shaumyan 1 970 .6. :244 ). That signifying essentially entails meaning is confirmed by Rulon Wells' observation that "Semiotic has two groups of affinities . . . communication, and . . . meaning" (1971 .6. :95), and it is again reflected in Paul Ricoeur's statement that "meaning is the noematic correlate (the 'what') of all that which deserves to be called experience (1979 .6. : 1 71 -2). However, as Jerzy Pelc (1970 .6. : 1 09) explains : "relativiza­ tion only with regard to meaning, which may sometimes be adequate in pure ,:,emantics and also in descriptive semiotics of artificial �-language, gives a static image when we confine ourselves to it in the analysis of natural language : it discloses the expressions of that language as isolated lexical items and fails to disclose the actual functioning of words, phrases and sentences in language context and extralinguistic situation." Linguistic method, which in the first half of the twentieth century is strongly influenced by the Pra­ gue and the subsequent Bloomfield phoneme construct, turns in rapid succession by the second half of the century to the development of ''syntactic theory (1957, Noam Choms­ ky's Syntactic Structures), then to ''semantic theory (1963, Jerrold J. Katz and Jerry A. Fodor's "The Structure of a Semantic Theory"), and finally to ,:-pragmatic theory (1 969, John Searle's Speech Acts). The way is thus open for a linguistic return to the text. �Text and �-Action Theory Language is "prototypical of all culture" (cf. above, Leach 1 976 .6. : 8 8 ) and "culture consists of a plethora of texts" (Silverman 1979 .6. :258). ''Text in the broad sense, ac­ cordingly, equates with culture, with language, with man himself (Peirce 1 960 :5.§ 3 1 4 wrote "the word or �· sign which man uses i s the man himself"). In a more narrow sense ,:•text refers to the "abstract theoretical construct underlying what is usually called dis­ course " (van Dijk 1 977a :3). Nevertheless, Teun van Dijk's delimitation is tenuous in view of the fact that "discourse is systematically related to communicative action" (Joe. cit.), and accordingly recycles into cultural behavior (cf. above, Mathiot 1 979 .6. : "Dis­ course is not a part of language"). The transdisciplinary status then that �·text enjoys calls for a transdisciplinary model, which is currently *action theory. Linguistic ,:•pragmatics is exploiting an •:•action theory of its own with a basis in •:•speech act theory. Special focus is placed on the category nar­ rative because of its generel nature. Thus A. J. Greimas writes (1 977:23) : "narrative structures can be identified outside of the manifestations of meaning that occur in the natural languages: in the languages of cinema and of dream, in figurative painting, etc." In fact, Greimas' narrative appears to parallel van Dijk's discourse which contains such universal features as propositional macrostructures, modalities, and inferential struc­ ture (cf. Rauch, 1 98 1 .6. ). Not surprisingly then, "conventional organizations of dis­ course . . . [parallel those] of the corresponding global speech acts, and of action in gen­ eral" (van Dijk 1977a:245-6 ).

Semiotics and Linguistics

919

Semiotic Linguistics Given the trend in development of linguistic method in the twentieth century, it can hardly be unexpected that the awakening of a semiotic linguistics has been delayed until the final quarter of this century. Traditions of influences such as the Prague School, which contribute so much to the refinement of feature analysis, at the same time have stunted the Saussurean discernment that all language is a semiotic, the existence of which depends on the interrelationships of sound with meaning. In other words, a lin­ guistic form of any sort without a meaning correlation is paramount to non-existence. The fuller acceptance of Saussure's semiology is accompanied by the incorporation into linguistic method of the Peircean division of ,:•signs and his phenomenological cate­ gories. Peirce's theory of 'cabduction is found to account for ,:•language acquisition strategies and �·language change, that is, the creative aspect of language. ,:•Icon, ''Index, ,:-symbol Linguistic iconicity, whether in speech, writing, or sign-language gives a deceptively simple impression when viewed in isolated data, for example onomatopoeic images as English click, clack, cluck (Wescott 197 1 ). However, diagrammatic schemata involving entire systems of a language are extremely intricate and sophisticated, for example the analysis of the morphology of the Russian verb in terms of the fit between the phonolog­ ical representation and the meaning of the morphemes (Shapiro 1 969 t:,_ , 1980 t:,_ ). The main principle at work in determining the fit is that of markedness, whereby feature values of an opposition in one component of the grammar correlate with feature values of an opposition in another component of the grammar. Whereas the morphology and the syntax of a grammar are strongly diagrammatic, the lexicon is predominantly symbolic (cf. Jakobson 1965 t:,. , Valesio 1969). Other than iconic elements in the vocabulary, indexical relations compose deictic words such as pronouns and morphemes of time and place ; the remaining symbolic component then comprises the largest mass of dictionary items in a language. A language can thus be called arbitrary only in certain parts of its lexicon. Saussure's oft quoted statement (1959:67) that "the linguistic ,:•sign is arbitrary" has been misinterpreted and applied with­ out restriction to the whole of language structure. As Saussure himself cautions (1 33) : "Everything that relates to language as a system must, I am convinced, be approached from this viewpoint, which has scarcely received the attention of linguists: the limiting of arbitrariness." ,:•Analogy, ,:•Abduction, Growth One of language's most powerful strategies, if not its single most effective universal ma­ neuver, is ,:•analogy. In its multifaceted functionings ,:•analogy displays a strongly iconic character, but also an indexical one. A basic mechanism of ,:•analogy is well expressed in the "iconic principle whose ideal is 'one meaning, one form'" (Anttila 1972 !::,. : 1 00), which clearly checks arbitrariness. The indiscriminate use of ,:•analogy in linguistic explanation, however, is curtailed by requiring the identification of discrete relational strategies operative within the ana­ logical processes. Thus, for example, if Secondness is determined as the main function in a given linguistic ,:•analogy, contrary to an ,:•analogy based on Firstness, the highly

920

Sender

valued predictability factor in linguistics as a science is strengthened (cf. Rauch 1978 t:,. and 1982 t:,. ). This is not to say that ,;analogy is necessarily less widespread than former­ ly thought; it does say that analogic explanations, in order to be valid, must be held more accountable. Indeed, *analogy's universality stems from the fact that the analogi­ cal process is an �-abduction, and hence it results from those laws of inference which are responsible for language creation or growth. Growth, comprising �-language change and acquisition, proceeds by a model "which recognizes, on the one hand, that the verbal output of any speaker is determined by the grammar he has internalized, and on the other, that any speaker's internalized grammar is determined by the verbal output from which it has been inferred" abductively, but also inductively, and deductively (Andersen 1973 t:,. :767; cf. also Savan 1980b ). Ultimately, it is the teleology of *language change, of language acquisition, of why "symbols grow" (Peirce 1 932 :2.§ 302), which semiotic method imparts to linguistics as a haunting legacy and challenge for the twenty-first century. LR.

Sender The agent responsible for an act of ''communication. As individual and empirical subject, external to the process of •·semiosis proper, the sender is generally excluded from the province of ,;semiotics theory and analysis (see Eco 1976c). As a factor evoked by or immanent within the actual *message or �-text produced, however, the category of the communicating agent has been amply restored particularly with reference to linguistic communication. •·Buhler (1934) proposes as one of the three major communicative ,;functions of speech that of �·expression (Ausdruck), namely, the orientation of discourse toward the sender himself (specifically, toward the speaker), in manifesting his emotional, psycho­ logical, or moral attitudes by means of the particular •·intonation patterns and linguistic modalities adopted. The fully emancipated semiotic status of Btihler's "expression" the constituents of which are governed by the rules of •·language independently of single senders and their particular attitudes - is further emphasized by Jakobson (1960), who rebaptizes the sender-oriented role of language the "emotive" function, identifying as its purest manifestation in discourse the interjection. The communicative range of the sender is extended by Wunderlich (1971b) to include the "representative" (or informa­ tive) function, the "provocative" function, and the "evaluative" function. An analogous restriction of the semiotic role of the sender to intradiscursive elements is found in ,;Benveniste's treatment of subjectivity in language (1958). Indeed, the very capacity of the sender to constitute himself as speaking subject or ego is essentially bound to the presence in language of deictics or shifters, in particular first-person personal and possessive pronouns and adjectives and "nominative" verb inflections, whose •·referen­ tial force depends on the presence of the sender at the moment of the •:•utterance. The indexical self-constitution by the sender in the very act of communication may also be realized or reinforced by paralinguistic or kinesic means : through, for example, what Birdwhistell (1970) terms "proximal" kinesic markers, whereby the communi­ cating subject indicates himself as such gesturally. Indices of the sender's self-manifestation with the message become institutionally ambiguous in the case of �-literary texts, and a great deal of attention has been devoted,

Sentence

921

in particular, to the status of the speaking or writing subject in narrative. In the act of narration, up to four more or less distinct agential components may be discernible: (1) the so-called (and extratextual) real '�author, original sender of the text; (2) the author's "voice," manifested as intrusive textual presence in direct first-person commentary; (3) the fictional narrator, to whom the "I" of the narration proper purportedly refers; (4) a virtual or ideal figure, inferred by the ,:•reader from the norms apparently expressed in the narrative - and often differing from the values expounded by the narrator - bap­ tized by West (1957) as the "official scribe" and by Booth (1961 b) as the "implied author." A second form of agential "doubling" in communication - again primarily literary - regards instances of what Lotman (1973e) describes as the "I-I" model : cases, that is, where the sender is at the same time the self-elected receiver of the message (the reading by an author of his own text when published; annotations made by a diarist by way of clarification for his own benefit; cryptographic marginal notes in a manuscript, etc.; see also Corti 1978 ). K.E.

Sentence A sentence as distinguished from a ,:- proposition belongs to an uninterpreted calculus, while the latter is endowed with a ,:•semantic interpretation. A semantic system is con­ structed (Carnap 1961) in two main steps: first, the syntactical classification of �-signs and the '�rules of formation are given; second, the semantical rules of designation and of truth are laid down. A sentence in a semantic system is defined by the syntactical rules of formation which determine admitted forms of sentences; such a definition is formal in the sense that it makes no reference to the designata of expressions but to their form or shape only. U.N. Sentence Logic Sentence logic (propositional logic, �-truth-function logic) is a theory of those inferences whose logical form is determined by certain logical connectives called truth-functors. This theory, as presupposed by the other logical theories, is the preliminary part of modern formal logic. Connectives of sentence logic are called truth-functors, since they are defined in terms of the so-called truth-functions, i.e., functions that take either truth or falsity as their values; these are called truth-values, and are denoted by the symbols "1 and O" respectively. Let a one-argument function called negation and a two-argument function called conjunction be taken as examples of truth-functions. The former is symbolized by ~ p and read "it is not the case that p," the latter is symbolized by p & q and read "p and q." The letters p, q represent any sentence whatever; such symbols are called sentential (propositional) variables. The negation ~ p is true if and only if its argument p is false, and it is false if and only if its argument is true. Conjunction is true if and only if both its arguments p, q are true; otherwise it is false.

922

Sentence

All the other connectives can be defined in terms of negation and conjunction. Those which most often occur in inferences are as follows: disjunction p v q (either p or q, or both), implication p ::) q (if p, then q), equivalence or biconditional p q (p if and only if q, for short, p iff q). Their respective defintions in terms of conjunctions and negation are as follows:

=

p V q = df ~ ( ~ p & ~ q) p ::) q = df ~ (p & ~ q) p = q = df (p :) q) & (q :) p) It can be read from the above defintions that: p v q is true iff at least one of the arguments is true; p ::) q is false iff p is true and q is false, otherwise the implication is true; p = q is true iff each of the arguments implies the other, hence either both are true or both are false. It has been proved that not only the truth-functors mentioned above but also all the other truth-functors can be defined in terms of " ~ " and "& "; owing to this property, the set { ~ , &} is said to be functionally complete. Other functionally complete sets of truth-functional connectives are { ~ ,v} and { ~ , :) }. Moreover, there are two connectives such that each of them suffices to define all the remaining ones ; these are: p, q (neither p nor q), called joint denial; and pl q (for which no reading exsists in English), called alternative denial. For example, ~ p can be defined in terms of the latter as pl p, and p ::) q as pl(ql q). The number of all possible one-argument truth-functional connectives is 22, and the number of two-argument truth-functional connectives is 22 ; however, some of them are not used, as being less convenient than those discussed above. A formula of sentence logic is said to be a tautology if it is true under all possible assignments of truth-values to variables occurring in it. There are a number of methods of deciding whether a formula is a tautology or not. The most usual is the truth-table method. Take truth-tables, listing all the possible value assignments, for negation, conjunc­ tion, disjunction, implication, and equivalence. p ~p

1 0 0 1

p

1 1 0 0

q

p & q p v q p:::,q

1

1 1 0

0 1 0

1 0 0 0

1

1

0 1 1

P = ~ 1) (~ l vo) = (~ 0 => ~ 1) (0 v 1) = (0 ::J 0 ) (0 v0) = (1 ::J 0 ) 0

(~ Ovl) :: (~ l v~ 0) (1 vl) = (0 v 1 )

923

(~ 0v ~ 0) :: (~ 0 => ~Q) (1 V 1 ) ) :: (1 ::J 1 )

0

The formula in question proves to be true under each assignment, hence it is a tautology. The truth-table method can also be used to prove or to disprove the validity of an inference by transforming this inference into the implication in which the conjunction of the premises forms the antecedent, while the conclusion is identical with the conse­ quent. If this implication is a tautology, then the inference represented by it, as de­ scribed above, is valid (correct). Sentence logic can be constructed as a deductive system. This means that appropriate axioms and rules of inference are stated, so that every tautology can be proved in the system; hence sentence logic is semantically complete. It is consistent as well: it has been demonstrated for every system of logic that only those formulas which are tautologies are provable in the system in question; and in a set of true sentences, as are tautologies, none can contradict any other. Here is an example of the deductive system of sentence logic which is a classical one, as it was stated in Prinicpia Mathematica (a fundamental work on logic) and improved in Hilbert and Ackermann 1928 (another fundamental work) . The primitive, i.e., undefined, expressions of the system are " ~ " and "v." The set of well-formed formulas is defined by the following formation rules: (1) all sentential variables are well-formed formulae; and (2) if A and B are formulae, so are ~ A and

A v B.

Rules of inference are as follows: Rule of detachment ( modus ponens rule): from the formulae A and A :::) B infer B. Rule of substitution: from a formula A in which occurs once or more times a sentential variable v infer what results from substituting any formula B for v throughout A. Rule of replacement: for any formulae A, B, C, D: if D is a part of A, and C= dfD, then from A infer Bwhich is like A except that C has replaced D in one or more places. Let C from the above rule be p :::) q, and D be ~ p v q. Let us make use of this replace­ ment when stating the axioms of our system which should be stated in terms of negation and disjunction as primitive terms; however, appropriate replacements by means of the symbol of implication will make the axioms more readable, e.g., instead of ~ (p vq)vp we put: A1 A2 A3 A4

(p vp) :) p. The other axioms are as follows : (p:) (p vq) (p vq) :::) (q vp) (p :::) q) :) ((r vp) :) (r vq) ) .

This set of axioms, together with the rules o f inference, i s sufficient to prove all the theorems of sentence logic. W.M.

92 4

Settlement Space

Settlement Space The semiotic study of settlement space (urban space when the settlement is characterized as a city) belongs to the domain of the * semiotics of space and is closely linked to the areas of the semiotics of '�rchitecture and the semiotics of regional (in the wide sense) space, which belong to the same domain. The semiotics of space shows a considerable over­ lapping with the semiotics of ,:•objects (of the environment - Baudrillard 1972 :7-58), and is in principle part of the (general) topological semiotics of Greimas (1974) or of the proxemics (in the wide sense) of Hall (1971) ; it is also related to * visual semiotics, and may be correlated to the semiotics of the natural world of Greimas (see 1 970: 49-56). Authors agree in defining the object of study of the semiotics of settlement space as the analysis of the settlement environment as a system of signification. We need to distinguish clearly between the settlement environment in its semiotic aspect and the same environment in its non-semiotic aspect. More specifically, settlement space is primarily a social system of ''use which is also a system of signification (Bok­ lund-Lagopoulou and Lagopoulos 1980 ; cf. Eco 1972b:25-30) - which incorporates a system of feeling (cf. •·Piaget 1970b). It is thus in an important respect different from natural ''language, which is a system of a signification - and the basic system of *com­ munication - used within the framework of social relations. The semiotics of settlement space - like every semiotic phenomenon - is inextricably linked to the conceptual sphere of a social subject. It follows that spatial semi­ otic ''structures are characterized by cultural relativity and related to ''ideology (in the wide sense). Ideology is the ,,;ediator between the "objective" social structure and the semiotic spatial structure (see also Lefebvre 1968:54-55, 62-63, 71-75 ; Castells 1972b; cf. Piaget 1970b:58-60 ; Bourdieu 1971; Goldmann 1971). The form of the •·content of semiotic systems is codified ideology. It consists of two levels, that of connotation and that of ''denotation, and includes a set of signifieds, sememes (cultural units, according to Eco 1972b :28-29, 63-65, 1979d:66-68). The connotative level is the ideological level par excellence (cf. •:•Barthes 1964b :131) and includes the (cultural) value system. The general framework of signification •·sender production •·text and text con­ sumption receiver is of course valid for the semiotics of settlement space ; within this framework one can also situate the model of communication, sender -. ''message -. receiver. However, we need to clarify the following points : (1) The *semantic aspect is not the dominant aspect of the system (see above), though it may be culturally felt as such. (2) The sender is not necessarily an individual or •·synchronic subject, and does not necessarily have the intention to communicate. (3) The receiver is a collective synchronic or diachronic subject, either known or unknown to the sender. (4) For these reasons, the content of the message may differ to a greater or lesser extent for sender and for receiver, more pronouncedly, on the whole, than is the case in linguistics. (5) Sender and receiver are psychologically, but also (and primarily) socially defined. (6) The sender produces a message characterized by spatiality - not temporality as in natural language - while the receiver reads the message mainly temporally which does not necessarily imply that the message is temporally organized.

Settlement Space

925

A first approach to the semiotics of settlement space is the analysis of the semiotic production of space: (*sender --+ ) �·message. A second approach is the analysis of the semiotic consumption or conception or reading of space (the "mental image"): message

(--+ receiver). Two other, parallel approaches are the analysis of space from the point of view of *pragmatics or a broader one, which can be concerned with either production or consumption. The first two approaches maintain the principle of pertinence, excluding the study of the substance of the *expression. But though the substance of the expression is not included in formal semiotic analysis, it is useful to be aware of it, because it places con­ straints on the process of ,:•semiosis. The last two approaches mentioned above are concerned with the message within the conditions of communication. Certain authors consider these conditions as having only semiotic aspects, while others see them as also having non-semiotic aspects (see Martinet 1975:43-48; Eco 1972b:56-57). In the latter case, the analysis may obviously be limited mainly to the psychologically pertinent aspect of the situation, in which case we have to do basically with a psychosemiotics, or it may extend to the whole of the situa­ tion, including the underlying social structure, in which case we have basically a sociose­ miotics. The latter permits an essential and final explanation of the message, as opposed to its description. The semiotic analysis of settlement space is based on oral and written discourses, and, secondarily, on visual systems (e.g., drawings, �·painting). The discourses include semiotic data referring to the settlement as a set of objects and to its relationships to the five senses, primarily sight. The semiotics of settlement space is one of the new areas of semiotics, and because of this there is not a large bibliography in the area; it is characterized by a lack of inter­ influences and criticism among the various approaches, a lack of a (relatively) general theory, and a vagueness concerning certain basic questions. Thus our presentation of the area is organized mainly around the approaches of individual authors and not around some more general "schools." In addition, for reasons of clarity, the presenta­ tion is organized according to the two groups of approaches which refer to the produc­ tion and the conception of space, although the material of each group does not always belong exclusively there, since some authors do not distinguish the two or treat the sub­ ject globally.

The Semiotic Production of Settlement Space Approaches The most comprehensive theoretical approach to the semiotic production of settlement space is that of Greimas (1974). He starts from the opposition 'extension' versus 'space,' which he equates with the opposition 'substance of the expression' versus 'form.' Signi­ fying space emerges from extension and is constituted as a 'place' (lieu, topos), which is a text. One of the two approaches of urban semiotics, a provisional branch of topological semiotics, is the analysis of the process of the production of space, which analyzes the sig­ nification, for the subject, of 'doing' (faire), and of which the object is place as enuncia­ tion. The sender of the process of communication is the producer of space, the 'enuncia-

926

Settlement Space

tor,' who possesses the �·competence of the *modalities of 'being able' (pouvoir) , 'know­ ing' (savoir), and 'wanting' ( vouloir) 'to do' (faire). The enunciator is a collective actant of whom the urban planner - when there is one - is an actor. The functioning of the actant brings about the mechanisms of decision-making and the mixture of different cul­ turally defined values, from which is constituted the ideological model of space. The semi­ otic process of spatial production starts from this deep-level semantic model which is sit­ uated on the level of the form of the content of the spatial language; corresponding to this level is the (relatively) autonomous level of the form of the expression ('phonologi­ cal' level), the level of the spatial forms, which constitute surface-structure realizations. The analysis of spatial production is done with the help of a generative '=grammar, while the analysis of the 'doing' of the actant leads into the field of sociosemiotics (in the narrow sense). Because the substance of the spatial expression is long-lasting, several grammars coexist simultaneously. The question posed by Greimas' interesting approach concerns the validity of the transfer of his narrative model to the semiotics of space. The macro-mechanisms of the production of architectural space are the concern of Agrest and Gandelsonas ( 1 973; Agrest 1 977). They believe that the theoretical object of architectural - and by extension of urbanological - analysis is a configuration-de­ limitation system (C-D), which includes architectural objects, related verbal texts, and drawings. The C-D system is connected with external semiotic systems; it includes cer­ tain characteristics determining its specificity, and also characteristics shared with other semiotic systems. Its boundary is subject to opening and closure, controlled either by the specificity of the C-D system or by external factors. The opening and closure of the boundary creates an ideological filtering and is due to metaphorical or metonymical ,:•op­ erations. Through ideology the C-D system is related to the (social) mode of produc­ tion. On the subject of this latter process there are certain studies by Lagopoulos, following structuralist-marxist lines and of a sociosemiotic (in the wide sense) nature (1 978, 1980). The organized or organized-and-constructed actual space of a settlement is seen as part of the superstructure of a social formation. This superstructure also includes ideology; both space and ideology are fundamentally determined by the base of the social formation. The production of space includes non-semiotic processes and one semiotic process, rooted in the codified ideology as a set of connotative *codes (see also Eco 1972b:259-31 7). Lagopoulos observes that there is a correlation between modes of production and the complexes of connotative codes of settlement space, and that the dominant connotative codes derive from the dominant level of the social formation. A strong identification of the settlement with the world appears in the framework of modes of production where religion is part of the dominant structure. This identifica­ tion begins to give way to *analogy with the appearance of the political level as main dominant, and even analogy disappears with the dominance of the economic level. On the other hand, he remarks, the correlation of the geometric forms of the models of set­ tlement space with modes of production shows no progressive line of development, with the exception of the industrial mode of production, and certain ideological, semiotic, and geometric complexes pass from one mode of production to another. According to Eco ( 1 972b :259-31 7), space - the level of the form of the �·expression, which for him is equivalent to the whole of architecture and urbanology - is subject to a binding typology of messages. The producer of space possesses this traditional typology but lacks a grammar; the grammatical rules are imposed on space from out-

Settlement Space

927

side. Eco considers this fact a characteristic of space which differentiates it, in practice, from natural languages ; however, to the extent that he is right, it seems rather to diffe­ rentiate natural languages from all non-artificial semiotic systems. Choay, who also disagrees with Eco on this point, has drawn up a typology of the discourses concerned with the production of settlement space (Choay 1973, 1 974). She distinguishes between these discourses and "commentary" discourses which refer to already constructed space and also include the discourses referring to the conception of space. The discourses of production belong to two categories. The first category is that of "connotatory" discourses, which are concealed - as Lagopoulos has shown behind the space of settlements of pre-Renaissance societies, a space which is not directly significant, as Levi-Strauss believes. This space is iconically related to the form of the content, and engages the whole of society and all the senses of the individual. The second category comprises "institutive" discourses, specific discourses on space which is not (in principle) iconically related to the form of the content - characteristic of the last five centuries of Western history and with their origins in classical antiquity. These discourses have resulted in space being read differently by different social groups. The Form of the Content In our presentation of the views concerning the form of the content seen from the point of view of the semiotic production of settlement space, we will proceed from the more general to the more particular. Studies by Pinxten (1974, 1976b, 1977) are concerned with the analysis of denotative spatial �·semantic characteristics, �·universal and �·culture-specific. Pinxten, who follows a psychological and not (only) a linguistic approach, attempts to identify universal semantic characteristics and in particular "epistemic universals," semantic characteris­ tics linked to man's interaction with his environment and belonging to his "*cognitive map" or "natural knowledge system" (semantic level). He uses as base the three main conceptual (not perceptual) spatial systems of *Piaget, topological, projective, and metric (Euclidean) space, of which topological space is most fundamental. Each succes­ sive system incorporates elements of the former system or systems together with new ele­ ments (see also Piaget 1972b :9-10). Pinxten shows that there are universal spatial charac­ teristics and that they are identical to all of the topological and certain of the projective concepts of Piaget's systems. Epistemologically, these universal characteristics are, for the analyst, metalinguistic in nature. They transpose culturally recognizable significa­ tions belonging to a semantic (or, more narrowly, "epistemic") deep •:structure, the cultural manifestations of which are invested with connotations. The various cultural spatial systems are layered, and contain basic concepts (without spatial constituents) and non-basic concepts. From the discussion by Eco (1972b:259-317), who is in principle concerned with architecture and with its conception, it follows that the form of the content in the pro­ duction of the architectural and the urbanological system is composed of two levels: the first level consists of a system of values, among which are the functional requirements (that is, the ideology of the functions), and the second comprises a system of functions which fulfill and express these requirements. The first level is constituted by the (symbolic) connotations of the system, which refer to cultural "utilities," that is spatial '�functions - whence the term second functions - as do the per se functions of the

928

Settlement Space

second level, the denotations of space - first functions (see also Agrest and Gandelsonas 1 973: 257-59, 264-65; cf. Barthes 1 9646: 1 06 ). The per se functions, in the process of the conception of architectural space, are bio-physiological (cf. the "�·utterances of doing" in Greimas; see below), that is, the behaviors called for during the carrying out of a function, and const;uctional in nature (cf. the "utterances of state" in Greimas). Thus, the denotation of architectural space is not its spatiality or its aesthetic significance (see also Eco 1 9726: 1 00, 1 05, 1 1 1- 1 3 ). Eco's two types of func­ tions seem less adequate to the form of the content in the production of settlement space, which, for example, does not seem to refer to micro-behaviors. The connotative level of the production (and the conception) of space in precapitalist societies (a production that corresponds to Choay's "connotatory" discourses), has been studied by Lagopoulos (1972, 1 973: 9 1 - 1 09, 1 975, 1977: 58-67, 1 978, 1 980). He observes that these societies share a common ideological view: that the house, the settle­ ment, and the region are identified with and analogous to the cosmos, and that the model of settlement space is constructed based on this view. Due to this model, both the form and the substance of the expression in the spatial semiotic system are in an ,:. iconic relationship to the general spatial organization of the world, as it is conceived by the society, which is a part of the connotative level of the system. The cosmic organization follows logico-algebraic rules based on relations derived from the structure + IOI- or + I-. It forms part of a '�model for the settlement space which has the characteristics of a structure as defined by Piaget. The iconic representations of the cosmic organization on the level of the form of the expression belong to an extremely limited domain of twodi­ mensional geometric forms, with two main categories, a "centric" and an "orthogo­ nal," which appear to have their origins in (non-semiotic) socioeconomic conditions. Choay (1965, 1 973, 1 974, 1 977) draws up a typology of "institutive" discourses which follows the principle of pertinence for the level of discourse, but which distorts the level of the semiotics of space with non-pertinent elements: elements of text analysis (presence or absence of first-person discourse, verb tenses) and elements of an �·articula­ tion of ideology with space which is simply causal (a produces b) and not also semiotic (b signifies a). The two poles of "institutive" discourse are the "utopian" discourses, governed by binary �·oppositions - of which one term imposes itself in space - and proposing atemporal, closed spatial models, and the "semiogenetic" discourses, gov­ erned by a combinatorial referring to generative spatial elements, and based on semantic taxonomies. The "utopian" discourses give two main models for the organization of space: a "culturalist" (nostalgic) and a "progressist" (modernistic) model. According to Greimas ( 1 974), the modern city, in both production and conception, is articulated, on the semantic level, on the basis of three axiological isotopies: �·aesthetic, sociopolitical, rational, and two semic categories: "euphoria" versus "dysphoria" and "society" versus "individual." This grid, which he sees as provisional, is at the base both of the �·generation of the "regular" forms of the city, actual or potential, and of our reading of it. The Form of the Expression Although Eco ( 1 9726:259-3 1 7 ) does not separate the analysis of production and con­ ception of space and is concerned principally with architecture, he formulates positions which are useful in a discussion of the •�signifiers in the production of settlement space.

Settlement Space

929

He discusses in detail the relationship between the form of the expression and geometry, and arrives at the following conclusions : (1) geometry is not exclusively a property of architecture; (2) the form of the expression in architecture is not composed of geometric elements; and (3) geometry is a metalanguage of the form of the architectural expression. He reviews the bibliography on the subject of the architectural signifier, and tends to support the existence of a double articulation. His position on the subject of geometry is equally valid for settlement space. On the contrary, Greimas (1974) considers the spatial signifiers as coinciding with geometric forms. After formulating a series of simplifying assumptions, Greimas posits that the form of the expression of the language of the settlement (and of the architectural whole) consists of a set of levels. The simplest level is that of elementary spatial catego­ ries (e.g., "straight line" versus "curved") which follow an elementary grammar. The second level is that of autonomous elementary structures (e.g., a square), spatial figures, which follow a phrase grammar. Finally, there is the level of configurations, which follow a text grammar; only this grammar refers to the whole of settlement space. Pas­ sage from one level to the next is by means of derivation (and �·transformation) rules. The elements of each level are bearers of signification. A parallel approach, for all its differences, is that of Castex, Panerai, and Depaule (Castex and Panerai 1974; Castex, Depaule, and Panerai 1978: 101-33), who are con­ cerned to analyze the logic of architectural space, which they consider an absolutely autonomous logic. They define three categories of architectural "signifiers," which they consider fairly generalizable and to which they ascribe immediate spatial "signi­ fieds" without specific cultural content, confusing a metalanguage on the signifiers with an arbitrary connotative semiotic. In their view, their categories, through operations, combine to produce the architectural whole, and can be decomposed into three systems of spatial models (contrast Ostrowetsky [1974:194], and Raymond [ 1 977], who considers that it is only the whole of the architectural type which is the bearer of signifi­ cation). The orientation toward an absolutely autonomous logic of space is also that of the work of Boudon (1973, 1977, 1978), who believes that the logic of settlement space is subordinate to a wider "topic" grammar - covering a range from buildings to cosmology - which is in its turn part of a very wide grammar of places - also including drawings - and also that space appears to be analyzable in topological terms. Boudon thinks that spatial forms can be studied independently, and attempts to find a "significa­ tion" which would have its source in their articulation. His goal is the formulation of a grammar of places and the artificial simulation of settlement space. He, like Greimas, takes the position that the spatial organization of the settlement is composed of three levels of increasing complexity, which together constitute the surface structure, and for which he attempts, on the basis of certain monographical examples, to formulate ele­ ments and rules. Passage to the surface structure is from deep structures, which he also situates on the level of forms, and which include generative elements. These latter are organized topologically, for the whole of the settlement, on the basis of three types of series. Boudon also believes that a spatial grammar can be constructed, based on the (ab­ solutely) autonomous analysis of space, and once constructed, related to other, cultural grammars, on which points his views differ from the perspective of Greimas. Hillier and Leaman (1973, 1975; Hillier et al. 1976) also believe in the possibility of an absolutely autonomous analysis of space. They distinguish between logical space

930

Settlement Space

and real, "spatial" space, which they consider to be related as langue to parole and deep structure to surface structure, and as signification to the vehicle of signification (they seem to consider these relationships as homologous). The same structures of logical space lie behind both production and conception of "spatial" space, and the purpose of structural analysis is to search for and formulate them. Such a view does not take into accout the differences between connotative semiotics and metalanguage, between the form and the substance of expression, or between semiotic production and conception, nor does it consider the social relativity of conception. The authors wish to formulate a "space syntax model," a theory of "morphic languages" and of the possible, socially realized, architectural and urban patterns. Their actual object is of the level of a mor­ phology of constructions. Their "morphic language" includes "elementary objects" and their relations, and eight operations, each one of which produces a pattern of settle­ ment and/or architectural complex. Hillier and Leaman consider real space as shaped on the basis of its own logic, and of "social variables" which constitute an exogenous source of its organization. They relate their spatial patterns to certain patterns of the organization of society - which they formulate through a similar approach, a very dubious undertaking - and conclude that space is in some cases a direct expression of social organization, with a parallel decrease in the expression of its own logic, but that usually it is a means and substitute for social organization, with a parallel increase in the expression of its logic. The global analysis of space is complemented, in the authors' view, by the analysis of the relationship of space to nature, and is based on the formula­ tion of the "morphic language." Analysis of urban plans can be of interest for the production of space. The plan may be a transcoding of the spatial discourse, or an anterior iconic representation, of a metalinguistic type, of actual space. An analysis of the urban plan as such in undertaken by Groupe 107 (1 973b), who, as a first approach, leave out the form of the content. They believe that the plan must correspond to a langue, and analyze a lay-out plan, using concepts from topology and from projective and Euclidean geometry, and the unsupported hypothesis that the mathematically invariant characteristics are signifiers. The Semiotic Conception of Settlement Space Approaches The theoretical approach of Greimas (1974) is part of the same theoretical whole as his analysis of the semiotic production of settlement space. His analysis of the signification of space concerns the spectator of "doing" and his reading of society through space. This reading probably utilizes codes different from those of the producer of space. The spectator of "doing" is the "user" of space and the receiver of the message-space, which is the place of the "utterance" and consists of subjects (users) and the objects which sur­ round them. There are two forms of utterances : utterances of " state" and utterances of "doing." The former belong to a set of isotopies resulting from the formalization of the relationships of the user with the sensible qualities of objects (see also Hall 1 971 ). The lat­ ter, which are presupposed by the utterances of "state," refer to the production and the transformation of states through action on objects by social subjects having syntactic roles with culturally determined axiological content. (This kind of approach is used by Groupe 1 07 [1 973a) and by Hammad et al. [1 977) for the analysis of interiors.

Settlement Space

931

Hammad et al. - who do not use questionnaires, appear to postulate for all users a com­ plete and identical set of "places" and relations, and who weaken their analysis with behavioral data - confine their analysis to utterances of "doing" and conclude that the content has a deep structure, identical to a narrative program of spatial use, and a sur­ face structure, and that there are two spatial structures corresponding to these.) The semantic representation of the way of life in the settlement permits a partial for­ mulation of the structure of the content of the user's conception of space, partial to the extent that it is supplemented by a conception of the imaginary, "absent" city. A typology of users leads to a typology of spatial conceptions. While the analysis of "do­ ing" remains on the level of the message, the typology of users enters the field of socio­ semiotics (in the narrow sense); the text grammar of spatial conception is constructed within a sociosemiotic framework. A weakness of Greimas' approach is that the analysis of objects and subjects together, which does not break the principle of pertinence for text theory, in the case of space results in a double semiotic analysis, of space and of society. For Ledrut (1973a), who adopts a sociosemiotic (in the wide sense) approach, the city signifies through a set of visual and non-visual semiotic systems and their connota­ tive codes which are integrated into the framework of a global urban connotative sys­ tem, a world view, of social origin. Pre-Renaissance settlements were not produced with the intention of the whole social group, but by certain representatives of it; there is thus a quasi-intention and a pseudo-sender. For the modern city, there are one or more senders for sections only of the city, and the message is not the result of an intention to communicate. This city is produced by society; it is at the same time a pseudo-sender and its message, which is a pseudo-text; for this latter reason, there is no general urban code. Eco (19726) attempts to lay a theoretical foundation for the analysis of the process of conception of architectural objects, which leads to the possibility of componential analysis. He applies the method of Katz, Fodor, and Postal which he extends to include both connotation and context. He considers that the signification of the form of the architectural object incorporates the signification of the pre-architectural spatial rela­ tions as defined by proxemics, and that the spatial signifier of proxemics, the choreme, has as its denotation a physical function. The architectural morpheme is complex and consists of a set of morphological features, each one of which includes one or more mor­ phological markers (e.g., distance; static function, material: here Eco goes beyond prox­ emics). To the architectural morpheme corresponds a sememe which consists of a set of semantic markers, semic features (semes) belonging both to the denotative and to the connotative level, which may correspond to the morphological markers. That Eco's approach is of interest also for the analysis of settlement space is clear from the similar approach of Fauque (1973). For Fauque, to the spatial signifiers which the subject conceives through all his senses - correspond semes; the sensible nature of the signifier does not determine the seme. The urban image includes "ur­ bemes," corresponding to signifiers (e.g., "square," "road"), which consist of "individual features" (see the semantic markers of Eco), which can be grouped into pairs of oppositions - e.g., "wide" versus "narrow," "dense" versus "sparse;" the transformation of a feature into its opposite produces a change in signification. Fauque, like Eco, underlines the importance of the context, which he also considers as dependent on pairs of features, for the signification of the urbeme.

932

Settlement Space

The Form of the Content We have already touched on some of the views concerning the form of the content in the conception of settlement space, in our discussion of the same aspect in its production. One such case is the view of Pinxten. For the conception of space, Pinxten aims at con­ structing a universal "minimal base," which he applies to the conception of regional space. It consists mainly of topological concepts, among them concepts of "border," "area," and "path," which correspond to three of the five categories of Lynch, and the concepts center-periphery (Pinxten 1974). He thus attempts to create a metalanguage for the denotative conception of space. Of particular interest is the analysis of Ledrut (1970, 1 973b), which is based on the structural semantics of Greimas and thus related to the componential analysis of Eco discussed above. Ledrut's analysis is on the semantic level much more complex than Eco's, and extends to statistical elaborations based on factorial analysis. (For the analo­ gous semantic Anglo-American approaches to urban space, mainly the "semantic differ­ ential" and "repertory grid" methods, derived from psychology and used in environ­ mental psychology and geography of perception, and also complemented by one of the types of factorial analysis, see Bailly 1977 : 151-63, 173-85; Pocock and Hudson 1978: 69-77.) Ledrut studies the axiological and non-axiological semic categories and the other pairs of opposition in the conception of the city, and their relationships - which may extend to the construction of semantic micro-universes - and the corresponding "structures;" he also studies the axiological semic categories involved in references of the preferred city, and the "systems of preference" or (urban planning) "models" to which they correspond (cf. Raymond et al. 1971, who find a world view consisting of semantic pairs of opposition as the connotative level of the single-family house). A sim­ ilar analysis is found in Fauque (1973), according to whom the semantic level of the urbeme consists of one or more paradigmatic tree (-structure)s. Ledrut finds that his "structures" and "models" are closely and systematically related, and that by grouping them he can arrive at two models which are in opposition without, however, being in contradiction : a functionalist or practical model, less widespread, and a hedonistic, "vi­ tal" model which concerns the individual's personal relationship to the city and is more widespread. The first model is related to operational urban planning, while in the second a dominant role is played by historical monuments referring to a city which does not coincide with the core of the actual urban organization, with the result that the city acquires a mythical historical meaning through its ruins. These two models are systemati­ cally correlated with different social groups, a fact which points to the alienation of these groups. While the studies of Ledrut stress connotative signification, Lynch (1965), who uses the Anglo-American approach, focuses on the analysis of the denotative level, in partic­ ular of those characteristics which concern the legibility of urban space and orientation within it. A peculiar view on the source of the signification of settlement space is expressed by Barthes (1971b). The analysis of functions as signifieds is elementary and precarious, because of the transformation of signifieds into signifiers and of historical changes. The signification of the city has its source in the signifiers and their relations, and consists of marked and unmarked elements. Barthes gives particular emphasis to the "eroticism" or "sociability" of the city, which he attributes to the unlimited metaphorical nature of

Settlement Space

933

urban discourse (for a scientific and ideological critique of this view, see Eco 1972b: fn., pp. 285-86). The Form of the Expression Ledrut (1973a) if of the opinion that the form of the expression of the modern city cannot be analyzed into *distinctive (urbemes) or significative units which would be common to all social subjects, basically because the city is a pseudo-text. An urban lan­ guage would exist only if the urban planners were to realize in space recognizable logo­ technique. We should note that, on the one hand, this view does not apply to pre-Renais­ sance settlements, and, on the other, it does not exclude a common mode of organiza­ tion of the signifiers of the modern city, nor the existence of generally the same signifiers within a particular social group. The approach of Fauque, which we already mentioned, could be applied within such a framework. Ledrut also believes that we cannot consider as a minimal unit of the urban discourse a section of space, e.g., a road, which is in general correct because such units are part of a metalinguistic analysis of space (for such an attempt, see Castex and Panerai 1970-71). The signifiers of the city are its "objects," which can be related to the elementary spatial structure "center/non­ center," derived from a global and fundamental conception of space. The signifiers of space can be classified into three paradigmatic categories, those of points, lines, and surfaces (Lagopoulos 1973: 91-113, 130; see also Lynch 1965 : 46-83; Norberg-Schulz 1971: 17-27), which, however, appear not to be exhaustive. The Anglo-American approach indicates that the relative dimension and distance of the signifiers are influenced both by the semantic level and by non-semiotic factors (see Pocock and Hudson 1978: 57; Downs and Stea 1977 : 1 43-44 ). Apart from the existence of urbemes, Fauque (1973) gives some hints as to spatial syntax. He believes it consists of spatial relationships (e.g., "above/below," "behind") and that it is the bearer of signification. Lynch (1965: 88-89) identifies four kinds of syntax based on spatial relationships (e.g., relative direction, relative distance, sequence) between urban signifiers; the two extremes are the tendency towards non­ existence of syntax and the coherent syntax. The preceding discussion has revealed some of the principal axes of analysis in the semiotics and sociosemiotics of settlement space. We will conclude with a brief presenta­ tion of the principal general and interrelated axes of the semiotics and sociosemiotics of space, accompanied by some comments aimed at completing a first impression of the domain. The relationship between non-semiotic factors and the semiotic spatial systems. The non-semiotic factors which influence or are influenced by the semiotic systems are social, psychological, or bio-ecological in nature. According to Hall (1971 :129-42), the cultural signification of settlement and regional space, which are included in one of the three types of spaces of proxemics, is founded on the biological substratum of man. It should be noted that in the production of space, the semiotic system influences the sub­ stance of expression. On the other hand, the mental image is influenced primarily by interaction with the environment, based on individual practice, and by psychological fac­ tors in the case of the immature subject (see, e.g., Downs and Stea 1977:187-208), while in the case of the mature subject by a series of factors among which are the settlement

934

Settlement Space

space and the mode of movement within it, and of which the socioeconomic factor or social praxis is dominant (see Klein 1967 ; Prokop 1967; Castells 1972 ; Lagopoulos 1973 : 115-20, 1980; Fauque 1973 : 19-20; Bailly 1977: 29, 90, 99, 106, 113, 183, 187; Pocock and Hudson 1978: 17, 65, 107, 134, 143 ). The mental image influences behavior and practices, which in turn alter space (see, e.g., Downs and Stea 1977: 12 ; Bailly 1977: 31, 173 ; Pocock and Hudson 1978 : 13, 128-29). This alteration is very limited in precap­ italist societies and much greater in industrial societies. The relationship between production and conception. There is no doubt that produc­ tion and conception are related both as processes and as content. In precapitalist socie­ ties, the spatial model of the mental image, which is accepted by the whole society, coin­ cides with the model of production, with the result that the latter corresponds to the expectations of the user and there is a strong tendency to permanence of actual space. This is not the case in industrial societies (see Lagopoulos 1980: 2.5 ), even though we find an analogous phenomenon: there is a close relationship between Choay's models of production and Ledrut's models of conception. The definition of the semiotic spatial systems. The semiotic systems of settlement space show a relative autonomy (Lagopoulos 1980: 3.5), while their message (as well as the architectural message), according to Eco (1972b:fn., p. 271), shows all six of the func­ tions of communication based on '�Jakobson. Many of the views of semioticians discussed above on the subject of the signifieds and signifiers of the mental image are supported by the views of the geographers of per­ ception and the psychologists of the Anglo-American approach (see, e.g., Downs and Stea 1977:41-55, 78, 79, 91, 117; Pocock and Hudson 1978:3, 32, 68, 77-80). The relationship between architectural and urbanological signifieds of denotation and their shell-signifiers, both in production and in conception of space, is conventional and has its source in cultural codes (see also Eco 1972b :271-72; Agrest and Gandelsonas 1973: 257); but the relationship between the former and their spatial location-signifiers is motivated during conception (see also Ledrut 1973a: 3-4). The signifiers in the pro­ duction of space originate from the functional signifieds and the connotative signifieds to which the functional signifieds correspond, as well as from denotative and connota­ tive signifieds of another nature (see, for architecture, Ostrowetsky and Bordreuil 1978). The mental image is composed of a series of interrelated levels, which correspond to approaches to the settlement space on different scales (Lynch 1965: 85-86; Lagopoulos 1973: 93-102, 134). This set of levels belongs to a wider series of interrelated images, which starts with the image of objects manipulated by the hand and goes as far as the image of regional space (see also Norberg-Schulz 1971: 27-33 ; for the denotative and connotative signifieds of regional space and their relation to non-semiotic factors, see Gould and White 1974). The characteristics of the semiotic spatial systems. An important number of general, spe­ cific, and comparative characteristics have been formulated for the mental image (see, e.g., Lynch 1965; Lagopoulos 1975, 1977 ; Downs and Stea 1977: 103-04, 117). One of these is the alteration in history of the denotative and connotative significations of the mental image of the same space (for a typology of these alterations, see Eco 1972d: 279-87). Choay (1970, 1970-71), dividing settlements into two great categories - pre­ and post-Renaissance - observes that the former engage the global personality of the individual and are hypersignificant, while the Western city became hyposigni/icant and,

Settlement Space

935

in the nineteenth century, monosemic, due to the new process of production, a fact which should not be evaluated as negative. The signification of modern urban space is supplemented by semiotic systems of supplementation (e.g., graphic systems) and, in conception, by imaginary constructions (see also Greimas 197 4:13). If we clarify that it is the semiotic production of the Western city which is characterized by hyposignifica­ tion, then this view does not contradict Ledrut's view (1973a: 8-1 3) that the mental image of the modern city-as-instrument is no less signifying, but signifies quite differently to the extent that the signs have become autonomous from the objects. Semiotic spatial systems and urban design. Urban design, a part of urban planning, is a synthesis of non-semiotic factors and a semiotic system of spatial production, to a greater or lesser extent metalinguistic ; contemporary urban designers have not yet con­ structed a strong metalanguage. This system is still today alimented by the mental image of the producer, and may be alimented by the mental image of the future users; since the Renaissance, producers are the creators of logo-techniques. A particularly interesting example among the few approaches to semiotic design is the approach of Bonta (1973), based on Buyssens and Prieto, which, although its object of application is the design of the pieces of a chess board, lays the foundations for an applied semiotics on all spatial scales of design. The author distinguishes between physical and significant form: the latter is that part of the former which constitutes a veh­ icle of signification. The significant form is constituted by a set of material features, the signification by a set of values (cf. Eco 1972b). The articulation between values and fea­ tures is accomplished when there is bi-univocal correspondence between them. One, or several hierarchically organized, "articulating systems," characterized by the existence of articulation, produce an "articulated system," also based on articulation. The forms of an articulated system can result from a greater or lesser number of features, which number corresponds to the "significant richness" of the forms. Bonta observes that in any semiotic system both features and values are characterized by themselves, i.e., by a ''lexical dimension, and by a syntagmatic and a paradigmatic context (cf. Eco 1972b ). A design proposal is the result of three primary variables - the "significant richness," the number of articulating systems ("degree of articulation"), and the context - which determine the "significant structure," the deep structure of the proposal. The choice of the relative weight to be given to these variables is an exo-semiotic matter, and depends on the designer. He has the possibility of choosing among an enormous variety of values, and must organize them according to a classification system, e.g., a binary one. The pas­ sage from values to features is bound by cultural factors. It must be accomplished on the basis not only of existing, but also of possible, codification, but if, in order to be under­ stood, it requires apprenticeship on the part of the reader, it is inappropriate. The authors we will refer to below have concerned themselves not with metho­ dological questions but with other aspects of urban design. Choay (1970-71: 10) con­ siders that industrial societies have two levels of history, a level of rapid changes and one of a "neo-archaic" nature, and that for the first level we need to produce a space which would correspond to the new networks of communication, while for the second - to which correspond the "institutive" discourses - we need a symbolic space. This space, then, would be the object of semiotic urban design according to such a view. For Fauque ( 1 973: 27), it would be possible to produce cities similar to an existing city by using the characteristics of the syntax of signifiers of the mental image of it, as well as pleasing cities by eliminating relations which produce dysphoria.

936

Sign

Many Anglo-American authors have observed that certain actual spatial organiza­ tions aid the legibility of the city, but we must emphasize that this latter is characterized by cultural relativity. Also, many such authors believe that the complexity - but not the labyrinthine quality - of urban space is essential as an optimal stimulation and for the identity of the subject. While Norberg-Schulz (1971 : 19, 35) emphasizes the need for the relative stability of space so that the subject can stabilize his mental image of it something which the author considers essential - Eco (1 972b :259-317) emphasizes mobile and open signifiers - products of a generative approach - in order to achieve variety in denotation and freedom in connotation. A.-Ph.L.

Sign Definitions The most common definitions of sign are summarized by the scholastic formula aliquid stat pro aliquo 'something stands for something else,' in which the first term is the signans 'signifier' and the second one is the signatum 'signified.' Commenting this formula, �Jakobson (1975a :16) says: "tout signe est une relation de renvoi." According to ''Augustine, signum est enim res prater speciem, quam ingerit sensibus, aliquid ex se faciens in cogitationem venire 'a sign is a thing which causes us to think of something beyond the impression the thing itself makes upon the senses' (De Doctrina Christiana 2,1,1) and, according to *Peirce (1931-66 : 2.228), it is "something which stands to somebody for something in some respect or capacity." Such a relationship of standing for (variously labeled as "representation," "reference," "''meaning," "signification") can be taken as a correlation linking the sign to a given representative mental state (idea or concept), to an external entity, *object or state of affairs, to another sign (*interpretant). These possibilities have originated, throughout the history of semiotics, different "triangles," more or less comparable, even though reflecting different epistemological views : semainon verbum vocis terminus name Locke: Zeichen Frege: sign or representamen Peirce: sign Carnap: sign Ogden and Richards: symbol sign vehicle Morris: signifiant Saussure: express10n Hjelmslev: Stoics: Augustine: Ockharn:

semain6men6n verbum mentis conceptus nominal essence Sinn immediate object sense intensional object reference significatum signi/ie content

tygchanon res res

thing Bedeutung dynamical object nominatum extension or designated object referent denotatum (outside language) matter or content continuum

As it is evident from this incomplete list, some authors call "sign" the material entity standing for something else, while others give this entity a specific name and presuppose or explicitly say that the sign is the bifacial entity represented by the left side of the trian­ gle. The medieval tradition often used the term dictio to indicate such an entity. The fact

Sign

937

that Saussure (and in general the structuralistic tradition in linguistics) does not consider the third corner of the triange means that they are interested in a merely inten­ sional description of sign systems or the ,:•codes correlating a given expression with a given content. The authors insisting on the third corner are mainly (or also) interested in an extensional *semantics and are concerned with the processes of *communication along which signs are emitted in order to denote, to indicate, to designate, to refer to, things or states of the real or of some ''possible world. These different ways of looking at the semiotic triangle distinguish meaning-oriented semantics from ,:•truth condition­ oriented ones. Among modern semioticians, the most widely accepted models of semiotic correla­ tion are those of Peirce and Hjelmslev. Peirce speaks of a Representamen (motivated by the Dynamical Object), which stands for an Immediate Object, which represents in some aspects the Dynamical Object and is interpreted (or can be translated) by an Inter­ pretant. Hjelmslev considers two planes, expression and content, both subdivided in form and substance: the form of the expression makes pertinent and organizes into a system of differential units the still unshaped elements of the continuum of the expression. In its turn, the form of the content (or the form of the universe of semantic properties) makes pertinent and organizes into a system of differential units the still unshaped ele­ ments of the continuum of the content. The correlation between the two functives (ex­ pression and content) constitutes the sign *function (in the sense of "semiotic relation­ ship," not to be confused with similar expressions such as "function of signs"). According to Hjelmslev, a previous sign function can become the expression of a further content, thus constituting a new sign function (relationship between ''denotative and connotative semiotics). It must be clear that, in Hjelmslevian perspective (as well as in many other structural approaches), denotation does not have the same sense it has in extensional approaches. An expression unit denotes a given content portion, and this relationship concerns only intensions. Not very much dissimilar is the phenomenon described by Peirce : a sign (or representamen) is defined and enriched in its meaning by an interpretant, this interpre­ tant being interpreted by another interpretant, and so on and infinitum (unlimited ''semi­ osis). Classification of Signs Many authors establish different typologies of signs. The classic tradition has distin­ guished signs as artificial versus natural, intentional versus unintentional, conventional versus motivated. Among more complex typologies, see the one of Morris, the classifi­ cation by ''channels and sources of Sebeok ( 1 976a), and the typology of modes of sign production by Eco ( 1976c). The most complex classification was proposed by Peirce who, starting from three basic trichotomies (signs in themselves, in relation to their object, and in relation to their interpretant) elaborates ten classes; later, he foresaw six­ ty-six classes as theoretically possible. The Crisis of the Concept of Sign All authors who have explicitly spoken of "semiotics" have defined it as the ,:•doctrine of signs (Locke, Dalgarno, ''Lambert, *Bolzano, �-Husserl, Peirce, Saussure, Morris,

938

Sign

Jakobson, and others). Many philosophers have written of the sign as a specific object of theoretical research (from �·Aristotle to Augustine, from Roger to Francis �·Bacon, from �·Hobbes and '�Leibniz to Carnap, etc.). However, contemporary semiotics has variously cast into doubt such a notion, at least in two ways : ( 1 ) by judging it too com­ prehensive, therefore replacing it by either an analysis of minor components (/igurae) - be they units of the expression or of the content - see Hjelmslev 1 943; Greimas 1 966a; Prieto 1 966; and all the representatives of lexical analysis, componential seman­ tics, semic analysis, and so on - or privileging the signifying face (Lacan 1 966; Derrida 1 967a; and a general French post-structuralism); and (2) by judging it too narrowly, therefore substituting for it the notion of ,:•sentence, '' enonce, �-seme (for instance, Buyssens 1 943), the notion of �·text or discourse (Ducrot 1972; Petofi 1 976a; van Dijk 1 977c), or the notion of production of sense (Kristeva 1 969), semiotic practice, semi­ osis. As a matter of fact, many of the problems that the ancient tradition dealt with under the heading de signis are today discussed under such entries as enonce/ enoncia­ tion, sentence, *utterance, '�index, *icon, symbol, semiosis, signification, ,:•speech act, etc. Nevertheless, it seems a too radical decision to get rid of a notion that, throughout the whole history of human thought, has proved to be so powerful in producing tre­ mendous theoretical developments. It would, of course, be useless to keep sign as a mere very general hyperonym (such as the concept of and the term thing), or to employ it met­ aphorically and equivocally. Some authors suggest that to label as signs radically different phenomena such as a word, a meteorological ,:•symptom , or an algebraic sym­ bol, is a pure analogical license. Yet, a semiotic theory, in order to exist, must build up a theoretical object of its own, at the same level of generality at which ''mathematics built up the concept of space and physics the concept of force: so that this general object can be singled out in different phenomena such as a vocal emission, a pointing finger, an image, a facial expression, an object used by �·ostension. The Linguistic Fallacy One of the stronger objections raised against the unified concept of sign is that it is the result of an undue extension of a category belonging to linguistics (where signs are con­ ceived of a intentionally emitted and conventionally coded expressions, to be analyzed in terms of minor articulatory components and to be syntagmatically disposed according to a linear sequence). If so, it is clear that many other phenomena labeled as signs do not share these properties. If, however, one reconsiders the whole history of the concept of sign, one discovers that it followed a rather different "evolutionary" or "phylogenetic" pattern. Only lately has a general semiotic notion, posited in order to define many natural phenomena, been used also to designate linguistic phenomena like nouns and verbs. Once the sciences of verbal languages, undoubtedly the more deve­ loped among other semiotic researches, had "captured" the notion of and the term sign, they brought the concept to a precise and rigorous level of definition, such that, when a renewed interest in a broader range of semiotic phenomena led some scholars to re-use the term outside the linguistic domain, such an adoption appeared be an illicit adapta­ tion. It is thus necessary to attempt an "archeology" of the notion of sign so as to redis­ cover its original sense, thus reversing, or turning on its feet, the linguistic paradigm that has dominated the semiotic approaches of this (but only of this) century.

Sign

939

Towards an Archeology of Signs

Words and Signs The pair semeion and/ or tekmerion (often interchangeably translatable as : proof, sign, index, symptom, and so on) appears in the Corpus Hippochraticum as referred to natural facts (the symptoms of modern medicine) which, be inference, lead to diagnostic con­ clusions. It must be clear that in this sense a sign does not entertain, with its own meaning or with its own referent, a relationship of equivalence (p = q), but a relation­ ship of inference (if . . . then, p ::J q). As a matter of fact, ,:•Hippocrates, challenging the current view of the medicine of his time, does not think of an elementary code (a given symptom stands for a given illness) but of a complex, contextual interpretation of co-oc­ curring data involving the whole �·body of the ill subject along with many aspects of his environment (air, water, weather). This inferential nature of signs is important in order to understand the position of Parmenides, who asserts that verbal language, with its on6mata 'words or names', pro­ vides us with false knowledge, based upon the illusion of experience, while true knowl­ edge of the Being is made possible by semata 'signs' or 'clues'. Thus, words seem to be deceptive tags (equivalents to equally deceptive perceptions), while signs are the right point of departure for a true reasoning about the real nature of the One. Aristotle seems to be equally reluctant to consider words as signs. In '�Rhetorics, he speaks of signs as natural facts able to reveal a possible consequence. He distinguishes between "necessary" signs (tekmeria) (where the antecedent entertains a necessary relationship with the consequent: "if one is feverish then one is ill") and "weaker" signs (for which no technical term is given), where the relationship is not necessary ("if one pants then one is feverish," yet one could pant for other reasons). These two sorts of signs represent two different cases of inference (p ::J q). While the necessary signs are sensitive both to the modus ponens and the modus to/lens, the weaker ones allow weaker inferences, to be used for the purposes of rhetorical persuasion, where the negation of the implicatum is not sufficient to deny the truth of the implicans. On the other hand, words do not seem to permit inferences, but entertain a relationship of equivalence with their '�definition: "man = rational animal." It is true that Aristotle (De /nterpretatione 16 a) states that even alphabetic letters are the semeia 'signs' of verbal sounds, and these are signs of the affections of the soul. But Aristotle means that linguistic utterances (which he properly calls symbola) can also be taken as symptoms of the fact that the one who utters these sounds has some thought on his/her mind. This oscillation or opposition between words and signs is detectable even in Stoic semiotics. The triangular relationship semainon-semain6menon-tygchanon 'signifier-sig­ nified-referent' always concerns verbal expressions. On the other hand, when the Stoics speak about a visible antecedent revealing a not immediately detectable or otherwise unknowable consequent, they speak in terms of semeion and lekton. The lekt6n is one of the as6mata 'incorporeals' (like the void, time, space); it is an dicible or a dictum (it is con­ troversial as to whether to translate it as "what can be said" or "what is said"). It seems, however, that between the linguistic pair semainon/ semain6menon and the semeion there is a relationship of connotation (cf., for instance, Rey 1 973 :39). Linguistic expressions convey lexical contents (incomplete lekta), which are articulated into complete lekta, or, in modern terms, �·propositions. The relationship of signification the Stoics attribute to

940

Sign

the semainon is the one occurring between two complete lektd (if antecedent then conse­ quent). In this sense, the antecedent proposition is the sign of the consequent one. In other words, verbal language is the most appropriate vehicle of a "natural semiotics," which is expressed by inferential schemes. Since for the Stoics signs are logical schemas, their "if . . . then" relationship does not hold between tokens but between �·types. It is irrelevant whether for the Stoics the inference was from cause to effect, form effect to cause, or between causally unrelated events, since they seem to accept the Philonian concept of material implication. The examples they give of ypomnestik6i 'commemorative' signs, in which a detectable antece­ dent stands for a transitorily undetectable consequent (for instance, if there is smoke then there is fire), and of endeitik6i 'indicative' signs, in which a detectable event stands for a definitely undetectable one (for instance, the alterations of the body which reveal the alterations of the soul), seem to be based upon an effect-to-cause inference. But when ,:•Quintilian elaborates both upon Aristotle and the Stoics' notion of necessary signs, examples such as "when there is wind on the sea then there must be waves" clearly refers to a cause-to-effect inference. Quintilian, in fact, speaks of signs whose consequent is a future event (cause-to-effect) (Institutio Oratoria V,9). Later, *Aquinas ( Summa Theologiae III,62, and even I,70,2 ad 2) says that material causes can be the sign of their possible or actual effect. Since this notion of sign will be largely utilized by rhetoric, the link os physical necessity between antecedent and conse­ quent was overwhelmed by a more "sociological" (so-to-speak) idea of a connection recognized by virtue of current opinion, and the inference "p :::> q" was frequently rati­ fied on the grounds of a socially acceptable verisimilitude. The point is, however, that the post-Stoic tradition, and mainly Sextus Epiricus (Adversus Mathematicos 8,11) matched semain6menon with lekt6n, thus unifying the theory of language with the theory of signs (even though in order to challenge both). Such a unification was definitely performed by Augustine (De Magistro and De Doctrina Christiana), who listed verbal terms as species of the more comprehensive genus "sign." It was, however, not clear how to subsume under a unique category both inferences and equivalences. Throughout the further development of semiotic thought, this problem continuously reappears. Even our present everyday language suffers from this ambiguity : sometimes one means by sign only conventional marks (road signals, inscribed boards), sometimes symptoms, sometimes *nonverbal devices (such as the so-called ''sign language of the deaf). Seldom do people recognize words as signs. But when Saussure speaks of le signe linguistique 'the linguistic sing', he is following an ancient tradition, and the word sign definitely means the relationship signans-signatum, that is, the correlational phenomenon that Hjelmslev called sign function. As this point of linguistic sign had become the paramount example of sign function.

The Sign Function Revisited Apparently, the most common interpretation of a sign function based upon the linguistic model takes the format of an equivalence, be it by mere ,:•synonymy (p = q) or by defini­ tion (p = a + b + c . . . ). However, not even the Hjelmslevian model establishes that to a given expression unit must correspond a given content unit. Hjelsmlev speaks of an

Sign

941

abstract relationship between two planes but does not exclude that it can take the form of a series of entailments: "if bachelor, then human male adult." Coming to Peirce, the representamen would be "equivalent" to the interpretant if the latter was only "the idea to which it (the former) gives rise" (1931-66 : 1 .339). But Peirce states expressly that the interpretant, in so far as it "is revealed in the right understanding of the sign itself" (1931-66: 4.536), makes it possible that a rheme (a *term) is interpreted by a dicent (a proposition), and a dicent by an argument. The meaning of a logical term is a rudimentary assertion (1931-66: 2.342), while a proposi­ tion is a rudimentary argumentation (1931-66 : 2.344). Even the definitions are trans­ formed from equivalence into inferences : a term is "the blank form of a proposition" (1931-66: 4.560), and in order to represent the verb to marry one should foresee a com­ ponential spectrum such as "marries to" (1931-66 : 2.379). This is the same idea that governs the contemporary representation in terms of �·case grammar and even the �-pragmatics and semantics of *presuppositions : if x cleans y then it must be presupposed that y was dirty. Naturally, any a semantics in the format of inference must rely upon a semantic encyclopedia able to take into account even a the­ saurus of intertextual ,:•frames and contextual operators (see Eco 1979c), an idea already proposed by Augustine (De Magistro 2) when, analyzing the meaning of a syn­ categorematic term like ex, he says that every time one finds in a text this preposition one should suppose (or infer) that it expresses the separation from something else. An intensional semantics in the form of an encyclopedia will thus be an lnstruktionsseman­ tik, where a sememe appears as the rule for the production of a given co-text (Schmidt 1 976). Such a semantics would also involve pragmatic conditions as coded restrictions or contextual selections (see Eco 1976c). If such an inferential structure is typical of verbal language, this is more reason to see it as a strict requirement for every other sign function. A stop sign should be inter­ preted as : "if this visual expression is inserted into a road co-text x, then (if a car) stop ; if stop, then look carefully left and right and ( 1 ) if no danger then go on, or (2) if do not stop and look then possibility of fine." In this sense the definition by Morris (1946 :87) becomes clear : "if anything A is a preparatory stimulus which in the absence of the stim­ ulus-objects initiating response-sequences of certain behavior-family cause a disposi­ tion in some organism to respond under certain conditions by response-sequences of this behavior-family, then A is a sign. "

The Theoretical Object of Semiotics It is the abstract notion of sign function based upon an inferential mechanism (sometimes strictly coded, sometimes open to contextual interpretations) that is instantiated and can be studied at different levels of complexity (terms, sentences, texts) throughout different continua (auditory, visual, plastic, etc.). In this sense the notion of sign implies the notion of semiosis. The construction of such a theoretical object is the main task of a gen­ eral semiotics as a philosophic discipline. Specific or applied semiotics will then be inde­ pendent disciplines which are more or less rigorous depending on the mode of correla­ tion they study, the degree of coherence of the systems organizing certain types of sign function, the complexity of the communication processes they give rise to, etc. Every specific code, from the most simple (like *traffic signals or secret alphabets) to the most

942

Sign

complex (like natural verbal languages, •:•painting, etc.) will then be studied according to their own "logic" (if any). The nature of the different systems and the sophistication of the metasemiotic instruments will characterize the scientific degree of each approach. U.E.

Conventionality of the Sign As is true of most basic concepts in the humanities, the term conventional is vague and polysemic, and the boundaries of its meaning are so fuzzy that its analysis inevitably turns into partial reconstruction. The history of semiotics shows that conventional tended to be used in three different senses: ( 1 ) Conventional as "produced by convention," i.e., non-natural. This definition proposes that there is an inherent, genetic dualism of signs. Some of them are simply fragments of the world and have been discovered by man, others are entirely man-made (we have created them by means of an informal contract which provided that certain objects would be treated as representative of certain contents). The origin predetermines the properties of signs. Those of the first type, i.e., natural signs (e.g., smoke as a sign of fire) are always a part of a concrete context and cannot be eliminated therefrom. They are independent of our will. We cannot decide what they will represent, what shape they will have and what they will refer to. We cannot produce them as we please, nor can we modify or replace them. They are inflexible and not sus­ ceptible to manipulation. On the contrary, the signs of the second type, i.e, artificial signs (e.g., arrow as a symbol of implication in logic), are fully at our disposal. Here we are not limited in their selection or the range of objects to which they shall refer. Indeed, we have a free hand to construct them and make them stand for whatever we choose: abstract notions, values, ficticious entities, etc. Artificial signs can transmit all conceivable contents, they can be replicated at any moment, and according to our purpose; we can create them, modify or reject them. Making use of convention is a distinctive trait of the human spe­ cies. (2) Conventional as "referring to an object by virtue of a convention," i.e., non­ motivated. This approach is usually connected with typologies based on Peirce's tri­ chotomy and puts less stress on genetic circumstances. The nature of sign is said to con­ sist in the bond with the referent. Some signs - indexes and icons - are motivated. Their bond with objects is founded upon spatial and/or temporal proximity (e.g., demonstra­ tive gestures) or upon the principle of resemblance (e.g., onomatopeia). It is of second­ ary importance whether this bond has been discovered or posited, or whether it is purpo­ sive or aimless. Its substantial character makes free choice impossible or imposes certain restrictions on the shape of signs. On the contrary, the sole foundation of symbols (e.g., a national flag) lies in the convention observed by all members of a community which has created them arbitrarily. They are indifferent and contingent as to the object they refer to, and, since their bond to referents is in a sense immaterial, existing only in human minds, they are hypothetically replaceable.

Sign

943

(3) Conventional as "interpretable on the grounds of a convention," i.e., non◄ universal. The last approach does not involve any typology of signs. Conventionality is a standard characteristic of semiotic entities. Every object and every phenomenon can mean something. This is not to say, however, that everything always has some meaning. We can neither deduce which entities are meaningful, nor what they mean if they are known to be meaningful. We must learn the meanings, or, to be more exact, we must be informed, directly or indirectly, by the community we want to communicate with, what contents are to be associated with a given object according to the agreements that they have reached - what selections they have made or what habits they have legitim­ ized. The knowledge of a convention is normally a sine qua non condition of the inter­ pretation of signs. Thus, only those very simple phenomena of nature (e.g., sex traits in animals) which everyone encounters in personal experience and which have the same meaning to all because they are believed to be universal will be classified as non-conven­ tional. Their interpretation does not require having recourse to socially accepted agree­ ments, and does not involve learning. Conventionality in this sense ceases to be a classifi­ catory instrument and has become a criterion of identification of the entire semiotic uni­ verse. Contrary to appearances, the three concepts described above are not equivalent. A sign may be conventional in the first sense, but non-conventional in the second one (e.g., a pointer painted on the wall of an underground passage). If a sign is conventional in the first or in the second sense then it is necessarily conventional in the third sense, but not vice versa (e.g., a strong skin reaction interpreted according to a neurological con­ vention). Since most semioticians usually do not differentiate the three variants of con­ ventionality, our discussion is more like an analysis of particular standpoints in their idealized formulation than a review of the positions found with particular authors or schools in semiotics. T.E.H.

Sign Function Hjelmslev (1943:13) proposed this notion in order to challenge the "old tradition" according to which a sign is first and foremost a sign /or something. Following Saussure, he maintains that a sign is not an expression that points to a content outside the sign itself, but is an entity generated by the connection between an expression and a content. This connection is a function which establishes a solidarity between its two functives. As such, it is inconceivable without its terminals, just as its terminals are inconceivable without it. This means that if one and the same entity contracts different functions in turn, in each case it is a matter of two different functives, of two different objects, depending on the point of view that is assumed. Thus, two homonymous expressions are two different functives of two different sign functions. Expression and content necessarily presuppose each other, and an expression is an expression "only by virtue of being an expression of a content . . . If we think without speaking, the thought is not a lin­ guistic content . . . if we speak without thinking, and in the form of series of sounds to which no content can be attached by any listener, such speech is an abracadabra, not a linguistic expression and not a functive of a sign function" (1943 :49).

944

Sign

Form and Substance In this sense the notion of sign-function is still not so different from the one of linguistic sign (signifier/signified) as proposed by Saussure. However, Saussure said that both the thought and the phonic substance are unshaped clusters, and that language organizes these two amorphous masses by producing a form. Hjelmslev maintans, on the contrary, that "the substance depends on the form to such a degree that it lives exclusively by its favor and can in no sense be said to have independent existence" (1943:50). Saussure's "substance" becomes, in Hjelmslev's framework, the mening 'purport', a common uni­ versal factor that every language has in common with every other language. The purport of the content is "the thought itself," an amorphous mass that can be analyzed by many points of views and which is ordered, articulated, and formed in dif­ ferent ways in different languages. For instance, "behind the paradigms that are fur­ nished in the various languages for the designation of color, we can, by subtracting the differences, disclose such an amorphous continuum, the color spectrum, on which each language arbitrarily sets its boundaries" (194 3:52). Such a subdivision and organization of the continuum does not take place only at the lexical level: morphemic paradigms show a similar state of affairs. (Hjelmslev recalls cases in which the zone of number is dif­ ferently analyzed by languages, which either distinguish between singular and plural or add a dual or a trial ; as well as cases of different arrangements of the zone of tenses.) Thus, the purport remains, each time, substance for a new form, and in the linguistic content we recognize the content-form, which stands in arbitrary relation to the purport and forms it into a content-substance (1943:52). The same happens as far as expression is concerned. Zones in the phonetic sphere are subdivided differently by different pho­ nological systems by producing a different number of "figurae" or phonemes. Hjelms­ lev stresses the parallelism between form and content by calling purport also the amor­ phous continuum in which each language cuts out its expression forms by producing expression-substances: "By virtue of the sign function and only by virtue of it, its two functives exist, which can now be precisely designated as the content-form and the expression-form . . . and only by virtue of them, respectively, do the content-substance and the expression substance exist, which appear by the form's being projected onto the purport, just as an open net casts its shadow down onto an undivided surface" (1943 :57). Hjelmslev says that even though the word ring is a sign for a definite object on one's finger, that thing on one's finger, in so far as it is recognizable as such, "is an entity of content-substance which, through the sign, is ordered to a content-form and is arranged under it together with various other entities of content-substance (e.g., the sound that comes from my telephone)." Likewise, "the sound sequence [r111] itself, as a unique phenomenon, pronounced hie et nunc, is an entity of expression-substance which, by virtue of the sign and only by virtue thereof, is ordered to an expression form and classified under it together with various other entities of expression-substance (other possible pronunciations, by other persons or on other occasions, of the same sign)" (1943:57-58). Semiotic and Philosophical Problems The notion of sign function is important for many reasons, and has influenced further semiotic theorizing. First of all, it shows that a sign is not a fixed entity but a formal rela-

Sign

945

tion. In this sense it can be applied not only to linguistic signs. If what Morris ( 1 938 ) said is true, namely, that something is a sign only because it is interpreted as a sign of some­ thing by some interpreter, and semiotics is not concerned with a specific kind of object but with ordinary objects insofar as they participate in semiosis, then every object can be viewed as a semiotic expression insofar as it is posited as the functive of a sign function. Hjelmslev's sign function can be interpreted in a more catholic way, since it can hold not only between discrete units but also between larger portions of the continuum (there can be sign functions correlating units to units, units to clusters, clusters to units, and so on). Secondly, the main result of Hjelmslev idea is that the procedures implemented by classical and contemporary phonology for analyzing the expression plane are thus extended to the analysis of the content plane. If there are figurae of expression then there ought to be figurae of content. Hjelmslev aims at identifying semantic invariants "which are basic to the system, the entities of which we must be able to demonstrate that all the other entities are constructed." The number of these "ultimate entities" must be "as low as possible" ( 1 943:60). Following the double principle of economy and reduc­ tion, given complex content entities such as "ram," "ewe," "stallion," "mare," "sheep," "horse," "he" and "she," then one can eliminate from the inventory "ram," "ewe," "mare," and "stallion" if they can be explained as relational units combined, on the one hand, by "he" and "she" and, on the other, by "sheep" and "horse." Thus, "ram" = "he sheep" will be structurally opposed to "ewe" = "she sheep" and to "stallion" = "he horse." In this sense Hjelmslev was a forerunner of the componential analysis of meaning, even though he did not say that the componential inventory must be given by a finite set of primitives, or that these primitives are metatheoretical constructs. On the contrary, he said that definitions of these entities are formulated in the same language and in the same plane as that to which the entity itself belongs (thus suggesting in some way that the content analysis proceeds by a process of '�translation from original signs to other signs used as interpretants). He also denied that there are semantic universals, since the elements of the inventory are cultural units differently produced by different languages insofar as they arbitrarily segment the purport, or continuum, or the content. Hjelmslev calls purport both the continuum of the expression and the continuum of the content. This makes one think that the continuum should be unique. A language organizes into expressions a given portion of this continuum (be it sound, space, human body, and so on) in order to express other portions of the same continuum (as the mate­ rial source of human experience). Thus, the unique continuum would schematically appear as a continuous extrasemiotic ring that surrounds its own semiotic organization and representation (see p. 946 ). This continuum is at the same time shaped by languages and studied by nonlinguistic sciences (Hjelmslev mentions mainly physics and social '�anthropology, thus recalling Locke's tripartite division of knowledge into physics, ethics, and semiotics). What is striking is that Hjelmslev, after having labeled the continuum of the content as mening 'sense' (purport), extends the same term to the continuum of expression, which should, on the contrary, be viewed as a lump of raw material. The cause of symmetry is not suffi­ cient to explain such a terminological choice. It makes one think that this "universal principle of formation" (1943 :76) is perhaps not so amorphous, but rather displays some tendencies toward privileged ways of formation. However, on the same page

946

Sign

Continuum

Substance Form � Si gn Expression Form � Function Substance

Continuum

Hjelmslev stresses that it is susceptible of "any formation whatsoever," is in itself inac­ cessible to knowledge since the prerequisite for knowledge is an analysis of some kind and every analysis is an arbitrary organization of it. Hjelmslev thus outlined a sort of idea of a thing-in-itself that at the same time recalls Peirce's Dynamical Object, without, however, answering all the philosophical questions that his model of the sign function inevitably (and fruitfully) raised. U.E.

Sign Typology It is generally considered difficult and controversial to distinguish typology from taxonomy (and classification). Taxonomies consist in terminological distinctions which can be described by ,:•graphs, showing their hierarchical order. In semiotics, these distinc­ tions are either meant as pure nominalistic suggestions (distinctio rationis), underlying discussions of sign conceptions (''models), or designed as a reconstruction of relations determining their mutual dependence. The crucial criteria for setting up taxonomy how­ ever, ultimately stem from a particular research interest. A typology, on the other hand, may also be conceived in its original meaning derived from typos. In this case, it is the intuitively conceivable ideas (archetypes) which partition the area of the subject (the ontological region) "sign, " without the ideas being neces­ sarily connected. It is, for instance, still controversial if natural signs (e.g., index) and conventional signs (e.g., symbols) do have anything essential in common besides being both media of noticing. Most of the proposed sign typologies - almost every author in semiotics up to now makes his own distinctions - stem from a mixture of taxonomic interests and an intuitive conception of ideas. Very often the question whether the class distinction is dual (e.g., natural versus conventional signs) or trichotomic (e.g., index, icon, symbol in Peirce) hinges upon underlying general philosophical assumptions. Hence there are two possible extremes: ( 1 ) a sign typology may claim to just register all perceivable kinds and forms

Sign Languages of the Deaf

947

of signs in a purely empirical and inductive manner; or, (2) every single semiotic concept (as it is found in this book) may become the basis of a typology (and if so, it seems diffi­ cult not to identify sign typology with semiotics in general). According to this second way, signs would be distinguished as to their different aspects according to which they are object- or metasign, elementary or complex sign, whether they are types or tokens. From this it is almost self-evident that the existing typologies overlap and show cross-classifications. It is, however, remarkable that the sum of the above-mentioned empirical typologies and the sum of all theoretical and philosophical typologies gradually seem to converge. Whereas it seems rather obvious that there exist different signs, it nevertheless seems difficult to explain why there are so different kinds of signs. The first question, in this context, is whether this plurality of different kinds of signs does or does not have an upper limit in particular, whether their number is finite or infi­ nite. In the latter case, a higher-dimensional semiotic process space might suggest itself as a suitable representation, and each of its points would correspond to a certain kind of the semiotic process. These kinds of processes would continuously merge into each other without having sharp boundaries, as in taxonomies. A typology, however, which arbitrarily distinguishes many types of signs would be uninteresting. Only typologies containing a limited number of types are important. Hence the essential and important types constitute a smaller set than the class of all pos­ sible types. This reasonable assumption calls for treatment by *information-theoretical terms. Information is the quantitative relation between a given set of distinctions and the set of all possible distinctions. The assumption made above is therefore equivalent to the assumption of '�redundancy. The existence of types of signs means that semiotic pro­ cesses show an effect of repeatability, i.e., redundancy. Information-theoretical redun­ dancy implies, under the aspect of production of signs, an invariance of the semiotic pro­ cess. Hence the definition: sign typology studies the typical invariants of the semiotic process. If the concept of typology is extended from a single sign to systems of signs, as in language typology, higher-level invariances will appear. Typical invariants in the system will be repeated in different places and thus there is a redundancy of invariants, or an invariant-invariant. Types which are invariant in all signs or systems of signs represent semiotic universals. H.v.d.B.

Sign Languages of the Deaf It is important to note at once the plural nature of the subject. Contrary to popular belief, sign languages of deaf people do not constitute a single universal ,:•language. National and regional groups of deaf people - even local groups if they are isolated each have their own sign language. What these sign languages have in common is the use of vision to receive and bodies to transmit the •:•signals expressing their *structures. They differ, just as spoken languages do, in �·lexicon, in '�grammar, and in the selection and use they make of the common fund of visible actions that constitute the "phonetic" inventory, in this case, of sign languages, gestural systems, and much of what is termed ' ":•nonverbal •:•communication." Sign languages differ from spoken languages in numbers of users ; about one to one thousand, respectively. They differ quite obviously in transmission systems; sights, not

948

Sign Languages of the Deaf

sounds furnish the sign vehicles for signed languages. (The term signed not only comes closest in form to its counterpart spoken but also avoids possible ambiguity with the semiotic term •�sign.) Signed languages of the deaf also have a characteristic pattern of acquisition as first language. For more than a century, demographers have reckoned the proportion of deaf per­ sons to persons in large populations as one to one thousand, when the term deaf denotes inability to hear speech from birth or early childhood. Counting those deafened after acquiring spoken language and those who have lost hearing acuity with aging makes the fraction larger (Schein and Delk 1974), but it is likely that the number of speakers is about one thousand times the number of signers. Although early profound deafness and membership in the population of signers destines a person for lifelong sign language use (in the sense that hearing people are lifelong speakers), only about one out of ten such persons will have a deaf parent or parents and so have ab natu exposure to signed instead of spoken language. Nine out of ten deaf signers consequently acquire their sign lan­ guage not from parents at home but from other signers at school or elsewhere and at a later stage in life than usual for first language acquisition. In semiotic perspective, sign languages of deaf people may be seen as sign systems with the familiar semiotic organization into sign vehicles, sign denotata, and sign rela­ tionships. The sign vehicles of signed languages used by deaf groups are not different in raw material from several other systems of signifying and communicating. Sign languages, gestures, gesticulation, kinesics, �·body language, coenetics, face-to-face interaction, and nonverbal communication all have sign vehicles composed of bodily actions. How much or how little of the sign producer's body is involved, how large or extensive or abrupt or striking the changes in appearance may be, and how unobtrusive they may be made and still serve as signs are matters far from determined. The combination of these actions is also something not yet thoroughly studied. Early investigations of sign lan­ guages of the deaf focused on actions of the hands. The arms are, of course, always involved, but the literature now speaks almost entirely of hands (see, e.g., Friedman 1977; Klima and Bellugi 1979; Wilbur 1979). More recently (Baker 1980a, 1 980b; but cf. Stokoe 1960, 1972), the actions of eyes, faces, and heads have been explored as sign vehicles in signed languages. The most general characterization of these sign vehicles (Stokoe 1980) is that in them something having distinguishable structure acts, and that structure-in-action, what is perceived as a sign language, " . . . can exist nowhere except in the imagination . . . " (Shands 1979 :117). The characterization of sign language signs takes account of principles at once phy­ siological and semiotic. As Shands points out, different receptor cells in the human retina receive different kinds of visual information. �·Action or motion is perceived as it occurs by peripheral cells, but seeing structures, especially fine details in structure, requires an image focused on the central foveal cells and so requires that the object be motionless. The human organism's intimate (proprioceptive, �·visual, •:•tactile, kinesthetic) knowledge of its own signaling structures enables a signer to say what signed language sign another produced, and to describe it. Only in the imagination, as Shands says, is it possible to see structure-in-action at once. Therefore, in a general "phonetics" of signed languages - something much to be desired but not yet available features of the sign vehicle will be found either to belong to the structural aspect (called dez for designator in Stokoe 1960) or to the kinetic aspect (sig, for significant action :

Sign Languages of the Deaf

949

Stokoe 1 960). If the sign is a unilateral nose wrinkling, an actual sign in American Sign Language, no specification of structure is needed other than that it is the part of the face nose denotes. If the sign involves a hand-arm unit, as a great many signs of American Sign Language and other signed languages of the deaf do, then it becomes necessary to specify both the details of the handshape (i.e., the degree of Hexion/extension of all the fingers and of adduction/abduction in the thumb) and the hand, wrist, forearm, and upper arm positioning. (Attempts in the literature to treat this positioning as "orienta­ tion" have been signally unsucessful, because it is the state of the natural system at its joints, and not the directions in abstract space that the palm or fingers are pointing, which determines these features of the structure acting.) The sign denotata of signed languages of the deaf are typically linguistic; i.e., they are whatever persons use their language to talk about, ask about, speculate about - in short, put into language. Consequently, the locus of the denotata is neither the individual consciousness nor the external universe, but the shared �·culture. In this case, it is the cul­ ture those deaf persons who share a signed language share as competent signers and members of a community. Differences between what the linguistic signs of spoken and signed languages denote have therefore to be sought in ethnography more than in psy­ chology. Because deaf persons use a different perceptual system - vision instead of audi­ tion - to capture information cast in linguistic form, it is most likely that there will be dif­ ferences between the cultures and subcultures of deaf and hearing groups. Information about the culture of deaf communities is beginning to appear (e.g., Baker and Battison 1980c, Proceedings NSSLRT 1 977, 1 978, 1 980), so that the universe of discourse, or Umwelt, of sign languages of the deaf may soon be compared with that of spoken lan­ guages and language. It is in the relation of vehicle to denotatum, of signal to �·meaning, that sign languages of the deaf show the most extensive semiotic identity with spoken languages. Both are marked by duality of patterning (Hockett 1978). Both have phonetic level and feature level, as has already been pointed out. Like words also, the signs of these languages are distributed into different form and function and �·semantic classes. Supalla and Newport (1978) first discovered the subtle but regular differences in the action aspect of signs that makes otherwise identical signs nouns or verbs in American Sign Language. Pedersen and other associates of Klima and Bellugi ( 1 979) showed how verbs of American Sign Language could be modified in meaning, "modulated" by changes in the brachial-manual action (not by entering phrase structures with other lexical units, as in many spoken languages). In the same study, they also describe the involvement of other than brachial-manual actions in the production of the modulated verb sign. At the level of phrase composition, signed languages make a striking departure from the norm of spoken languages. Two lexical signs of a sign language can be, and often are, executed at the same time by both arm-hand structures acting independently (e.g., right hand signs "never" as left hand signs "see," to give "I never saw that"). Even more frequent are signed "phrases" of an arm-hand sign executed simultaneously with a non-manual sign, as in many negative and emphatic constructions of American Sign Language. Apart from the lexical simultaneity, which is impossible for unaided speech, the �·syntax of a signed language of the deaf proves, as research continues, to be quite similar to the syntax of spoken languages. Liddell ( 1 980) showed that subordinate structures in American Sign Language could indeed be distinguished from coordinate sequences of American Sign Language ,;sen-

950

Sign Languages of the Deaf

tences. He thus refuted a long-held belief that sign languages could not have grammar or syntax. Part of the reason for the negative estimate of signed languages is doubtless the circumstance that in English there are markers to show that a sentence has been transformed in the embedding process (e.g., "The man who phoned is here" ; "The dog that chased the skunk came home"), while in American Sign Language the markers are non-manual and consequently overlooked by observers ethnocentrically expecting signing to be exclusively manual. Baker and Padden (1978 ) found that eyeblinks virtually always appear at a clause boundary within complex American Sign Language sentences. Padden (1981) has dem­ onstrated that different kinds of complement structures in American Sign Language result from the working of syntactical rules that can be made explicit. Research on the syntax of American Sign Language, and the sign languages of deaf populations in Bri­ tain, France, Denmark, Norway, Sweden, and other countries proceeds, even as courses in American Sign Language are being taught at more than one hundred United States colleges and universities, many of them courses in the structure and grammar of the language. In addition to complying with semiotic criteria for language (sign vehicle, denotata, and relations), sign languages of the deaf furnish information of value to a number of other sciences. The earlier use of gestures than voices by infants to point at and otherwise signify objects and events, and the much earlier use. of symbolic (context-free) signs than words by children in sign language surroundings supports modern anthropological searches for language origins (Hewes 1974, 1975, 1977a; Hamad et al. 1975; Hockett 1 978). Infor­ mation about sign languages of the deaf and the communities in which they are used also assists anthropologists investigating language evolution (Washabaugh 1981 ). Stu­ dies of the use of signed and spoken languages are shedding more light on the involve­ ment of left and right cerebral hemispheres in language operations (D.M. Armstrong and Katz 1981 ). Social psychology, ethology, and other sciences of behavior are finding new and valuable information in descriptions of sign language structure. Language is the key or master system in the matrix of cultural systems analyzed by Hall and Trager (Hall 1959). Language description and linguistic theory in the twentieth century have contributed both information and new paradigms to the study of other cul­ tural systems. There have been brilliant practitioners of linguistic science, of course, but surely this fostering role of linguistics owes much to the nature of language as the most intricate, extensive, and highly organized of cultural systems. Now that sign languages of the deaf are being investigated more and more thoroughly, it is likely that a whole group of st1:1dies will profit, again because a sign language used as primary language by a community of deaf persons is the most intricate, extensive, and highly organized system that uses visible behavior as sign vehicle material. Thus, studies of nonverbal communica­ tion, gesture, gesticulation, facial expression, face-to-face interaction, proxemics, kines­ ics, etc. will look to descriptions of signed languages (especially of their "phonetic" lev­ el), and to the linguistic theory they entail, as models to explain phenomena not so fully linguistic. In the realm of science applied to human needs, sign languages of the deaf have to contribute, if possible, even more. As the world's population grows larger, the needs of one one-thousandth of the people can hardly be ignored or denied or perversely disre­ garded, as they have been in the past. The social and economic well-being of the mil-

Signal

951

lions of deaf persons in the world is inextricably linked with the languages they use both by preference and out of necessity. This truth seems to be grasped intuitively in small, isolated communities that have both larger than usual incidence of deafness and a simple economy and social structure (Kuschel 1973; Woodward 1978). Technologically advanced societies developed in the nineteenth century the nai:Ve faith that science would eliminate all ills. This became doctrine, and improvements in medicine and surgery, in telephony and electronics, and in speech pathology and therapy were confi­ dently expected to extirpate deafness and its effects. As part of this doctrine, however, sign languages were banned as a pernicious obstacle to the goal of having deaf people learn to speak, to lipread, and (in the recent rise of the "unisensory method") to hear. While some progress in preventing and treating deafness itself has occurred medical science reduces the deafness that once resulted from childhood diseases, and surgery corrects some hearing disorders of certain specific kinds - genetic research has shown that increasing population growth inevitably brings increased numbers of con­ genitally deaf persons. Since 95 out of 100 children born deaf do not have even one deaf parent, the eugenicists of the nineteenth century, including A. G. Bell and E. M. Gal­ laudet, and their modern fellow travelers, like surgeons in Britain performing secret ster­ ilization of deaf women in hospitals, are backing a false hope. On the positive side, interest in sign languages of the deaf and research supported by the United States National Science Foundation, National Institutes of Health, and private foundations have resulted in several advances. Interpreters able to translate competently from American Sign Language to spoken English and from English to American Sign Language are being more effectively trained and increasing in number. (Their numbers are still small, however, compared with those able only to put English they hear into signs for words or to fingerspell the words.) At the same time, medical, legal, and educational situations in which deaf persons find themselves are becoming more and more amenable to the participation of skilled interpreters. Thus, all kinds of opportunities for deaf persons to participate more fully in the culture of hearing persons around them are increasing (Stokoe 1979). Even in education, that most conservative social institution, there are changes related to sign languages of the deaf. A few schools in the United States are beginning to plan and implement bilingual, bicultural programs for deaf children. These use written English as the second or target language and American Sign Language in all its many varieties, natural and contrived, as the source of first language. This movement, if prop­ erly implemented, calls at once for the employment of more educated deaf persons as teachers. Schools in Denmark and Sweden began such bilingual educational experi­ ments with a spoken and a signed language some years earlier (Hansen 1 980; Bergman 1981 ). w.c.s.

Signal Signal is an expression now widely used in technology, biology, the humanities, and social science. Sometimes signal appears as a mere synonym of ,:• sign, '' symbol, and ,:•mes­ sage. More often, however, signals are identified with specific interactional units, which are described as elementary, substantial, and operative. These cause immediate reac-

952

Signal

tions in men as well as in animals and automata (cf. �-cassirer 1 944). There is no unanimity, however, as to further peculiarities of signals and their place within a semiotic typology. In post-Bloomfieldian linguistics (which commonly ignored the sign concept), sig­ nal, defined as a physical (e.g., acoustic) event that delivers and discriminates •:•meanings (Hockett 1 958 : sec. 1 3.3; Pei 1 966), is seen as a universal ''communication medium. In the Saussurean tradition, signals are seen as particular realizations of abstract elements of a *code (Prieto 1 966: ch. 3 ), or as highly idealized units reduced to minimal ''opposi­ tions within a code (Russo 1 958). In the Bi.ihlerian sense, signal (as opposed to ''symptom and �·symbol) relates to one of three "sematologic dimensions" of a sign. According to *Bi.ihler's organon model of language, a sign serves to denote "Things," but the manner in which it has been produced may also reflect emotions experienced by the Speaker and provoke certain responses in the Hearer (Bi.ihler 1 934 : 28-30, 35 ff.; Sebeok 1 962: 433-34 ). When the latter tendency is dominant, the sign becomes a signal (Bi.ihler 1 934 : 3 1-32, 46 ). This "triggers some reaction on the part of a receiver" (Se­ beok 1 97 Sb). While a tradition based upon everyday speech use suggests that signals are a kind of accidental signs which, in a given situation, have been chosen by a closed group of men to control activities of its members in a manner unintelligible to others (Schaff 1 964 : sec. 2, ii, 4, A), or they are signs which pertain to asyntactic codes Qakobson 1 970: 1 3-1 4), etc. From a mechanistic point of view, signals are feeble impulses which control large amounts of energy (Haldane 1 955: 385), or they are physical phenomena homomorphous with their sources (Resnikov 1 968 ), etc. In technology, the signal is regarded as any physical event which can propagate through some medium (Spataru 1973 ). Signals must be �-information-bearing entities which have either discrete (digital) or continuous (analog) form (W.R. Bennett 1 970). Signals vary in space or in time (Carlsson 1 968). Signals are capable of inducing excitation in some system. Such changes, once induced, may be, in turn, treated as (output) signals to be recorded by the observer (Stewart 1 960; Lynch and Truxal 1 962; Marshal 1 965). Sometimes signals are conceived as cer­ tain abstract patterns which may undergo various modulations (Goodyear 1 97 1 ). The mathematical theory of signals ordinarily deals not with the original physical phenomena but with their symbolic representations, such as ''/unctions (Oppenheim and Schafer 1 97 5) or vectors (Milen'kij 197 5), which are called signals as well. Signals are frequently discriminated on biological grounds. A classical physiological approach views them as temporary stimulators which release conditioned reflexes (Pavlov 1 94 l b : 1 23, 2 1 4). After Morris (197 1 : 1 25, 1 27), signals are preparatory stimuli that provoke "overt behavior" in the �·addressee. Bloomfield (196 1 :sec. 2.2-2.5) holds that (speech) signals are speech stimuli (e.g., the exclamation water!) that stand for prac­ tical stimuli (thirst). In �·Benveniste's opinion (1966a), signals belong to the sensorimotor aspect of human activity, while symbols pertain to its conceptual and representational aspect. Signals may be described in etiological terms (Pazukhin 1 972). Here signal is thought of as a specific stimulus, which provides for no cause-and-effect link between it and the automatic reaction it releases. Such signaling is based upon the occasioning, a mechanical action which is mediated by special "sluice" -transducers and which thereby constitutes neither simple causation nor data transmission. R.P.

Signifier I Signified

95 3

Signifier/Signified Signifier and signified are the terms introduced by ,:-Saussure (1916:101 = 197 4 : 1116f.), to indicate "sound image" (image acoustique) and "concept" (concept) as elements of a linguistic �·sign. They are reminiscent of the Stoic terms semai'non and semain6menon, differing from them, however, by the fact that they do not refer (as the Stoics' notions do) to the third element (the Stoics' tugkhdnon 'object'). On the other hand, the unity of the sign - fundamental for Saussure - does not appear in the Stoic model. The mode and interrelation of these terms make it possible to outline the essential points of the Saussurean theory of sign : ( 1) Signifier and signified exist only reciprocally as elements of the sign ( signe) in relation to signification. They are elements of the science of signologie (or semiologie [ = semiotics ] ); (2) Signifier and signified are mental entities determined by a sign and not fixed with respect to any material substance: thus, the substance of a signifier is accidental and sub­ stitutable ( e.g., sound may be substituted by writing); and it is not the positive form but only the negative difference from other signifiers that is determinant. Signified does not represent an object, . but classifies it within a sign, and its range is determined differentially by the co-occurrence of other signifieds; (3) Sign, signifier and signified are thus system-bound and, moreover, are entities in a network of relationships (oppositions of signs as positive entities, differences in signifiers and signifieds as negative entities). They are values which may be compared to values in '�economics; ( 4) Sign, signifier and signified are semiotic, i.e., they are applicable for all systems of signs and not only to language. There are a number of further characteristics and criteria concerning the delimita­ tions of signifier and signified and the principles by which they are related; (5) qualification of the relation between signifier and signified as natural, or symbolic, when the signifier in some way reflects the signified by means of expressiveness of sub­ stance or by structures of form; as arbitrary, when no such correspondence exists; as motivated, when a sign recurs in other signs of the same system. In each case the relation­ ship is understood by Saussure as conventional, i.e., subordinate to social sanction; (6) Differentiation of the mode of existence of signifier as unispatial (linear) as in the case of auditory signs, or as multispatial (multidimensional), as in the case of visual signs or gestures; (7) The assignment of sign to the virtual, social code (langue), whereas in performance (parole), a mere ,:•signal - subordinate to signifier - is transmitted (Engler 1975c); (8) Attribution of sign to synchrony, whereas in diachrony a dissociation and shift of the relation between signifier and signified may occur; these may originate in changes in the determination of signifier, a different identification of signified or a new disposi­ tion of values carried by a sign, the result of multiplication or reduction. In Hjelmslev's terminology, signifier corresponds to "form of �·expression," signified to "form of content." For Prieto ( 1966), signifier is a class of signals which have the same signified, while signified is a class of bits of information transmitted by one signal, i.e., the utility of a signifier. The application of signifier and signified to material sign­ carrier and object, as sometimes occurs in paralinguistic *structuralism, is a misuse of the terms. R.E.

954

Skeptics

Skeptics Classical Greek skepticism arose in reaction to the conflicting claims of dogmatic philo­ sophers to knowledge of the true nature of things. The Skeptics tried to show that: ( 1 ) for any arguments supporting any philosophical thesis thus far advanced, equally good countervailing arguments against that thesis are available; (2) therefore, it is best to suspend judgment (epokhe) and neither affirm nor deny theo­ retical explanations of the sensory presentations, feelings, and involuntary reflex pre­ sumptions which make up our ordinary daily experience; and (3) the goal which the Skeptic seeks is peace of mind and freedom from disturbance (ata­ raxia), and it has been his experience that freedom from disturbance does in fact super­ vene upon suspension of '�judgment. Pyrrho of Elis (ca. 368-270 B.C.) was the first to make doubt concerning both per­ ception and reason the basis for morality and a way of life. However, Pyrrho left no writ­ ings and founded no school. The major arguments for theoretical skepticism were devel­ oped by Arcesilaus (ca. 315-240 B.C.) and Carneades (ca. 219-120 B.C.), both of whom were heads of the Platonic Academy. Arcesilaus directed his arguments chiefly against Zeno and the early Stoics, attempting to show that each of their major theses leads to self-contradiction. He claimed, and the claim itself appeared dogmatic to later and more radical skeptics, that '�truth cannot be attained, but that error can be avoided if we make suspension of judgment our prime goal. Carneades developed a more mitigated skepticism by formulating an important, original, and subtle theory of probable '�belief. Aenesidemus (last century B.C.) gave up the mitigated Academic position in favor of a radical skepticism, which he called Pyrrhonism. He formulated ten tropes or modes of skeptical argument, and applied skeptical modes of argumentation to dogmatic claims to knowledge of causes. It was probably Aenesidemus who systematically organized the counterarguments to the Stoic and Epicurean theories of signs reported by Sextus Empiricus. No writings of the major Greek Skeptics have survived, and Sextus Empiricus is a major source for our knowledge of classical skepticism in general and the critique of '�sign theory in particular. Other important sources include �-Cicero's Acade­ mica, and Diogenes Laertius' account of Pyrrho in his Lives of Eminent Philosophers. The skeptical critique of signs is searching and ingenious. Sextus Empiricus is careful to point out that the arguments are directed primarily against the Stoic theory of indic­ ative signs, since it is these that are claimed to give knowledge of what is naturally non­ evident and non-observable. However, the Epicureans also were included in the skeptical critique, since they held that signs were sensible, and that from the sensory evi­ dence of motion the existence of the non-evident void could be inferred. As to the signs which the Stoics called commemorative (smoke as a sign of fire, lactating breasts as a sign of recent pregnancy), these are essential to the normal conduct of our daily lives, and in that non-speculative and non-theoretical function the Skeptics do not reject them. The following is a summary of some of the most important skeptical arguments against signs as supporting claims to theoretical knowledge.

Skeptics

955

Argument 1

If something is a sign, it is so only as it is related to something of which it is a sign. Sign and signatum are correlatives like right and left, lighter and darker. Now, either the sig­ natum is apprehended along with its sign or it is not. ( 1 ) Suppose the signatum is not apprehended along with its sign. Then the sign cannot be apprehended as a sign, for correlative terms exist as such only as both are appre­ hended. Without the right there is no left, without the darker there is no lighter, and without the signatum there is no sign. (2) Suppose, then, that the signatum is apprehended along with its sign. But a sign exists as such only to reveal what is otherwise not evident. If the signatum is already appre­ hended it needs no sign to be revealed. So whether or not the signatum is apprehended along with its sign, the sign cannot exist. Argument 2

According to the Stoics, a sign is something evident which reveals what is non-evident. It is also generally agreed that a sign is either a sensible or an intelligible. Sextus Empiricus offers as examples of sensibles white, black, bitter, sweet, hot, and cold. Examples of intelligibles are fair, foul, lawful, lawless, pious, impious. ( 1 ) Suppose, now, that a sign is something evident. Then it will be evident whether a sign is a sensible or an intelligible. But there is an old and apparently irresolvable con­ troversy as to whether signs are sensibles or intelligibles. What a sign is, then, is not evi­ dent. (2) Suppose that the sign is not evident. Then, to know whether something is a sign a further sign will be needed. Since, according to the present hypothesis, that further sign will itself not be evident, we are caught in a vicious regress. Hence, whether a sign is or is not evident, a sign cannot exist. Argument 3

Since it is generally agreed that signs are either sensibles or intelligibles, let us see first whether they are sensibles. ( 1 ) Suppose signs are sensibles. But there is an old dispute among philosophers whether sensibles exist independently of human sensation. Evidently we do not know whether there are independent sensibles. If signs are independent sensibles, a fortiori we do not know whether signs exist. (2) Suppose independent sensibles do exist. They are simply apprehended, without teaching. But signs are taught and learned. Hence, signs are not sensibles. (3) Again, suppose signs are sensibles. Sensibles affect normal observers in similar cir­ cumstances in similar ways. Signs do not affect normal observers in similar circumstances in similar ways. For example, some physicians do not recognize a state of the �·body (hot, cold, flushed, pallid) as a sign at all, while others disagree about what these sensible states signify. So signs are not sensibles. ( 4) Suppose signs are sensibles. Then they cannot be of use as signs. For consider that any sensible sign signifies several different possible signata. For example, a flushed face will be a sign of several quite different possible bodily states. It cannot signifiy all its sig-

956

Skeptics

nata simultaneously, since some are incompatible and cannot co-exist. But if it is a sign of one rather than another of its possible signata, the sign will require yet another sign to single out its particular signatum. But that further sign will itself signify several different possible signata, and will require yet further signs, and so we are involved in another infi­ nite regress. Hence, if signs are sensibles, they are of no use as signs. Argument 4

Suppose signs are intelligibles. The Stoics maintain that signs are lekta, the intelligible communicable content of what is said, and lekta are non-corporeal intelligibles. (1) There is an old and irresolvable dispute over whether there are non-corporeal lekta. Not only the Epicureans deny their existence, but even a leading Stoic by the name of Basileides says that lekta are corporeal. Hence, if signs are non-corporeal intelligibles, we cannot know that they are. (2) Suppose it is granted that signs are known to be incorporeal intelligibles. According to the Stoics, what is incorporeal can produce no effects. But the Stoics also say that in indicating something and making it evident a sign does produce an effect. The Stoics then contradict themselves in saying that a sign is an incorporeal intelligible. (3) Suppose it is granted that a sign is an incorporeal intelligible proposition. But according to the Stoics, what is signified (semainomenon) is incorporeal, while the '�signi­ fier (semainon) is corporeal, like the physical vibrations in utterances. Hence a sign (se­ meion), if it is a signifier of anything, is corporeal. W. and M. Kneale (1962: 142) point out that Sextus ignores here the difference in terminologies reflecting differences of �-theory: "The relation between semainon and semainomenon is that between language and what it expresses, while the relation between semeion and semeioton is that between what is known first and what is known through it." (4) The Stoics say that a sign is a true antecedent in a true conditional �-proposition. (For example, if the skin perspires, then the skin has invisible pores.) But among Stoic logi­ cians there are at least four quite different theories concerning the truth of a conditional proposition. If there is no agreement about when a conditional proposition is true, there is no agreement and no knowledge concerning signs. (5) While the conditions under which a conditional proposition is true are disputed, there is no disagreement that a conditional is false when its antecedent is true while its consequent is false. Now, according to the Stoics, the consequent of a true conditional is the signatum of the sign, and the signatum must be non-evident. Since the signatum is non-evident the consequent (i.e., the signatum) may be false. But if the consequent is false and the antecedent (i.e., the sign) is true, the conditional proposition as a whole is false. In that event, there is no sign, for a sign is a true antecedent in a true conditional and any consequent (i.e., any non-evident signatum) may, for all we know, be false, thus making the conditional as a whole false. (6) Finally, either animals recognize signs or they do not. It cannot be maintained that animals do not recognize signs. Even the Stoics agree that when a dog tracks his prey and when a horse jumps at the crack of a whip these animals are recognizing the scent and the sound as signs. On the other hand, the Stoics cannot consistently say that animals do recognize signs. Signs are antecedents of conditional propositions, and to understand a conditional proposition one must understand transitivity, logical connec­ tives, and logical consequence. Only human beings and no animals recognize these

Sociology

957

logical factors, and hence (for the Stoics) signs follow from the nature of man but not from the nature of animals.

Argument 5

It should be added that since signs cannot be sensibles and cannot be intelligibles, they cannot be both sensible and intelligible. D.S.

Sociology

The domain of sociology ranges from the conduct of everyday life through the structure of bureaucracies and villages, to nation states and, recently, to the interaction of nations in the modern world system. The social principle at the base of this vast domain is arguably a sociosemiotic principle in that no social life of the human sort is possible if we deny humans the right to create, exchange, and interpret •�signs. Other claims, which base society on biological or economic necessity, assume the priority of socio-'�semiosis. There can be no marketplace, for example, outside communities which can create and exchange values, a purely semiotic capacity. Sociologists tacitly recognize the importance of '�language, �·communication, exchange, expression, symbolism, value, etc., as essential components of the human social order. But they prefer to study the gross morphological features of society and association (population dynamics, power, dominance, coercion reciprocity, deviance, the effects of technological change and class conflict), excluding from their analyses consideration of the sign character of these phenomena. [Several noteworthy excep­ tions to this tendency are discussed below, e.g., symbolic interactionism and *ethnome­ thodology.] Sociology (along with •=·economics and political science) is among the last of the human sciences that have not been declared subfields of •�semiotics. At about the same time that modern academic sociology came into being, around 1 900, Ferdinand de *Saussure and Charles S. �·Peirce claimed linguistics and logic to be subfields of semiot­ ics and backed their claims with extensive scholarship. By the 1 950s and 1 960s, Levi­ Strauss, •�Barthes, •�Lacan, and Foucault had performed the same operation for �·anthro­ pology, literary criticism, *psychoanalysis, and �·history. But there has been no equivalent movement within sociology.

The Nonintellectual Base of Sociology

In the 1 950s, Talcott Parsons ( 1 95 1 ) attempted to unify sociologists under the banner of ' "�structural functionalism." Since that time there has been no general effort within sociology to strive for intellectual unity. Subfields of interest and effort within sociology center on particular areas of empirical study, e.g., rural society, small group research, social problems, race relations, etc. Some powerful insights and distinctions recur in the subfields, and many useful concepts (role, norm, class, institution, etc.) have been pro­ duced by these insights. But there has been little institutional follow-through in the the universities, professional journals, and associations to build a consensus around the definition, importance and meaning of sociological concepts and observations. Today,

958

Sociology

sociology is not unified after the fashion of a science or a scholarly discipline. Its unity is more like that of one of the complex systems it studies. It is held together at a structural or unconscious level, and thrives on the interplay of internal differences between sub­ fields, special theories, and topics of inquiry. Two oppositions internal to sociology are consequential for the determination of the place of semiotics in social inquiry and theory: (1) European versus American approaches; and (2) individualistic versus structural explanations of social phenomena. The Europe/American Opposition In the United States, sociology was developed as a "science of society" to produce increasingly accurate empirical descriptions of social behavior and organization. Western European critical and Marxist versions of sociology produce their interpreta­ tions of social life in opposition to prevailing political '�ideologies and historical forms of popular and/or class consciousness. The difference between European and American sociology has been a main source of the integration and continued development of the field. The opposition of the two approaches provides ongoing insight and perspective that is external to the national '�culture but internal to the field. In each recent genera­ tion of scholars a minority obtained training on the other side of the Atlantic, or other­ wise developed a keen appreciation for what was sociologically current on the other side. For example, Talcott Parsons ( 1 937) built "American" structural functionalism on a foundation of his German graduate education and his reading of European sociological classics by Vilfredo Pareto, Max Weber, and Emile Durkheim. The novelty of structural functionalism is in its requirement that Americans view their national cultural situation the way Europeans have long viewed their own, as a social system. Sim­ ilarly, ethnomethodology is a combination of European *phenomenology and American ethnographic empiricism. In making this combination, ethnomethodology raises European questions of consciousness and existence in the context of close studies of everyday behavior. Other examples which could be cited include Max Weber's field visits to rural American Protestant communities, and Levi-Strauss's fascination for New World ethnography. Given the more advanced state of development of semiotics in Europe, and sociology in America, any awakening of semiotics in sociology will neces­ sarily continue this pattern of combination. The Individual versus Structural Opposition In addition to the European/American difference, which has produced a stream of soci­ ological insight, each sociological school and subfield is internally divided along the same axis: the individual versus social structure. Sophisticated sociologists object that the individual versus group or structure is a false dichotomy or a pseudo-problem, but all would agree that it is still the basic opposition that operates within sociology. While it may be theoretically artificial, this particular opposition between the individual and the group produces a form of intellectual exogamy between the subfields of sociology which operates to keep the entire general field together. Individualist approaches locate the causes of social change in the psyche, motivation, and intention, while structural approaches locate cause in institutions, the relations between social classes, rural-urban dynamics, center-periphery, and so on. An individ­ ual/structural opposition can be found in even the most rigorous sociological schools

Sociology

959

which, on the surface at least, claim to embrace only one (structural or psychological-be­ havioral) position. For example, Marxist sociology is structural. Still, some Marxist soci­ ologists advance essentially individualistic perspectives against the grain of their own theory. This deviation from the logic of a particular school leads some scholars to look outside their school for methods, alternative lines of theory, and collegial understand­ ing. In so doing, they contribute to the creation of higher unities out of originally disparate subfields. Both the United States/European connection and intellectual exogamy produced by the individual/structural split are evident in the studies of Mar­ cuse (1964) and Adorno (1 950, 1 967). Marcuse and Adorno combined American-style social psychology and European Marxism in their critique of modern culture, the authoritarian personality, and other analyses of the individual in advanced capitalist sys­ tems. On the other hand, sociologists such as Homans (1950), who have embraced psy­ chological explanation, make use of social concepts (group, class, community, etc.) to distinguish their work from that of their colleagues in the field of psychology.

Semiotics and the Individual/Structural Opposition The division of sociology into individualist versus structural approaches is important from the standpoint of semiotics in that the structural side of sociology always and of necessity contains a semiotic of social relations, at least in prototypical form. Individual­ istic sociologies may contain a semiotic, and many do, as a supplement, but, insofar as they base their idea of society on assumptions of psychic unity or biological unity of humankind, no semiotic is strictly speaking required. The basic assumption of structural sociology is that society is something "more than" or "other than" a mere aggregation of individual tendencies and characteristics. When developed in detail, "more than other than" are forms of communication and interaction and associated patterns of values. Semiotics refer to the same forms and patterns as semiosis. As suggested above, several sociological schools have formed around a semiotic conception of their subject matter, and have consistently interpreted social phenomena in terms of its sign-character. These include dramaturgical sociology (Goffman 1 959; Burke 1 969), symbolic interactionism (Mead 1 934 ; Blumer 1 969), phenomenological sociology (Schutz 1 964; Berger and Luckmann 1 966), ethnomethodological (Garfinkel 1 967), and structural symbolics (Young 1 970). Oddly, these schools have tended to re­ strict their analyses to everyday life, remain peripheral to the general field of sociology, as it conceives itself, and are negatively evaluated by it. The relationship of semiosis to social structure on a macro-level has eluded sus­ tained sociological investigation. The restriction of considerations of sociosemiosis to micro-studies of face-to-face interaction follows from a view of language, communica­ tion, thought, consciousness, �·belief, etc., as essentially individual-level phenomena which are important sociologically only insofar as they are involved in the formation of the social character and personality of individuals. In holding to this belief, sociology is merely participating in the broader system of Western quasi-philosophical ideas which equate humanism with individualism. While each draws from it different conclusions, the idea that collectivities, social structures, and communities are anti-individualistic or anti-humanistic is basic to both vulgar Marxist and Western bourgeois thought in the twentieth century.

960

Sociology

Interestingly, the great semiot1c1an Peirce, and the great sociologist Durkheim, devoted much of their energies to a critical attack on the false opposition of individual/ structure. Both searched for the highest achievements of humanity within the community. It is a paradox in the history of ideas that sociologists have largely ignored (even ridiculed) Peirce's and Durkheim's solutions, which involved the displacement of thought and consciousness to the community level. Sociologists have toyed with the idea that language, mind, and social structure may not be distinct realms. But there is no sys­ tematic research designed to link them, with the possible exception of some studies in ethnomethodology. However, at the present time, a latent parallelism between lan­ guage, mind, and social structure has become so strong that concepts such as differentia­ tion, network, involution, etc., have complementary and non-exclusive definitions in lin­ guistics, sociology and phenomenology. Still, few scholars, operating within discipline boundaries, are aware of the conceptual relationships which have evolved naturally and spontaneously in studies that are unrelated except via mutual participation in the grossest movements of modern thought.

The Pre-Semiotic Debate with Individualism Structural or proto-semiotic sociology originated in continental critical philosophy in the writings of •:•Rousseau, •:•Kant, and �·Hegel, and was eventually developed as sociology by Marx and Durkheim. According to this perspective, there are essential dif­ ferences between communities, classes, historical types of society (e.g., primitive and modern), and cultures. These are not conceived as mere differences but as the most basic dimensions of social structure and personality, and the causes of social change. For example, according to one of the more influential early hypotheses of this school, Marx claimed that a division of society into classes can take the form of a structural con­ tradiction which plays itself out in revolutionary and other activities leading to the crea­ tion of new forms of society and consciousness. Eventually, Levi-Strauss (1962b ), would specifically combine Marxism and semiotics, and generalize Marx's formulation to all social divisions (not merely that between the classes) and all types of society : primi­ tive as well as modern, capitalist, industrial, etc. According to Levi-Strauss, structural oppositions within society produce •:•myths and *symbols which interact with social structure, sometimes serving to reinforce it (e.g., the myth of white supremacy) and sometimes undermining it (e.g., the myth that there is a correlation between effort and social reward). In short, according to Levi-Strauss' famous slogan, "myths are good to think with," or structure expresses. Within individualistic sociologies, structural expressions, myths, etc., are held to be less important than individual efforts to maximize the fulfillment of needs and desires through behaviors that are conceived as either egoistic or altruistic from a utilitarian perspective. Within structural sociology, on the other hand, questions of how individuals and groups communicate, express themselves, arrive at agreements, organize their con­ flicts, understand themselves in relation to others, etc., is problematical and central. The structural problematic presuppose a semiotic analysis or some equivalent under a different name. In short, semiotics occupies the same place in structural sociologies as assumptions of psychic unity occupy in individualistic sociologies: only semiosis is less an assumption than an invitation to additional analysis. One of the first things structural

Sociology

96 1

sociologists discover is that social communication and expression have requirements of their own which go beyond the social functions they are serving. This is the central dis­ covery of the several sociological schools (symbolic interactionism, ethnomethodology) which have systematically studied the sign-character of social phenomena, for example. In the historical progression of ideas, the structural-semiotic stream of sociology emerged in opposition to the individualist paradigm. With all their political and metho­ dological differences, Herbert Spencer, Durkheim, and Marx are in agreement on one matter: an intellectually rigorous study of society cannot be based on a metaphysic of desire and individual progress. Such notions of "human nature" that are uninformed by cross-cultural knowledge are easily shown to be unconscious projections of prevailing (Western) economic and religious systems. Perhaps the simplest way to sloganize this structural insight is to say that "morality cannot be reduced to utility." Only recently have these principles, which originally rested on a purely intellectual base, received pow­ erful backing and justification in the rise of the Third World as an important force in global social relations. The failure of the Western individualistic paradigm to be general­ izable to the rest of the world is now fully exposed. It is noteworthy, in this regard, that Third World sociology faculties were among the first to be completely subsumed within the semiotic movement. While structural and critical thought helped to establish sociology as an intellectually independent and vital modern field of study, and continue to help align it with the cur­ rent general framework of Third World development and semiotics, these ideas are not replacing earlier individualistic thinking. As already suggested, the modern field of soci­ ology remains divided along structural versus individual lines. In one early �·hermeneutic moment, and by application of an almost superhuman effort, the great German sociologist, Weber (1 958), fully integrated both streams in his classic study of The Protestant Ethic and the Spirit of Capitalism. One could argue that the subsequent development of modern academic sociology has been based on an effort to prolong this moment. Weber himself, in his later studies, eventually Parsons and, some would claim, Harold Garfinkel, have attempted to advance the sociological synthesis of individualism and structuralism. There is growing concern within the intellectual community, however, that the individual/structure split and all its variants cannot continue to be the opposition that generates sociological knowledge, except in some mythic form. An alternative appears to be emerging within structural sociology in the form of a clearer understanding of the opposition between semiosis and social structure. New approaches to sociological subject matter are forming around the question of which is dominant, semiosis or structure: do groups produce communications, expressions, symbols, etc., or do expressions symbols, etc., produce groups? (See, for example, D. MacCannell 1976, 1979; Lemmert 1 979.)

Marx and Durkheim - The Proto-Semiotic Stage

Marx's theory of economic and social values is still the best example of analysis which links semiosis to social structure, holding structure to be the most basic or causal. According to Marx, in a capitalist society, a person's labor (and, by association, the per­ son) can have value only when it is invested in the creation of a commodity and then, only when the commodity is entered into the social system of exchange. This system,

962

Sociology

which determines the meaning of life in modern capitalist societies (i.e., social rank, eval­ uation of types of work, quality and desirability of different communities, the appropri­ ateness of behavior, etc.) is entirely "unnatural" in that it is fully capable of generating social and economic values without reference to any external standards. In short, the system of work, commodity production, exchange, and the measure of human worth under capitalism in most essential respects resembles a language. Both the social system of commodity production and a language are arbitrary and capable of generating •>meaning in and of themselves. Marx (1965a:74) wrote: "value . . . converts every pro­ duct into a social hieroglyphic ... .(W]e try to decipher the hieroglyphic; to get behind the secret of our own social products; for to stamp an object of utility as a value, is just as much a social product as language." Even as he laid this groundwork for a first semiotic of social and economic values, Marx struggled to subordinate his semiotics to his analysis of social relations, i.e., the contradictions inherent in the class system under capitalism. "Struggle" is an appropriate word choice in this instance. In all of Marx's writings, and in some of the Marxisms that followed, this particular intellectual struggle parallels the struggles for self-determination they posit for the international proletariat. The products of sociose­ miosis (the opiate of the masses, the fetishism of commodities, false consciousness, the superstructure) are perceived by Marx and Marxists as ascending over the social order while materialist doctrine suggests it ought to be vice versa. Marx's general theory of alienation is a by-product of the way he conceived the opposition of semiosis and social structure: human beings produce things, commodities and images, e.g., the image of the holy family. Then we subordinate ourselves to the things and images we have created. To Marx, this seems a useless subjugation. In sum, in Marxist analyses the semiotic order ascends over the social order, but this ascendancy is seen as alienating, antirevolu­ tionary, and ultimately inexplicable. In the later writings of Durkheim (e.g., The Elementary Fonns of the Religious Life, 1912 and 1965) we find a similar basic opposition of the semiotic and social orders. But Durkheim accepts the ascendancy of the semiotic over the social as inevitable and bene­ ficial rather than inexplicable and alienating. According to Durkheim, social differences recognized at the group level, as between men and women, classes, age grades, clans, political leaders and followers, etc., are the models for our symbolic forms. All our capacity to categorize and distinguish in terms of fact and fiction, truth and lies, good and evil, sacred and profane, etc., is, according to Durkheim, linked to the basic social divisions and ultimately designed to maintain these divisions and the functional fit that exists between them. Durkheim (1965 :264) wrote: "social life, in all its aspects and in every period of its history, is made possible only by a vast symbolism." According to this thesis, not merely magic and religion, but rationality, logic, *mathematics, the various conceptions of time and space, what Durkheim called "our priceless instruments of thought," are by-products of social structure.

Recent Socio-Semiotic Contribution - Mead, Goffman, and Barthes George Herbert Mead claimed that behavior, the mind, the human personality, and society itself originate in semiosis. According to Mead's radically semiogenetic perspec­ tive, the symbolic meaning a gesture has for others is what m ot ivates its production.

Sociology

963

Mead's followers have had some difficulty linking their symbolic interactionism to the concerns of mainstream sociology. Goffman was the first writer to integrate social organizational and explicitly semiotic concepts. Although he makes little reference to the semiotic scholarly tradition, the analytical units in his respected (1959) study of The Presentation of Self in Everyday Life are "signs," "sign vehicles," "signs given," and "signs given off," etc. Goffman's contribution has been to make semiotic and/or dra­ matic norms co-equal with social norms in his analyses of behavior. For example, in his studies of deviance, Goffman shows the way criminals and other miscreants often embed or attempt to embed their socially abnormal behaviors into dramatically normal routines as illustrated by the behavior of a bank robber waiting his turn at a teller's win­ dow. He goes on to show us how the concrete ways in which humans do their deviating (as well as the ways they do their normal routines) is possible only under conditions where there are two codes operating on behavior: (1) the social-moral code and (2) the dramatic-semiotic code. In addition to behaviors which are dramatically correct while being morally incorrect (waiting in line to rob a bank), there are behaviors which are normal from a social standpoint but dramatically weird, e.g., adjusting a contact lens. In Goffman's later writings, in particular Frame Analysis (1974 ), he suggests some possible forms for the relationship between social and symbolic orders, specific ways, for exam­ ple, that one can become a model for the other. The larger movements of structuralism and semiotics have already begun to produce sophisticated studies which openly address the displacement of structure, consciousness, the emotions, language, the self, etc., which is bound to occur within an advanced sociosemiotic perspective. Foremost among these studies is the sociological writing of Barthes, in particular his analytical statement "Myth Today" and the other essays in Mythologies (1972b). A schematic outline of the relationship of signs to social structure, and an implied program for future sociosemiotic research, can be glimpsed in the writ­ ings of Goffman, Barthes, and others. Conclusion

Signs and social structures have a parallel evolution in the course of which each influences the other in specific ways. Signs in their original form are metaphorical, and they establish a fictional order alongside the social order, a fiction which resembles social life in some ways while diverging into the psyche and the imagination at key points. In the first stage of the co-evolution of signs and structure, social relations are disorganized, unsettled, and potentially dangerous. Under these conditions, signs and language express the passions: love, fear, hate, revenge. As society becomes progressively more complex, organized and orderly, language and the other symbolic forms become more regular, rational and literal. We develop systems of signs which stand for ideas and are specifically de-signed not to express the emotions. The language of modern science is an example. But the expression of the emotions never completely disappears. It is only radically transformed. Complex social structures separate true social intimates (e.g., family members) and at the same time forge a widespread pseu­ do-intimacy of strangers in bureaucracy, other workplaces, at public events, through the media, etc. Under these pressures, expressions of love and fear and of the imagina­ tion become non-referential aspects of the conceptual order: images which are pro­ duced and manipulated in systems that have no connection with their roots in simpler

964

Somatosemiotics

social situations. Our social condition is determined by the operation of this double dia­ lectic of social simplicity and complexity, and the semiotic opposition of psyche and con­ cept. There is a clear program for social research in the sign/structure relationship, but it is far from assured that this research will take place in the field which calls itself "soci­ ology." Sociology "contracts out" virtually all the research it might be doing on sociose­ miosis to community studies, ethnic studies, women's studies, folklore, popular culture and urban ethnography. And there is little evidence from within sociology of any move­ ment to regain centrality among the sociological subfields under a semiotic or any other banner. D.M.

Somatosemiotics Somatosemiotics is a term suggested for a discipline which should describe the sign pro­ cesses which combine *endosemiotic sign processes with sign processes of higher com­ plexity. Somatosemiotic sign processes are, for instance, received during the discharge of reflexes which are responsible for our bodily posture and its integration in the environ­ ment. They are the basis of what originally was termed by Head ( 1 9 1 1 ) and Schilder ( 1 923) bodyschema. "We must assume that the single sensations (received from the dif­ ferent parts of the body) are not retained without any order in the memory, but are com­ bined into a comprehensive image of the body, into a Schema, after the term of Head" (Schilder 1 924 : 65 ). This body-schema can be observed most impressively in patients after the amputation of a limb. They experience the missing limb as if it were still a part of their body (phantom limb). Later on, the term body-schema has been replaced by the term body image and has acquired a more psychological meaning. Fisher and Cleveland understand the body image as a psychological experience . . . The assumption is that as each individual develops he has the difficult task of meaningfully organizing the sensations from his body which is one of the most important and complex phenomena in his total perceptual field. It is a phenomenon relatively more complex and difficult to organize perceptually than other pheno­ mena "out there" because of its unique simultaneous role as a participant in the perceptual pro­ cess and also as the object of this same perceptual process . . . ( 1 968 :x)

This statement underscores a quality of somatosemiotic signs which is of central impor­ tance : the reflexive attitude of the receiver. This reflexive quality is also demonstrated in the concept of the "Impulse-to-operation-sign" by J. von *Uexktill. This sign informs the subject about his own impulses to his muscular organs (effectors) . The underlying principle has been described on the basis of neurophysiological research by von Holst and Mittelstaedt ( 1 950) as the Reaf/erenzprinzip. T.v.U.

Speech Acts When *Morris proposed his trichotomy of *Syntax, �·semantics, and �·Pragmatics, he defined the latter as "the study of the relation of signs to interpreters" ( 1 938 : 6 ), but

Speech Acts

965

then (1938: 29) generalized this to "the relation of signs to their users." One year later, *Carnap (1939: 4) proposed to "call pragmatics the field of all those investigations which take into consideration . . . the action, state, and environment of a man who speaks or hears [a linguistic sign]." This tradition continues. Both linguists and philo­ sophers (see Akmajian, Demers, and Harnish 1979; Gazdar 1979a) have taken it to cover the study of *language •:•use in relation to context, and, in particular, the study of speech acts. What is a speech act, and what are the different kinds of speech acts? On these ques­ tions current theorists disagree. •�Wittgenstein, reacting in part to his own earlier views on the nature of language, asks, rhetorically : But how many kinds of sentence are there? Say assertion question, and command? - There are countle55 kinds : countless different kinds of use of what we call "symbols," "words," "sentences." And this mutiplicity is not something fixed, given once for all; but new types of language, new language-games, as we may say, come into existence, and others become obso­ lete and get forgotten. ( 1 953 : sec. 23)

Wittgenstein's views on language have been, as we will see, enormously influential. The first serious, systematic answer to Wittgenstein's question seems to be the work of *Austin (1962a). In his series of William James Lectures delivered at Harvard in 1955, Austin was initially concerned to distinguish •·utterances which count as stating some­ thing from those that count as doing something - the so-called performative utterances. As Austin remarks (1962a :6), "When I say, before the registrar or altar, etc. 'I do,' I am not reporting on a marriage : I am indulging in it." However, Austin noticed that the distinction between stating something and doing something collapses, since stating is a kind of doing and can also be marked with the performative prefix "I (hereby) state that . . . " Consequently, in the later lectures he began to work out "the senses in which to say something may be to do something, or in saying something we do something" (Austin 1962a :91). He distinguished three main types of speech act: acts of saying some­ thing with a definite sense and reference (Locutionary acts); acts performed in saying something (Illocutionary acts); and acts performed by saying something (Perlocutionary acts) : *UTTERANCE: I 'll pay you back LOCUTION: He said that he would pay me back ILLOCUTION : He promised ( warned me) that he would pay me back PERLOCUTION: He convinced me that he would pay me back In his last lecture Austin proposed a taxonomy of illocutionary acts consisting of five categories. Although he tried (Lectures VII-X) to theoretically define and distinguish these various speech acts, he was never successful. As Searle (1968, 1975a) noted, locu­ tions carry some broad delimitation of illocutions with them, and no consistent princi­ ples for categorizing these as acts was provided. Furthermore, as Strawson (1964) noted, by overly emphasizing the conventional and ritual character of some illocutions, Austin was unable to formulate the distinction between illocutions and perlocutions in terms of the intentions, *beliefs, etc. of speakers and hearers. The first theorist to articulate both the conventional and the intentional aspects of speech acts was Searle (1969), who distinguished four major types of speech act: utter­ ance acts, propositional acts, (with subsidiary acts of referring and predicating), illocu-

966

Speech Acts

tionary acts, and perlocutionary acts. Utterance acts are simply acts of uttering expres­ sions from a language, and perlocutionary acts are acts in which the speaker has a cer­ tain effect on the hearer (such as persuading them). Searle focuses most of his attention on propositional and illocutionary acts. For Searle, speaking a language is engaging in a form of intentional, rule-governed behavior (think of Wittengenstein's language games), and a theory of speech acts must exhibit these *rules, such as the following ones for promising, in uttering some expres­ sion Pr (such as / promise to A): PROPOSTIONAL CONTENT RULE : Pr is to be uttered only if A is a furture act predicated of the speaker PREPARATORY RULE: Pr is to uttered ony if (1) A is in the hearer's interest, and (2) the speaker can do A SINCERITY RULE: Pr is to be uttered only if the speaker intends to do A ESSENTIAL RULE: uttering Pr counts as undertaking an obligation to do A. He also shows that this pattern of analysis can be extended to many different kinds of illocutionary act, as well as to the propositional acts of referring and predicating. On the basis of this analytical scheme, Searle (1975a) proposed a five-fold classifi­ cation of illocutionary acts : REPRESENTATIVES : represent some state of affairs (asserting, classifying, etc.) DIRECTIVES : attempt to get the hearer to do something (ordering, requesting, etc.) COMMISSIVES: commit the speaker to do something (vowing, promising, etc.) EXPRESSIVES: express a psychological state of the speaker thanking, apologiz­ ing, etc.) DECLARATIONS: bring extralinguistic states of affairs into existence (vetoing, appointing, etc.). Although Searle ( 197 Sa) does suggest that there is a correlation between some of these kinds of illocutionary acts and certain syntactic structures, so far no one has investigated the connection between form and function in any detail using Searle's framework. Such work was carried out in the related framework of generative semantics and went by the name of the Higher Performative Analysis (HPA). In the late 1 960s, a number of linguists (see, e.g., Ross 1970; Saddock 197 4) noticed that simple declarative sentences such as ( l a) share many syntactic properties with so-called "explicit perfor­ mative" sentences, such as (lb) (called "explicit performatives" because the main verb explicitly indicates the speech act being performed in the literal utterance of the sen­ tence): (1) (a) Prices slumped (b) I state to you that prices slumped Since it was also thought that such pairs of sentences were synonymous, these facts could be easily explained by deriving ( 1 a) from ( 1 b) via a Performative-Deletion *transforma­ tion. It was then noted that nondeclarative sentences such as (2a) and (2b) share syntac­ tic, semantic, and pragmatic relations with explicit performative sentences such as (3a) and (3b ), respectively:

Speech Acts

967

(2) (a) Pass the salt! (b) How old are you? (3) (a) I (hereby) request you to pass the salt (b) I (hereby) ask you how old are you. Consequently, it was proposed that sentences (2a) and (2b) also be derived from sentences such as (3a) and (3b). In the light of these results it was proposed that every sentence has a performative clause in its underlying representation, and this is what is meant by the Higher Performative Analysis. Such a view would give an interestingly uni­ fied picture of the syntax, semantics, and pragmatics of these sentences, but one may seriously doubt wheather pairs of sentences such as ( l a) and ( l b) are synonymous. Syn­ onymous sentences have the same entailments, but (4) is entailed only by ( l a) and not by the explicit performative ( 1 b). ( 4) Prices fell

Both Searle's theory and the Higher Performative Analysis have serious problems (see Harnish 1 977, 1 979; Gazdar 1 979a), and the root of many of these problems is the tendency to tie ''meaning too directly to (illocutionary) use. Both theories assume that ,..communication is successful by virtue of sharing a language, and, if the speaker and hearer both know the meaning of the expression, this will suffice for successful com­ munication. But this is not so (see Akmajian, Demers, and Harnish 1980). According to a more recent view, performing speech acts and thereby communicating successfully is basically a form of intended inference (not constitutive rules), and Bach and Harnish ( 1 979) propose various inferential strategies, shared by speaker and hearer, which, if fol­ lowed, will allow the hearer to recognize the speaker's communicative intent and so result in successful communication. These strategies not only allow for '-·ambiguity and inderterminacy of reference, but they also give a systematic and homogeneous account of various ways of performing speech acts. For instance, we can speak literally (meaning what we say) and directly (not doing anything indirectly), as in examples ( l a), (2a), and (2b). But we can also speak nonliterally, illustrated by (Sa) and (Sb), as well as indirectly, as illustrated in (6a) and (6b). (5) (a) SARCASM : That's really a brilliant idea ! (b) METAPHOR: Sleep that knits up the ravelled sleeve of care (6) (a) LITERAL: I want ten gallons of regular (directly a report, indirectly a request) (b) NONLITERAL: I bet the cat likes being poked (directly a nonliteral statement, indirectly a request). Bach and Harnish ( 1 979: ch. 1 1 ) attempt to find psychological evidence in favor of their system of communicative strategies and also build a taxonomy of illocutionary acts com­ prising over one hundred cases. An attempt is also made to connect the speech acts underlying language use with a systematic analysis of basic social notions. But, as the topic of speech acts is barely twenty years old, we can expect considerable development in the next few years. R.H. All that we do in and by speaking is a speech act. The units of communication are not sentences or even ''texts, but speech acts. This means that analysis of expressions should be abandoned in favor of speech acts. Thus, with the most essential features of

968

Speech Acts

human communication brought into focus, the theory of speech acts comes to stand in apparent opposition to the prevalent referential and ideational theories of meaning. The latter assume that the most important trait of the basic means of communication (lan­ guage) is its relatedness to extralinguistic and extramental entities or its relatedness to mental ideas. Speech act theory stresses the activities of the speaker and de-emphasizes the relatedness of expressions to extralinguistic entities. The beginnings of the new approach to the meaning of expressions are to be found in Wittgenstein's Philosophical Investigations ( 1953 ), where the meaning of a word is viewed as determined by its use. Strawson ( 1 950) points to the difference between an expression and its utterance, coming close to the notion of the speech act. The utterance of an expression and not the expression itself becomes the real bearer of meaning. Austin asks what in fact we do in and by uttering an expression, thus founding the theory of speech acts. Austin's views can be roughly summarized as follows.

Parts of Speech Acts

In uttering and by uttering an expression we perform a speech act which consists of three components : a locutionary (propositional in Searle 1968) act, an illocutionary act, and a perlocutionary act. Consider the uttering of (1) (2) (3)

Will John leave the room? John will leave the room. John leave the room !

What is common to all three utterances is the locutionary act. It consists of the phonetic act of uttering sounds, the phatic act of treating the produced noises as ordered by grammatical rules, the rhetic act of referring to John, and the act of predicating his leaving the room. By contrast, asking ,;questions and ordering are illocutionary acts. The differences are due to different illocutionary forces. The perlocutionary act con­ sists in producing some effects on the hearer; e.g., (3) may frighten John and ( 1 ) addressed to John's host may embarrass him. Perlocutionary force i s the effect, here : fear or embarrassment.

Delimitation of Illocutionary Acts

Different authors offer varying, and at times conflicting, characterizations of the three components of the speech act. Austin identifies some locutionary forces (reference and predication) with meanings which, however, differ from illocutionary forces. Locutionary acts, in his view, are only necessary conditions for illocutionary acts. Thus, depending on circumstances, one locutionary act may be associated with many illocu­ tionary acts. Searle ( 1 968 ) reduces indirectly illocutionary forces to locutionary forces, i.e., to meanings. He adopts the so-called expressibility principle to the effect that "what can be meant can be said . " In addition, he assumes that an utterance of a sentence has only one illocutionary force (Sadock 1974). Illocutionary forces may be expressed with the help of performative verbs, e.g., warn, advise, order, ask, promise, etc. All three assumptions amount to the claim that sentences whose illocutionary force is not repre­ sented in surface *structure may be embedded in sentences whose main verb is a suitable performative verb. The whole clause is equivalent to the embedded sentence; e.g.,

Speech Acts

969

"Leave the room !" is equivalent to "I order you to leave the room" ! This conclusion has been combined with generative semantics, which assumes that all semantic compo­ nents of an expression are represented in their deep or logical structure. Thus, every illocutionary force of an utterance of an expression has its representation in a suitable logical structure (Ross 1 970). The paraphrase or logical structure I order you to leave the room! of the order "Leave the room !" has no truth value, according to Austin, but D. Lewis (1 972) puts forward the opposite opinion. Illocutionary forces are thereby reduced to locutionary forces of their performative expansions, which are treated in the frame of intensional semantics. This performative analysis has at least one merit. The order "Leave the room" differs in meaning from the question "Who leaves the room?" According to the generative hypothesis, they have different underlying logical struc­ tures, and hence transformational rules can be meaning-preserving. If only the locutionary force of both examples had underlying structure, the rules transforming it into two different surface forms would change meaning.

Per/ormative Analysis The main arguments which support this view are grammatical in nature. One of a set of arguments is closely connected with the analysis of pronouns. Reflexive pronouns can be used meaningfully when they have an antecedent. But we often use sentences in which there is no visible antecedent to the pronoun : "This is a picture of myself." When this sentence is embedded in a higher clause with a performative verb, then the antecedent I becomes visible: ' 'I say this is a picture of myself." This expansion has the logical struc­ ture of the surface sentence devoid of the antecedent. The hypothesis of the existence of abstract performative higher clauses explains the correctness of sentences exhibiting reflexive pronouns without antecedents, and is thus corroborated to some extent. The performative analysis faces some criticism (Stampe 1 975). An utterance of a sentence is to be synonymous with its performative expansion (it does not matter whether on the surface or in the underlying structure). Thus, the statement "This snow is grey" is to be synonymous with "I state that this snow is grey." But "This snow is grey" may be false in the case it is actually white, and "I state that this snow is grey" may remain true when the speaker is convinced of the truth of the embedded sentence and when additionally we admit that explicit performative sentences can have *truth value. How can two sentences remain synonymous when they have different truth values?

Delimitation of Perlocutionary Acts Perlocutionary acts are taken to be intended or unintended, while illocutionary acts are always intended. Perlocutionary acts are unconventional. Different utterances may have the same perlocutionary force in the same circumstances. Thus, perlocutionary force does not depend on linguistic conventions. Contrarily, illocutionary acts are conven­ tional. One utterance may have more than one perlocutionary force; an utterance may both frighten and prepare the hearer for the danger. An utterance of an expression can have only one illocutionary force. What is more, perlocutionary forces cannot be made explicit, just as is the case with illocutionary forces. Thus, "I'll break your neck" may frighten the hearer, but this force cannot be equivalently rendered in "I frighten you that I'll break your neck."

970

Speech Acts

In spite of these differences, the delimitation of illocutionary and perlocutionary acts and forces is by no means clear. This deficiency is not only due to the vagueness of the notions of convention and intention. Let us consider the notion of suggestion. The acts of warning, promising, advising, stating, asking, ordering, etc. imply the act of saying. If one warns he must do it by uttering words. Is it always so? If all illocutionary acts imply saying, then suggesting is not an illocutionary act. As a rule, suggesting is indirect, without explicitly saying what is suggested (Stampe 1975). No wonder, then that Sadock (1 974) assumes that the set of illocutionary forces is a subset of perlocutionary forces. Both kinds of forces are effects of utterances on the hearer.

Speech Act and Pragmatics The theory of speech acts falls obviously into the domain of pragmatics as defined by Morris (1946 ). Speech acts are produced by speakers, and the theory of speech acts has to take users of language into consideration. Pragmatics, in Morris's sense, is the domain where the relations between language and its users are studied. However, the relation of speech act theory to pragmatics, in *Montague's view, is not obvious. Montague takes intension to be the reconstruction of the prescientific notion of meaning. Intension, in turn, is the function from circumstances of utterance (speaker and �·possible worlds included) into extensions, that is, sets of objects, relations or truth-values. Montague does not consider the acts of the speaker. However, this function may be conceived of as an act creating or determining the extension, since action can be treated as a function which, in the circumstances given, yields only one object. Thus, there is a possiblity of reconciling intensional semantics and pragmatics with speech act theory, at the cost of enlarging the set of the arguments of the intensional or pragmatical function. Within this pragmatical framework, the differ­ ence between the two basic conceptions of the speech act becomes visible. Thus, according to Austin, illocutionary forces depend on locutionary acts and forces. Hence, the speech act may be conceived of as a function, one argument of which is the locutionary force (reference and predication). The outcome, i.e., the value of the func­ tion, is illocutionary force. In contrast, Alston( 1964) seems to be suggesting that meaning is dependent on illocutionary forces. In this case, illocutionary forces are among the arguments of the function (here the speech act), with reference and predica­ tion making up the value of the function. Although Alston's view can be cast in pragmatic terms, it calls for a general reconstruction of the ways meaning is determined.

Other Problems Speech acts pose many problems which have been left out of account here. Austin recog­ nized about one thousand performative verbs in English. Are there as many illocutionary forces, or are there just a few of them, which can be combined and thus make up one thousand complex illocutionary forces or more (Sadock 1974)? There is also the problem of classifying illocutionary forces (and thus speech acts). An attempt in this area has been made by Searle ( 1975), who introduced twelve inde­ pendent principles used for the division of illocutionary forces. Another problem is the question of cirumstances in which illocutionary forces fail or do not fail, i.e., in what circumstances a speech act is felicitous or infelicitous. If a

Spet

971

little boy says : "I declare the Olympic games open," this utterance is infelicitious, since the boy is not entitled to open the games (Austin 1962a). A vast area of problems is connected with the relation of the speech act theory to different conceptions of meaning. Only one of them - Grice's theory - has been partially exploited by the speech act theory (e.g., Searle 1965). L.K.

Spet, Gustav Gustavovic (1879-after October 1937) A Russian philosopher and phenomenologist, born and educated in Kiev, Spet came to Moscow University in 1907. In 1910 and from 1912 to 1913 he studied in Gottingen, under *Husserl. In 1916 he defended his thesis, "History as a Problem of Logic." In 1918 he became professor at Moscow University, in 1921 a member, and from 1924, Vice-President of the Russian (State) Academy of Sciences of the Arts (until 1929). In 1935, Spet was arrested and exiled; he was arrested for the second time in 1937. In his analysis of intentional acts (acts of perception directed toward the ,:•object which was posited by them), Spet, following Husserl, shows that the very relating of the act to the object is realized through the sense which constitutes an element of the struc­ ture of an object perceived. Developing this idea further, Spet affirms that an intentional act is "thoroughly permeated" with theoretical comprehension ; a thetic act is "devoid of soul" unless it is accompanied by the *hermeneutic act, given that, if the positing is correlated with the sense of an object, the hermeneutic action would be corre­ lated thereby with the entelechy of the object, which defines (determines) its teleology. Hence: The being of reason consists in the hermeneutic functions which establish the reason-motiva­ tion emanating from the entelechy as "carrier" of the object's being, understood as the "spirit of the object. " The features of this spirit are to be found in the logos "its expression", which penetrates the object and makes up the manifestation, "the revelation," "the incarnation" of its spirit. Its "objectification," being rational . . . is an organizing directedness of the various forms of the spirit taken in their social essence: ''·language, religion, art, technology, *law." (Spet 1 9 1 4 : 209-10)

In this way, semiotic ,:•structure is ascribed to reality itself in its manifestation and expression, and language is revealed as the primary model of this structure. Spet rejected the triad "thing - concept - word," which presupposes a "wordless" logic of concepts separated from language, and treats the signified of the verbal sign (the inner form of the word) as the actual object with its ,:•meaning, not, however, in the full­ ness of this meaning but with a certain selection of object-content, a selection which con­ ditions the specific character of languages and ,:•cultures. In place of conceptual logic Spet proposes the "dialectic of realized cultural meaning, " the "hermeneutic dialec­ tic, " that is, the study of the relationships of object to cultural �-signs. From this point of view, the word is "the universal sign which can substitute for any other sign," it is "the ontological proto-image of any cultural-social thing." For this reason too, language is the primary sign system. "Manifestations of cultural consciousness such as art, science, law, etc., are not new principles but modifications and forms of a single cultural con­ sciousness which has in language its archetype and origin" (Spet 1927a:37). The princi-

972

Sports

pies of sign organization are the basis of living philosophy (cf. Dilthey). Hence, the study of various sign systems (in art, ''literature, pedagogy, etc.) is the main factor in the study of both the structure of consciousness and the structure of the cognizable, since these systems mediate between man and the reality he conceives. On this point Spet is close in his views to •·Cassirer, with the difference, however, that the symbolic sphere for Cassirer is the domain of philosophical reflection, while for Spet it is the layer of meaning which links consciousness and reality. Spet was the first of the phenomenologists to speak of the historical nature of mea­ ning-structures (cf. the later works of Husserl). This approach follows logically from the notion of the cultural and social nature of comprehension. Sign systems are endowed with an immanent dynamism, while their evolution is regarded as a creatively teleological process (of comprehension). Hence ''history too, as a factor in the develop­ ment of cultural consciousness, can be looked at as the evolution of sign systems. Among those deeply influenced by Spet are "-Jakobson, V.V. Vinogradov, and G.0. Vinokur. Spet took part in the Moscow Linguistic Circle, and his ideas find reflection in the notions of •·Prague •:•structuralism and of the branches of •·semiotics connected with it (the relativity of the opposition •·synchrony/diachrony, the linguistic •:•compe­ tence of the linguist as the basis of his scholarly work, notions of poetic language, of language in relation to other semiotic systems, etc.). V.M.Z.

Sports The Symbolic Character of Sports As studies of the subject concerned indicate, there seems to be general agreement that all the •·actions of sport are symbolic in character. Sports competitions are seen above all as representative of events, states or ideas, as the case may be; as symbolizing the prin­ ciple of individual performance, man's struggle for excellence, mankind's existence, the manner in which society functions, or modern '' myths ("·Barthes 1 957; Moser 1 960; Slusher 1 967; Weiss 1969; Vanderzwaag 1 972; von Krockow 1 972, 1974; Lenk 1972, 1 979; Adam 1 975). Yet another approach emphasizes the *aesthetic character of the sign system in the world of sport : it is a world that, within a specific ''structure of space and time, creates its own reality and thus its own ''meaning ("·Merleau-Ponty 1 942; Bouet 1968; Magnane 1 964). This meaning is interpreted in various ways: one reading sees it as taking the form of myths that are over and above the world of sport as such - myths that provide a uni­ versal rationale within a popular •·culture (Magnane 1 964), or all-embracing myths of reconciliation and deliverance (Barthes 1 957) - while the other sees it as a meaning which reaches its full development in sport alone, this being brought out, for example, in its dramatic elements (F. W. Keenan 1 975; Arnold 1 978; Masterson 1 98 1 ) and in its principle of supremacy. Taking the latter approach, Frayssinet has put forward a radi­ cal, thought-provoking interpretation : sporting activity gives rise to an aesthetic pro­ duct which in itself has the character of a work of art - the athletic cruvre characterized by a specifically aesthetic perception, a specially created world, a mode of existence at once sensual and transcendental.

Sports

973

All the above-mentioned approaches see sport as a means of ,:•communication or expression. Up to the time of writing, however, no attempt has been made to establish whether, if at all, sporting actions are a suitable means for communication or ,:•expres­ sion. Does sport, whatever form it takes, actually posses the kind of *sign structure that makes communication or expression possible? By making use of the conceptual system built up by Goodman (1968), it is possible to answer this question.

Does Sport Constitute a Language ? Each form of sport presents a system which, viewed as a sign system, is syntactically characterized as follows: (1 ) the use of marks common to more signs than just one in other words, the signs are not disjunct; (2) signs that are not finitely differentiated, it being in principle always possible to range a sign between any two others. It is not pos­ sible to decide whether a mark belongs to one particular sign or any number of others. Owing to these two characteristics, then, the sign structure of each form of sport is "syn­ tactically dense." It has the following �·semantic characteristics: ( 1 ) it is not unambigu­ ous, presuming the sporting action concerned is not "semantically separated" ; (2) it has no finite semantic differentiation, it not being possible to decide whether any one action is compatible with a particular sign but not with others, and (3) there is an overlapping of the various classes of compatibility. Owing to these characteristics, then, the actions of sport are "semantically dense," fulfilling none of the conditions stipulated by Goodman for a notation system. It is true that, in contrast to the actions of sport, verbal *language is semantically dense but syntactically disjunct and finitely differentiated. Only a musical score fulfills all five conditions. The actions of sport can be compared not with *music or speech but with *painting, which is also syntactically and semantically dense. Like a painting, a sporting action is irredeemably bound up with its own realization through the perfor­ mance of a human �·body. The sensuous quality of the body's actions in sport is a constit­ uent part of the sign system of sport. Verbal language is syntactically disjunct and finitely differentiated precisely because it involves no bodily activity. Of its very struc­ ture, the written word precludes bodily action. The constituent link between sport and body, with its resulting syntactic density, has a bearing on the sporting sign's qualities of communication and expression: as it is not possible from the sign structure alone to decide which of the body's actions can be considered as signs, it has been left to the interpreter to stipulate which of the aspects of the action are to be taken as meaningful. The arbitrariness of such a procedure strongly detracts from the value of any interpretation arrived at in this way. By its very structure, the system of sporting actions cannot of itself produce a series of purposeful, successful or unsuccessful performances. But such exist and are, in fact, decisive in the perception of sporting actions. The ability to interpret sporting actions in a form acceptable to all concerned is not, however, something that is generated of itself from the sign system of sport. The actions of sport do not constitute a *text in the way, for example, that dramatic events do in the world of drama (cf. F. W. Keenan 1 975; Arnold 1 978 ; Masterson 1981 ). It only becomes actuality in performance. For this reason, such actions are essentially subject to the hazards bound up with bodily movement of any kind. On the other hand, most of the

974

Sports

factors involved in sports competitions are in themselves independent of hazards (con­ stitution, talent, mental attitude, physical condition, physiological, and anatomical effi­ ciency). However, the necessity of such factors is continually called into question by the above-mentioned hazards to which a body is subject when actually engaged in a sports competition. It is the play between hazard and necessity which creates the tension so characteristic of any competitive sporting events, a tension that bears no comparison with that to be found in drama or in games of chance.

Discourse on Sport Sporting actions exist in interpreted form at the level of a description or account of the same (a report, a commentated recording, excerpt or photographs). Such representa­ tions - including �·film presentations of the actions - bear the fundamental stamp of a discourse on the actions in question. The intention of the bodies vying with each other and the success that follows are defined within and through the discourse. The World of Sport as Created by the Discourse The problem of defining which of the actions is to be considered as a sign is solved at the level of discourse: it is the verbal language that defines the signs of sporting actions. In this way, semiotic analysis of, for example, competitive sport actions is confined to oral or written reports or commentaries, in other words, the events concerned are repro­ duced in linguistic form. This does seem justified however, not only because direct expe­ rience of sport lends itself but little to semiotic considerations, but also and above all because it is sport as reported through the form of discourse built up from a series of meaningful individual events that is absorbed into everyday culture. A competitive sporting event does not develop its special spatiotemporal structure, its "own reality," except when it is constructed as such in a discourse (cf. the notion of sports activities as construction: Gebauer 197 3; Franke 1978). Barthes bases his analysis of the Tour de France exclusively on the verbal descriptions of the event; the discourse that he reveals at this level is in the form of a literary •=·genre - the epic. The discourse on sporting actions transforms the physical aspect of signs in sport into verbal ones. In this respect, sport is not a language of the body. A parallel is to be found in any attempt to analyze art : just as the use of language to describe or reconstruct syntactically dense signs makes the elements of a painting meaningful without being able to replace them, the use of a discourse to express one's perceptions of sporting actions is bound to leave untouched a significant "surplus" of sensous impressions. Bodily Expression There have been many special discourses on sporting actions, but it is possible to discern a number of factors common to them all. In the discourses, the contending bodies are described from the point of view of what is expressed by those bodies. The observation of bodily expression is a fundamental factor for the spectator. Competitive sporting actions are to the highest degree organized in a "panoptic" (Foucault) manner (the indoor arena, the stadium, the athletic circuit, the indoor swimming pool complex, and,

Sports

975

above all, television coverage), allowing the spectator to keep the contending bodies under constant observation from various sides. The basic expressive functions of a body involved in sporting action are as follows (Gebauer 1 98 1 b) : ( 1 ) the individual body's realization through its physical activity of the •�rules of the sport concerned, of social norms and conventions (or a transgression of the same) ; (2) the individual body's actions in respect to the partner/adversary; (3) the specific treatment by the individual of his own body and that of his adversary; and (4) the individual's assertion of his personal identity over that of his adversary. Body Roles On the basis of the modes of expression attributed to him in the discourse, the individual competitor receives a definite body role. He can influence the nature of this role; there are bound to be certain reciprocal effects, but he has no control over the formation of this role. It is not determined biologically - although, wherever possible, physical factors are reinterpreted as actual attributes of that role - but only by the competitive qualities of the body and the power of the discourse: the underdog can prove the stronger, the lamb becomes the destroyer, the weakling humbles the proud one, etc.

The Discourse on Competition and Personal Identity The body roles become apparent in competitive interaction. The form taken by the interaction is determined by the structure of the discourse on competition. In the dis­ course, the competition is presented above all as a conflict situation involving the expres­ sion of personal identity. Compared with the results of the competition, the three other forms of expression are of secondary importance. There are not a great many structures available for such discourses, nor are they variable. It is possible to make a list of the most important features, those that are to be found again and again in nearly all dis­ courses on competitive sport: (1) Discourses are, by their very nature, limited by perspective. They present a com­ petitive situation from the perspective of one of the competitors. (2) This competitor is presented in such a way that he finds himself in a contest against at least one opponent (and against unfavorable conditions). (3) The contest in which he is taking part consists in his exploitation of the role of his body to overcome any resistance offered by others or arising from the situation, and in this way to establish his own personal identity. His opponent's intention is to prevent any such role enforcement or identity establishment. (4) The contest as presented in the discourse is the struggle of one body against another to realize its role and to gain the power to define its personal identity. Defining Personal Identity in Competition The power of the one to define is the other's loss. The victor is able to exercise power over the loser, because, from his newly gained position of strength, he can prevent the latter from developing his body role and thus define his own personal identity against the other's resistance. The extent of his power determines the influence he can exercise on his opponent's attempts to define his identity. He is able to characterize the other at least as a loser. With increased power, he is in a position to bear negatively upon the

976

Stoicism

other's identity in respect to more than just one characteristic (also making him appear weak, non-descript, awkward, or tired, for example). In a discourse, a competition is not only a struggle for victory but far more a contest to decide who will be able to exercise the power of defining the other's identity for as long as possible. Such a discourse does not, in fact, present the constitution of ego and alter as the result of a common �-dialogue (as stipulated by F. Jacques (1979) in his theory of dialogue), but instead as the result of usurping interaction. In the place of a common attempt to define identity, there is the use of force by the ego over the alter. The Sportsman's Symbolic Power The power to define one's own identity and that of the other is not to be confused with true power. Its nature is symbolic and bound up with competitive sport. At the end of a contest, it ceases to exist. Professional sport and show sport, however, are subject to special conditions. In tennis, skiing, football, or golf, for instance, individual contests are continued in a series of meetings that last a whole season. In such a series of contests, it is possible to gain a monopoly of the power to define one's own identity and limit that of the other, this also being accompanied by the acquisition of economic power. Such a discourse as that analyzed above has the structure of a drama. One can assume indeed, that it has been strongly influenced by the literary drama form (rather than, as Barthes argues, by the epic form). In a structural sense, it is largely identical with certain notions on the constitution of an authentic, uniquely personal identity (cf. Lionel Trilling's critical essay on sincerity and authenticity). The establishment of per­ sonal identity in a dramatic contest against the resistance of another is rooted in the antagonistic manner of thinking of an industrialized civilization; it is also revealed, for example, in the worlds of law and politics (Eichberg 1981 ). The competitive sport's dis­ course does not duplicate the world, but creates its own world through and through (Lenk 1979), one in which a basic *ideology of industrial society is expressed in a sensuous, concrete from. G.G.

Stoicism Stoicism was an organized and all-embracing system of philosophy, founded in Athens by Zeno of Citium (ca. 336-264 b.c.). Zeno's greatest follower was Chrysippus (ca. 280-206 b.c.), a major logician and a profound and comprehensive philosopher. Both Zeno and Chrysippus wrote extensively on the theory of *signs, but unfortunately only fragments and quotations have survived. For our knowledge of Stoic sign theory we must rely primarily upon Sextus Empiricus (ca. 200 a.d.), a Skeptic and opponent of Stoic philosophy. Among the important figures of the Greek period of Stoicism were Cleanthes (ca. 331-222 b.c.), Panaetius (185-110 b.c.), and Posidonius (135-151 b.c.). Many Romans were attracted by the moral teachings of the Stoics, and by the conception of a rationally ordered cosmos with which mankind can achieve harmony through understanding and self-control. Large portions of the writings of Seneca (4 b.c.-65 a.cl.), Epictetus (ca. 50-138), and the Emperor Marcus Aurelius (121-180), have sur­ vived. However, for our knowledge of the important contributions of the Greek Stoics

Stoicism

977

to logic, ''grammar, *rhetoric, theory of ''truth, theory of the emotions, and cosmology we depend on the reports of Sextus Empiricus, Diogenes Laertius, �-Cicero, Plutarch, '' Galen, *Philodemus, and others. The Stoic theory of signs is an integral part of their conception of the logos or reason which governs the cosmos and all its parts. Their view was that there is no empty space. The cosmos is a material plenum. Matter varies in density, tension, and movement. The ultimate source of variation and movement is a pair of complementary principles, dynamic activity, and recipient passivity, pervading all matter everywhere. The origina­ tive and formative activity is perfectly rational and yet entirely material. In its pure form, it is a divine fire. Intermingled with a more passive moisture it forms pneuma or breath, which is the soul of all things. The soul, both cosmic and human, is a material and rational thing. Its activity in thinking is therefore a material process. Rationality or logos meant for the Stoics order, regular movement toward good ends, and means perfectly adapted to achieve those good ends. Everything that happens in this rational cosmos is the effect of prior reasons or causes and is itself a reason and cause of further effects. The course of nature is thus fully determined, and yet this does not absolve human beings from responsibility for their acts. This is so because the nature of the human agent, along with external factors, is part of the causal process. Every living organism seeks to maintain itself in existence, and is innately so disposed as to seek out what is most conducive to its own well-being. What is most completely human, what mankind seeks as most conducive to human well-being, is such understanding of nature and of our own nature as will enable us to live in perfect harmony with the cosmos. This moral end requires that we be able to interpret natural phenomena as *symptoms or signs. The Stoics appear to have had the medical model in mind. The interpretation of nature itself depends upon the unique human power of meaningful speech, through which we are able to distinguish and separate aspects of what is physically a seamless continuum. Equally important, ''lan­ guage enables connections and relations to be formed. Words joined may form simple *propositions, and such propositions joined through connectives like and, or, and if. . . , then . . . , form complex propositions. An illustrative example is in order. Consider a walking man. As a simple material activity, this is movement without parts or aspects, continuous with its material environ­ ment. Consider next utterances like Dion, walking. These too are physical events, vibra­ tions. Any relation between the two material events, the walking man and the sounds, is coincidental, a matter of chance. But now if "Dion" is used to discriminate one aspect of the matter as human, male, etc., and "walking" is used to discriminate another aspect as movement of a certain kind, the utterances become words, speech. The Stoics distin­ guished the physical noise, the physical inscription, and, in the case of the letters of the alphabet at least, the name of the sound or inscription (e.g. "Alpha") from the sound or inscription itself ("A"). Further, they distinguished from these three the signifying physical sound, i.e., the word or the �·sentence. This they called the semainon. The semantic content which is signified by the word or sentence is the semainomenon. The physical movement (e.g., the walking man), which can be picked out as a referend by a demonstrative like this, is the tynchanon. Semainon and tynchanon are corporeal; the semainomenon - what is signified - is not. A barbarian, aware of the same physical events as a Greek, may not understand the Greek sentence.

978

Stoicism

The content of what is said in speech is not corporeal, but neither is it a thought, if by "thought" we intend something purely psychological. Thought requires language, even if that language is an internal speech. The words of that internal speech are physical states of the material soul. What is said is something of a logical nature, and so considered the Stoics called it a lekton. If what is said must be supplemented to convey •:•information, the lekton is incomplete. Walks signifies an incomplete lekton. Dion walks signifies a complete lekton. The kinds of lekta studied most fully by the Stoics were propositions, both simple and complex. Among other types of lekta worthy of study they recognized *questions, commands, prayers, and suppositions. A r_ational sensory presentation (logike phantasia) is one which speaks to us; it is expressible in speech, whether inner or outer. The Stoics held that in some way there is a natural similarity between words and rational presentations. It is not clear just how they meant this, since they were fully aware of the diversity of languages. Chrysippus is reported to have argued that the ruling principle of man is in the chest because in saying "ego" the lower lip and jaw move first toward the chest. This suggests that he may have counted gestures and related behavior as part of the natural language of mankind. Words. gestures, and empirical presentations are all material entities and could be somehow similar. It would be much more difficult to argue for any kind of similarity between material things and the non-material lektawhich words signify. The Stoics were among the first to make a systematic study of many of the areas now grouped together as grammar. They probably had the conception of a set of *rules governing the production of well-formed sentences in a given language. The first half of what they called dialectic was the study of the voice as it forms rule-governed meaningful ''utterances. This included the study of euphony and •:•music, poetic diction, verbal ''ambiguity, the parts of speech, and written language. The theory of signs (se­ meia) proper falls within the second half of dialectic, the theory of the signified semaino­ mena. This included the study of the logical content of presentations, subject and ''predi­ cates, propositions simple and complex, arguments, syllogism, mood, demonstration, and fallacies. A particularly important kind of complex proposition is the conditional, in which antecedent and consequent are connected by "if . . . , then . . . " If the antecedent of such a true conditional is true and evident, and if the consequent is non-evident, then the antecedent is a sign (semeion) and its consequent is its signatum. In this definition of a sign the terms evident and non-evident require some explana­ tion. Something is evident if to an observer at a given time and place it is present and may be known directly; that is, it does not need to be inferred from prior premisses. Some things are evident of themselves to a competent observer. No additional factor is requisite in order that they be directly known. Such things the Stoics called pre-evident, and their favorite example is light. Light is evident of itself to a present and competrnt observer. To such an observer, therefore, light can never be a signatum. It never require� an additional sign to make it known. Other things are evident but not pre-evident. They may, therefore, sometime he signs and at other times signata. For example, the color of a building is evident, but n,,t pre-evident. It is evident to a competent observer only if an additional factor, light, i s present. The color may sometimes be a sign, but in darkness the color may be the signatum of some further sign, such as the oral report of someone familiar with the buil d ­ mg.

Stoicism

979

As to the non-evident, some things can never be revealed or made known by any sign at all. Such things can never be signata, and they are called absolutely non-evident. The Stoics' favorite examples are the number of grains of sand in Libya, or whether the number of the stars is even (or odd). Both what is pre-evident and what is absolutely non-evident can never be signata, but for opposing reasons. Some things are temporarily non-evident. For example, the death which will follow an evident wound to the heart, or Athens for someone who is away on a trip, are tem­ porarily non-evident. Such things can then be the signata of commemorative (hypomnes­ tikon) signs, as in conditional propositions like, "If there is a wound to the heart, then he will die," and "If this Roman plans to study in Athens, thens Athens exists as a center of learning." A commemorative sign is a true antecedent in a conditional whose conse­ quent is based upon something temporarily non-evident. Some things are by nature non-evident. For example, although the soul is, on the Stoic view, material, its matter is of such a nature as not to be evident to our senses. Or again, the pores of the skin are of such a nature as not to be evident to our senses. Although what is naturally non-evident can never be directly observed by us, it can be the signatum of an indicative (endeiktikon) sign, as in the following conditionals: "If an organism moves in such and such ways, then it has a soul," and "If there is perspiration on the skin, then the skin has pores." An indicative sign is a true antecedent in a condi­ tional whose consequent is based upon something naturally non-evident. This classification of commemorative and indicative signs shows clearly that the Stoics recognized that sign and signatum exist only as relative to one another. The char­ acter of a sign as either pre-evident or dependently evident is used by the Stoics to show that certain kinds of things can never be signata, while other kinds of things may be sig­ nata. Conversely, it is the character of the signatum as temporarily non-evident or natu­ rally non-evident that is used to classify the sign as commemorative or indicative. Sign and signatum exist as such only correlatively. According to the Stoics, sign and signatum must both exist at the same time. If they were material things or events, like smoke and fire, or wound to the heart and death, this would be an obvious error. But sign and signatum, since they are antecedent and consequent in a true conditional, are logical entities. In the true conditional, "If there is a wound to the heart, then death will follow," the consequent ("then death will fol­ low") is true at the same times as the antecedent. Demonstrations in which the premisses are more evident than the conclusion are a kind of sign, and for the Stoics at least part of the theory of proof comes under the theory of signs. The reason is that a valid demonstration can be formulated as a true con­ ditional proposition in which the antecedent is a conjunction of the premisses, and the consequent is the conclusion of the demonstration. The correct analysis of a conditional proposition was of prime importance for the theory of signs, and the stoic proposals are a major contribution to logic. On one analysis of the conditional, that of Philo, a true conditional could have a false antecedent. The Stoic sign theory was careful to point out that in Philonian conditionals it is not the case that every antecedent is a sign. Only those antecedents which are true and evident, and are followed by consequents which are non-evident, are signs. Chrysippus in particular combined the logical analysis of signs with a strong claim concerning the semiotic abilities of animals. Chrysippus argued that a dog follows a scent because he takes it to be a sign of his prey. Even more, however, if the path divides

980

Structuralism

into three the dog is able to smell the. two roads by which the quarry did not pass and then to follow the third without taking the time to smell it. In short, according to Chry­ sippus, the dog, like other animals, has an internal type of language which enables it to infer that, given that its quarry must have followed paths A, or B, or C, if it did not follow A or B, it must have followed C. We do not know just how the Stoics conceived of this internal speech. Perhaps it included imagined gestures and similar behavior. As has been mentioned, Chrysippus treated the motion of the lower jaw toward the chest in the pronunciation of ego as sig­ nificant. In any case, it is clear that the Stoics did not require that all signs be external­ ly-uttered antecedents. For Chrysippus and his followers, it can be said that something is a sign if the behavior of an animal (or man) can be plausibly expressed in a suitable con­ ditional proposition. It is most likely that this behavioral criterion of a sign holds only for commemorative (hypomnestikon) signs. For indicative (endeiktikon) signs, where the signata are naturally non-evident, distinctively human powers are needed. Only human beings are able to speak of signs as such, since human beings can conceive of a conse­ quence relation as synthetic and as transmitting (metabatike) something from anteced­ ent to consequent. Verbeke (1978) and Eco (1980a) have written very good comprehensive accounts of the Stoic theory of signs. Most of the discussions of this topic, however, are to be found in writings on Stoic logic or on Stoic grammar. For the former, the work of Frede (1974, _1978) and the Kneales (1962) is authoritative. For the latter, the work of Barwick (1975), Frede, and Pinborg (1975) is basic. Other excellent papers and books on aspects of this subject have been published by Brochard (1954 ), Brunschwig (1978), Gould (1970), Graeser (1978), Hamelin (1908 ), Lloyd (1971), Long ( 1 971), and Watson (1966 ). D.S.

Structuralism Structuralism in Linguistics It is difficult to say precisely when structuralism in linguistics started. Commonly, the Cours de Linguistique Generate ( 1 916) is considered to be the watershed. It was published by Charles Bally and Albert Sechehaye, three years after its author, �· Saussure died. The Cours emphasizes the opposition between langue ":•language', the underlying linguistic system, and parole 'speech', the individual concrete speech manifestations. It also gives priority to •:synchronics over diachronics, preferring a complete description of language �·structure at a given moment to the description of isolated elements as they evolve in time. Another important notion introduced by Saussure is the linguistic sign consisting of a '1signifier, or acoustic image, and a signified, or concept. Mounin (1968) gives his monograph on Saussure the subtitle "Le structuraliste sans le savoir" (the structuralist without knowing it), which would suggest that the real start of structuralism as a disci­ pline falls after Saussure. On the other hand, structuralist ideas can be found much earlier, in work by the American Whitney (1867), and a contemporary of Saussure, Bau­ douin de Courtenay (1845-1929) made very important contributions to the develop­ ment of phonemics, the cornerstone of ,:-Prague School structuralism. (For a survey of

Structuralism

981

Baudouin de Courtenay's life and work, see Schogt 1 966.) As an official discipline and branch of linguistics, structuralism made itself known for the first time at the First Inter­ national Congress of Linguistics, in the Hague, in 1928, where a number of papers on phonological analysis were presented, by Trubetzkoy, Karcevskij, and *Jakobson. The Geneva School, following the tradition of Saussure's teaching and the Prague School, from then on closely cooperated (see Matejka 1 976d:XI). The Hague Theses were pub­ lished in revised form in the first volume of the works of the Prague Circle ( Travaux du Cercle Linguistigue de Prague, 1 929, pp. 7-29). Prague structuralism shares with Geneva linguistics the preference for a synchronic approach. Both also stress the fact that each unit in a system is defined with respect to all other units. Moreover, an important aspect of both schools is that they posit a level of ,:abstraction and generalization behind the immediately observable linguistic data. It is not surprising that those principles of analysis and description are more easily applied to a set of units belonging to a closed, or limited inventory, such as the phonemes of a particular language, than to an unlimited, or open inventory, e.g., the ,:-Lexicon. The vowel phoneme /a/ is defined in opposition to all other vowel phonemes, the vowels as a group being opposed to the consonants. By abstraction and generalization is meant that, in a language with only one /a/ phoneme, all phonetic realizations [a], [a], etc., correspond t o the abstract, general phoneme /a/. The features that serve to distinguish / a/ from / e/ and / o/ are called pertinent; the feature that distinguishes [a] from [a] is not pertinent. Instead of pertinent, the term distinctive is often used. It is also possible to describe the features from a functional point of view. A feature that distin­ guishes two phonemes is functionally relevant. As phonemes can be structured quite neatly according to their ,;distinctive features, it is no surprise that the term system is often replaced by structure. Saussure, who was more interested in meaningful units than in phonemes, always used the term system, while the more phoneme-oriented Prague lin­ guists introduced the terms structure and structuralism. There are several offshoots of the Prague-Geneva structuralism, some more linked with Prague and others more connected with Saussurean linguistics. They all fall under the general heading of structuralism, even though some are known under other names. Some have developed so many new ideas that they are only marginally connected with the mainstream of structuralism. The Saussurean emphasis on the linguistic sign consisting of a concept or signified, and an acoustic image, or signifier, and the Prague preoccupation with systematic analysis of phonological systems, combine in the works of �-Hjelmslev, the founder of the Copen­ hagen School or School of Glossematics. Hjelmslev's most influential work (1943) claims to be only an outline of a theory of language. Hjelmslev introduces the notions of thought-substance and sound-substance, for which he also uses, somewhat surprisingly, the term purport. In the thought-substance, an articulation into units of content-substance, or concepts, takes place. Similarly, in the sound-substance ,;expression-substance is cut out. Content and expression level combine into a ,:•sign that consists of a content form and an expression form: Hjelmslev tries to apply the same procedure to the content form splitting it into smaller meaningful units - as to the expression form, which can be broken down into phonemes. Although the endeavor was not successful, it is worth men­ tioning, as it foreshadows componential analysis. Hjelmslev underlines the conventional, arbitrary character of this combination, as well as of the •:articulation of both purport levels, i.e., the amorphous substance. This leads

982

Structuralism

to the stressing of interrelationships and the importance of relative value arrived at nega­ tively. (A form, at a certain level of analysis, is what all the other forms of that level are not.) Language analysis should therefore have an immanent character. (For a more detailed discussion of Hjelmslev's theories, see Siertsema 1965.) Whereas Hjelmslev moved away from the Prague School in its relatively early days (around 1935), Martinet's Structuralist-Functionalist School of Paris is rather a direct continuation. The principle of functional analysis, forcefully presented by Trubetzkoy in his landmark study on the principles of phonology (1 939), is further strengthened in numerous works of the Parisian School that forms the core of the International Society of Functional Linguistics/Societe Internationale de Linguistique Fonctionelle. Martinet brings the articulation principle into sharper focus than Hjelmslev. Instead of stressing similarities in the expression-substance and the content-substance, Martinet introduces the notion of double articulation. Concepts divide a continuum into discrete units. To name these concepts, discrete phonic units, or phonemes that articulate the continuum of possible sounds, are used in different combinations. Phonemes are therefore called units of the second articulation. In Hjelmslev's work one observes a growing tendency toward purely deductive reasoning, while Martinet remains faithful to inductive methods and bases his theoretical considerations on direct observation of linguistic data (Martinet 1960). For a long time British linguistics followed a rather independent course. The London School of Linguistics did not concern itself with the theoretical problems that preoccu­ pied the other schools. Jones (1918) can be linked to Baudouin de Courtenay, with the latter's student Scerba (1958) as an intermediary. Practical language description and adequate orthographic systems for languages without a written tradition, became more and more important within the British Commonwealth. This fact explains the pragmatic orientation and the interest in anthropological linguistics one finds in the works of Firth (1964), a combination reminiscent of that found in Boas' (1911) and Sapir' s (1921) writ­ mgs. The link between Baudouin de Courtenay and the London School of Linguistics, via Scerba, is very tentative. The same is true for Guillaume's connection with Saussurean linguistics, with Antoine Meillet as an intermediary. Guillaume was discovered by Meillet when he had already elaborated, on his own, an intricate system to account for the use of the article in French (Guillaume 1919) and of verbal tense and mood in French, Latin, Greek, German, and Russian (Guillaume 1 929). His highly original system - based on the idea that a time-image is formed in the mind just before the speech act, and on levels of concretization for the article - remains controversial. Out­ side France and Quebec, where Valin (1955) and his team continue the ideas of Guil­ laume, the psychomecanics of language is not very widely known. Opponents claim that mental processes should not serve as a basis for an explanatory theory because of circu­ larity, but it is undeniable that the theories are thought-provoking and provide new ideas on language. Except for Guillaumism in Quebec, so far only European structuralism has been dis­ cussed. Two major currents in North American linguistics must also be included in this survey, one founded by one of the initiators of Prague structuralism, Jakobson, and the other genuinely North American, started by Bloomfield. Jakobson, one of the most prominent members of the Prague Circle, moved to North America in 1 942. He founded the journal Word in New York in 1 945, with Mar-

Structuralism

983

tinet, thus providing an outlet for European ideas. Martinet returned in the mid-fifties to Paris, and Jakobson left New York for Massachusetts, where he was a professor of linguistics at Harvard, later combined with a position at MIT. There are some major dif­ ferences between the two most important continuators of the Prague tradition. Whereas Martinet stresses functionalism and direct observation, Jakobson's approach is more the­ ory-oriented, with special emphasis on general features (see, e.g., Jakobson 1 936 and Ja­ kobson and Halle 1 956b). Phonemes are described using a small set of distinctive fea­ tures that account for all possible phonemes in human language; the acquisition of pho­ nemes by a child follows a fixed order of growing complexity of the features involved, and this order is the same for all languages (Jakobson 1 941). Martinet rejects this abso­ lute universality. Another point of contention is Jakobson's thesis that all linguistic '�oppositions are binary. If there is a choice between three or more units, the speaker makes a series of binary choices, until the wanted unit is selected. So, if from /a/, /el, and Ii/, the middle one has to be chosen, the first opposition is /a/++/el + I ii or open ++ not open, the second one being /e/ ++ Iii or not closed ++ closed. Martinet, though not denying that many oppositions are binary, judges binarism as a general principle to be an oversimpli­ fication for the sake of theoretical neatness (see Martinet (1955: 73). However, the gen­ eral validity of binarism and other theoretical constructs had a great appeal for many lin­ guists and mathematicians in the earl� days of the computer era. Jakobsonian structuralism had a great influence on the linguistic theories of Chomsky, whose trans­ formational and generative grammar (1957) forms the starting point for a new school that will not be discussed here. In spite of structuralistic elements, transformational-gen­ erative grammar is too far removed from the original Prague theses to qualify as a member of the structuralist-functionalist family. The introspective approach characteristic of transformational-generative linguistics is a reaction against structural­ ism, especially North American structuralism and its major proponent, Bloomfield. Bloomfield (1 933) proposes a set of procedures for language analysis, concentrating on formal criteria and distribution patterns. His followers, often known collectively as Neo-Bloomfieldians, continued to refine these procedures, which are always explicitly accounted for. Bloomfield excluded ,:•semantics from his investigations, considering lin­ guistics not far enough advanced to tackle problems of ,:•meaning. The methods of inves­ tigation tended to lead to fragmentation, each occurence of a unit chraracterized by spe­ cial distributional circumstances. Some North American works attempted to surmount this fragmentation, especially one by Harris (1 947), a work that announced in many ways the new direction Chomsky, Harris's student, whould choose ten years later, and had a great influence on that choice. Structuralism has changed considerably over the years. The sharp division between diachronics and synchronics, which Saussure stressed in an era of comparative, dia­ chronic linguistics, was abandoned in later years. The notion of dynamic synchronics was introduced (see, e.g., Walter 1 976; Martinet and Walter 1 973). Structuralism continues, however, to deal with discrete units and stays at the ,:•cognitive level. '�Intonation, associ­ ation, and connotation, however important their role may be in the '�communication process, fall outside structuralist language models. Structuralism in linguistics has had a tremendous impact on many disciplines, espe­ cially *anthropology and literary analysis. The trend to use procedures of structuralist linguistics and to borrow its terminology was particularly strong in France in the 1 960s

984

Structural Linguistics and Semiotics

and 1970s. Inevitably, many concepts underwent considerable changes in the borrowing process, and one should never assume that behind the identical terminology used in different disciplines there are also identical concepts. H.G.S.

Structural Linguistics and Semiotics The development of structural linguistics was one of the direct stimuli for the advance­ ment of �·semiotics. The interconnection of these disciplines is, however, complicated by the fact that under the heading "structural linguistics" are gathered together a variety of tendencies the main unifying feature of which is a negative attitude to the preceding linguistic tradition. The common points stem from the ideas of Saussure : the notion of language as a system in which all elements are interconnected, and also the idea of the dual articulation of language, on the expression and content level, and hence the idea of linguistic signs as arbitrary combinations of elements from these two levels. The most restricted view of language, for instance in American descriptivism, reduces it to a closed system of elements on the expression level. The interdependence of elements within a language, when these elements are defined exclusively by their rela­ tionship to other elements, makes it impossible to compare one language with another. Since semantic articulation in language is to be fully and specifically understood only through the comparison of languages (semiotic systems), the idea that languages cannot be compared leads to the exclusion of semantic relations from the language system, and this hinders the juxtaposition of language with other semiotic systems on the level of con­ tent. Moreover, the idea that languages cannot be compared destroys the totality of the object of linguistics; and if the totality of the science is not to be found in its object, it is sought in the method. Linguistics becomes a descriptive methodology and the influence of this branch of structuralism on semiotics is thus reduced to the transfer of descriptive methodology from languages to other semiotic systems. Such a transfer of methodology if successful only in the case of those semiotic systems where the articulation of the con­ tent level is already given and needs no analysis (for example, artificial systems of signali­ zation, e.g. �·traffic lights, the "language of flowers," etc.), i.e., in the case of systems which are simpler than natural language. When applied to more complex systems, de­ scriptive methodology turns out to be inadequate : however, this inadequacy has stimulated a reappraisal of understanding both of language and of other semiotic sys­ tems. A comparison of contemporary works in semiotics with those of the early 1960s makes this clear. In other branches of structuralism, the problem of the articulation of the content level has been more or less explicitly posed (Sapir, Whorf, Trier, Weisgerber, ,:•Prague structuralism). This approach stimulated the analysis of other semiotic systems, since language was in this way understood as one means of making sense of the world along­ side other sociocultural "•:•modeling systems," such as mythology, ritual, etc. (and hence arose the problem of the interrelationship of language and �·culture). The study of how reality is represented in language is naturally connected with the study of how the world is modeled in literature and art. From this point of view poetic language is espe­ cially significant, since it systematically manifests the latent semantic capacities of the lin­ guistic system.

Structural Linguistics and Semiotics

985

In the study of modeling systems, stress may be put either on their semiotic nature or on their functioning, that is, either on the "sign" or on the "language." When the stress is on the sign, those systems in which a minimal unit combining a signifier and a signified is clearly evident are the most studied. This was how Saussure understood semi­ ology when he compared language with "':•writing, ''sign language of the deaf, symbolic rituals, forms of etiquette, military *signals, and so on" (Saussure 1 9 1 6: ch. 3, par. 3) . The task of the researcher, in this case, is to show how the signs combine in a system, their *syntactics and ''pragmatics. When stress is on functioning, systems are studied which serve to exchange *information, that is a ''message with a content. This branch of semiotics (*Jakobson, Levi-Strauss, Greimas, the *Moscow-Tartu School) shows a clear dependence on �-Prague strcuturalism, according to which language was defined as "a system of means of expression intended for a particular purpose" (Prague School Theses 1929). The communications model, when applied to semiotic systems other than natural language, lays stress on the message, that is the ,:•text, as the basic object of study. In the communicatory process, a content is encoded by the addresser and decoded by the ''addressee of the message; the semiotic system, then, is envisaged as a *code (or codes) used in the process of communication. With this approach the separate sign is of interest only as a minimal unit of the code, and moreover it need not be given, but must be discovered (for instance, in the languages of ''painting and cinema). From this point of view, semiotics is concerned with any "meaning" activity in its functional aspect, although language activity is basic and primary. The notion of text is transferred from linguistics and with it the notion of text analysis as a means of discov­ ering the communicative code (the language) standing behind it. Hence, any cultural phenomenon may be studied from the semiotic point of view. For instance, a painting from the point of view of art history can be analyzed with regard to its composition, coloring, texture, technique, and so on, while from the semiotic point of view what is important is the means by which the content, what the viewer has to decode, is transmit­ ted, i.e., perspective, foreshortening, the distribution of what is depicted and what is to be implied, the significance of the levels, symbolic and allegorical features; all these fea­ tures are looked at from the point of view of their systematic correlations. Thus, it is the model "text - mechanism generating it" that is taken over from linguistics, and hence also the concept of system as a totality of mutually dependent elements which coordi­ nate the expression level and the content level (dual articulation). The analysis of nonlinguistic semiotic systems has transformed structural linguistics. The study of cultural "languages" has made it clear that in one social group, there usu­ ally function not one but several interrelated systems (subsystems), and so a text may correlate not with one but with several codes (subcodes). The correlation of these sub­ systems is determined historically and thus may be regarded as diachrony in synchrony. The study of analogous phenomena in language leads to the revision of the Saussurean opposition of synchrony and diachrony. The choice of a subcode may depend either on the content or the situation of the message, and thus is partially conditioned by factors of a social and ideological nature. This explains the importance of the attitude of the user (including his conscious attitude) to the languages he uses (the importance of the inner point of view). Changes in the system that result from the interaction of subsys­ tems are also dependent on the (conscious) influence of the speaker. This is obvious in the sphere of culture; and it becomes obvious also in the sphere of language when this idea is applied to language. In this way, semiotics has brought about the revision of one

986

Structuralism and Post-structuralism

more thesis of structural linguistics, that of the independence of the linguistic system from the consciousness of the user. V.M.Z.

Structuralism and Post-structuralism The Structuralist Perspective It is only with hindsight that "structuralism" has taken on the aspect of a unified move­ ment or common activity of thought. Such handy perspectives are often misleading in their tendency to reduce a complex and multiple phenomenon to the manageable com­ pass of a single, comprehensive outlook. Most of the pioneer "structuralists" have felt obliged, at some stage, to denounce this communal myth and insist on the specific nature and bearing of their own activity. More often than not, the structuralist label was applied by those who wished to attack what they saw as a threat to the values and disci­ plines of humanistic study. With the advent of "post-structuralism" (espcially in the heady climate of present-day literary theory), it is clear that there was always something shifting and provisional about the concept of structure, a term defined more by its nega­ tive than its positive attributes. Yet it is equally true that these latest developments are dependent on - and implicated in - that structuralist prehistory which brought them to light. There is no question here of a total shift of paradigms which would render struc­ turalism obsolete or merely irrelevant. What is at stake, as Derrida has argued (1978), is a different understanding of structuralism, one which perceives its moments of inherent blindness and self-limitation but which also recognizes that these problems can only be seen for what they are through the critical lens of structuralist thinking. It follows from Derrida's argument that this is not a topic to be dealt with straight­ forwardly as an episode in the history of ideas. The concept of structuralism resists sum­ mary analysis and refuses to settle down historically as if its significance could now be placed and assessed. This is to say that structuralism questions all the forms and catego­ ries of knowledge by which such histories are normally written. It rejects, to begin with, the notion that ideas exist in some transcendental realm of their own, independent of lan­ guage and the various cultural codes which determine the limits of thought at any given time. Structuralism takes the opposite view, that concepts are possible only insofar as they exist within a certain intelligible structure of meaning, the effects of which are mostly unconscious but none the less decisive. What counts as "knowledge" will always be determined very largely by selective codes and conventions which work to validate certain kinds of discourse and to exclude others (Foucault 1971). This applies to the whole range of human cultural activity, from "primitive" myth and ritual to the current paradigms of acceptable scientific knowledge. In this respect structuralism has much in common with the work of recent historians and philosophers of science like Thomas S. Kuhn. The quest for knowledge reveals an alternating pattern between periods of patient collaborative effort and moments of drastic upheaval, when even the most basic assumptions are radically revised. Structuralism marked precisely such a moment in the human sciences. It proposed not only a fresh methodology but a new perspective on all methodologies, its own included. Structuralism resists the temptation of exempting itself from this general relativity of

Structuralism and Post-structuralism

987

knowledge and meaning. The structuralist "sciences of man" are able to accommodate the fact that all such discourse must ultimately be a product of the codes and constraints which map out their domain. Traditional philosophy was aimed at an order of absolute �·truth, which it assumed to exist outside and above the local accidents of language, cul­ ture, and convention. This belief is systematically questioned by the structuralist argu­ ment that all ideas - even the most seemingly self-evident - are bound up in a complex economy of thought which alone determines their significance. Language in this sense precedes thought, rather than serving (as most philosophers would like to believe) as a more or less adequate means of expression. The structuralist "revolution" can best be grasped as the outcome of this funda­ mental shift of priorities. Starting from the structures of linguistic representation, it avoids any hint of a premature absolute in its own theoretical grounding. Structuralism, as '�Barthes insists, must always be regarded as an activity of thought, a work carried out on the texts and products of culture, rather than a self-sufficient theory of meaning (Barthes 1 9776 ). Only thus can it escape the charge of setting up its own brand of relativism as yet another philosophic system. Structuralism is aware of the ceaselessly transformative activity of language, the fact that no ultimate meaning (or "transcen­ dental signified" ) can ever foreclose the play of signification. This awareness stems from a particular view of language and its workings, elaborated in the early years of this century by Ferdinand de Saussure. Saussure pro­ posed that linguistics concern itself ( 1 ) with language in its synchronic, as opposed to its diachronic aspect, and (2) with the structural characteristics of langue as distinct from the individual speech chain of parole. This program turned out to have immense implications not only for linguistics but for many other disciplines where traditional methods increasingly seemed open to doubt. Nineteenth-century scholarship had taken it more or less for granted that phenomena were best explained by describing their various stages of historical change and evolution. Philologists had tried to reconstruct the (often hypothetical) origins of language, just as Darwin had done for the history of species and ,:•Marx for the socio­ economic changes leading up to modern capitalism. This historicist configuration of knowledge is documented by Foucault in his book The Order of Things (197 1 ) . He goes on to show how the paradigm was challenged and eventually gave way to a new configu­ ration, one devoted (as Foucault sees it) to language as the structure and image of self­ understanding in the human sciences. As man comes to recognize the omnipresence of linguistic mediation, so history loses its hold as a privileged order of discourse. For Foucault, this signifies a break with the humanist ideology which underpinned the nineteenth-century faith in historical progress and meaning. Man "composed his own figure," his illusory self-image, during that period when language "lost its privi­ leged position and became . . . an historical form coherent with the density of its own past" (Foucault 1971 : 352). Modern man, aware once again of language as the "funda­ mental form of finitude," finds himself not at the point of some new self-discovery but "at the brink of that which limits him . . . " Man has "come to an end," his image dis­ solved in the structures of language where subjectivity finds no place. Foucault's pro­ nouncements bring out the extent of this break between structuralist thinking and the main tradition of modern humanistic scholarship. Structuralism rejects the idea that men create their own meaning, whether in the act of speech or in the process of forging a humanly intelligible history. Just as language precedes and articulates the instance of

988

Structuralism and Post-structuralism

speech, so history presents a site of innumerable structures and orders of event, the meaning of which cannot be grasped by a simple act of "human" identification. These issues were raised by Levi-Strauss, whose work has done much to extend the implications of structuralist thinking from language to other, related systems of cultural exchange. In "History and Dialectic" Levi-Strauss poses the problem of historical method, or what it is to "understand" the past from a present-day standpoint. His remarks are directed more specifically against the idea of history as possessing some ulti­ mate telos or meaning which the mind can grasp by somehow reenacting the logic of events. As opposed to this "diachronic" mode of understanding, Levi-Strauss puts the case for a "synchronic" approach which would come to terms with the gaps and discon­ tinuities created by the passage of time. In Sartre's philosophy he detects a confusion between the order of historical events and the means by which thought makes sense of its available data. It suffices for history to move away from us in time, or for us to move away from it in thought, for it to cease to be internalizable and to lose its intelligibility, a spurious intelligibility attaching to a temporary internaliry. (Levi-Strauss 19666: 255)

The preference for diachronic history is caused by this illusory projection of an inward "continuity of the self," a fiction sustained in turn by "he demands of social life." To abandon this humanist position, as Levi-Strauss requires, is to recognize history for what it is: a plurality of structures in which "classes of dates each furnish an autonomous system of reference," and "events which are significant for one code are no longer so for another" (1966b:255). Fields and Applications Structuralism can thus be seen as the result of applying a synchronic framework of analysis to various regions of cultural and symbolic activity. Levi-Strauss has adapted the insights of structural linguistics to his own ambitious project of comparative anthro­ pology. He treats a vast range of myths, customs, and social constraints as pointing toward an underlying "language" of operative contrasts and distinctions, the logic of which can best be understood by structural analysis. Superficially, there may seem little in common between myths which derive from very different cultures and embody diver­ gent *narrative themes. But to see them as related to the totality of myth - as parole relates to langue in Saussurean terms - is to bring about a radical change of perspective. This is precisely the object of Levi-Strauss's structural anthropology. Myths are subjected to a form of combinatorial analysis which looks beyond the isolated narrative instance to a generalized logic or grammar of *functions. The same technique is applied to the structures of kinship and marriage regulations, as well as to the various ritual procedures (like those surrounding the preparation of food) by which cultures maintain their identity. Levi-Strauss is quite explicit about the analogy with language as a model of all such symbolic activity. "Phonemic patterning is an intervention of culture in nature, an artefact imposing logical rules upon the sound continuum," (Levi-Strauss 1958b: 83 ). As language depends on distinctive oppositions of sound and sense, so cul­ tures erect their own systems of relationship and difference, whether in the form of myths or elaborate social constraints. In literary studies also, the effect of structuralist thinking was to concentrate atten­ tion on synchronic aspects of style and device. This shift had already been signalled by

Structuralism and Post-structuralism

989

the *Russian Formalist School of the 1920s. The Formalist movement effectively came to an end in the late 1920s, but their texts were rediscovered and translated by Western scholars at a time (the early 1960s) when the influence of structuralism was also gaining ground. This connection is strikingly exemplified in the career of Roman Jakobson. From his early involvement with Russian Formalism and the Prague Linguistic Circle, Jakobson went on to settle in post-War America and to play a prominent role in the field of structural linguistics, as it thenceforth developed in the West. His contributions have ranged over diverse areas of method and theory, including some notable essays on literary style and poetics. Perhaps most important is Jakobson's famous distinction between ,:•metaphor and ,:•metonymy as the two main axes or ordering principles of lan­ guage. These two terms provide a descriptive framework with which to classify various phenomena, from types of •:•aphasia to aspects of literary �-style. In France, the advent of structuralist theory reflected a growing disenchantment with the methods and assumptions of traditional criticism. Barthes was chief spokesman for the Nouvelle Critique, which challenged what it saw as the covert ideology of old­ style positivistic scholarship. Barthes' Elements of Semiology (1967a) combined an expo­ sition of Saussurean thought with a programm for the total transformation of cultural and literary studies (Barthes 1967a). With Mythologies (19726) this program had been worked out in practice, by applying the structuralist method to diverse contemporary topics and themes (Barthes 1967a). The uses of "myth" in a bourgeois culture are treated as a second-order signifying system, investing the object with an adventitious value which comes to seem part of its "natural" meaning. This confusion between nature and culture is at the root of all ideology and can only be exposed - Barthes argues - through a structural critique of the mythic sign. Semiotics becomes an applied science of demystification, stripping away the pretences of bourgeois sentiment. Among the topics of Barthes' brief essays are the latest Citroen, the brain of Einstein, Parisian strip­ tease, the image of Garbo, margarine, and the Michelin Guide! The result is a rare com­ bination of exuberant fancy and concentrated intellectual purpose. In Barthes' later writing this concern with method gives way to an allusive deploy­ ment of structuralist themes, teasing out the significance of texts (his own included) while shrewdly undermining the claims of systematic theory. SIZ (1970c) takes a short novel by Balzac and subjects it to what might seem the reductio ad absurdum of structuralist analysis. The text is broken down into 561 "lexies" (or minimal significant units), which are then exhaustively glossed according to the various narrative "codes" which determine their meaning (Barthes 197Sc). However, the fascination of S/Z lies not so much in its top-heavy method as in the passages of interleaved commentary, where Barthes finds the codes almost buckling under the strain of a "realist" narrative up against the limits of its own dominant conventions. Though it seems on the surface a straightforward instance of fictional verisimilitude, Balzac's story is treated as a "limit­ text," one which reveals the contradictions and deep-lying crisis of bourgeois ideology. The narrative deconstructs its own ruling assumptions, just as Barthes' commentary subtly questions the formalities of structuralist method. Barthes remained an idiosyncratic thinker for whom the exercise of style, or the pleasures of a well-turned paradox, often outweighed the strict call of theory. It was left to others, like Tzvetan Todorov and Gerard Genette, to develop French structuralism in the direction of a worked-out narrative poetics and �-rhetoric of style. Their program reflected the revitalizing impact of structuralism on a tradition of analysis going back

990

Structuralism and Post-structuralism

to the Russian Formalists. For these critics, structuralism holds out the hope of providing an orderly, systematic framework for the classification of literary genres and styles. Jon­ athan Culler's Structuralist Poetics (1975) is a defence of this broadly conservative posi­ tion. On the other hand, there are those in the post-structuralist camp who regard such a program as inherently reductive or conformist, and who press their critique to the point of rejecting Saussure's "metaphysical" concept of the sign. Derrida's influence has been paramount here, along with that of American deconstructionists like Paul de Man and J. Hillis Miller (see Hartman 1979). ,;Deconstruction in this vein is one example of the radical applications of structuralist thinking. Others have taken a more overtly political direction, including the theories put forward by the French Marxist philosopher Louis Althusser. Althusser calls for a return to Marx's major texts and a rigorous practice of reading which should bring out the articulated structure of concepts that governs their production. This involves drawing a definite line (or "epistemological break") between Marx's early formulations and the "scientific" status which his thinking achieved in texts like Das Kapital. The business of Marxist philosophy is to isolate this structured economy of themes and to separate science from ideology in the various practices and representations of political thought. Along with this rigor goes the strain of theoretical anti-humanism which, as we have seen, characterizes most forms of structuralist thinking. Althusser is totally at odds with the Marxist-humanist tradition represented by a critic like Lukacs. Althusserian Marxism exerted a powerful influence on theoretical discussion during the 1970s, but has lately been criticized for its conceptual dogmatism and excessively abstract rigor. Equally controversial have been the writings of ''Lacan, the Parisian psychoanalyst whose claim to a unique understanding of •:•Freud is based on Lacan's well-known dictum that "the unconscious is structured like a language" (Lacan 1977a). Lacan argues that this truth is everywhere implicit in the Freudian text, and that Freud was only prevented from spelling it out by the fact that Saussure and structural linguistics had not yet made their appearance. The resultant history of misinterpretation is exempli­ fied at its worst (for Lacan) in the American school of ego-psychology. To recover Freud's insights is a matter of translating the topology of unconscious drives and desires into a framework of linguistic description. Most important, from this point of view, are Saussure's formulation of the arbitrary sign (the "bar" between signifier and signified) and Jakobson's metaphor/metonymy distinction. Lacan's arguments are couched in a tortuous, mandarin style which seems almost calculated to baffle any but the most devoted reader. In principle at least, the object is not merely to "explain" the uncon­ scious but to represent its workings through the puns and involuntary twists of sense which constantly signal its activity. The analyst is drawn into a complex reciprocal exchange with the patient's discourse, rather than assuming a superior diagnostic stand­ point. Unfortunately, Lacan's imperious style has the opposite effect on many of his readers, tending to block comprehension at every turn. Nevertheless, it is significant that Lacan rejects the idea of a masterly or objective discourse, standing outside and above the patient's utterances. Structuralism has wit­ nessed a general shift in this direction, especially among those (like Barthes) who were quick to perceive its reductive effects when applied in a mechanistic fashion. At the outset it seemed that structuralism could legitimately set itself up as a "•:•meta-language," or scientific discourse, explaining the various "natural" languages of culture and society. As this dream of method receded, so the concept of "structure" itself under-

Structure

991

went a process of critique and displacement. For Barthes and Derrida, it comes to figure as a species of heuristic metaphor, one which resists the powerful enticements of system and method. As Barthes puts it, in a passage of quizzical self-address: You keep the notion of "meta-language," but in the category of image-reservoir. This is a con­ stant procedure in your work: you use a pseudo-linguistics, a metaphorical linguistics ... these concepts come to constitute allegories, a second language, whose abstraction is diverted to fic­ tional ends ... (Barthes 19776: 124).

In this sense, structuralism can claim to represent the destiny of all self-critical semiotic thought. It offers a reflective methodology capable of grasping and interrogating its own presuppositions. Hence the force of Derrida's paradoxical argument: that structuralism was always "post-structuralist" in its moments of greatest lucidity and insight. C.N.

Structure

The Various Meanings Of The Term "Structure"

Derived from the Latin verb struere 'to construct,' the term structure is lexically syn­ onymous with construction, composition, ordering, in short with organization, and essen­ tially refers to the system of connections which "organically" link up parts within an organized whole. As such, it occurs in many domains as a more or less technical term: the structure of a building or of an urban zone, the structure of a mechanism, the struc­ ture of a work of art, judicial and social structures, the structure of a firm, the structure of a biological organism, the structure of a molecule or of a macromolecule, the struc­ ture of a physical milieu, of a crystal, of a stellar object, of a texture, the structure of a sentence, of a narrative, of a myth, the structure of a phonological system, the structure of perception, cognitive structures, kinship structures, the structures of the imaginary, etc. However, according to the domain considered, the term can have a very different problematical content and epistemological value. In the case of a technical mechanism, a construction or a work of art, one generally knows how to describe the structure of the whole by means of a plan of composition. In the case of physico-chemical systems (crystals, macromolecules, etc.) one also knows (at least in principle) how to derive the structure from the interactions between components. For example, spectacular progress in molecular biology and in microbiology has enabled a decisive breakthrough in the understanding of the structure (the stereo-chemical morphology) of the DNA chains, of proteins, of enzymes, of membranes, etc. Of course, experimental difficulties are con­ siderable. To resolve them one must apply to highly sophisticated technologies of obser­ vation and reconstruction techniques. But by right, even if the structure is not completely mistaken for the morphology observed, even if the former is made the organ­ izational principle of the latter, one is nonetheless in the presence of phenomena and material objects whose ontological status is non-problematical. In other domains, on the other hand, such as, for example, in natural biology (taxo­ nomic, anatomic, morphogenetic, that is not strictly physico-chemical), in perception, in ,:•anthropology or in semiolinguistics, one comes up against immanent non-material structures, against abstract forms of organization which cannot be reduced to systems of interacting components. This simple fact raises considerable theoretical problems

992

Structure

insofar as structures, without further inquiry, can no longer be considered as objects of experience having a predi/ined ontological status. The objectivity of the structures there­ fore must be constituted as such, and that is why, in every one of the cited cases, in an effort to get to the bottom of things, it has been necessary (1) to make the organizational concept of structure an authentic comprehensive and explanatory catego­ ry, (2) to critically interrogate its objective value and (3) to attempt to formalize its cate­ gorical semanticism. This is how the new methodology and new rational unity called structuralism have been progressively developed. In the natural descriptive sciences, as well as in the human sciences, the theory and the formalization of structuralism are based on a rationalist attitude. Its point of view is opposed to that of reductionist "atomism" as well as that of evolutionary historicism. The passage from atomistic psychology to ''Gestalt theory, from the neogrammarians' historical, comparative and philological linguistics to Saussure's structural linguistics, from "behaviorist" linguistics to Chomsky's generative •:•grammars, from biographical and socio-psychological literary criticism to structural criticism, etc., represents various aspects of a general theory of systems conceived as (self)-regulated totalities. From this perspective the horizon of structuralism is that of the phenomenological description and then of the formalization - of the a priori constraining the formal relations of dependency linking- "organically" and systematically- parts in wholes. Thus, concerned with the general dialectic of parts and wholes, structuralism, in one way or another, and this irrespective of the aspect considered, is indeed concerned with the relation between substance and form, and so with the Greek problematics of morphe, eidos, and ousia . . . As the ideal form of the organization of a substance, structure is not a perceptible phe­ nomenon. As such it is "invisible," though its substantial realization and its effects can be clearly experimentally observed and analyzed. This is why every structure is a theoretical object (and not a fact), a theoretical object which is at the same time ideal and real. (Paraphrasing what •:•Kant said about space and time as forms of intuition, one could say that the object "structure" has both empirical reality and transcendental ideal­ ity.) If naive idealism is to be avoided, the difficulty is therefore to understand how it can

emerge, by a process of epigenesis, from the organization of the substratum where it is real­ ized. This is in a way the "founding aporia" (to refer back to Rene Thom's expression),

of all structuralisms. This is what Deleuze insists on: "real without being actual, ideal without being abstract," pure "virtuality of coexistence which preexists things," (1973: 313), a structure "is embodied" in its substratum, is "expressed" in it, but is never actualized as such in it. The perceptible expression of a structure is always, in essence, the negation of its formal being. As Pomian writes: This is why, for the object as it initally appears, two objects, whose ontological status is different from that of the first one, are, while remaining different from each other, substituted [in every structural approach]: speech and language (Saussure), sounds and phonemes ("Jakobson, Trubetsko"i), substance and form (*Hjelmslev), kinship systems and elementary structures (Le­ vi-Strauss), performance and •=·competence (Chomsky), empirical morphologies and underlying dynamics (Thom) are examples of this. All of the first terms of these couplings (which can be called "realizations") can be reached by means of perceptual experience or by observation, and this is in what their reality consists of. As far as the second terms are con­ cerned (except for Saussure's language, to be exact: system, or the function of structure, is taken up by the "mechanism of limitation of arbitrariness") structures, by definition, cannot

Structure

993

be perceived or observed. They are given a reality by means of demonstration, which is, according to the case, more or less rigorous. Relations between realizations and structures are variable but it is always the latter which make the former stable and intelligible. This leads to the following definition of the notion of structure: sets of rational and interdependent relations whose reality is demonstrated, whose description is given by a theory and which are realized by a visible or observable object whose stability and intelligibility they insure." (1981 :758)

Because of their ideal status, which is nonphenomenal in the classical sense (perceptible) of the term, structures are therefore ontologically ambiguous. As Eco asks: "Is the structure an object, insofar as it is structured, or is it the set of relations which structures the object but which can be abstracted from the object?" (1972c:322). As eidos, a structure cannot be detached from the substance in which it is actualized, from the ousia where it becomes substance. It is at the same time intelligible framework and structured object (Eco 1972c:323). But must it be considered as given or posited? In the first case an ontological (realist) conception of structure will be developed, and, in the second case, an epistemological (nominalist) one. The influence exercised on the contemporary episteme by positivism and logical empiricism, through their dogmatic will to eradicate the ontological question from the theory of knowledge, has lead to a privileging of the epistemological interpretation by reducing the concept of structure to a mere metalinguistic operational one whose reality is non-ontological but purely methodological. Yet, it should be noted here that all of the "great" structuralists (Saussure, Jakobson, Tesniere, Hjelmslev, *Piaget, Levi-Strauss, Chomsky, Greimas, and, more recently, Thom) were, or are, "realists" even if they refuse to get involved in philosophical dispute. In addition, realism in no way claims to detain the "final" say about phenomena but simply affirms the need for the theoretiza­ tion of phenomena (and in particular their formalization) to be in agreement with the "thing itself". Finally, as such, realism is the only rational position able to transform the concept of structure into an explanatory concept, that is to say into a category of expe­ nence. Indeed, from an epistemological, methodological, and "nominalist" perspective, the concept of structure can be only a descriptive concept which, although empirically founded, is without proper objective value. Although operational, it is then only an intel­ lectual artifact (a metalinguistic construct) and cannot therefore become the source of an authentic theoretical progress. In particular, it cannot lead to progress in the mathem­ atization of phenomena. From a realist perspective, on the contrary, it becomes a theo­ retical object which, although initially problematical, must acquire, beyond its empirical validity, a categorical content, an objective value and a constitutive thrust. Since it does not subsume perceptible phenomena it therefore leads to new and fundamental prob­ lems. First it leads to a re-examination of the transcendental problematics (Kantian, Hus­ serlian) of the constitution of objectivities so that (1) one can found rationally the new class of phenomena (non-perceptible) which are the structural phenomena, and (2) one can determine the ontological thrust of the category of structure. Next, and perhaps especially, for an "authentic" �·formalization of the former, it leads to the substitution of an adequate and explicitly constructed mathematical content for the semanticism of the latter. This last requirement meets with considerable difficulties which until now were insurmountable. It is dramatically lacking in purely epistemological perspectives. If one wished to undertake a "spectral analysis" of structuralism, either historically or systematically, it would be necessary to analyze at least the following trends and their interferences:

994

Structure

(1) Dynamical structuralism, of biological origin and of vitalistic nature, which, beginning with Naturphilosophie and Dynamismus in German speculative idealism, has cut across the history of biology before leading up to Waddington's concepts of "mor­ phogenetic field" and "chreod." This dynamical structuralism centers upon the out­ standing problem of morphogenesis. (2) •:•Phenomenological and Gestalt structuralism, which was elaborated at the beginning of the century from the works of •:•Brentano (Stumpf, �·Meinong, Ehrenfels, ,:-Husserl, Kohler, Koffka, Wertheimer, etc.). (3) Linguistic structuralism which, stemming from the Saussurean "epistemological rift," became one of the fundamental paradigms of the human sciences whether in pho­ nology, with Jakobson, in anthropology, with Levi-Strauss, or in linguistics, with Chomsky. This structuralism itself split up into two streams, (3a) phenomenological "realist" structuralism (linked to Jakobson) which bears a close relationship to dynamic structuralism and Gestalt theory; and (3b) formal, "methodological" and "epistemo­ logical" structuralism (linked to Hjelmslev, Levi-Strauss, Chomsky, and Greimas) which considers structures as "axiomatizable" theoretical objects and resolves the ques­ tion of their ontological reality by rooting them in biologically (therefore genetically) determined *cognitive capacities. (4) Piaget's epigenetic and cognitive structuralism. (5) "Catastrophic" structuralism developed by Thom. This one (last but not least) is an original and fundamental synthesis of the concepts of morphogenesis and structure, and therefore of "vitalistic" dynamical structuralism and semiolinguistic rationalism begun with Saussure. Thom is the first to have succeeded in mathematizing, in a sophis­ ticated manner and according to the "things themselves," structures as theoretical objects. As such, his contribution constitutes an authentic renaissance of structural thought, and leads to recasting it on radically new, mathematical bases. In order to be more or less complete, it would also be necessary to make explicit a number of general problematics indissolubly linked to the rationalist structural project. Among these, five seem to be essential: (1) That of experimental methods making the access to structures possible: Since these are immaterial and ideal they cannot be directly observed. A first method of access (that, for example, advocated by Levi-Strauss) consists in analyzing by variational procedures the •:•transformations of a structure. If indeed this latter can be identified with the global, internal, and regulated cohesion of a system of relations, then all local varia­ tions must bring about global transformations which manifest the structure. A second method (that, for example, of Chomsky's "native speaker," conceived as an automaton receiver of language) consists in submitting old introspective practices to experimental control. (2) That of the relation between structure and '1unction: Indeed the latter is the raison d'etre and adaptive signification of the former. Since the historical debate between Geoffroy Saint-Hilaire's principle of connection and Cuvier's principle of func­ tional correlation, a natural dialectic opposes, in secular misunderstanding, two comple­ mentary aspects of the same reality. As the author of the article "Structure and Func­ tion" in the Encyclopaedia Universalis brings up, this dialectic opposes to a naturalistic attitude (a) inspired from a vitalistic holistic conception, (b) based on observation, (c) founded on morphogenesis, and (d) linked to Lamarck, a strictly biological, mechan-

Structure

995

1st1c, and materialistic attitude (a) drawing upon a "micromeric" conception, (b) stemming from active experimentation, (c) focusing upon physiology and (d) of essen­ tially Neo-Darwinian inspiration. But the debate is rather fallacious, since phenomena of adaptation (and in particular those of adaptive convergence and parallel evolution) show that there exists an irreducible correlation between these two aspects (cf. Delattre et al 1973). The problem is rather- and it is a most formidable one at that- to reach a theoretical comprehension of this correlation. (3) That of the relation between structure and finality: One of the main reasons for the disfavor that the concept of structure fell into is that, dealing with the general problem of the systematic organization of the parts in a whole, for a very long time it covered only a finalistic content (a teleological idea) and could be expressed only by the vitalistic concept of entelechy. Before becoming an operational scientific concept, it had therefore to be ''de-finalized." This was only possible through formalization. (4) That, precisely, of the formalization of structures: It has become commonplace to say that it is the combined development of the general theory of systems, of cybernet­ ics, and of a formal logic of interdependency relations, which, by favoring the ' 'axiomat­ ization" of the concept of structure, established its claim as science. But one must recog­ nize that these mechanistic and formalist approaches are quite inadequate. Indeed, they are possible only if the structures are reified beforehand and do not agree any more with the "thing itself." They do not permit an accounting for immaterial structures as organ­ ized and self-regulated forms emerging epigenetically from the material structure of their substrata. In other words, they do not help to resolve the traditional question of the relation between form and substance. As such, the problem of the formalization of struc­ tures is still largely open and has begun partially to be solved only by means of catas­ trophe theory. (5) That of levels of organization: The correlation structure-function operates at every level of integration and observation. The central question therefore becomes that of the definition of the objective reality of these levels as well as the understanding of their hierarchical dependency and, at the same time, of their autonomy. These points will be raised here in an extremely lacunary fashion. In particular, "classical" structuralism, which has become part of common scientific culture, will not be accentuated (Saussurean structuralism, Parsons' structural-functionalism, Bloom­ field's, Harris', and Chomsky's structural linguistics, the structural analysis of economic equilibria etc.). For an introduction to structuralism, see, for example, in addition to the works cited here: Almansi 1970; Bach 1965; Badcok 197 5; ,:•Barthes 1966a; Bastide 1962; Benoist 1975; ,:•Benveniste 1966a; R. Boudon 1968, 1973, 1981; Broekman 1974 ; '"Cassirer 1946c; Chomsky 1965, 1966a, 1968a; Damisch 1973 ; Delattre 1971; Ducrot et al. 1973 ; Eco 1963; Ehrmann 1966; Gandillac, Goldmann, and Piaget 1965; Gluck­ mann 1974 ; Greimas 1966a; Harris 1951, 1970; Hawkes 1977; Hjelmslev 1968, 1971; Jacob and Francone 1970; Jakobson 1971j ; Jakobson and Levi-Strauss 1962; Katz and Fodor 1963 ; Laughlin 1974; Leach 1976; Levi-Strauss 1949; Macksey and Donato 1970; Maranda and Maranda 1971 a; Marin 1977; Raccani and Eco 1 969; Robey 1973 ; Saussure 1916 ; Sebag 1964; Sebeok and Osgood 1965b ; Segre et al. 1965; Viet 1965. We prefer to emphasize: (1) the theoretical difficulties, still largely unresolved, of structuralism; (2) its unpopular aspects, whether phenomenological, or vitalistic, or

996

Structure

Gestaltist; and (3) the content and contribution of the new "revolution" brought about by the catastrophic paradigm. However, we shall begin by saying a few words about structuralism in mathematics, since it often serves as a support to justify formalistic conceptions. For further explana­ tion cf. Bell and Slomson 1 969; Bourbaki 1 969; Dieudonne 1 977, 1 97 8 ; Lautman 1 977; Luxembourg 1 969; Machover and Hirschfeld 1 969; Manin 1 98 1 ; A. Robinson 1 966.

Structuralism in Mathematics

In mathematics, the concept of structure was worked out and progressively imposed during the nineteenth century, and from the Hilbertian axiomatic project and Bourbaki's Elements ( 1 969), took modern achievements by storm. Its relevance stems from the more and more important role ,:abstraction played in mathematics. For the longest time men reasoned on "concrete " mathematical objects which were supposedly naturally and intuitively given ("concrete" in the mathematical sense, that is to say objects which are of course ideal but individuated: numbers, geometric spaces, functions, equations, etc.). But it was quickly realized that, in the same way as natural language enables the concep­ tual analysis of concrete natural objects, so too it was necessary to develop a "conceptu­ al" analysis of "concrete" mathematical objects as syncretisms of abstract moments. This original "conceptual" analysis evidently cannot be simply reduced to the truism about the existence of general concepts in mathematics (the concept of triangle, polygon, polynomial, set, etc.). It is structural in the following sense: Mathematical objects are explicitly constructed (by procedures of construction formalized first within the context of set theory, and then, more recently, within that of the theory of catego­ ries). Therefore (contrary to given, non-constructed objects), they present opacity, and all of their properties are analytically contained in their construction. Demonstrations are used to make these properties explicit. But when such demonstrations are analyzed, generally one finds that only a part of the properties characterizing the object have been used. For example, in a demonstration bearing on the set IR of real numbers, only the algebraic operations of addition and/or multiplication, or the relation of order x � y, or again the possiblity of measuring distances will have been used. With Bourbaki, through abstraction, different levels of structure (that of algebraic structures, that of order struc­ ture, that of metric structure) will then be distinguished, and the "concrete" object will appear as a synthetically intertwining syncretism of some of them. Obviously, the choice of "good" base structures has demanded a considerable work of comparison between "concrete" objects. This work can now be considered mostly finished. Once the fundamental concept of abstract structure and of level of structure was established through finer and finer intertwining structures, a concrete object could be specified. For example, in the case of IR one can say that :

( l ) As a set with the operation of addition IR is a commutative group. (2) Multiplication being compatible with addition (i.e., distributive in relation to it), it defines a ring structure (commutative) on IR. (3) Since for multiplication all non-null elements of IR (i.e., those of R* = IR (0)) are reversible, IR is also a commutative group. This multiplicative group structure being compatible with the additive structure, the ring IR is in fact a (com­ mutative) field.

Structure

997

(4) As there exists no number x £ IR such that x E O and nx = 0, IR is a field of

characteristic zero. (5) Provided with its natural relation or order, IR is a totally ordered set. (6) This order has the property of density, that is to say, for every pair (x, y) £ IR x IR so that x < y, there exists z £ IR so that x < z < y. (7) This order is Archimedian : for every number x > 0 as small as can be and for every number y > 0 as large as can be there exists a (large) integer N such that Nx > y. (8) Order being compatible with the algebraic structure, IR is a totally ordered Archimedian field. (9) Since there exists a distance in IR, IR is a metric space. ( 1 0) For its natural metrics IR is a complete metric space (i.e., "without holes" con­ trary to its totally ordered Archimedian subfield of rational numbers). ( 11) Metrics proceeds, in fact, from order and algebraic structures, since distance d (x, y) between two numbers x, y £ IR is the absolute value Ix - yl of their difference and since the absolute value is defined from the order and addition (Ix! = x if x � 0 and Ix! = -x if x :5 0). Because of the compatibility between the algebraic and the order levels, it remains invariable through translation. ( 1 2)Therefore, considered as a field, as a vectorial space on itself, IR is a normed vector space and, following ( 1 0), a complete one, etc.

The importance of the structural approach to mathematical objects is considerable. Indeed, it permits us to substitute pure formal criteria of identity for the substantial ones. Two isomorphic structures, that is to say, indistinguishable as far as their structure is con­ cerned, will be considered equivalent even if the elements of their underlying sets are of different types. For example, the set P of the polynomials ax2 + bx + c of the second degree of a variable x with real coefficients, has a natural vector space structure over IR, isomorphic to IR3 , although such a polynomial (which is a function of IR to IR) and a point of three dimensional space are heterogenous entities. This approach also permits us to consider transformations between structures of the same type (what are called mor­ phisms of the structure, of which isomorphisms are particular cases), and thus opens up onto the theory of categories. It then permits us to treat the objects axiomatically, which leads to those delicate problems concerning relations between '}syntax and ,:•semantics in the logical theory of models. Given a structure of type S, for example the group structure, it can be considered in two ways, either syntactically or semantically. Syntactically speaking, one chooses a formal language adequate to the formulation of formal expressions about S (for example a first order predicate calculus having a symbol of operation for the law of the group and a symbol of constant e for the neutral element). S is characterized in L by a finite system of axioms L. The theory of S relative to L will then be the deductive closure Th(L) of L in L. Semantically speaking, on the other hand, one considers in a universe U of set theory, the category (L) of objects having a type S structure (here the category of groups), in other words the category of the models of L. Given a model G (here a group), the set of expressions of L which become true once interpreted in G, constitute the theory Th(G) of G. Whereas Th(L) is a syntactic entity (based on the syntactic notion of deductibility or of derivation), Th(G) is a semantic entity (based on the semantic notion of validity). Clearly Th(L) C Th(G) since every expression deduced from a valid one is

998

Structure

valid and since expressions of :E are by hypothesis valid in G. The ideal would then be that: (1) in increasing :E in :E 1 , while maintaining it finite, the situation Th(:E 1 ) = Th(G) could be reached, which means that the theory Th(G) of G relative to L would be axiomatizable; (2) if two objects G and c· have the same theory relative to L (Th(G) = Th(G')), they would be isomorphic, which means that the theory Th(G) of G would be cate­ gorical. As a matter of fact, this is not generally the case, and most theories are neither axio­ matizable nor categorical (existence of non-standard models). This is due to a collection of subtle logical phenomena whose analysis is the object of model theory (for further explanation cf. Petitot 1979a). But without a doubt, the principle interest of the structural approach is that it enables us to reveal a rational unity within the diversity of mathematical theories. This is what Manin emphasized when he wrote: the usefulness of this approach stems from the fact that, in diminishing the number of initial properties granted for the objects, the number of objects which satisfy them is increased and that, by freeing ourselves from the hypothesis of the particular properties of a concrete situa­ tion, we begin to perceive a deep generality and analogy between apparently independent mathematical objects. ( 1 98 1 :774)

This being true, it is necessary, however, to understand that, contrary to what is claimed by logicism (logical formalism) its transformational mechanisms and its algo­ rithms constitute the least important part of the axiomatico-structural method in mathe­ matics. For what is essential is not the syntactic form of language but rather the qualita­ tive and integral characterization of a domain of objects. This point was remarkably developed by Lautman: Logicians of the Vienna School seem always in full agreement with Hilbert's School. However, nothing is more questionable. Following Russell, in the Logistic School an attempt is made to find the atomic constituents of all the mathematical propositions . . . Hilbert's and his school's axiomatics . . . tend, on the contrary, to isolate, for each domain studied, a system of axioms such that, at the same time, a domain and the operations valid in this domain, arise from the set of conditions implied by the axioms . . . The consideration of a purely formal mathematics must give way to the dualism of topological structure and functional properties in relation to this structure . . . The object studied is not the set of propositions derived from the axioms, but organized, structured, complete beings, having their own anatomies and physiologies . . . The point of view to be retained here, is that of the synthesis of the necessary conditions and not that of the analysis of the primary notions. (Lautman 1 977 :282-83)

Critical Points of Structuralism When one passes from mathematics to empirical sciences, structural methodology changes completely insofar as one is no longer dealing with explicitly constructed (ideal) objects but with natural "opaque" phenomena (a priori non-intelligible) which must be no longer logically but ontologically understood and explained. Then the problem is no longer to abstract levels of structure but to theoretize the natural phenomena of (self)or-

Structure

999

ganization. Then the concept of structure no longer corresponds to properties of objects but to a conceptual category whose content must be founded, whose objective value (ontological thrust) must be (transcendentally) deduced, and whose explanatory force must be legitimated. In all domains (biology, psychology, anthropology, semiolin­ guistics) where structure has a real empirical value and is necessary for understanding (if not for explaining) specific phenomena, its theoretical function is to get beyond the con­ flict between on the one hand, objectivist-reductionist explanations, which consider organized wholes as systems of interacting components, and, on the other hand, holis­ tic-idealist comprehensions which consider organized wholes as amorphous material substrata imparted with essences foreign to them. It is to make possible a relational and epigenetic doctrine of organization. This intention is obviously not without problems, and brings us back to one of the oldest aporia of modern scientific rationality. The Aporia of Organization in the Kantian Critique of Teleological Judgment One can find the origin of modern structural problematics in Kant's Critique of Tele­ ological Judgment, where he deals with biological organization in terms of finality (what he called internal finality of natural ends) and shows that the theoretical comprehension of organization must, of necessity, resort to two principles, to two antinomic "maxims" of •�judgment, one reductionist, the other holistic. Let us briefly retrace the various steps in his argument: (1) Granted the a priori structure of every possible experience, one cannot posit an objective finality in nature. Objectively speaking, nature is necessarily mechanistic. In other words, in more modern terms, reductionism is the only objectively admissible thesis. (2) Nonetheless, it cannot be denied that there exist "natural ends" in nature, that is to say, things that are "their own cause and effect" (Kant 1790: 190), in short organized living beings. The fundamental characteristics of these natural biological ends are, according to Kant, regulation, reproduction, and adaptation to environ­ ment (external finality). (3) Yet, although Kant did admit that progress in physics could lead one day to a reductionist ( "mechanistic") explanation of these, he also, in a decisive remark heavy of consequences, noted that such an explanation would always be incomplete, and this for a priori reasons. Indeed it cannot account for the contingency of the form of organized beings. For Kant the contingency of form is a major "characteristic property" of natural ends. Being far beyond the laws of geometry and physics, form is comprehensible only in a reflective way by means of the rational concept (and not the category) of finality. ( 4) Internal finality is not only organization but self-organization. In a natural end, there exists a reciprocal determination between the parts and the whole. A structure is not a mechanism but the effect of the Idea of whole, which determines its syste­ matic unity as form. Thus organization depends on the "formative force" (bil­ dende Kraft) which, not being mechanically explainable, is not objective. This is why it is an "unfathomable quality," an "immeasurable abyss" where reduction­ ism, although being the only objectively valid attitude, must necessarily come to grips with the holistic concept of finality. But finality is a non-determinative (non­ objective) concept which is only regulatory for the faculty of reflexive judgment.

1000

Structure

(5) The maxims of judgment, respectively reductionist and holistic, are contradictory and lead to a natural "dialectic." For Kant, this conflict is not an antinomy since it only bears on maxims; that is to say on prescriptions which the sub­ ject must follow to gain knowledge (on heuristics for the understanding of phenom­ ena). There is an antinomy only if, going dogmatically from the reflexive to the determinative, the regulatory Idea of finality is used as a constitutive concept, as an objective category. But this does not hinder the rational concept of finality, even though being transcendent relative to the faculty of determinative judgment, from being empirically conditioned and valid "with the same necessity for our human faculty of judgment as it could have if it were an objective principle" (Kant 1790 : 2 1 8). (6) This possibility that a regulatory Idea can have for us the same value as a consti­ tutive concept, comes essentially from the finite and the discursive (non-intuitive) nature of our understanding. Although things seem to have drastically changed since Kant, this is in the main illu­ sory. The epistemological obstacle brilliantly circumscribed in the Critique of Teleological Judgment (that is to say the impossibility of physically explaining morpho­ genesis, (self)organization and regulation) is still far from being resolved. Obviously, decisive and joint progress has been made. On the one hand, reductionist (molecular and neoDarwinian) biology, and, on the other, techniques of cybernetic simulation have been at the origin of important achievements in the elucidation of such phenomena. But we are still far from understanding how stable and self-regulatory struc­ tures can emerge from a physico-chemical substratum. The difficulty is not so much experimental as theoretical. The facts are not lacking but the concepts are. It is only recently that in elementary (non-biological) cases, it has become possible to begin to explain how a physico-chemical substratum can spontaneously self-organize, either tem­ porally (oscillating chemical reactions), or spatiotemporally (spatial patterns of the Belousov-Zhabotinsky reaction, Benard cells, etc; cf., e.g., Prigogine 1980). For this it necessary to use deep and sophisticated mathematical progress in the analysis of bifurca­ tions of non-linear dynamical systems (cf. below). From this point of view, biology still truly remains, as Piaget affirmed (1968), "the key to structuralism." Structuralism in Biology In biology, the dynamical aspect of the structuralist paradigm imposed itself several times, in fact every time the problem of morphogenesis arose. But it has always been linked with the speculative, vitalistic concept of entelechy, which explains its rejection by tenants of the reductionist paradigm. It all begins with the principle of spatial connection of parts in an organic whole, a principle introduced by Geoffroy Saint-Hilaire and then taken up by Goethe. In his long and patient meditation on vegetal morphogenesis between 1770 and 1832, Goethe did not attempt to understand the physico-chemical and mechanistic processes underlying the formation of organisms but rather the principle by which an organism is what it appears to be. He quickly came to the conclusion that what separates organisms from machines is that, in the case of an organism, the external appearance is dominated by an internal principle producing the spatial (external) connection of its parts. The under-

Structure

1001

standing of this principle constituted for him the central theoretical problem of biology. Yet, although rooted in indisputable empirical realities, the concept of connection is only, as we have seen with Kant, a "suprasensible," "noumenal" Idea (and not a deter­ minative concept, a category). Resolutely transgressing the argument of the third Cri­ tique, Goethe nonetheless hypothesized that there existed a scheme for it with infinite possible concrete variations. Instead of making this scheme, which is what Darwin did later on, the result of evolutionary chance, he attempted to imagine its characteristics. To understand the organisms' reactions to internal and external pressures, he attempted to understand their ideal constituting principle, in other words their formative laws. He progressively recognized this principle in the spatiotemporal unfolding of an internal organizing force taking on existence by the very act of manifesting itself spatiotemporally. Such is the a priori entelechic principle which according to Goethe commands the forma­ tion of "natural ends." But one of the central results of the Kantian critique is precisely that an Idea is by essence irremediably dissociated from the pure intuitions of space and time. Goethe therefore had to transgress the Kantian doctrine of objectivity and to think of entelechy as an intuitive concept. Whereas in physics the concept is abstracted from the perceptible world and permits only discursive expression, here the concept is self-de­ termining. Its unity gives itself its own form. It is no longer abstract but, like Hegel's Concept, real, concrete, and perceptible. Such is entelechy: an intuitive concept and an efficacious idea, which, by unfolding itself spatiotemporally, governs morphogenesis. Goethe's response to the aporia of form in biology was therefore speculative. It was at the origin of vitalism. Yet it nonetheless possesses an eminent epistemological value insofar as the whole of dynamical structuralism (whether in biology, in psychology, or in semiolinguistics) has remained more or less implicitly bound to it. To see this better, first in the case of biology, we can apply to a recent defense of structuralism proposed by the biologists B. Goodwin and A. Webster (1981, 1982) in the Waddingtonian style (Waddington 1956, 1957). Goodwin and Webster analyse historically and epistemologically the classical and recurring conflict between the structuralist and Neo-Darwinian points of view (the latter being a synthesis of the Darwinian theory of evolution and of the molecular theory of heredity). For them structuralism is opposed to Neo-Darwinian historical empiricism not at the level of facts, but as a rationalism according to which a priori con­ cepts, categories, and principles control the theoretical explanation of empirical data. The central problem they raise is that of morphogenesis, and they seek the type of cate­ goriality necessary to render the concept of form intelligible. Indeed, the Neo-Darwinian paradigm is a conceptual system whose apparent "evidence" precisely renders unintelligible morphological phenomena. It attributes these merely to evolutionary chance and negates any necessity in the order of form, any "laws" of form. This is essentially due to the fact that, in this paradigm, the concept of control and the category of cause are surreptitiously identified. The genome controls form and development. Its mastery and its control therefore enable the control of its effects. But this does not imply the nonexistence of autonomous, specific, and ideal (platonic) mor­ phological constraints. By equating genetic control with determining cause, it is postulated without further inquiry that there is nothing more to explain in what is being controlled. As Jacques Monod claimed, form is causally reducible to the primary struc­ ture of proteins; the remainder being dependent only upon thermodynamic processes.

1002

Structure

As a historicist and dualistic conception, Neo-Darwinism rightly postulates the exist­ ence of a material principle of organization but negates any formal causality. It is at the same time a materialistic holism and a reductionism which posits the primacy of the func­ tional, which reduces the structural connection and the positional organization of the parts to a mere spatial contiguity, and subordinates "internal finality" to "external finali­ ty" (to adaptation and selection). According to it, structure is reduced to heredity. It is historically given and its only necessity is that of its evolution. It is only an artifact of its own control, the expression of its genetic program. Structural rationalism denounces in this conception the inconsistency of making �·history not only the cause of evolution but also that of the stability and invariance of species (cf. also Gould 1977a and Gould and Eldrege 1977b). It conceives organism not only as a complex genetically-controlled physico-chemical system, but also as a structure, that is to say as a totality organized by a system of internal relations satisfying formal laws and universals. The universe of organized beings is a constrained one which mani­ fests a certain necessity. The structures are neither irreducibly diverse, nor the arbitrary results of evolution. The fundamental hypothesis of structural rationalism is that the expression of the genotype by the phenotype remains incomprehensible until one introduces positional information controlling cellular differentiation. In organized beings there would be an efficacy of position, position selecting certain local metabolisms by triggering certain genes. And it is the understanding of such positional information, of such efficacy of posi­

tion, that constitutes the central theoretical problem of structuralism, not only in biology, but in all the other structural disciplines. For example, whether in Driesch's vitalism or in

Waddington's theory of morphogenetic fields, the principal characteristics of organiza­ tional structures are:

(1) dynamical genesis, self-regulation and structural stability; (2) equipotentiality, that is to say the fact that structures cannot be reduced to the interaction of components but include a reciprocal determination of places, of posi­

tional values;

(3) equifinality and homeorheses, that is to say the fact that development is itself structurally stable as process, its final state being in large part independent of its initial state; (4) the drastic limitation of the set of elementary structures and the existence of constraints, of "laws of form"; (5) the "generativity" of forms, the opening up of this limitation toward complexi­ ty. These concepts are in fact truly regional categories for the region of morphological phenomena. They subsume fundamental properties of self-organization and ultimately refer back to this efficacy of position according to which a system of connections is a relational and dynamical network of positional values. Their categoriality which, as we shall see later, owes more to linguistics than to physics, determines the type of theory that is able to render the morphological and dynamical concept of structure intelligible. The main problem is to (transcendentally) deduce their objective value, enabling them to go beyond their reflecting status and thus permitting dynamical structuralism to go beyond the non-scientific stage of speculative vitalism.

Structure

1 003

Gestalt Theory and ''·Phenomenology In the domain of psychology, structuralism is exemplified by Gestalt theory where the same themes, the same problems, the same criticisms of reductionism and the same regional categoriality as those which have just been briefly sketched concerning biological morphogenesis can be identified. In his classical introduction, recently reedited, Guillaume (1979) underlines that Gestalt-theory is a rational monism which puts forth the category of structure at the same time in the physical, biological, and psy­ chological regions. In psychology it proceeds first with a critique of the atomism of sen­ sations and associationism and maintains close links with Husserlian phenomenology. The concept of pure sensation is only an experimental artifact, a hypothetical explanatory concept, insofar as a sensation cannot exist outside of perceptual organiza­ tion. Granted, there do exist peripheral excitations (for example, retinal) produced by external stimuli. But these are only controls of global percepts which integrate them cor­ tically and are not their determining cause. The fundamental hypothesis of Gestalt theory is that it is impossible to treat an immediate perception as a system of relations between atomic sensations, for such a system is the product of analysis and analysis is a real transformation of mental states. Neither terms nor relations have therefore an actual psychological existence, and this is why perceptions must be conceived of as "com­ plexes" (to take up a term borrowed from Meinong), as Gestalten, as "organic unities which individuate and limit themselves in the spatial and temporal field of perception or of representation" (Guillaume 1979). These organized and internally articulated struc­ tures are the result of an original formative activity. What distinguishes them from sys­ tems of components in interaction is again the existence of connections determining pos­ itional values. They are non-summable totalities, indecomposable into elements and whose "moments" do not possess the status of independent parts detachable from the whole (cf. ''·Merleau-Ponty 1948). Rather than seeking for a phenomenological description of structures, the Gestaltists tried to theoretize them dynamically as natural phenomena. They put forth the hypothesis (brilliantly confirmed since then, cf. below) that "the principles of dynamics, because of their generality, extend beyond their purely physical applications" (Mer­ leau-Ponty 1948:56 ). As Guillaume remarks: "the explanatory value of the Gestalt depends upon the systematization which it will succeed to establish between the various forms, upon the construction of a sort of Dynamics which will make the laws of their transformation appear" (1979 :251-252). In other words, Gestalt theory concieves of organized beings, whether physical, biological, or psychological, "as subject to very general dynamical laws, those of organized wholes, which are neither specifically physi­ cal, nor specifically psychological, but common to physics and psychology" (Guillaume 1979 :253). Concerning these, Kohler speaks of Eigenstruktur governed by a principle of functional proximity. Thus, at the level of natural phenomena (physical and biological), as well as at that of perception and of language mediating perception and world, before the structuralism of the 1950s and 1960s the category of structure was worked out in detail, at the beginning of the century, within the context of Austro-German philosophical rational­ ism, where the following bodies of ideas were closely interlinked: (1) the conceptions of the founders of structural psychology, Stumpf, Meinong, and Ehrenfels (all students of Brentano, Carl Stumpf being himself the teacher of

1004

(2) (3) (4) (5)

Structure

''·Husserl and of the Gestalt theoreticians of the Berlin School: Wertheimer, Kohler, and Koffka); Husserlian phenomenology (that of Logical Investigations); Hilbertian axiomatics; ''·Wittgenstein and the Vienna School's reflections on language; through Jakobson, the '�Prague Circle's reflections.

Following the intellectual collapse of Germany during the Nazi period, the rational unity of this context was irreversibly disjointed. There now remain only mathematical

structuralism, logico-combinatory structuralism in linguistics and anthropology, and its misrepresentation by analytic philosophy of language and logical empiricism. In particu­ lar, the foundations of modern structuralism, rooted in biological vitalism, phenomenology and Gestalt theory, have been completely clouded over. It is therefore necessary to say a few words about them. (I shall here follow the important updated work on this point recently carried out by Kevin Mulligan, Barry Smith, and their col­ leagues ; cf. Smith 1982.) All of the classical debates about the relations of parts in wholes, from their Aristote­ lian origin up to the empiricists and passing through the Scholastic, were taken up again in a new vein from Brentano onward. A good many of these problems can be considered conceptually - but not necessarily scientifically - solved (for example, the physical content of the relations of causality and of interaction, the set theoretical notions of an element belonging to a set or of inclusion of a class in another, the nature of spatial con­ nections in a space, the relations of syntactic dependency in a logical formula, etc.). But others, considered crucial by phenomenologists and Gestalt theoreticians, are still very open. Let us menti_on a few of them:

(1) That of the objective correlate of the classificatory relation between genus and species, whether in biology or in other domains (i.e., that of the objective value of classifications, or, in other words, of a realistic, ontological conception of abstrac­ tion). (2) That of the objective correlate of relational accidents, whether static (such as con­ tractual or kinship relations), or dynamic (such as actantial relations). This central problem (which shall be discussed below) is that of the states of affairs, which, without being objective in the strict sense of the term, are nonetheless the objective

correlates of their linguistic description.

(3) That, as we saw, of the general problem of organization, whether of biological organisms or of perceptual Gestalten. (4) That of what Husserl called the non-separable parts of a whole, or again that of the dependent "moments." For example, in the perception of an object, a perceptible quality such as a color is non-separable, except by abstraction, from spa­ tial extension. In the same way the apparent contour of a form is non-separable. It cannot exist in an independent manner.

This last problem, that of dependency relations (also called relations of foundation) between a moment and the whole of which it is a non-separable moment, was investigated in detail not only by Stumpf and Meinong but also by Husserl (see Smith 1982). It can be looked at in two ways: either as a psychological problem, which opens up onto the field of Gestalt theory, or as a general ontological one. This last one is the

Structure

1005

route chosen by Husserl when he conceived of the concept of dependency as a formal concept and attempted to work out an "axiomatized" formal ontology of it. It is extremely ambitious, since it presupposed (a) the realist hypothesis that the relations of foundation (also called "metaphysical connections" by Husserl) are not simply psycho­ linguistic (i.e., valid only for mental contents and their expression), but a priori valid for every realm of objects and (b) that they therefore possess an objective content (within the framework of what later on would become for Husserl the correlation noese-noeme). Its importance should not be underestimated since ( 1) it played a basic role in Gestalt the­ ory; (2) applied to syntactic units, to what Ehrenfels and Meinong called objects or com­ plexes of superior order, it decisively influenced the Polish school of logic (in particular *Lesniewski and �-Ajdukiewicz) and, through this school, everything done since in the realm of ' 'purely logical" and ' 'categorial" grammars; and (3) it is at the origin of Jakob­ sonian phonology, *distinctive features typically being dependent moments: pho­ nemes are neither equivalent classes of allophones nor descriptive artifacts, but formal units constituted by relations of foundation, relations that are real in the sense of an onto­ logical autonomy of the phonological level (cf. below). The State of Affairs (Sachverhalte) Before bringing up issues related to structuralism in semiolinguistics, it is necessary to sketch a crucial notion (which has generally been misinterpreted) linking linguistic structuralism to Gestalt theory, and originating in the Hegelian realist conception of dependency relations. We mean the famous notion of "state of affairs" (etat de choses, Sachverhalt). In order to understand, beyond ''denotation, the descriptive relation between language and external reality, one must postulate a third term. If one considers a ''proposition describing an external fact (for example, an actantial interaction), then it is necessary to suppose that its syntacticosemantic structure reflects an objective corre­ late, in other words that there is an "objective " structuration of the fact - a system of real structural connections - which is linguistically expressed. Now such an "objec­ tive" structuration is not of a physical nature. It does not exist materially. It subsists ideally. Yet, though it can be expressed linguistically, nor is it of a linguistic nature. It constitutes a third, phenomenologically real term between expression and reality, a medium called a "state of affairs." Granted its nature of third term, the concept of state of affairs can be looked at from two opposite sides. The first way, and this has become the dominant point of view in the *philosophy of language, by identifying it with the external fact whose occurrence guaranties the *truth-value of the proposition. Whatever sophistication is brought to it (for example, by intensional logics to explain the opaque contexts and the de dictolde re gaps, etc.), this denotative conception reduces the duality between syntax and semantics to the one developed by the logical theory of models. The second way is by attempting to understand how, as an "objective" structure, as a phenomenological invariant, as a reality which is neither physical nor logicolinguistic, it can emerge from the external fact. This point of view, which is much more restrictive, enables one to constrain linguistic structures by these of reality. It has, therefore, no more to rely on innatistic hypotheses (for example, of a Chomskian nature) to explain lan­ guage *universals. This is the point of view adopted by Thom.

1006

Structure

Can we not admit . . . that factors of phenomenological invariance, which create in the observer the sense of singification, come from real properties of objects of the external world, and manifest the objective presence of formal entities linked to these objects, and about which it could be said that they are the bearers of signification (Thom 1980a: 170 ).

Obviously, such a position is tenable only if - and this has never been the case until now for lack of theoretical tools - one can (1) attain a synthesis ofphenomenological appearance and objective reality; (2) mathematically define, from the point of view of the mathematics

ofphysics, the ''lormal entities" which are the ''lactors ofphenomenological invariance"; and

(3) therefore objectively understand the qualitative structuration of reality in states of affairs. As shall be shown below, the concept of catastrophe provides the first synthesis between phenomenology and physical objectivity. It permits us for the first time to justify (from a non-idealist perspective) Husserlian realism by saying, with Thom, that The proper geometrico-topological analysis . . . enables us to associate with every spatio-tem­ poral process certain invariants of a combinatory nature [catastrophes] . . . that one can reason­ ably think play an essential role, by their fundamental character, in the verbal description of the process. According to us, this is the source of the original schematism which governs the lin­ guistic organization of our vision of the world. (Thom 1 980a :24) The primary function of language [being] to transcribe in a form communicable by our organs the phenomenological catastrophes of the external world, ( . . . ) the message bearing an au­ tonomous signification inherits the structure ofthe external catastrophe that it is supposed to signify. (Thom 1 972 : 329).

In the absence of such a theoretical tool one is obliged to postulate, with Husserl or with Wittgenstein, that physics, even if it were perfect and complete, could not attain an adequate description of phenomenological states of affairs and that these are there­ fore apprehensible only through their linguistic expression. But then they become indistin­ guishable from it and consequently one is led to postulate that a proposition simulates pictures - through its logico-linguistic structure a real (non-logicolinguistic) state of affairs, without however being able to say anything about the autonomous reality of the latter. Since Wittgenstein's Tractatus this theoretical stumbling block is manifest in all con­ temporary linguistics and analytical philosophy which, as Ouellet (1982 : 2) noted, "were constituted only by eliminating from their object and their method any phenome­ nological and ontological preoccupation which could have founded the concept of representation otherwise than on its accepted history," and where "the relation of lan­ guage to the world - a relation which is language- is not explained but simply relayed by the relation of a metalanguage, conceived as a set of formulae . . . , to a language per­ ceived as a formal system" (1982 : 1). In the absence of a synthesis of phenomenology and physical objectivity one cannot overcome the following alternative : "Is it language which gives being its place, creating in the world those discontinuities which we call states of affairs . . . those states to which we refer in thinking and in speaking; or is it rather these states of affairs which, perfectly constituted as phenomena in the being, lend themselves to discourse, to which they give their place in thought and in language? (Ouellet 1982:10). In the Tractatus, the nature of the logical image (the proposition as "picture") is dual. On the one hand, it concerns the structural unity of the '�proposition, and, on the other, it establishes a homology between that structure and the structure of the correlative state

Structure

1007

of affairs, more precisely between the syntactico-semantic connections of the proposition and the real connections of the state of affairs. From this point of view, the logical image (Form der Abbi/dung) is, in its form, the very form of the manifestation of the state of affairs (Form der Darstellung). This is why the logical form (logische Form) becomes identified with the form of reality (Form der Wirklichkeit) (Ouellet 1982:47-48). For Wittgenstein, it is this relationship of pictorial similarity that enables us to identify the meaning of the proposition with the evaluative conditions of its truth-value. The possibility for an object to occur in a state of affairs (its logical form) and that of a proposi­ tion to have a truth value (its form of representation) are included in the possibility that a logical image has of being structured in its own way in accordance with the reality it represents (that is to say in its form of reproduction) (Ouellet 1 982 :52).

In other words, contrary to what happens in the logical theory of models, the comprehension of the relations between language and world presupposes the elucidation of the ways in which the qualitative structure of a state of affairs can emerge from objective reality. For, if this were not the case, what meaning could the supposed homology between a propo­ sition and a correlative state of affairs have? For Wittgenstein, the qualitative structura­ tion of reality into states of affairs corresponds to the way we think it. By thinking reality according to such and such a state of affairs, we apply to it the correlative proposi­ tion, this projection constituting the form of �·meaning (Form der Sinn). In other words, for Wittgenstein, there exists an equivalence between the way in which a state of affairs (conceived as a system of real connections) manifests itself, and the way in which the meaning of the proposition which supplies the logical image is thought. But, unless one wants to return to a traditional logic of representation, then, within this equivalence, it is necessary to go from manifestation to meaning and not from meaning to manifestation. It is necessary to make explicit objective "formal entities" by means of which the "fac­ tors of phenomenological invariance" exist, or, in other words, to root propositional meaning in the qualitative structuration (among other things perceptual) of reality. Structuralism in Phonology As far as the sciences of language are concerned, the structuralist conception can be traced back to Saussure and, in particular, to the elaboration of the fundamental concept of paradigm (in the sense of the �·opposition between the paradigmatic and the syntag­ matic axes). The fundamental contribution of Saussurean structuralism is, as we know, to have substituted in the study of language relational criteria of identity to the classical substantial ones. The identity of a linguistic unit in a paradigmatic system is a value, that is to say, a purely positional identity. By using a "geographical" analogy, a paradigm can be viewed as a categorized domain D, as a domain which is decomposed into sub-do­ mains Di by a system K of boundaries. Each sub-domain Di is defined by its •:•extension, in other words by the categorization K. It is in this sense that the structure does exist : the global organization K determines, by being implicitly present in each of them, the local units Di. A paradigm is therefore not a system of relations between terms defined before­ hand. As for their value, the terms of a paradigm (which can obviously possess in addition a substantial identity) do not have an autonomous existence. They can be defined only through their reciprocal determination. As specific interpretation of the third Kantian

1 008

Structure

category of relation, the category of reciprocal determination is the fundamental cate­ gory of structuralism. (Obviously one should not mistake category here in the philoso­ phical sense for category in the sense of a sub-domain of a categorized domain. In the same way paradigm in the Saussurean sense should not be mistaken for paradigm in the Kuhnian sense.) It gives meaning to the famous structural "axiom" according to which difference ontologically precedes identity. Saussure is very explicit on this point (Du­ crot 1 968 ). According to him, there are no intrinsic boundaries which delimit the phonic and semantic zones covered by the *signifiers and the signified of language. Each term of a paradigm tends to "invade " the whole paradigm, its domain (its value) being limited only through its conflict with the other ones. The determination of a positional value is therefore a purely negative one through a process of limitation, limitation consti­ tuting here a specific interpretation of the third Kantian category of quality. Since they are subsumable under the categories of reciprocal determination and of limitation, the relations between the terms of a paradigm are structural connections, relations of dependency or of foundation in the sense used above. It is in this very sense that, for Saus­ sure and for semiolinguistics in general, language is a form and not a substance (Coquet 1982). The concept of paradigm was worked out by Jakobson in a phenomenological perspective as the founding concept of phonology. Whereas allophones are substantial units of an audio-acoustic nature and content (units of the substance of expression in Hjelmslev's terminology), on the contrary phonemes are abstract, distinctive and dis­ criminatory units, of a functional and linguistic nature and content (cf. the classic Preli­ minaries to Speech Analysis by Jakobson, Fant, and Halle ( 1 952); also Jakobson and Waugh 1 979 and Malmberg 197 4), describable as bundles of dinstinctive features and subject to formal laws. The theoretical problem is therefore to understand the link between phonetics and phonology, between the organization of the substance of '1expres­ sion and the articulation of the form of expression. How phonological categorizations and stratifications, whose description is the very object of phonology, can emerge as struc­ tures from the phonetic substratum, from the audio-acoustic flux? For a long time this problem has acted as a sort of antinomy within the area of pho­ netics. It has led to a conflict between, on the one hand, the "substance-based" reduc­ tionist conceptions, which reduced phonological descriptions to epiphenomena! arti­ facts without objective value, and, on the other hand, the "form-based" structuralist ones, which asserted the ontological autonomy of the fqrm of expression. For these the phoneme is "a unique differentiating unit without concrete qualities, but manifested, in speech, by an allophone having physical qualities (physiological, acoustic, perceptual) which translate their differential qualities in the world of physical realities" (Malmberg 197 4 :220 ). In other words, the form of expression is an abstract system, comparable to the Aristotelian morphe, which is realized in the substance of expression, in the concrete process of speech. If one admits that it governs phonetic perception, then one is forced to proceed from the abstract to the concrete : "description proceeds . . . from the abstract and the functional to the concrete and material, from form to substance" (Malmberg 1974:30). But the substance of expression is not a sort of undifferentiated hyle which would be informed by an ideal form, an essence, an eidos, foreign to it. It is highly organized. To avoid the speculative pitfall of idealistic holism, and in order to become an authentic structural doctrine (epigenetic and relational), phonological structuralism must there-

Structure

1009

fore make clear how the articulation of a discrete form can emerge from the organization of a continuous substance. But this leads to a highly complex problem. As Ladefoged noted: "one of the major difficulties in achieving this kind of description is in relating the essentially continuous nature of speech with the essentially discontinuous nature of linguistic description" (1972:276). And Didier Pisoni adds: "[The] lack of corre­ spondence between attributes of the acoustic signal and units of linguistic analysis has been, and still currently is, one of the most important and controversial issues in speech perception" (1979:334; cf. Massaro 1972). The key to solving this difficulty can be found in the very structure of phonetic perception. As Derwing noted: Linguistic "structure" . . . , if this term refers to anything real at all, must refer to representa­ tions or interpretations imposed upon the speech signal by language users, normally as part and parcel of the *communication event itself. In short, psychological reality is not merely a convenient luxury which linguistic theory may or may not choose to be concerned with, but is rather the sine qua non for any linguistic construct which aspires to anything more than an epi­ phenomena! or artifactual status, and hence for any linguistic theory which can justifiably claim to go beyond the bounds of an arbitrary taxonomic system. (1972:86)

As we shall see, the fundamental character of phonetic perception is to be categorical. The important works on the morphology of phonetic sounds (the analysis of sonograms) developed since the early 1940s, particularly by the Haskins Laboratories (in particular Pierre Delattre, Alvin Liberman, Franklin Cooper, Michael Studdert-Kennedy, and Kenneth Stevens) have shown that these sounds depend on a small number of para­ meters called acoustic cues. Modern methods of acoustic synthesis enable the continuous variation of the cues. If subjects are subjected to tests analyzing their perception of such continua, for example of a continuum W going, by continuous variation of the voicing cue, from the syllable [ba] to the syllable [pa], one notices two things:

(1) As far as the identification of the stimuli is concerned, there is, as could be expected, the formation of an interface K, of a threshold, of a boundary, of a frontier, of a catastrophe set, categorizing W into two domains which respectively correspond to [ba] and to [pa]. (2) As far as the discrimination of two adjacent stimuli is concerned, there is no intracategorical discrimination: subjects discriminate two neighboring stimuli only if they are situated on different sides of the interface K, in other words only if they are identified as being different. Discrimination is therefore subordinated to identifi­ cation: '' categorical perception refers to a mode by which stimuli are responded to, and can only be responded to, in absolute terms" (Studdert-Kennedy et al. 1 970 ; cf. also Pisoni 1979). It is this latter characteristic which defines phonetic perception as categorical. Categorical perception is drastically different from continuous percep­ tion (such as in the case of colors), where discrimination is essentially independent of categorization.

Discovered in 1957 by Liberman, the deep phenomenon of categorical perception enables us to understand how perception can spontaneously render discrete the audio-acoustic flux, in other words how the discontinuous can emerge from the continu­ ous. As Delattre showed in his famous article of 1968, "From acoustic cues to distinctive features" (cf. also Jakobson and Waugh 1979), as well as Liberman, Cooper, Schank-

1010

Structure

weiler, and Studdert-Kennedy, in their equally well-known "Perception of the speech code" (1961), it permits us to link the audio-acoustic level of phonetics (the organiza­ tion of the substance of expression) and the linguistic level of phonology (the abstract relational content of the form of expression): the phonemes encoded in the audio-acoustic flux being categorical, they have a psychological reality as discrete units. Since the pioneering work done in the 1960s, the study of this phenomenon has greatly progressed. In particular, it was shown that it is for a large part innate: the learning of a specific phonological system consists essentially in deforming (and not in forming) an initial categorization K0 through operations of displacement, erasing, and bifurcation of its thresholds (cf., e.g., Eimas 1980). But debate is still open concerning its explanation. In our opinion, the most interesting and most plausible hypothesis was put forth by Stevens, in his 1972 article "The quanta! nature of speech." Stevens considers percepts as Gestalten whose qualities are global spectral patterns of a superior order and are controlled by acoustic cues (whose content, according to Stevens, is articulatory) var­ ying in a control space W. One then has only to suppose that phonetic perception is qual­ itative in order to understand its categorical character. Indeed, when control (acoustic cues) varies in a category delimited in W by the catastrophe set K, the percept is stable and remains qualitatively invariant. Perception being qualitative, no intracategorical dis­ crimination can occur. On the contrary, in crossing K, one of the qualities of the percept becomes unstable and the qualitative type of the percept changes radically, and so does its identification. As Stevens notes, "there are certain articulatory conditions for which a small change in some parameter describing the articulation gives rise to an apparently large change in the acoustic characteristic of the output; there are other conditions for which substantial perturbations of certain aspects of the articulation produce negligible changes in the characteristics of the acoustic signal" (1972 :52). The import of this type of explanation is to make the phenomena of categorical per­ ception perceptive cases of critical phenomena, analogous to phase transitions in thermo­ dynamics. We can therefore apply to it (and to phonological categorization) the universal models of critical phenomena which are actually the catastrophe theory models (cf. Petitot 1982a, 19826, 1983a). Actantial Structures and �·case Grammars In the syntactic domain, the structural approach goes back to the work of Tesniere (1959). For Tesniere, a •:•sentence is first of all a system of connections wich are, "incor­ poreal" (non-perceptible) and can be apprehended only by "mind." These oriented and hierarchized structural connections (relations of dependency in the sense used above), are not of logical essence and constitute an "organic and vital" principle of organiza­ tion (this is the way Tesniere reformulated the Humboldtian innere Sprachform). With his so-called stemmas, Tesniere gave a graphic representation of them which is at the origin of the syntagmatic trees found since then at the basis of all formal descriptions of language. "The stemma clearly shows the hierarchy of connections. It schematizes the various nodes that link them into networks, and thus visually materializes the structure of the sentence . . . The structural scheme of the sentence is none other than the visual representation of an abstract notion" (Tesniere 1959:sec. 3-9-3.10). For Tesniere, the structural connections define the functions; that is to say, the roles assigned to words "in the mechanism of expression and thought" (1 959:sec. 19.4). They are projected on the

Structure

1011

linear order of syntagmatic concatenations: "Structural syntax is based entirely on the relations that exist between structural order and linear order' (19 59). The modern development of generative-transformational grammars and of genera­ tive semantics could lead us to believe that the Tesnierian point of view has been ade­ quately /onnalized, and therefore superseded. But this is far from being the case. Actually, these developments stem from a static, taxonomic, formalist, and logico-com­ binatory conception of syntactic structures and lead to a more or less sophisticated alge­ braization of them. From this point of view they are far removed from Tesniere dynami­ cal, "vitalistic," Gestaltist perspective. Indeed, Tesniere always insisted on the fact (1) that syntactic structures are self-regulated organizations analogous to biological ones, (2) that structural syntax is neither a logical grammar nor a psychological semantics, (3) that it is functional and dynamical and not categorical (in the sense of grammatical cate­ gories) or static. His conception is an actantial- and therefore a "scenic" - one cen­ tered on the concept of verbal valence. "The verbal node . . . expresses a complete mini­ drama. Indeed, like a drama, it obligatorily comprises a process, and generally actors and circumstances". (Tesniere 1959: sec. 48.1.) After the development of the generative-transformational grammars one had to await the case grammars of Fillmore (1968, 1969, 1971, 1972b), Chafe, and Anderson (1971, 1975), and, more recently, the relational grammars of Keenan (1972), Comrie, and Johnson (1974, 1977; cf. Cele and Sadock 1977) for this aspect to be again taken into consideration. Fillmore, for example, reverted to a scenic and actantial conception of syntactic structures. Based on a semantic interpretation of deep syntactic structures, classical case grammars, in spite of their success, encounter deep difficulties in the definition of case universals. Their fundamental hypothesis is that there exists a finite list of case universals (or functional categories such as Agent, Dative, Instrumental, Loca­ tive, Objective, Beneficiary) whose notional content is definable. These deep cases sup­ posedly select semantic roles (actantial roles), which, although semantic, can be unco­ vered and justified by syntactic criteria (the thesis of the primacy and centrality of syn­ tax). But as soon as different languages are compared, a conflict arises between, on the one hand, the proliferation of cases, brought about by their conception as sentence dis­ criminators, and, on the other hand, their limitation, brought about by their conception as universals. If we want to attribute a discriminatory notional content to deep cases, we must become involved with the lexical semanticism of verbs, and therefore cases begin to proliferate. If, on the contrary, we wish to attribute to them a general enough notional content for the list not only to be finite but also limited, then this content imme­ diately becomes too general. As Willems has noted, the case roles "end up by losing all semantic value" (1978 :247). In short, the major difficulty, clearly formulated by Fill­ more, "is the observation that nobody working within the various versions of grammars with 'cases' has come up with a principled way of defining cases, or principled proce­ dures for determining how many cases there are, or for determining when you are faced with two cases that happen to have something in common as opposed to one case that has two variants" (1977:70; our emphasis). A first solution to this difficulty has been proposed by Anderson (1971, 1975). Parallel with the analysis of phonemes into distinctive features, it consists of : (1) treating case contents as complex contents which can be decomposed into case features (multicase analysis of actantial roles); (2) identifying case features with a limited number of universals by taking up a fundamental hypothesis, the so-called localistic

1012

Structure

hypothesis, according to which the positional relations between spatiotemporal actants serve as scheme for the actantial relations in general (on the importance of the localistic hypothesis in structural syntax, cf. Petitot 1979a, 1979b, 1982a, 1982b, 1985); (3) positing that verbs select the case features; (4) elaborating a "generative grammar" of these selections. Fillmore's solution is different. It consists in : ( 1 ) noting that several different semantic fields can serve as substratum to the same abstract scheme of actantial connec­ tions, and therefore, (2) distinguishing in case semanticism the part coming from the spe­ cific semantic field considered and the part - purely positional - defined by the "stem­ ma". As Potts suggests : "wherever there are isomorphisms between sets of inter-rela­ tionships we should . . . recognize only one set of semantic roles. The differences between the roles in the two situations will be accounted for by the different semantic fields in which they lie, the semantic field working upon the role to modify it" ( 1 978 : 454 ). Fillmore called these semantic fields "scenes. " Each scene is lexico-syntacti­ cally organized by a limited number of specific constructions which select relative cases "presenting precise syntactic and lexical characteristics" (Willems 1978 :247). Hence the slogan "meanings are relativized to scenes" (Fillmore 1 977 :59). In order to develop this scenic conception which, according to us, is truly structural, Fillmore resorts to the notion of case *frame, the function of which is to mediate the descriptions of the situations and their underlying syntactic representations. The case frame assigns semantico-syntactic roles to the actants of the process expressed by the sentence, and this assignation constrains the choice of a perspective selecting, according to case hierarchy, one of the actants as the grammatical subject. In so doing, Fillmore maintains a conceptual definition of case. He puts it at the interface of thought and lan­ guage and takes up the Wittgensteinian idea of a "pictorial" similarity between the syn­ tactic structure of a sentence and the scene (the state of affairs) it describes : "such de­ scriptions [are] in some sense intuitively relatable to the way people thought about expe­ rience and events that they [are] able to express in the sentences of their language" (Fill­ more 1 977 :62). This point of view shifts semantic analysis towards •}cognitive analysis. "The study of semantics is the study of the cognitive scenes that are created or activated by utterances" (Fillmore 1 977:73). We choose and understand linguistic expressions by "exciting" in our minds scenes, memorized prototypical situations relative to which expression has a descriptive and classificatory function. In other words, when set into perspective, an expression evokes the global ground from which it detaches itself. "It is as if descriptions of the meaning of elements must identify simultaneously 'figure' and 'ground"' (Fillmore 1 977 :7 4 ) . Thus, i n structural syntax, we find once more the same Gestaltist and phenomeno­ logical problems raised above. Language universals must be rooted in the perceptual and cognitive organization of the state of affairs. As Osgood noted : "It seems perfectly rea­ sonable to think that much, if not all, that is universal in human language is attributable to underlying cognitive structures and processes . . . Perceptual and linguistic signs and sequences must, at some level, share a common representational (semantic) system and a common set of organizational (syntactic) rules, cognitive in nature" ( 1 97 1 ) . "We must look for the roots of basic linguistic structures in the relations between the active subject and reality and not in the mind itself, " according to A. R. •:•Luria ( 1 975 ) . The most delicate o f these problems consists i n achieving a con/igurational definition of case semanticism. Indeed, even if case contents are relativized to scenes, nevertheless

Structure

1013

it remains true that basic case semanticism is a purely positional one. Notional (actantial) contents of cases cannot be defined as autonomous contents, independent of one another. They can be defined only in terms of positions in schemes of actantial connec­ tions, positions determining and presupposing each other as values in a paradigm. The problem of structural syntax is a dual one :

( 1) How can the schemes of actantial connections emerge, as self-regulated struc­ tures, as dynamical morphologies, as syntactic Gestalten, from the phenomenolog­ ical organization of reality in states of affairs? (2) How do these schemes permit the definition of case semanticism in terms of positional values?

To solve this problem it seems necessary to return to the localistic hypothesis and to posit that the possible structural connections between spatiotemporal actants have served as the matrix for the syntactic schemes in general. (For an historical presentation of the localistic hypothesis, see Hjelmslev 1 935.) Such a hypothesis has been convincingly confirmed by Thom's interpretation ( 1 972, 1 980a) of case universals in terms of elementary catastrophes. This interpretation makes possible the resolution of nearly all the theoretical difficulties raised here (cf. Petitot 1 979a, 1 979b, 1982a, 1 982c, 1 98 3). As Wildgen noted ( 1 98 1 , 1 982), this constitutes a decisive step which brought into question many dogmas of generative-transformational grammars, of generative semantics, and even of case grammars. The structure of the elementary interactions which are derived from paths in the bifur­ cation space of elementary catastrophes, defines different roles which can be roughly compared to the 'schemas actantiels' proposed by Tesniere and to the 'case frames' clas­ sified by Fillmore. The basic difference between the structures and the [catastrophe] semantic archetypes consists in :

( 1) The preverbal character of archetypes. The structures proposed by Tesniere, Fill­ more and others are only generalizations of linguistic structures found in natural languages. (2) The foundations of the classification of archetypes in a formalism which is sup­ posed to be basic for many biological systems. It is therefore universal in a very deep sense and it is of interdisciplinary relevance. (3) The semantic archetypes are irreducible gestalts. They are not composed in a sin­ gle combinatorial way. This fact constitutes a major difference in Thom's theory against all theories proposed up to now. Some of these have tried to describe field­ like structures, but as no tool for consequently doing so was available they all drove away irresistably attracted by the static-logical paradigm. (Wildgen 1 98 1 :264-65) Semio-narrative Structures

Actantial structures do not exist only at the sentence level. They do not have only a lin­ guistic reality but also a narrative one. When narrative corpa are examined, for example '�myths, fairy tales, tragedies, novels, etc., it can be seen that at the (apparent) superficial level, these narratives relate plots between characters (actors) localized and spatiotem­ porally evolving, defined by thematic roles, linked by civil or contractual or emotional relations, and interacting together through conflicts, gifts, contracts, separations, unions, etc. Beyond this proliferation of superficial discursive structures, deep struc-

1014

Structure

tures, called semio-narrative by Greimas, can be identified. The interest of this proce­ dure (standard in semiolinguistics) is dual. It shows that: (1) abstract (non-figurative) discourses, whether philosophical, political, or scientific are also organized according to analogous structures, and (2) deep semio-narrative structures are actually expe­ rienced in passions, *ideology, action, dreams, etc. These two facts lead to the idea that, using an expression borrowed from Gilbert Durand, semio-narrative structures are anthropological structures of the imaginary, and that there exists a "grammar" for them. One of the major interests of a semio-narrative grammar is to lay bare a relation between syntax and semantics which is very different from that ordinarily transferred from logical models to linguistics. We mean the projection (or conversion) of the para­ digmatic axis onto the syntagmatic axis. Its understanding constitutes one of the central problems of structuralism. The identification of a syntactic component in the deep semio-narrative grammar goes back to •:•Propp's work on the Russian folktale: The Morphology of the Folk Tale. Underlying the various plots of the characters, Propp isolated /unctions (typical actantial relations) canonically linked, expanding in a determined way as in a process of morpho­ genesis, and expressed in typical sequences: establishment of an initial lack (for exam­ ple, the transgression of social norms, betrayal, etc.) contract between an Addresser and a hero, succession of tests, first qualifying tests by which the hero acquires the modali­ ties of wanting, knowing, and/or being able to (for example, the acquisition of a magical weapon through the intermediary of a helper), then the decisive test where the hero accomplishes a performance liquidating the initial lack (for example, killing a drag­ on), finally the glorifying test where the hero's performance is sanctioned by the Addresser. In other words, Propp identified and formulated invariant, stable and universal actantial structures. These are governed by an actantial syntax which syntag­ matizes, along what Greimas has called a global narrative schema, an actantial paradigm constituted by the relations Addresser/'�Addressee, Subject/Object value, Subject/An­ ti-subject, Helper/Opponent (Greimas 1970; Greimas and Courtes 1979). After Propp's work, credit for identifying a semantic component in the deep semio-narrative grammar must go to Levi-Strauss' work on myth (1958; 1964-1971). Without doubt this is the most delicate and most difficult part of the structural analysis of narrative. Indeed, this deep semantic component is very different from the superficial, discursive and figurative semantics, which spreads the lexical semanticism in the superfi­ cial syntactic structures. If one remained at the simple discursive-figurative level of the deep actantial structures, then obviously one could still analyze narratives as myths. But one could not understand their anthropological /unction. It is precisely this function that Levi-Strauss attempted to define. To do so he had to lay bare the logical coherence of myth and therefore to adopt a semantic and non-morphologico-syntactic approach. Hence his critique of Propp. By reverting to the classical conception of the paradigma­ tic-syntagmatic relation (according to which the paradigmatic dimension is reduced to a dimension of substitution which can be laid bare by means of commutation tests), Propp could uncover only syntagmatic invariants and, thus, was able to elaborate only an actantial syntax (cf. Coquet 1982). Inspired by Jakobson's work in phonology, and borrowing the Hjelmslevian principle of parallelism between the plane of expression and the plane of content, Levi­ Strauss was thus led to introduce, in the theory of deep semionarrative structures, the

Structure

1015

second fundamental aspect of the paradigmatic dimension, not substitution but categori­ zation (cf. above). His famous and exemplary analysis of the Oedipus myth ( 1 95 8 :236) is well known and, whatever the critiques made of it since, it had the great merit of showing that the syntagmatization of narrative was governed by semantic homology: the overevaluation of kinship relations (Cadmus seeks his sister Europa, carried off by Zeus; Oedipus marries his mother Jocasta; Antigone buries her brother Polynices in spite of the ban) is to their underevaluation (the Spartoi mutually exterminate each other; Oedipus kills his father Laius; Eteocles kills his brother Polynices) what the dene­ gation of autochtony (Cadmus kills the dragon; Oedipus slays the Sphinx; the dragon and the Sphinx being chtonian figures) is to its affirmation (affirmation manifested by the proper names, Labdacos [father of Laius] signifying "lame," Laius "askew," and Oedipus "swollen foot"). The fundamental idea is therefore that, underlying narratives, and particularly myths, there do exist deep semantic categorizations articulating values, constituting unconscious codes (kinship, natural, cosmological, economic, culinary, etc.), and projecting themselves syntagmatically. Although they can be designated within an ade­ quate metalanguage, these deep values cannot be identified with the superficial semanti­ cism. They operate in a contextual and global fashion. They select in the (highly overde­ termined) superficial lexical figures specific semantic features Csemes). But the con­ tents they articulate (Life/Death, Nature/Culture, Man/Woman, Divine/Human, etc.) are not referentially definable significations. They are like psychic "forces" (drives and/or ideals) "giving meaning to life" and whose meaning is never grasped as such but only through conversion into actantial structures. More precisely : ( 1 ) The deep semantic structures are imaginary universals. (2) Unconscious, they exist only when axiologized and when ideologically invested in object values whose quest governs actions (narrative programs in Greimasian ter­ minology) of the subject actants. (3) Only the circulation of object values governed by actantial syntax enables their apprehension. In other words, they are not subjectivable as such but only through a logic of action. ( 4) The function of actantial syntax is thus to convert into a narrative doing the fun­ damental semantics determining the deep anthropological meaning of the narrative. It is to solve, by simulating a "drama" staging them, the paradoxes of subjectivity. Thus the elaboration of a semio-narrative grammar must deal with three problems :

( 1 ) What is an elementary narrative structure (we already know that it is an actan­ tial one)? (2) What is a semantic categorization (we already know that it is one analogous to those found in phonology)? (3) What is the nature of the conversion of the fundamental semantics into syntax? What about the projection of the paradigmatic axis onto the syntagmatic axis?

These are (among others) the three questions Greimas' theory deals with. That is why in concluding this presentation of some aspects of structuralism, we would like to make a few remarks about it.

1016

Structure

In what concerns fundamental semantics, it is necessary first to define /onnally the morphologies of categorization constituting the /onn of the content (in Hjelmslev's sense). This is the aim of the celebrated semiotic square (for more detailed reflections on the semiotic square, see B.G.R.S. 1981 and Nef 1976 ). Of a morphological nature, the semiotic square is a universal scheme of articulation ensuring, according to Greimas, the minimal conditions for grasping the meaning. "First nucleus of an elementary mor­ phology", (Greimas 1970 :164 ), "first taxonomic instance" (p. 163 ), it operates at the "immanent level . . . where narrativity is situated and organized anterior to its manifesta­ tion" (p. 158). As such it is "susceptible of being transformed into a constitutional semiotic model" (p. 161) which is none other than "the elementary structure of significa­ tion utilized, as form, for the articulation of the semantic substance of a micro-uni­ verse" (p. 161). As a taxonomic form, as a scheme of articulation, as an elementary struc­ ture anterior to any semantic investment, it develops a semic binary category s/s2 which correlates, in a relation of junction (conjunction/disjunction) and of reciprocal '"presup­ position, two contrary semes, s 1 and s2 • This "development," this sort of morphogenesis of the opposition s/s2 proceeds from the relations of contradiction s/s 1 and s/s2 opposing the semes to their "negation." Hence the square: where s and s are respectively complex and neutral terms (in the sense given by Brnndal) associated with the category s1 /s2 • Three types of relations (structural connections) constitute the square : (1) Hyponymic relations s 1-+ s, s2-+ s, s 1-+ s and s2-+ s (as well as the inverse hypero­ nymic relations). These are relations of selection in Hjelmslev's sense. (2) Categorical relations of contradiction s/s p s/s2 and of contrariety s/s2 , s/s2 • (3) Hypotaxic relations of *implication (of unilateral presupposition) s 2-+ s 1 and s 1-+ s2 (as well as the inverse hypotaxic relations s 1 -+ s2 and s2-+ s 1 ) . These are also relations of selection. These various relations can be classified into three "dimensions" : (1) Two axes, the contraries s 1 - s2 and the sub-contraries s 1 - s2 • (2) Two schemas, the contradictories s 1 - S p and s2 - s2 , the first being (convention­ ally) positive and the second negative. (3) Two ,;deixes, the implications s 1 - s2 and s2 - s P the first being (conventionally) positive and the second negative. Handled as a simple logical form and formalized in terms of elementary Boolean set logic, the semiotic square becomes trivial and of little interest. It simply reformulates "logical squares" going back to �·Aristotle. But everything changes if one notices that it is, in fact, a structure in the strong sense of the term, that is to say, an "organic" and self-regulated system of relations of dependency and of foundation (in the sense used above) defining, as a phonological paradigm, positional values. The semiotic square is not of a logical nature. As Utaker (1974) has shown, the relations of contrariety and of

Structure

1017

contradiction are respectively qualitative oppositions and privative oppositions in the Jakobsonian sense. Therefore they must be treated as such. Now the notions of opposition and presupposition refer to the primitive notion of position which is primitively topological and not logical. In the same way, the notions of conjunction and disjunction refer back to the primitive notion of junction, which is prim­ itively topological and not logical. The formal essence of the semiotic square is therefore ulti­

mately dependent upon a dynamic topology of places and connections and not upon a static logic of terms and relations. It is within such a non-trivial framework that we must

formalize it and understand how a connection between two places (which, after the investment of the places by terms, becomes a qualitative opposition s/s2) can develop morphologically without any change in the number of places. It can be shown that this is possible by schematizing the semiotic square by means of elementary catastrophies (Petitot 1977, 1982a, 1983b). Concerning narrative syntax, Greimasian theory conceives of the elementary actan­ tial structures as the syntagmatization, by narrative programs, of the actantial paradig­ matic relations. These can be reduced to the reciprocal presupposition between Addressor and Addressee (dimension of contract), between Subject and Object (dimen­ sion of the acquisition of modal competences and/or dimension of performance), and between the Subject and Anti-subject (dimension of performance). Concerning the basic relation Subject/Object, Greimas in particular conceives of the subject actant as an intentional subject (a subject of lack, a subject of the quest, a subject· of desire) having as goal a semantic value (positively ideologized) invested in an object. That means that the basic narrative program is a program of conjunction between a subject and an object­ value, quite similar to the conjunction between a predator and its prey. This point of view raises a certain number of difficulties. For example, Ricoeur (1980) has emphasized that it is doubtful that the phenomenology of action, which if manifested narratively in all its richness, can be reduced to simple operations of conjunc­ tion and disjunction between subjects and object values. On the other hand, to conceive of semiotic subjects as intentional subjects obviously leads to the raising of questions about their *intentionality. The joint problematics of •:•belief (and of its avatars, seduc­ tion and manipulation) and of selection of object-values are linked to this problem (which would necessitate a reinterpretation of the connections between phenomenology and *psychoanalysis) (cf. Petitot 1983a, 1985). But the major difficulty remains the conversion of the fundamental semantics into actantial syntax, that is to say, the projection of the deep semantic categorizations onto the narrative programs. (For an analysis of certain aspects of conversion, see B.G.R.S. 1982.) In Greimasian generative theory, this difficulty is solved by the introduction of an intermediate level, called the syntax of ,:•operations, linked at the same time to the semiotic square and to the actantial structures. The main idea is that taxonomic relations constituting the semiotic square can be converted into logical operations on the terms they define. As Greimas remarks: An examination of the conditions of the manifestation of meaning clearly shows that if signifi­ cation, insofar as one attempts to find it in the object, appears as an articulation of fundamentally stable relations, it can at the same time be represented dynamically as soon as it is considered as a manifestation or as the production of meaning by the subject. Keeping in mind this dynamic aspect, a network of equivalences between the constitutive fundamental rela­ tions of the taxonomic model and the projection of these very relations, or operations, this time

1018

Structure

bearing on the terms already established in the elementary morphology, can be set up; the governing of these operations would be syntax. Thus, contradiction, as relation, at the taxo­ nomic level establishes binary schemas; as operation of contradiction at the syntactic level it consists in negating one of the terms of the schema while at the same time positing its contradic­ tory term. (1970: 1 65).

If, therefore, an equivalence is posited between, on the fundamental semantic level, a seme s and, on the actantial level, the conjunction S n O linking an intentional subject to an object invested with the semantic value s, then, through the mediation of the logical syntax of operations, the contradictory term s will correspond to the negation of this conjunction, that is to say to a disjunction S U O programming S to "recapture" 0. As operational as it may be, this answer to the question of conversion is still incom­ plete and continues to raise problems. Indeed, it does not permit us to elucidate either the nature of the intentionality, or the mechanisms by which an unconscious psychic "force" can be invested in an object and so confer on it the "aura" of an object-value. But it must be said that these phenomena extend far beyond semiotics and that to under­ stand them one must turn toward a "metapsychology" conceived in the Freudian sense or an anthropology of the imaginary conceived in the perspective of biological regula­ tion. (Cf., e.g., Brandt 1982, 1983 ; Petitot 1982a, 1982b, 1983a; Thom 198 1). The Formalization of Structures

This overview of some aspects of structuralism shows that, as a conceptual system and as a methodology, as a regional ontology in the Husserlian sense, it is intrinsically trans­ disciplinary (biology, anthropology, Gestalt theory, cognitive psychology, linguistics, semiology, etc.), and that in every domain where it can be said to be valid, it covers, as Piaget affirmed, "a common positive ideal of intelligibility" (1968). As an epigenetic and relational doctrine of organization, it constitutes, next to physics, practically the only domain where the rational unity of very different phenomena has been achieved. The question of its formalization is therefore particularly crucial. Now, concerning this most important question, one must recognize that structural­ ism, for lack of appropriate formal tools, has never been equal to its ambitions. Until recently, it has succeeded in developing only a formalistic, logico-combinatorial concept of structure. This is why, in conclusion, we shall attempt a critique of this and indicate an alternative. Here, the formalistic point of view is itself dual and is concerned with either the for­ malization of structural theories, or the formalization of structures conceived of as natural phenomena (as objects of experience). The ideal of formalizing the theories goes back to Husserl's attempts "to axiomatize," in a quasi-Hilbertian mode, a formal ontology of general dependency relations. ("Theory" is used here in the sense of con­ ceptual theory - developing the a priori constituants of a regional ontology - and not in the sense of mathematical theory.) It was taken up by Hjelmslev in the semiolinguistic domain and, for example, is the basis of Greimasian epistemology. From this perspec­ tive, the theory can be considered as a conceptual system (a descriptive metalanguage) which, through a definitional hierarchy, enables one to derive the relevant concepts from basic ones. These basic concepts are thus undefinable. They have the status of regional categories. In Greimasian theory such are, for example, notions as contin­ uous/discontinuous/ discrete, relation, difference, opposition, junction, transforma-

Structure

1019

tion, operation, etc. It is postulated that by giving them a formal expression, the descrip­ tive metalanguage can be "axiomatized" and converted into a formal language, into a "pure algebra" (Greimas and Courtes 1 979:225). The major difficulty encountered with this conception is that it can at best elucidate only the logical architecture of the theory and cannot, in any way whatsoever, lead to an adequate mathematization. In Kantian terms, it is based on a formal logic oriented towards the discursive form of knowledge and not on a transcendental logic oriented towards the objective content of knowledge. Here we meet with a philosophical and epis­ temological obstacle wich cannot be underestimated but wich is generally carefully shaded over. (This shading is essentially due to the importance of logical positivism in the contemporary episteme.) It can be solved only if an explicit mathematical content can be substituted for the semanticism of the regional categories. Taking up the Kantian term of schematism (i.e., of construction of a concept into a mathematically determined intui­ tion), we shall say that no mathematization of a theory (whether structural or physical) is possible without a schematization of its categories: the schematization of categories is the key to every constitution of objectivity (in the Kantian, then Husserlian sense of "con­ stitution"). We cannot help being struck by the fact that, in every domain we explored, the structural categoriality was always the same and that it had a content which, in the last analysis, was rooted in a topological intuition (position, junction, paradigmatic categori­ zation, connection, etc.). As we saw, every structure is, first and foremost, a structurally stable and (self)regulated system of connections between positional values. This crucial fact was deeply illustrated by Deleuze. Every structure is a "combinatory of formal ele­ ments which by themselves have neither form, nor signification, nor representation, nor content, nor given empirical reality, nor hypothetical functional model, nor intelligibility behind appearances" (Deleuze 1973 :303 ). Its elements "have neither extrinsic designation, nor intrinsic signification" (p. 304). They have "nothing but a meaning: a meaning which is necessarily and uniquely a 'positional' one" (p. 304 ). This is why the "scientific ambition of structuralism is not quantitative, but topological and relational (p. 305). This is why "structuralism is inseparable from a new transcendental philosophy, where places are more important than what fills them" (p. 306). But then, the schematization of the structural categories depends entirely on the possibility of mathematically determining "the positional intuition" playing the role of a "pure form of intuition" for structural phenomena. It depends hie et nunc on the elaboration of a geometry of position. Now, as Buffon had already noticed concerning embryology, such a geometry of position was in his time (and so it remained until recently) completely nonexistent: Everything dealing directly with position is totally lacking as far as our mathematical sciences are concerned. This art which •·Leibniz called Analysis situs is not yet born and nonetheless this art, which would make us understand positional relations between things, would be just as useful and, perhaps, even more so for the natural sciences than art which has as its object only the magnitude of things; for often it is more necessary to know form than matter. (Buffon 1744:vol. 4, ch. 9; p. 73 ).

This means that structuralism essentially depends, as far as its mathematization is con­ cerned, on the elaboration of a general mathematical theory of morphogenesis, and there­ fore on one of the most ancient and critical problems in the history of science (cf. Petitot

1 020

Structure

1 985a and b). In the abscence of such a theory, its concepts, although empirically valid, spontaneously drift toward a speculative vitalism which is scientifically unacceptable and can only be avoided by a formalism empty of content. This dialectical antinomy is the modem counterpart to the one developed by Kant concerning biological organiza­ tion (cf. above). This has clearly been operating in modem structuralism. On the one hand, speculative perspectives, under the umbrella of materialism and/or psychoanaly­ sis, have attempted to recover the obscure vitalistic power of Dialectic. On the other hand, formalistic perspectives were able to overcome their deficiencies only by appealing to precarious and uncertain innate hypotheses. (Concerning this debate, which took place during the 1960s, cf., for example, Wilden's classical System and Structure ( 1 972) and the dialogue between Chomsky and Piaget in Piatelli-Palmarini 1 979.) It is only recently, and in particular with the advent of catastrophe theory, that elements of a gen­ eral mathematical theory of morphogenesis have been available to make possible the schematization of structural categories. This is a decisive progress which for the most part renders the formalistic ideal obsolete. The situation is quite different concerning the formalization of structures as a par­ ticular class of natural phenomena. Indeed, structural methodology had led to many relevant models through systems theory and cybernetics, the development of '}artificial intelligence, the various theories of language acceptors automata, formal grammars, categorial grammars, intensional logic, etc. It should, however, be noted that all of this progress presupposes a generalized reification of structures. In biology, for example, no matter how relevant cybernetics may be, one should not hide the fact that its use raises, as Thom noted : "very serious problems as soon as one attempts to pass from abstract schematicism to a material realization in spacetime " (1980a:154 ). From a formalistic perspective, one cannot understand how the structures represented by such formal schemas can emerge from physico-chemical properties. Such a perspective is valid "only for partial mechanisms, already put together, and in full func­ tional activity . . . In no case it can apply to the global structure of living beings, to their epigenesis and to their physiological maturation" (Thom 1 972 :207). This is why, in biol­ ogy, it is pertinent to go back in a certain way to vitalism, but only after having trans­ formed it into a dynamical geometry of position explaining the emergent structures "uniquely by means of local determinisms, theoretically reducible to physico-chemical mechanisms" (Thom 1 9 8 3 : 167). Of the two points of view traditionally opposed in Biology, the vitalist one and the reductionist one, contrary to current opinion it is the latter that is metaphysical, for it postulates a reduction of vital facts to purely physico-chemical ones which has never been experimentally established. On the contrary, vitalism is based on the impressive array of facts of regulation and finality which cover the quasi totality of vital activity. (Thom I 983 : 1 66)

In the same way, in linguistics and in semiotics, the formalistic point of view comes up also against intrinsic limits. The current belief "that the strictly generative virtue of a structure, stemming from its very own form, must be admitted a priori and does not require any explanation" needs to be questioned, as for natural languages "it is the self-limitation of the generative capacities of syntax which must be explained" (Thom 1980a : 1 64 ). To understand this self-limitation, one must introduce dynamics underlying the formal cinematics described by formal languages. One must base structures no

Structure

1 021

longer on "the automatic iteration of certain operations, but on the contrary, on an intrinsic combinatories given by dynamic interpretation" (Thom 1980a: 1 06).

Conclusion

On February 2, 1706, Leibniz wrote to Reverend Des Bosses : If one posits the plenitude of things (as do Cartesians) and the uniformity of matter and if one adds only movement one always obtains a succession of equivalent things; . . . and thus, no one can distinguish one state of affairs from another, not even an Angel; and therefore no variety could be found in phenomena: so, to figure magnitude and movement, forms must be added by which the difference of appearances arises in matter, forms that, it seems to me, can be sought intelligibly only by means of the Entelechies. (Leibniz 1702)

From what has been sketched in this article, it ought to be clear that the mathematization of the concepts of entelechy and form constitutes the central difficulty of the structuralist episteme. Form is to structuralism what force is to physics, and therefore structural objectivity depends entirely on a general mathematics of forms. This theme appeared on several different levels: (1) the level of biological organization; (2) the level of qualitative structuration of the phenomenological states of affairs through which, as Leibniz said, "the difference of appearances arises in matter"; (3) the level of the perceptive organization simulating the states of affairs; and (4) the level of the constraints imposed by this simulation on the cognitive, linguis­ tic, and semiotic structures. The systematic working out of these intimately linked questions is part of a major tradition which has constituted one of the main streams of philosophical thought without nevertheless having ever succeeded, for lack of tools, in reaching the scientificity of experimental method. There was here an epistemological obstacle the elemination of which structuralism has contributed to by introducing, in place of philo­ sophical thought, new methods of objective determination. The question then arises of formalizing the structures according to the "things themselves."After having been formal­ istic, structuralism can, and must, be reborn as a new natural philosophy integrating organization and signification with purely physical explanations. Such are its actual epis­ temological value and ontological thrust. In the introduction of his pioneering article, "Biologie et structuralisme" ( 1 968), Thom asked the following question: "Can recent structural research in the human sciences, such as linguistics and anthropology, bring new methods for the enlightenment of an experimental science such as Biology" (1968). In other words, is it possible to attain a synthesis between dynamic structuralism, focusing on the problem of morpho­ genesis, and semiolinguistic structuralism, focusing on the form of signifying systems? It seems that a positive answer can now be given. For this, it is necessary first, as Thom pro­ poses : (1) to reduce every structure (paradigmatic categorization, actantial interaction, morphogenetic differentiation, etc.) to a morphology; and (2) to reduce every morphology to a system of qualitative discontinuities on a substratum space (physico­ chemical, semantic, or abstract). By distinguishing, on the one hand, the regulari­ ties, and, on the other, the variations relative to perturbations in the initial conditions,

1022

Style

one can attempt to describe the empirically observed morphologies as aggregations an_d as concatenations of a finite number of structurally stable submorphologies, "morphoge­ netic fields," "chreods" (Propp's analyses are a good example of this). The next step is to elaborate local dynamical models to generate the chreods. Finally, one needs to elabo­ rate global dynamical models which can, from a purely morphological point of view, account for the stable associations of chreods and therefore for notions of order and hier­ archy. Although vast, such a research program is no longer wishful thinking. Results have already been obtained which show that there exists a mathematically describable rational unity between phenomena of spontaneous organization of matter (dissipative struc­ tures), critical phenomena (phase transitions, optical caustics, elastic buckling, etc.), biological morphogenesis and semiolinguistic structures. This possibility of a rationally founded and mathematically legitimated matching of, on the one hand, formalisms of fundamental physics, with on the other, the ontological concepts of structure, organiza­ tion and event, conclusively confirms the intuitions of vitalists, phenomenologists, and Gestalt theoreticians. It leads to a rethinking of the rift, until now thought to be unbridgeable, between scientific explanation and hermeneutic comprehension. It uncovers an unexpected solidarity between the two main rational conceptions physics and structuralism - of the modern episteme. This is an outstanding theoretical event and a decisive advance in what Michel Serres so acutely called "the North-West passage" separating the exact sciences from the human ones. As was previously mentioned, this "passage" depends upon a new natural philo­ sophy based on the (transcendental) constitution of a new regional ontology. For the main it invalidates the various "Freudo-Marxist," Heideggerian, and Lacanian "motifs" which, at least in France, have been an integral part of the structuralist ideological drift. If, for a time, it seemed legitimate to accompany structural rationalism with such seduc­ tive and mystifying rhetorics on the Unconscious, on the Subject's Negativity, on Mate­ rialism, etc., it was essentially because this rationalism remained cut off from its real ontology and lacked adequate mathematics, so that its authentic essence could only be expressed in a reifying formalism. In other words, a purely epistemological, methodo­ logical, and formalistic structuralism leads to the almost magical evocation of the trans­ cendent problematic of the "originary place" (a noumenal one) producing structures (for example, the subject of the unconscious or the infrastructures). Unable to constitute its own objectivity, it subverts itself and drifts into a non-objective "ontology" since, as Eco has remarked, "an ontology of the originary place imposes, if all conclusions are cor­ rectly drawn, the destruction of the notion of structure" (1968 :363 ). Thanks to the new transcendental idealism and to the new natural philosophy animating catastrophe the­ ory, this non-objective "ontology of absence" - which is none other than a new guise of materialistic and existentialist gnoseology already "deconstructed" by Merleau­ Ponty in the Aventures de la Dialectique - can now be laid to rest without reducing structuralism to formalistic positivism. J.Pe. (P.P.)

Style Derived from the Latin form stylus etching needle, and having in classical Latin the same general meaning as in all modern languages, the word style presents a great variety of

Style

1 023

acceptations (the Oxford English Dictionary counts 27 of them) but has never been given semiotic definition. Nonetheless, it has been used - and still is - to refer to one of the main phenomena in the semiotic process, by playing the part of a "black box" in the '�epistemology of the human sciences. All classical definitions of style may be reduced to a constant: it always serves to characterize either the connection between the �·utter­ ance and its addresser - be it singular or plural - (style as "physionomy of the soul," as "image of public customs," or "signature of a school") or the connection between the same utterance and its *addressee - singular or plural (style as "means of action," or as "agreement between ends and means"). Consequently, reflection on style encroaches upon the semiotic of utterances, which is to be opposed to the semiotic of the *code in the same way as the linguistics of language is to be opposed to the linguistics of '�messages (�·Saussure's Jaits de parole versus Coseriu' s hechos de norma). Two other features of the classical definitions of style have contributed to make the concept unsuitable for immediate semiotic reinterpretation. First, the term may theoreti­ cally be applied to any semiotic occurrence, i.e., to Kinesics or to the '�semiotics of *cul­ ture (we talk about "someone's style," about a "style of living"), but it is more specifi­ cally used in *visual, musical, and linguistic semiotics. In this last field, the coinage of the term stylistics, as "analysis of the means of expression of a language" (Bally 1909), or "analysis of literary styles," reinforced the specialization of the word. Secondly, although it is theoretically a neutral expression, the word style can be associated with adjectives of the most varied kinds (Lausberg 1 960 mentions 175 Latin phrases, from stilus abruptus to stilus vulgaris, and 1 31 French phrases), and frequently has a connotation of qualitative judgment. In some cases, it may even be used with an absolute value: "to have style." These reflections show that inevitably the concept of style was to disappear from scientific discourse as knowledge conditions continued to develop. This evolution is due to: ( 1 ) the development of precise disciplines, each of them dealing, in its own way, with the relation between the utterance and the addresser-addressee dyad; nowa­ days we realize that the word style was used to refer to a very general class of phe­ nomena falling within the competence of �-rhetoric, psycholinguistics, sociolinguis­ tics, *pragmatics, etc. (cf. Enkvist 1 964); (2) the diversification of these approaches according to the various branches of semiotic: there must necessarily be a kinesic stylistics, a proxemic stylistics, and an ''·iconic stylistics in the same way as there is a linguistic stylistics, which has tradi­ tionally enjoyed an unduly privileged status; (3) the deletion of normative perspectives from poetics and rhetoric (cf. Klinken­ berg 1 975, 1 98 1 ).

Now the remaining question is to know how the different problems that formerly were the concern of sty!� analysis are to be classified in the various semiotic disciplines. In this respect, let us notice that style analysis has always given alternatively a pre-emi­ nent importance to one of the two aspects that characterize the relation between the utterance and the addresser-addressee pair. It either described the elements of the utter­ ance which are relevant to the setting up of the relation (stylistic features) or it centered on the actual correlations between the stimulus, i.e., the utterance, and the induced response or expressed intentions. These two approaches correspond to the distinction between stylolinguistics and stylobehavioristics proposed by Enkvist ( 1 973).

1 024

Style

The underlying principle in any stylistic description of an utterance is the following one: Style in an utterance can only be isolated if this utterance is performed in accor­ dance with a code providing for free variants or, in other words, if each sign produced is capable of several manifestations; this has been more particularly acknowledged in the field of phonostylistics (Leon 1 969). Style in an utterance is thus the result of several cho­ ices between the different possible manifestations of the same semiotic event (for exam­ ple, between what traditional linguistics calls registers or levels of language or, in the kinesic message, between the different possible gestures of "contempt" or "submis­ sion," or, in the visual message, between the various degrees of iconicity). These choices may be technically described at the various levels taken into account in each semiotic: phonemes, syntagms, suprasegmental structures in the oral message, plastic and iconic units in the visual message (Groupe µ 1 980). From the methodological point of view, these choices may be described in two different ways. On a quantitative basis, a given utterance might be described as an original compound of frequencies (this is the concern of stylostatistics; cf., for example, Herdan 1 956 and Muller 1 968). On a qualitative basis, this utterance might be described as a particular network of implica­ tions, contradictions, contrasts (Riffaterre 1 97 1 ), ,;redundancies, rhythms, etc. set up between the various stylistic features. We know, for example, that studies about parallel­ isms in phonic or �·syntactic structures, or about the different possible strategies in the production of polysemic sets, have proved to be particularly successful in poetics CTakobson 1 9606 ; Levin 1 969; Groupe µ 1 9776). But these different types of stylistic patterns are not only relevant to linguistic description; they also may be accounted for in a semiotic analysis. The basic concept here seems to be that of isotopy, which has already been discussed at the linguistic level (Klinkenberg 1 97 3 ; Groupe µ 1 976) as well as at the level of fixed image (Minguet 1 976; Odin 1 976) or of moving image (Wolf 1 979). Research on reciprocal actions between utterance and speakers has not yet achieved the status of a consistent discipline. Although some authors have tried to draw up a gen­ eral pattern that would integrate all these interactions (Benveniste 1 9666), most schol­ ars have considered style either as the imprint left out in the utterance in the ,;enuncia­ tion process, or as the effect induced by the utterance in the addressee. In the first case (for a general approach see Todorov 1 970a), traditional stylistics, and after it the various branches of visual and musical semiotics, had already developed a description - rather empirical, to tell the truth - of the different techniques used in the utterance layout, that was based on a distinction between direct speech, indirect speech, represented speech, etc. (M.J. Friedman 1 955). On the other hand, from a psy­ cholinguistic point of view, the addresser's behavior had already been the object of numerous classifications in lyric style, modalizing style, evaluative style, etc. The effect induced in the addressee, called ethos (Groupe µ 1 970), or pathos in classical rhetoric, has also been subjected to many classifications, most often without any consistent crite­ ria, from sociological, psychological, semiotic points of view. At best we may say that modern stylistics has concerned itself in establishing a dis­ tinction and a hierarchy between the different factors involved (Imbs 1957). Although some attempts have been made at the semiotic level, for example to define the poetic ethos (Groupe µ 1 9776), progress has mostly been achieved in the field of formal logics, rhetoric of argumentation (Perelman 1 976), and, more recently, in pragmatics and sociolinguistics, the latter being able today to provide a whole sociosemiotic with

Supersign

1025

various patterns. The common concept unifying all these approaches is that of context, be it intratextual, intertextual or situational, and in the latter case, of sociological or spa­ tial nature. Consequently, the object of stylistics ultimately comes to merge with that of semiotic as a whole and style ultimately dissolves itself in �·semiosis, as defined by '}Mor­ ns. J.-M.K.

Supersign The term supersign is a neologism in semiotics probably inspired by '}cybernetics, *infor­ mation theory and/or ,:•Gestalt theory (cf. the term supersymbole in Moles 1958). If used loosely, it applies to connections between *signs of a certain �·code forming a whole which constitutes a higher level sign in another code. It is possible, however, to define the term supersign in a more systematic way. The term code is used, in semiotics, synonymously with the term sign system and semiotic '}structure. Codes as semiotic structures consist of two correlated universes or classification systems. Each class in one of the two systems, containing objects alotted to it on the basis of certain distinctive properties (�·distinctive features), is consequently correlated to a class of the other system. Since, by virtue of this correlation, each class of one classification system contains information regarding its corresponding class in the other system (over and above the information regarding itself), the correlated classes are called indices, or signs. Indices, or signs are "double-faced" entities connecting a class from a universe of the *signifier with a class from a universe of the signified. There are codes which are said to possess a first *articulation, in which certain classes of the universe of the signifier may be grouped together according to dimensions featuring par­ ticular properties, where certain classes of the universe of the signified may also be grouped together into dimensions with particular properties, and where both dimen­ sions of the codes are correlated. According to this description, the correlated dimensions must be composed of signs. Such larger double-faced units of correlated dimensions articulated into signs have been called *semes (Buyssens 1943b, 1967; Prieto 1 966, 1968). The promotion of the term seme with this particular meaning has not been successful, however, since the same term is also used with other meanings (e.g., that of semantic feature) in *semantics (e.g., Greimas 1966a). It has therefore been proposed (e.g., Eco 1976c) that the larger double-faced units consisting of signs be called supersigns. This proposal has the advan­ tage of appealing to "common sense: " It is easy to agree that larger units composed of signs should be called supersigns. Apart from this pragmatic reason, this usage of the term supersign also has a systematic advantage, since it allows us to make a connection between the �·structuralist concept of the sign and *Peirce's sign concept. According to Peirce (e.g., 1931-66:2.43-2.53 ), an "elementary" sign is amalgamated, by an internal semiotic process, from three variables, which may also be considered subsigns. These subsigns refer to the material structure (qualisign, sinsign, legisign), the kind of relationship between the material structure and the '}object the sign represents (�·icon, '}index, •'symbol) and a degree of interpretation, i.e., a certain kind of *interpretant (rhema, dicent, *argument). An elementary sign is thus generated by the internal amalgamation of certain hierarchically permissible combinations of these

1026

Suppositio

trichotomized variables or subsigns, yielding ten classes of elementary signs. This internal generative process of elementary signs is a function of the state taken on by the interpretant variable. Once generated, elementary signs are subjected to external sign processes or ''semioses. These external sign processes produce non-elementary signs, which are again functions of their interpretant variable. If the variable takes on a rhematic value, sign connections with an open interpretation are produced by simple adjunction of elementary signs. Such an adjunction would be a mere assemblage of verbal terms or visual elementary signs allowing no interpretation of the connection as a whole. If the interpretant variable of the external sign process takes on the value of dicent, sign connections with a closed interpretation are produced. This external sign process is called superization, and produces supersigns. Supersigns are thus sign com­ plexes, such as *propositions, �·sentences, or configurations of visual elementary signs, allowing an unique interpretation as a whole. This is excatly the characteristic required by what used to be called semes in codes featuring the first articulation. The interpretant variable of external sign processes may also take on the value of an argument. In this case, connections of elementary signs with a complete (exhaustive) interpretation are produced by what has been called iteration. Such completely inter­ preted sign connections are exemplified by a logical conclusion or a mathematical proof. Since works of art are concerned with unique interpretations which are realized in a spe­ cific *medium-object relationship, supersigns, as interpretable sign complexes, are espe­ cially relevant in semiotic ,:•aesthetics. A superization process of elementary signs may apply to elementary sign classes featuring either iconic or indexical or symbolic object relationships or subsigns, thus yielding either a super-icon, super-index, or super-sym­ bol. Since the property of "uniqueness" is also frequently stipulated as a condition of works of art, their elementary signs are realized, in their material structure, as sin-signs. Thus, works of art tend to be unique superizations of different super-icons, i.e., of dicent (dicisign)-iconic-sin-super-signs. This condition of super-supersigns is realized, for example, in �-paintings and metaphorical sentences or verses, but also in musical pas­ sages and other artistic configurations which reflect iconically the general artistic principles of a work as a whole. M.K.

Suppositio Suppositio (English supposition) was the name given by the medieval terminist logicians such as Peter of Spain, William of Sherwood, William of ,:-ockham, and a host of others, to the referential function of ,:-signs. The word was borrowed from the terminology of �-grammar, where it used to designate the "putting" of a subject "under" a predicate (de Rijk 1962-67). By the thirteenth century, it had become the key technical term of the new �-semantics, the logica moderna. Supposition, then, was the property of standing for something, and the suppositum was no longer the grammatical subject of a verb, as it had been in Priscian, but the ontological subject of ,:-discourse, the thing which is talked about. Most writers, especially after the thirteenth century, considered supposi­ tion in this sense as a property of �-terms - predicates as well as subjects - in ,:-proposi­ tional contexts only. It was sharply distinguished from the logically prior and non-pro­ positional property of signification ( significatio): among realist thinkers, for example, a

Symbol

1027

general term was said to signify a ,;universal nature and, in its normal use, to "supposit for" (supponere pro) the individuals that shared this nature. Many kinds of suppositions were identified and variously classified by different au­ thors. The normal referential use of a proper ,;name or of a general term was called its "personal supposition" (suppositio persona/is), while the sui-referential use (like that of "man" in man is a word) was known as "material supposition" (suppositio materialis). In sentences such as man is a species, "man" was said to have "simple supposition" (sup­ positio simplex) : it was a matter of dispute between nominalists and realists to decide whether it then stood for a mental concept or for a universal nature. Personal supposi­ tion of general terms was in turn divided into different modes, according to the various ways in which a common name can stand for its ,:•referents : according to most authors, for example, "man" stands for all men in some man is running as well as in every man is mortal, but disjunctively, in the first case (suppositio determinata), and conjunctively, in the second one (suppositio distributiva). On the whole, the main point of supposition theory was to give a systematic account of the contextual variations in the ,:•denotations of terms. It thus provided a powerful tool for the philosophical analysis of arguments, and, among other things, allowed for pre­ cise formulations of the truth-conditions of many kinds of sentences. Recent detailed his­ torical work on the theory tends to suggest that its leading ideas could still be of great interest for modern semantics. C.P.

Symbol Language Usage in Ancient Times Already in ancient times ,;signs (semefa) and symbols (symbola) are either exchangeable or each of these concepts can convey the same meaning. While Homer used semata as a sign of recognition for spouses, after his time the word symbolon was used for that pur­ pose. First, the symbol was a sign of recognition consisting of two linking segments of coins, rings, or clay-plates which indicated the unity of the owners. Treaties between two parties, the omens in ,:-divination, as well as the passwords of secret societies were al­ so called symbola. For the Stoics, a symbol meant a veiled intimation of a philosophical or theological ,:•truth. According to Philo of Alexandria, in the symbol the reality of the created is brought into relation with that of the creator ; symbols mean sacred signs or signs leading to redemption. Bishop Cyprianus of Carthago was the first to use the sym­ bolum as a synonym for "creed." Sign and Symbol The word symbol is not always used in the same sense: sign and symbol can be used syn­ onymously; either term can be the superimposed concept for the other; or the terms are used to indicate opposites. One can say - as the majority of the symbol experts do - that all symbols are signs but not all signs are symbols. Consequently, there are two catego­ ries of signs: those merely being signs and those beyond that having symbolic signifi­ cance. ,:-cassirer (1923-29 ; 1953-57) differentiates between '-· signal and symbol: "A sig-

1028

Symbol

nal is a part of the physical world of being; a symbol is a part of the human world of meaning." As a composition (the Greek symbtillein means 'to combine', 'to confound'), the symbol is found in the center of intersection of two spheres of existence. Owing to that fact, the symbol is not only an indicating sign (directing from one sphere to the oth­ er), but also a sign which takes part in both spheres. It reveals in the exterior an interior, in the physical world a spiritual world, in the visible the invisible. The symbol represents the symbolized, it is a signum repraesentativum, thus being in accordance with Goethe's interpretation that true symbolism exists wherever the particular (das Besondere) repre­ sents the general (dasAllgemeine), not as an illusion or a phantom but as an existing, mo­ mentary revelation of the inscrutable. Contrary to the merely indicating or denominating sign (signum signi/icativum) there exists an inner connection between the symbol and the symbolized. For instance, the color red can be synonymously used for blood, the latter for life. Or the west sym­ bolizes the realm of the dead since the sun sets ("dies") in the west. Unlike the sign, the symbol is not accidental but ultimately part of the reality it represents; it participates in the reality of all it is a symbol for. All symbolism focuses on the poles of existence: birth and death, light and darkness, man and women, good and evil. Contrary to the simple sign denoting something definite, the symbol is *ambiguous. The antagonistic character of the world is expressed in the ambivalence of the symbols. Thus, a "snake" can indi­ cate life and death, the fall of man and his salvation, the origin of the material world and eternity. Despite this distinction between arbitrary sign/signals (as used, e.g., for chemical ele­ ments, in cartography, in the computing field) and the symbols which (by means of ''analogy) represent or make transparent the symbolized or take part in the reality of the symbolized, these two categories must not be regarded as being inflexible. Thus, numbers can be thought of either as merely numerical or as sacred or unlucky numbers. A flag produced in a factory originally as a significant emblem can be charged with emo­ tions and finally become a national symbol. Terms meaning nearly the same as symbol, such as allegory, archetype, cipher, emblem, simile, ''metaphor, motive, and �·type, can form a kind of "spectral line" where some of them, owing to their different value, can claim more room than others, and where their order of succession is optional.

The Symbol in the Arts and Sciences

While the intellectuals of the Romantic period (Novalis, Coleridge) supported the idea of an analogy between man and the universe and were convinced of a universal system of concordances and participations, those of the twentieth century more frequently con­ clude that in certain ways all forms of human ,:•culture are symbolic. This explains the in­ creasing interest in symbols by scholars in almost all disciplines. Every act of perception can be understood as a symbol (Whitehead), man himself can be understood as an anim­ al symbolicum (Cassirer), and symbolism can be understood as a clue to spiritual life (Langer). Symbols represent the essence of philosophy and religion, art and science (Urban 195 1). The German philosopher Karl Jaspers speaks of ciphers leading to the impenetrable depths of existence and thus enabling man to comprehend the correlation between human existence and transcendence. The theologian Paul Tillich discriminates between symbols (as in ''mathematics, logic, and science) and representative symbols,

Symbol

1029

or those partaking in the reality they represent. Independent of individual religious sym­ bols (for example, the tree of life, dove, or water), in theology symbolum means a norma­ tive (verbal) form, being used in �·liturgy and '}doctrine as a sign of initiation into the se­ cret of salvation. In the science of religion the works of Eliade (1952) are of greatest im­ portance. According to him, symbolic conception gives way to the immediate reality. Important contributions to the research in the field of symbolism in �-literature were made by Tindall (1 955), Gaston Bachelard, and Hermann Pongs (1927-73). In the arts (Bialostocki, Fingesten, Sedlmayr) and literature, emphasis has been placed on the au­ tonomous character of the symbol in relation to allegory and sign. Ferguson (1 954) des­ cribes the symbol : "It has acquired a deeper meaning than the sign, because it is more completely identified with what it represents, and its character is derived from what is known by it." In iconology, Erwin Panofsky asks for the "alter" which according to him should be symbolized in each work of art. In the field of ,:•psychoanalysis CFreud, I.A. Caruso, Erich Fromm), there are in the first place individually suppressed complexes - according to *Jung and his followers (M.-L. von Franz, J. L. Henderson) it is superindividual collective unknown - which find expression in words, deeds and dreams. Several branches of psychotherapy adopted the idea which originated in expressive psychology (E.A. Weinstein, R.L. Kahn) that all ex­ pressive movements symbolize inner proceedings. Behaviorism examines symbols and signs first of all with a view to their signal function. G.H. Mead establishes the theory of symbolic interaction so important for ,:•sociology: Every human being lives not only in a physical world but also in a symbolic world, symbolic '�communication being the condition for man's admittance to the systems of intellect and value. M.Lu.

Symbol and Semiotics

Linguists, philosophers of language, and semioticians use the term symbol with different senses. The whole of the various contrasting positions can be summarized according to the following alternative: either the symbol is the semiotic genus of which all the other semiotic phenomena are species, signs included, or there is a semiotic genus called "sign" of which symbols are one among the species. The followers of the first alternative identify the universe of symbols with the semio­ tic universe at large. For instance, Malmberg (1977:21) calls symbol any element repre­ senting something else and keeps the term sign to indicate "these units which, like the signs of language, have a double �·articulation and owe their existence to an act of signifi­ cation (where signification means intentional communication). The identification "sym­ bolic = semiotic" is explicit in Levi-Strauss (1950), since every human culture can be considered as an ensemble of symbolic systems, among which this author ranks verbal language, art, religion, science, kinship rules, and so on. The possibility of the mutual �·transformation among ,:•structures is permitted by the existence of a more profound symbolic activity of human mind which organizes the whole of our experience accord­ ing to the same modalities. The same position was held, in a different line of thought, by Cassirer : "the fundamental concepts of each science, the instruments with which is pro­ pounds its questions and formulates its solutions, are regarded no longer as passive im­ ages of something but as symbols created by the intellect itself" (1953-57:75). Symbolic activity does not "name" an already known world, but establishes the very conditions

1030

Symbol

for knowing it: "consequently all truly strict and exact thought is sustained by the sym­ bolic and semiotics on which it is based" (1953-57:85-86 ). '�Lacan (1975) distinguishes three registers of the psychoanalytic field: the imagi­ nary, the real, and the symbolic. The imaginary is characterized by the relation between an image and a similar object, but the similarity of which Lacan speaks is not the one of so-called iconic signs, but is a phenomenon that takes place within our perceptual mechanisms (for instance, the perception of mirror images, many cases of isomorphism, the so-called "real" images produced by curved mirrors, and so on). The psychic subject achieves his self-identity only through a "symbolic" experience. Language and other sorts of visual signs are defined as belonging to the "order of the symbolic. " Thus, the symbolic is identified with the semiotic in general. Alternative (2) can generate different and contrasting options. ,:•Peirce defines the symbol as "a sign which refers to the ,:-object that it denotes by virtue of a Law, usually an association of general ideas" (1931-66: 2.249). Peirce uses the term sign for the semi­ otic genus, and symbol for arbitrary and conventional devices, as opposed to icons and •:•indices. His choice, rather infrequent in semiotic literature, followed, however, a com­ mon scientific usage by which symbols are conventional signs standing for chemical, physical or mathematical entities (even though Peirce said that these symbols display many iconic qualities). On the opposite side one can rank both •:•Saussure (1916) and ,:-Hjelmslev (1943). Saussure called symbols those signs that for Peirce would have been labeled icons, and Hjelmslev defined as symbolic those semiotic systems that are "interpretable" but not "biplanar," since they are isomorphic with their interpretation, "like the Christ of Thorvaldesen as a symbol for compassion, the hammer and sickle as a symbol for Com­ munism, scales as a symbol for justice, or the onomatopoeia in the sphere of language" (1943 :113-14). Todorov (1977, 19786) identifies as symbolic the whole gamut of con­ notations, '�presuppositions, implications, ,:•implicatures, figures of speech, and so on, thus calling symbol every instance of indirect meaning. A narrower identification of sym­ bols with rhetorical devices is suggested by Freud when, in The Interpretation of Dreams (1899), he speaks of oneiric symbols, produced by "displacement" and "condensation" like •:•metonymies and '�metaphors. At the end ot his exhaustive survey of all the possible uses of the term symbol, Firth suggests that this term be used in the place of sign when there is a certain "ineffectuality" (in the sense that a so-called symbolic gesture does not attempt to get immediate and concrete effects), and proposes a provisional and "pragmatic" conclusion: "in the in­ terpretation of a symbol the conditions of its presentation are such that the interpreter ordinarily has much scope for exercise his own judgement . . . Hence one way of distin­ guishing broadly between signal and symbol may be to class as symbols those presenta­ tions where there is much greater lack of fit - even perhaps intentionally - in the attribu­ tion of the fabricator and interpreter" (Firth 1973:66-67). This seems a reasonable sug­ gestion for classifying as instances of "symbolic mode, " or of symbolic use of signs, those species of '�texts that elicit a free and uncommitted interpretation on the part of their addressee, by conveying vague and/or polyvalent meanings. In this sense, cases of symbolic textual strategies are the symbols of French Symbolism, Joyce's "epiphanies," Eliot's "objective correlative, " and so on.

Allegorical and Symbolic Theories of the Middle Ages

Symbol

1031

Scriptural Allegory

The whole medieval Christian tradition is influenced by a sentence of Saint Paul (II Co­ rinthians): " Videmus nunc per speculum et in aenigmate," 'presently we see only through a glass in a dark manner'. The final truth will be revealed only in the vision of God in the Heavens. During the course of our permanence upon the earth we can discover the truth only through obscure signs. The allegorical interpretation of the Holy Scriptures is fully theorized by Origen (third century A.D.). There was, however, the tradition of the Hebrew exegesis and, from immediate post-Homeric times (with Theagenes) until the Stoics, Greek culture had attempted a philosophical reading of the classical mytho­ logical poetry, trying to find in it "natural" truths, expressed under the form of allego­ ries. In the second century of the new era, Clement of Alexandria first posited the distinc­ tion and suggested the equivalence between both Testaments, against the Gnostic claim that only the New Testament was conveying the truth. Origen proposes the parallel reading of both Testaments: the Old is opposed to the New as the Letter is opposed to the Spirit. In other terms, the Old Testament is the signans of which the New one is the signatum. Since even the New Testament can be read allegorically, both Scriptures have three senses: literal, moral (psychic), and mystic (pneumatic). According to Origen, the moral sense was understandable even to infidels. Later, from Bede to ,:•Dante Alighieri, the senses became four and the moral one was based upon the allegorical one, so that there was no longer a moral truth independent of the Revelation. At the time of Dante the theory, universally accepted, considers four senses, literal, allegorical (or typologi­ cal), moral (or tropological), and anagogical. Thus, the verse of the Old Testament des­ cribing the people of Israel abandoning Egypt (in exitu Isral de Aegypto), can be inter­ preted literally as if it were speaking of the Hebrew people escaping from captivity, alle­ gorically as a discourse on the redemption of mankind, morally as an allusion to the pas­ sage of the soul from sin to Grace, and anagogically as a reference to the sanctified soul that, at the end of human history, will approach the Heavenly Jerusalem.

Allegoria in factis

In order to conceive of such a multifarious interpretation the Fathers of the Church had to recognize two main types of allegory: one concerning words (allegoria in verbis) and the other concerning the facts themselves (allegoria in factis). The allegoria in verbis, typ­ ical of poetic discourse, is the effect of a rhetorical manipulation of language. On the contrary, the allegoria in /actis considers things, facts, and events of the Sacred History as they were tropes, figures of a "material" speech written by God Himself, when He or­ ganized the events of the history so to make them expressive of some higher truths. The '�hermeneutic problem faced by ,:•Augustine in De Doctrina Christiana concerns this double interpretation of the Scriptures, where the facts narrated by the Old Testa­ ment are to be allegorically referred to the mysteries of the New one. Thus, the interpre­ ter must first understand the rhetorical strategy of Biblical language qua verbal language (and a propos of this Augustine revisits all the techniques of the Roman rhetorical tradi­ tion), and then must understand the allegorical meaning of the facts narrated by verbal language.

1032

Symbol

Augustine elaborates an interesting series of hermen.eutic rules in order to recognize when a given event has to be taken allegorically. First of all, everything said about faith and morals has to be taken as a literal precept. A given sentence should have a second meaning when (1) it says something against religious truths, (2) conveys something hardly verisimilar and contrary to our natural experience, (3) appears unnecessary, or (4) is semantically or literally "poor," as in the case of proper '�names, which should have a more profound signification. The whole Book 3 of De Doctrina Christiana is a paramount example of textual semiotics and contains hermeneutic rules that still hold in profane (poetical or philosophical) interpretation, including the idea of infinite inter­ pretation (cf. De Lubac 1959; Compagnon 1979; Todorov 1978b). An interesting side effect of the idea of allegoria in factis is that the interpreter must know the properties of things in order to ascertain their possible allegorical meaning when they are named by the Scriptures. Augustine advocates a careful knowledge of natural facts and encourages his readers to collect information about the nature of animals, plants, flowers, stones, numbers, astronomical phenomena, and so on. In doing so (and according to the state of natural science of his time), he relies upon Hellenistic encyclopedias such as the book of Physiologus (between the second and fourth centuries A.D.) nourished by many oriental influences and depending in some way on the Her­ metic tradition. The Physiologus will become the model for the various Specula Mundi, Books of Monsters, bestiaries and lapidaries and herbaries, from the Patristica era to the '�Renaissance, in which every item of the natural world is endowed with fabulous proper­ ties, in order to make it a suitable "coded" sign for the purposes of the allegorical inter­ pretation. Universal Symbolism Within the framework of Biblical exegesis allegorical interpretation is limited to the uni­ verse of Sacred History. However, in the early Middle Ages, this tendency overlaps with another line of thought influenced by the Neo-Platonic tradition through the works of the Pseudo Denys the Areopagite, definitely translated into Latin in the ninth century. The negative theology of Neo-Platonists represented the world as an emanative chain, from the unspeakable down to the elementary Matter. Christian Neo-Platonists tried to find a way to know something about the nature of God through the properties of His creatures, and encouraged an interpretation of the visible universe as a "teophany," a manifestation of God. Scot Erigene (ninth century) said that God is so-to-speak created in His creatures in an admirable and unspeakable way (In Coe!. Hierarchiam, 678). However, God can be made evident by the created world in two ways: metaphori­ cally or metonymically. These alternative interpretive options should be considered as radically different even though many medieval authors shift from one to another. The first option represents an extension fo the principle of allegoria in factis from the scriptu­ ral universe to the whole physical world, which is interpreted as though everything were a metaphor of some supernatural truth. In the twelfth century, Alanus ab Insulis de­ clared that every element of the furniture of the world is for us like a written word or an image that speaks of our destiny, of our life and our death ( omnis mundi creatura, quasi fiber et pictura, nobis est in speculum. Nostrae vitae, nostras mortis nostri status, nostrae sor­ tis fidele signaculum). Hugh of Saint Victor (Didascalicon) said the whole world appears as a book written by the finger of God. According to the Pseudo Areopagite, the more a

Symptom and Syndrome

1033

signifying relationship is incongruous (for instance, when God is named as lion or bear or panther), the more the metaphorical link becomes evident as such, since the gap be­ tween signans and signatum reminds us that the signans is a trope and not a literal expres­ sion. Since the allegorical codebooks frequently attributed to a given entity contrasting properties (so that, for instance, the lion was at the same time a symbol for Christ and a symbol for the Devil), this universal allegorism is basically polysemic, this polysemy was controlled by exegetical ,:•frames provided by the Auctoritates, that is, by the hermeneu­ tic tradition, so that an impressive series of contextual rules tried to control the dissemi­ nation of indiscriminate readings. According to the second option, the elements of the furniture of the world are no more viewed as metaphors but as metonymies: the world speaks of God by allowing a series of inferences from effect to cause, even though the cause was recognized as "hy­ persubstantially" different from its effect (cf. Scot Erigene, the School of Chartres, the twelfth century, and the Aristotelianism of the thirteenth century). ''Aquinas, with his theory of analogia entis, eliminates any possibility of intending the effect-to-cause rela­ tionship as a Neo-Platonic emanative connection. The effect becomes a sign for its cause by virtue of a more abstract proportional ratio. Naturally at this point it becomes difficult to decide whether such an analogical link is rooted upon the nature of things themselves or upon our conceptual organization of the world. In any case, Aquinas had a relevant impact on the decadence of the allegorical and symbolical mind, since he ex­ plicitly said that the allegorical interpretation of facts and things must be reserved only for the facts of Sacred History, as narrated by the Old Testament. As for poetic lan­ guage, Aquinas intended all rhetorical figures as instances of modus parabolicus 'parabol­ ic mode', which represents a case of literal meaning ( Q uodlibetales 8,6,16, ob.1 and ad 1). Therefore, rhetorical connotations are viewed as functions of natural language and do not imply any allegorical interpretation. U.E.

Symptom and Syndrome Symptoms are ,:•signs which in medical usage refer to conditions of health and disease of organisms according to the three temporal dimensions of present (diagnosis), past (eti­ ology), and future (prognosis). By metaphorical extension, symptom is often used generi­ cally as "sign." Syndromes are rule-governed groups of symptoms with stable ,:• referents. One can therefore speak of a "grammar" of symptoms (Buchanan 1938), which T. v. Uexktill (1979a), following Morris (1938), subdivides into ,:•syntactics, ,:•semantics, and •:•prag­ matics. Medical syntactics consists in collecting and ordering relevant symptoms; se­ mantics ( = diagnosis) deals with interpreting them by assigning them to tested syn­ dromes; pragmatics is concerned with therapeutic action. All three levels can be studied as paradigmatic and syntagmatic systems. In paradigmatic systems, symptoms function as diacritical signs within increasingly inclusive 'rcodes, from syndromes to larger medi­ cal taxonomies to sociocultural •:•belief systems. Syntagmatic systems are topical, as in the simultaneous control of parameters during a surgery (electrocardiogram, elctroence­ phalogram, blood pressure, respiration, etc.), and temporal, as in the daily auditory ob-

1034

Symptom and Syndrome

servation of the excrements of giraffes in zoos, where differences in the sound of the fall­ ing feces may indicate states of health and disease (Sebeok 1976a:127-28). Two major issues have dominated the semiotic analysis of symptoms from the be­ ginnings of symptomatology (or semeiology) in ancient Greece: (1) Subjective versus objective. Symptoms are generally different for adresser and *addressee with regard to what they signify as well as to the way they signi­ fy. Thus patients may experience stomach-aches and attribute them to stress ( = subjective symptom). Physicians, in addition to the patient's narrative, may in this case examine radiological and chemical signs ( = objective symptom) and conclude that the patient has an ulcer. (2) Natural (non arbitrary) versus conventional (arbitrary). While many semioti­ cians consider symptoms to be a subclass of natural ,:•indices (e.g., ,:•Peirce 1931-66:8.119), others have argued that they are in addition a subclass of cul­ tural and personal ,:•symbols. Both traditions can be integrated in the notion of symptoms as indexical symbols. In most �·cultures, symptoms were originally linked to oracular signs and were inter­ preted in the context of the overall ,:•meaning system (magical, religious, common sense, etc.) available to a particular culture. With the Prognostikon of the Hippocratic Corpus (ca. fifth century B. C.), symptomatology begins to develop its own code. Medical progno­ sis can be made either semeiotically, by interpreting bodily processes as signs of health and disease, or mantically, by recourse to a divinatory code. Geographic and seasonal differences have to be taken into account in the interpretation of symptoms ( = semeia), but they are not taken to be culture-relative in the modern sense. They have the same meaning in Lybia, in Delos, and in Scythia. With the rise of modern subjectivity during the ltalien ,:-Renaissance, the difference between subjective and objective symptoms is intensified and by the seventeenth century reflects a dualistic mind-';body ontology. Subjective symptoms are taken to refer to mental processes C:-Descartes' res cogitans), while objective symptoms are seen as signs of bodily processes (res extensa). In the medical literature since then, the term symptom is often reserved for the introspective subjective description by the patient; the intersub­ jective objective evidence, by contrast, is collectively referred to as signs. In the twen­ tieth century, a remarkable deepening takes place in the semiotic study of symptoms, due especially to biology, ,:•psychoanalysis, and ,:•anthropology. The new insights stem mainly from the fact that symptoms are seen as communicative acts, as attempts "to order the disordered pronoun, the unstructured "I," to define and communicate a self" (Staiano 1979:115). In biology, J. v.*Uexki.ill introduces, in the 1920s, the concept of the functional circle, in which subject-object relations are regulated by subject-specific codes. Accordingly, symptoms have to be interpreted within the subjective meaning sys­ tem enacted by a given organism. Uexki.ill's son Thure (T.v.Uexki.ill 1979a) applied his father's ,:•zoosemiotic views to medicine by (re)introducing the human subject as the main point of reference of all symptoms. He added to the univocal functional circle of animals the polysemic situa­ tional circle of humans, in which symptoms become multireferential symbols of the self. ,:-Freud's conception of the psychoanalytic symptom as a symbol of unconscious conflicts whose interpretation and cure is subject to interminable dialogic narratives is both ex­ emplary and too narrow for this person-oriented (as opposed to disease-oriented) con-

Synchronic I Diachronic

1035

cept of the symptom. Since the self is itself a sociocultural construct, the person-orient­ ed interpretation of symptoms is relative to cultural beliefs about self and illness studied by ethnomedical science, because "particular beliefs regarding the origin of the sign have become part of the symptom itself" (Staiano 1979: 117). The symptom as a "cultural­ ly ordered behavioral form" (Fabrega 1975) thus (re)assumes in medical semiotics the place within the overall meaning system of a given reference group which it originally had in the mythic past. E.B.

Synchronic/Diachronic (Greek syn 'with,' dia 'through,' and chronos 'time'; synchronic thus means "same time," i.e., "contemporaneous" (no time), diachronic "diffused through time." The terms were created to stress the fact that the functioning of a system is independent of its development.) The dichotomy is known primarily in linguistics, in the form that ,:•Saussure used in his Cours de Linguistique Generale (19 1 6). Saussure sought a counterbalance to the then dominant Neogrammarian interest in the historical mode of language study, which maintained that the historical approach is the only theoretically valid one. Saussure's ma­ nifesto states that the dichotomy is absolute and does not tolerate any compromise. At best, diachronic linguistics can be a supplement to synchronic linguistics, and it is possi­ ble only through collation of subsequent synchronic cuts. Saussure adds that synchrony is more difficult, diachrony more appealing ! The dichotomy and the primacy of syn­ chrony became a cornerstone of most varieties of ,:•structural linguistics in theory, al­ though leakage to varying degrees has been normal in practice (Seebold 1973 ; Vraciu 1979). Coseriu pointed out in 1958 that the true concept is not system and change, but sys­ tem in change. Every ''language state represents a systematic structure just because it is a moment in constant systematization, a concept which overcomes, in a fundamental way, the antinomy between diachrony and synchrony (Coseriu 1974 :236). The antino­ my pertains to linguistics, not to the facts of language, i.e., theoretically diachrony is primary, practically synchrony. A dynamic conception of structure must incorporate di­ achrony. Even Saussure developed his notion of system in historical linguistics, which provides an introduction to synchronic linguistics. The crucial element in �·communication and ''meaning is intention, which breaks strict synchrony (Hormann 1976 ). Furthermore, the very nature of the ''symbol is tele­ ological. Meaning is change (Shapiro 19806 ). ,:•Peirce (1 973-76 ) phrased this that there is a tabula rasa in the middle of the symbol in which ,:-interpretants grow out of the origi­ nal symbol. The symbol is a purpose that seeks to make itself definite. Symbols carry al­ so ,:-history with them, since the common stock of knowledge of utterer and interpreter is part of the sign. This anamnestic-proleptic character of the symbol is strong proof of the untenability of the dichotomy, but the matter waits fuller semiotic explication. R.A.

1 036

Synonymy

Synonymy Popular Views

Synonymy or synonymity - from the Greek syn 'with' and onoma 'name' - is colloquially referred to as sameness of �·meaning or similarity of meaning. Sameness of meaning Two words, phrases, in general *expressions, or, still more generally, '�signs are regarded as synonymous if and only if they have the same meaning, e.g., aim and purpose. Synony­ my is often contrasted with homonymy on the grounds that while synonyms have differ­ ent ,:•form, i.e., shape, and the same meaning, the homonyms, conversely, have different meanings and like form. This distinction, however, is not always tenable. According to one view, synonyms may also have the same shape, namely, it is sometimes claimed that synonyms are not only heteromorphic expressions with the same meaning, but also equiformal expressions with identical meaning, e.g., aim and aim, two tokens of the same expression in both cases occurring in the same meaning. The notion of sameness of meaning is imprecise if only because of the polysemy of the word meaning. Here we may distinguish two opinions:

( 1) Synonymy consists in the identity of meaning in the broadest sense, i.e., the entirety of semiotic functions of the sign, including, among others, the '�seman­ tic *cognitive meaning together with the *pragmatic emotive meaning, the '�dictionary meaning as well as the contextual meaning, with both the linguistic context and the situational context taken into account; and (2) Synonymy is merely sameness of cognitive meaning, the synonyms differing in emotive meanings, e.g., school teacher and beak, or environmental coloring, e.g., your good lady and your wife, lass, girl and flapper.

Similarity of Meaning According to some linguistis and ordinary-language-philosophers, synonymy in the sense (1) does not exist - a view already held in the fifth century B. C. by Prodicus of Keos, a Sophist - because two strictly identical situational contexts are impossible. This opinion leads to the replacement of the classificatory notion of synonymy as the identity of meanings by the typological notion, i.e., the degree concept of synonymy as the great­ er or lesser similarity of meaning. It seems that (2) is just such a degree concept of synon­ ymy. The popular concept of synonymy is sometimes subjected to the following restric­ tion: an expression E 1 of an ethnic language L 1 and its '�translation E2 into another eth­ nic language L2 should not be called synonyms. According to this view, the term applies only to the expression E 1 and its paraphrase E2 within the one ethnic language L 1 • Drawbacks of Popular Concepts of Synonymy The above restriction does not remove a number of drawbacks of the popular under­ standing of synonymy. One such shortcoming is that the word synonymy is explicated

Synonymy

1 037

in terms of meaning while the word meaning, apart from being polysemic, i.e., having different meanings, is in each of these meanings ambiguous and vague, and hence equally or perhaps even more incomprehensible than the word synonymy. Thus, defining synonymy with the use of meaning, we construct a ,:•definition burdened with the didactic error of ignotum per ignotum '(explaining) the unknown by means of the unknown'. Secondly, we often resort to an opposite course of action : to explain the meaning of some expression we provide its synonym. In doing so we rely on an intuitive conjec­ ture as to what is the synonym of the given expression. Such intuitions only seemingly precede the knowledge of that meaning. In reality, to be able to supply a synonym of the given expression we first have to know what it means, since the expression E2 is syn­ onymous to the expression E 1 with regard to the meaning M 1 of the latter, e.g., apart­ ment and flat, and may not be its synonym with regard to a different meaning M2 of its own, e.g., when flat is used as a musical term. This is why some philosophers ('�Ajdukiewicz 1978a :30-32), when defining the meaning of an expression E in the language L, by the principle of �·abstraction, as the class of all expressions in L synonymous with E, give empirical or even behavioral con­ ditions for the recognition of two expressions as synonymous, thus avoiding the vicious circle in their definition as well as defining ignotum per ignotum. In this lies the advan­ tage of the presently discussed direction of analysis over defining synonymy in terms of meaning. Using this last method, it is difficult to specify empirical conditions for accept­ ing something as the meaning of an expression because meaning is an abstraction. Another drawback of the popular notions of synonymy is the fact that as a rule they do not account for the important difference between synonymy and coextensiveness, that is the extensional equivalence of expressions. A result of this is that both grippe and influenza as well as the Morning Star and the Evening Star are given as examples of syn­ onyms, which leads to blurring of very important distinctions in logical and philosophi­ cal analysis, namely those of '�intension and �·extension, '�connotation and ,:•denotation, Sinn 'sense' and Bedeutung 'reference'. The efforts of logicians and philosophers are ac­ cordingly directed towards such an explication of the concept of synonymy that would preserve these distinctions.

Logical and Philosophical Views Ajdukiewicz's Synonymy and the Rules of Language Ajdukiewicz ( 1 978a :35-66) defined synonymy by referring to the •:•rules of language. A language rule specifies a relation between a �·sentence of a given language and an ex­ perience which, jointly with having the experience, is sufficient for a speaker to accept the sentence. For example, an axiomatic rule of language requires one to accept and pro­ hibits rejecting the sentence "Each square has four sides" ; a deductive rule of language tells a person who has accepted the sentence "London is bigger than Warsaw" to accept the sentence "Warsaw is smaller than London" ; an empirical rule of language obliges all those who look at snow in daylight to accept as true the sentence ' 'This is white." An expression E 1 is synoaymous with E2 in a given language if and only if: ( 1 ) each and only such an experience, or mental act, from which, by virtue of the rules of the language a sentence S 1 containing E 1 is immediately inferable in a manner essential for E l > is also an

1 03 8

Synonymy

experience from which in virtue of the rules another sentence S2 obtained from the former one by replacing E 1 by E2 is immediately inferable in a manner essential for E2 ; and (2) all and only those sentences which are immediately inferable from the accep­ tance of the sentence S 1 are also immediately inferable from the acceptance of the sent­ ence S1 • The necessary condition for the synonymy of two expressions of the same language (Ajdukiewicz 1 978a:56) is as follows: if such expressions are synonymous, then the total scope of rules of language is not altered if synonyms are replaced by synonyms in the rules. This last means that if there is a rule of language which requires the acceptance of a sentence S 1 or the inference from S 1 of another sentence S2 , and S 1 contains E 1 while S 2 contains E2 that is synonymous with E P then there is also another rule of language which requires the acceptance of S2 or the inference from S2 of another synonymous sentence S 1 • Analogously in the case of two languages: an expression E 1 of a language L 1 is synonymous with E2 of L2 if and only if E2 is a translation of E i , that is to say, if between E2 and the remaining expressions of the language L2, together with certain data of experience, there occur semiotic relations analogous to those occurring in the lan­ guage L 1 between E 1 and the respective expressions of that language together with the respective data of experience. According to the above view, two expressions are syn­ onymous in a given language if and only if the rules of the language say the same things about both expressions. Following oral criticism by Tarski, Ajdukiewicz weakened the above statement, leaving only the claim that the invariance of the rules of language called also meaning rules - is necessary for synonymy. The concept of language rules was associated with Ajdukiewicz's radical conventionalism, which he abandoned in the forties. Lewis' Sameness of Intensions or Equivalence in Analytic Meaning

,:•Lewis ( 1 943) believes that two expressions are synonymous if and only if they have the same connotation, i.e., intension, and that this intension is neither zero nor ,:•universal, or if their intension is zero or universal, they are equivalent in ,:•analytic meaning. In the above formulation, the intension of a term is delimited by any correct defini­ tion of it. For example, in the definition square Df. equilateral rectangle, the expression eq­ uilateral rectangle in the de/iniens specifies the connotation, i.e., the intension of the term The intension of a ,:•proposition includes whatever must be true of any ,:•possible world in order that the proposition should apply to or be true of it. For example, to the intension of the proposition that this figure is a square belong: that it is equilateral, that it has four angles, that it has two equal diagonals, etc. Two expressions are equivalent in analytic meaning ( 1 ) if at least one of them is elementary, e.g., satyr, i.e., it has no symbolized constituent the intension of which is a constituent of the intension of the ex­ pression in question itself, and they have the same intension; or (2) if, both being com­ plex, i.e., non-elementary, they can be so analyzed into constituents that (a) for every constituent distinguished in either there is a corresponding constituent in the other which has the same intension, (b) no constituent distinguished in either has zero inten­ sion or universal intension, and (c) the order of corresponding constituents is the same in both, or can be made the same without alteration of the intension of either whole ex­ pression, e.g., the winged horse and das ge/liigelte P/erd.

square.

Synonymy

1039

Carnap's Intensional Isomorphism ,:-Carnap (1947, 1949a) is of the opinion that the identity of the intension of two expres­ sions is in itself insufficient grounds for their being recognized as synonymous. The ex­ pressions ought also to be intensionally isomorphic, i.e., they should have the same inten­ sional structure. We say that the same intensional structure is exhibited by two expres­ sions if they are similarly built out of corresponding designators that have, respectively, the same intensions, e.g, 2 + 5 and II sum V These designators, according to Carnap, are expressions for which the relation of L-equivalence has been defined as follows: a and b are L-equivalent if and only if a = b is L-true; or, in other words, it is logically true and not in virtue of the real state of affairs according to which if something has property P it also has property Q, and conversely; e.g., in the above example, 2 and II are designators having the same intension, and 2 is L-equivalent to II Carnap's explication of the concept of synonymy involves the concept of interchan­ geability of expressions in all contexts salva veritate 'without altering truth-values' of sentences under consideration, and the notion of analyticity. The first is contained in the concept of L-equivalence, the second in the notion of L-truth, or logical truth, also called necessary truth or analytic truth. (See: Intension vs. Extension; Antinomy of Name Relation. ) Criticism of Carnap's Views It was pointed out (Linsky 1 96 7a:55) that the relation of intensional isomorphism is not transitive, which is indicated by the following example : flat is equivalent to apartment if and only if the following expression is true "For every x: if x is a flat than x is an apart­ ment, and conversely." Let us assume that (1) 'Ylat is equivalent to apartment " is L-true. Sentence (1) is intensionally isomorphic to: (2) 'Ylat is equivalent to flat " which has its definitional expansion in the form (3) "it is true for every x that if x is a flat, then x is a flat, and conversely." But, as we see, (1) is not intensionally isomorphic to (3). The non-transitivity of the relation of intensional isomorphism is inconvenient in that the relation of synonymy ex­ plicated with its use becomes non-transitive itself, i.e., from the synonymity of expres­ sions E 1 and E2 and also of E2 and E3 it does not result that E 1 and E3 are synonyms. The second objection concerns intensionally isomorphic sentences when these occur in belief sentences (Mates 1952). The sentence (S 1 ) "Equivalent terms have the same denotation" is intensionally isomorphic to: (S2) "Coextensive terms have the same denotation" hence, of course, the sentences (1) and (2) below are also intensionally isomorphic:

1 040

Synonymy

(1) "Whoever believes that S 1 believes that S i '' and (2) "Whoever believes that S 1 believes that S2 . " However, while the sentence (3) "Nobody doubts that (1)" is true, the sentence (4) "Nobody doubts that (2)" turns out to be false despite the fact that (3) and (4) are intensionally isomorphic. As we see, intensionally isomorphic sentences can differ in truth-value, while syn­ onymous sentences cannot. Hence, the identity of intensional structure of expressions cannot be regarded as the necessary and at the same time satisfactory condition of their synonymy. The remaining objections are aimed not so much at the concept of intensional isomorphism as at the requirement insisted on by Carnap, and many other authors, that synonyms be interchangeable salva veritate in all contexts. Interchangeability salva veritate This condition is often regarded as the criterion of synonymy of expressions by authors who wish to avoid explicating the notion of synonymy using the notions of intension, connotation, or meaning. They prefer to build on a more tangible property of synonyms, namely, the fact that if in any given sentence the given expression is replaced by its syn­ onym, the truth-value of the sentence remains unchanged. It turns out, however, that the principle of exceptionless interchangeability salva veritate of synonyms is open to criticism not only in the aforementioned belief-contexts but also in others, namely the so-called intensional sentences, or, more generally, inten­ sional expressions. For example, let us suppose that John asks whether aim is the same as purpose. In this case, despite the fact that aim and purpose are synonyms, the replacing of the word aim by the word purpose in a sentence may lead from a true sentence (1) "John asks whether aim is the same as purpose" to the false sentence

(2) "John asks whether purpose is the same as purpose." A different criticism comes from Goodman (1 952, 1953), who claims that even empty predicates, i.e., ones having null extension, e.g., unicorn and centaur, are not always in­ terchangeable salva veritate. Although their, as he calls it, primary extension is the same, their compounds, e.g., unicorn picture and centaur picture, have extensions - in this case called secondary extensions of the terms unicorn and centaur- that are different. Sim­ ilarly, the expressions unicorn description and centaur description have different exten­ sions and this explains the difference in the meanings to the predicates unicorn and cen­ taur. According to Goodman, any two predicates will be synonymous only if both their primary extensions are identical and likewise their secondary extensions. But this, ob­ serves Goodman, never happens: this (here concrete) printed inscription, unicorn that is

Synonymy

1041

not a centaur, i s referred to b y the compound predicate unicorn description and accord­ ingly constitutes a part of the extension of this expression. However, the inscription is not referred to by the compound predicate centaur description. Hence the two expres­ sions have different extensions and thus the simple predicates, unicorn and centaur, have different secondary extensions. Since unicorn and centaur are arbitrarily chosen pred­ icates, we may generally say that there are no two predicates having the same primary ex­ tensions and also all identical secondary extensions. From this Goodman draws the con­ clusion that no two predicates are synonymous and opts for the adoption of the degree concept of synonymy, that is synonymy as being similarity, to a certain degree, of mean­ ings. Lastly, the third objection (Alston 1967: 240) to the criterion of interchangeability salva veritate of synonyms consists in pointing to the existence of sentences to which this criterion does not apply, for the reason that they are neither true nor false. For example, no truth-value can be assigned to sentences that are �·questions, commands, or that are used as examples.

Cognitive Synonymy and Analyticity Not only in Carnap but in numerous other authors we find the opinion that cognitive synonyms, e.g., eatable and edible, differ from equivalent expressions, e.g., creature with a heart and creature with kidneys, in that the interchangeability salva veritate of the former is of necessary, analytic character arising as it does from the rules of language themselves, thus being in fact interchangeability salva analycitate 'with saving analyticity' (Quine 1953d), while the interchangeability salva veritate of the latter is contingent and factual in character, being due to the empirical state of affairs. We see, therefore, that cognitive synonymy is explicated here in terms of analyticity. But the converse also happens: analyticity is universally being defined in terms of meaning - namely, we call analytic sentences which owe their truth solely to the mean­ ings of the words they are composed of and not to the state of affairs they refer to. In more detailed analysis we abandon the notion of meaning in favor of the concept of cog­ nitive synonymy, and by analytic sentences we mean two kinds of sentences: (1) "No unmarried man is married" and (2) "No bachelor is married." (1) is logically true while (2) may be transformed into (1), that is into a logical truth, by replacing bachelor with unmarried man; we assume here that an unmarried man is one who has heretofore never been married, thus eliminating widowers and divorced men. The transformation of (2) into ( 1 ) is an instance of replacing an expression with its cog­ nitive synonym. Thus the analytic character of (2) has been explicated in terms of cogni­ tive synonymy. Quine (1953d) and M. White (1952, 1956) reject this explication because cognitive synonymy is a notion no clearer than analyticity. Moreover, Quine points out that the concept of cognitive synonymy acquires definite sense only when it is explicated in terms of analyticity, that is, when we accept the sentence

1042

Syntactics

(3) "All and only bachelors are unmarried men" as analytic. And we regard sentence (3) as analytic first because the sentence (4) "All and only bachelors are bachelors" is obviously necessarily true and hence analytic, and second because (3) arises from (4) by the replacement of the second occurrence of bachelors in (4) by the expression un­ married men. In this context the question arises whether this exceptionless interchangea­ bility salva veritate of the expressions bachelor and unmarried man is indeed of necessary, analytic character, or whether it is perhaps contingent, factual as in the case of creature with a heart and creature with kidneys. Hence the acceptance of (3) as an analytic sent­ ence also becomes questionable. The thing is that the word necessary is vague: namely, we may believe that not only necessarily all and only creatures with a heart are creatures with a heart, but, in another sense of the word necessarily, necessarily all and only crea­ tures with a heart are creatures with kidneys. In the former case the necessity arises from the rules of language, while in the latter it springs from the laws of nature. Grice and Strawson (1956) have observed that analyticity, necessity, and cognitive synonymy form a family of terms, and Quine refuses to accept as an explication of any one of them a reference to any other member of the family, believing, moreover, that when we venture outside the family, e.g., referring to the concept of extensional equivalence involved in the concept of interchangeability salva veritate, our explications are inadequate (Hamlyn 1967: 1 07). (See: Intension vs. Extension.) Stimulus Synonymy Quine also considers the concept of stimulus synonymy. This concept is applicable ex­ clusively to sentences, and particularly to the so-called occasion sentences, e.g., "John has the hiccups. " lntrasubjective stimulus synonymy is the sameness of stimulus mean­ ing for one speaker. The stimulus meaning of a sentence for a person is defined in terms of the class of kinds of stimulation which would prompt the person's assent to the sent­ ence (Presley 1967). Two sentences are socially stimulus synonymous if they are intra­ subjectively stimulus synonymous for nearly everyone who speaks a given language. It is only with difficulty that the concept of stimulus synonymy can be applied to the so-­ called standing sentences, e.g., "ice is lighter than water. " The above drawback notwith­ standing, it has the advantage that with its help one can attempt to analyze the concept of synonymy of '�nonverbal signs, heretofore undefined ; thus, for example, one can search for the necessary and at the same time sufficient condition for the synonymy of two iconic signs, or the synonymy of an iconic sign, a ,:•tactile sign, an ,:•olfactory sign and a musical sign. J.P.

Syntactics Semiosis Syntactics is that branch of •�semiotics that studies the formal aspects of *signs, the rela­ tions between signs, and the combinations of signs. In order to understand what is cov-

Syntactics

1043

ered by this characterization, one has to consider syntactics in the context of the other branches of semiotics, which are usually defined with respect to its subject matter, *semi­ osis. Semiosis is the process in which something functions as a sign. Different semioti­ cians distinguish different factors in semiosis; cf., e.g., the traditions of �·Peirce (1931-66), *Morris (1938, 1946, 1964), and �·Carnap (1939, 1942, 1954); of *Saussure (1916), ''Hjelmslev (1943), Buyssens (1943b), and Prieto (1966); of �·Ogden and ''Ri­ chards (1923 ), �-Buhler (1934 ), and *Jakobson (1960); and of *Uexki.ill (1940a). However, all such structurings of semiosis involve at least three factors, whose interrelation is described by phrases such as : "A takes B to stand for C," "B refers to C for A," "A takes account of C by virtue of the presence of B," etc. We therefore stipu­ late that a necessary and sufficient condition for something to be a semiosis is: (1) A interprets B as representing C. In this relational characterization of semiosis, A is the interpreter, B is some object, property, relation, event, or state of affairs, and C is the ,:•meaning that A assigns to B. These factors are connected by the triadic relation:. . . interprets --- as representing .-.-.-. The term sign can be used in two different ways with respect to this relation. While logic-oriented semioticians like Morris and Carnap use it as a term for B's, linguis­ tics-oriented semioticians like Saussure and Hjelmslev use it as a term for pairs of B's and C's. The latter usage is justified by the fact that what meaning is assigned to some entity in a specific semiosis depends in part on properties of that entity. However, it also depends on properties of the interpreter : It is true for all three ranges of entities that can fill the three slots in the triadic relation that they are dependent on each other. In the rest of this article we will therefore use a separate term for each of the three factors and refer to A as an interpreter, to B as a sign, and to C as a meaning.

Syntactics, Semantics, and Pragmatics

Because of the interdependence of the three factors it is not easy to devise a clear-cut division of labor for the investigation of semiosis. Nevertheless, a tripartite division of semiotics is now generally accepted. It is usually defined with regard to the three factors of semiosis. The conditions an entity must fulfil in order to be able to represent meaning for interpreters in semiosis is the subject matter of syntactics. The conditions an entity must fulfil so that it can be represented by signs for interpreters in semiosis is the subject matter of ''semantics. The conditions an entity must fulfil in order to be able to interpret signs as representing meaning in semiosis are the subject matter of ,:•pragmatics. The historical origins of these disciplines can be traced back to the artes dicendi, viz. ''grammar, �·rhetoric and dialectic, the teaching of which was organized in the so-called trivium in medieval European schools from the ninth century A.D. onwards. Peirce rein­ terpreted the artes dicendi as branches of "semeiotic" and systematized them as disci­ plines treating signs as Firstness, Secondness, and Thirdness, respectively. He distin­ guished between speculative grammar, critical logic - the successor of dialectic -, and methodeutic - the successor of rhetoric (cf. Peirce, 1931-66 : l .191ff., 2.93). It was Morris (1938) who coined the terms syntactics and pragmatics and introduced the di­ vision of semiotics into syntactics, semantics, and pragmatics. Although this trichotomy is related to that of Peirce, its introduction was motivated differently : Morris tried to show that the objectives of three leading philosophical movements of his time, logical

1 044

Syntactics

positivism, empiricism, and pragmatism, were not antithetical but complementary (cf. Morris 1 929, 1 937), since logical positivism studied the formal structures of the lan­ guages of science, empiricism the objects of research and their relations to the languages of science, and pragmatism the procedures and conventions governing communication among scientists. Thus, for Morris, syntactics could utilize the methods and results of logical syntax developed by the logical positivists (cf. Carnap 1 934 ), while semantics and pragmatics could proceed from the analytical achievements of empiricism and prag­ matism, respectively. Despite their general acceptance, the theoretical status, delimitations, and subdivi­ sions of the three branches of semiotics are still controversial today. The Morris of the Encyclopedia of Unified Science ( 1 93 8 ) treats the three branches in a parallel way. He iso­ lates the three factors of semiosis and specifies the subject matter of each semiotic disci­ pline on the basis of a dyadic relation between one of the factors and the sign : Syntactics studies signs in relation to other signs (B in relation to B), semantics studies signs in rela­ tion to meaning (B in relation to C), and pragmatics studies signs in relation to interpre­ ters (B in relation to A). Since these three definitions do not cover all aspects of semiosis (e.g., the triadic relation between A, B, and C) and of semiotics (e.g., the problem of the interrelation of the three semiotic disciplines), all remaining aspects are said to fall under semiotics proper. In this conception pragmatics cannot treat much of the relation between signs and interpreters without including meanings, i.e., without becoming semiotics proper. This and other criticisms made Morris give up the parallel conception of the three branches in favor of a hierarchical one proposed by Carnap :

If we are analyzing a *language, then we are concerned, of course, with expressions. But we need not necessarily also deal with speakers and designata. Although these factors are present whenever language is used, we may abstract from one or both of them in what we intend to say about the language in question. Accordingly, we distinguish three fields of investigation of languages. If in an investigation explicit reference is made to the speaker, or, to put it in more general terms, to the user of a language, then we assign it to the field of pragmatics. (Whether in this case reference to designata is made or not makes no difference for this classification.) If we abstract from the user of the language and analyze only the expressions and their designata, we are in the field of semantics. And if, finally, we abstract from the designata also and analyze only the relations between the expressions, we are in (logical) syntax. The whole science of language, consisting of the three parts mentioned, is called semiotic. ( 1 94 2 :9)

As Morris emphasized (1 946:21 8f. [ = 1971a:302], cf. 1 938: 1 6 [ = 1 971 a :30f.]), this exposition can be adopted if it is generalized in a number of aspects: 1 . Semiotics not only deals with linguistic expressions but with all kinds of signs ; therefore "language" has to be replaced by "sign system." 2. Pragmatics has not only to do with users and uses of signs in the sciences but with all kinds of sign users and sign uses. 3. Semantics not only treats the designative mode of signification but all ways of representing meaning and all kinds of meaning. 4. Syntactics not only comprises the logical syntax of the languages of science but is much wider, also including (a) phonology and morphology of the languages of science, (b) phonology, morphology, and syntax of other languages, and (c) the analysis of the formal aspects of nonlinguistic sign systems.

Syntactics

1 04 5

The result of this generalization was formulated by Morris in the glossary of his 1 946 book (352ff. [ = 1 97 1 a :365ff. ]) : Pragmatics is ' 'that branch of semiotic which studies the origin, the uses, and the effects of signs." Semantics is "that branch of semiotic which studies the signification of signs." Syntactics is "that branch of semiotic that studies the way in which signs of various classes are combined to form compound signs. It abstracts from the signification of the signs it studies and from their uses and effects [ . . . ]. " Whereas Carnap had made syntactics seem to be the basis of all semiotic studies in 1 934 and Morris had treated syntactics, semantics, and pragmatics as relatively independent of each other in 1 938, they later tended to see syntactics as embedded in semantics and semantics as embedded in pragmatics (cf. Carnap 1 939 : 1 6 ; cf. also Cherry 1 95 7 :219ff.). This and the fact that Morris ( 1 946) avoids speaking of �·abstraction in his characteriza­ tion of pragmatics (and semantics) have led some of his exegetes to speak of "Morris' pragmatically unified semiotics" (cf. Apel 1 973) and to advocate the identification of semiotics with pragmatics. However, this would substitute a maximalist conception for the minimalist conception of pragmatics in Morris 1 93 8 and was clearly not the inten­ tion of Morris or Carnap at any time of their lives (cf. Posner 1 979).

Delimitations and Subdivisions of Syntactics

It is not always easy to apply the term syntactics to the traditional problem areas of semi­ otics and to decide whether they are included within syntactics or not (cf. Morris 1 946 : 2 1 9f. [ = 1 97 1 a :303 ]). In many cases the answer will be different according to which conception of syntactics it is based on :

- Syntactics 1 as the study of the formal aspects of signs (Morris 1 929, 1 937 and 1 93 8 : 1 3ff. [ = 1 97 1 a :27ff.]), - Syntactics2 as the study of the relations of signs to other signs (Morris 1 937 and 1 938 :7ff.[ = 1 97 1 a :23ff. ]), or - Syntactics3 as the study of the way in which signs of various classes are combined to form complex signs (Morris 1 93 8 : 1 4 [ = 1 971a :28f.], 1 946 :354f. [ = 1 97 1 a:367]).

The areas of research defined by the three characterizations overlap but are not identi­ cal. In what follows we will therefore use the term syntactics without a subscript only when we speak of syntactics proper, i.e., the discipline whose subject matter is the inter­ section of the subject matters of syntactics I > syntactics2 , and syntactics3 • Formal Aspects of Signs

The question of what are formal aspects of signs has been given alternative answers by different traditions of semiotics. While logic-oriented semioticians like �·Wittgenstein ( 1 922b: 3.33), Carnap ( 1 934:1, 208; 1 93 8 : 1 6), and Morris (1938 : 1 3ff. [ = 1 97 1 a :27ff.]) equate the distinction between form and substance with the distinction between signs (B's) and meanings (C's), linguistics-oriented semioticians like Saussure ( 1 9 1 6 : 1 55-57) and Hjelmslev ( 1 943 :69-73, 98ff. [ = 1 963 :76-8 1 , 1 1 Off.]) use it to differentiate two kinds of aspects within signs (B's) as well as within meanings (C's). Yet the underlying conceptions of form are quite similar and are based on the theory of invariants as devel­ oped in geometry ; cf. the "Erlangen Program" of f. Klein ( 1 872 [ = 1 92 1 ] :463f.):

1046

Syntactics

A generalization of geometry raises the following problem: Let there be a manifold of elements and a group of transformations defined on them; now the configurations of elements in that manifold should be studied with respect to those properties that remain unchanged when trans­ formations of that group are applied to them. The task is to develop the theory of invariants for that group.

In such configurations of elements two different types of entities may vary. On the one hand, certain properties of the configurations can change while others remain constant, cf. movement in space under similarity transformation or mirroring, where the location of a figure varies but all its spatial proportions stay the same. On the other hand, the elements themselves can also change, as when geometrical figures are replaced by sequences of letters or numbers. What remains invariant here is the abstract structure of the figure (cf. Holenstein 1975 :30ff. and Wunderlich 1971a). The structure of a relation was extensionally defined by •:•Russell (191 9b:59ff.) and Carnap ( 1928: 13ff.) as the class of relations isomorphic with that relation. This approach was developed further in the mathematical theory of �·structures published by the group of French mathematicians Bourbaki (1935). It was applied to the theory of definition by Hilbert (cf. 1918, 1925), who proposed to define basic technical terms of a theory implicitly by specifying the axioms in which they occur. Discussing the develop­ ment of '�mathematics, Hilbert wrote in 1925 : In intuitive number theory formulas were always exclusively used for communication. The let­ ters stood for numbers and an equation communicated the fact that the two numbers coin­ cided. In algebra, on the other hand, we regard expressions containing letters as independent structures which formalize the material theorems of number theory. In place of statements about numbers, we have formulas which are themselves the concrete objects of intuitive study. In place of number-theoretic material proof we have the derivation of a formula from another formula according to determinate rules. (In Putnam and Benacerraf 1964: 1 45.)

Carnap (1934) applied this idea to the study of language in general. He was con­ vinced that for every language one can work out "a formal theory of the forms of that language" (1934 : 1 ). It was the task of that theory to specify the rules that determine the sign forms of a language. For Carnap, the formal nature of the concrete individual sign was no problem, as it was for Hjelmslev (see below). Therefore, he was content to char­ acterize the sign forms of a language by specifying "the classes and the serial relations" of their components (1934: 1 ). Analogous to Hilbert's program of metamathematics, Carnap's "formal theory" was to be formulated in a special �·metalanguage whose expressions refer to the object-language in question (1939:5). Morris (1938 :9 [ = 1971a:23f.], 1 946:1 78£. [ = 1971a:256f.]) generalized this conception for sign sys­ tems of all kinds with all their aspects. If we call a sign system under study an object-­ code, Morris (1938 :13ff., 21ff., 29ff. [ = 197la:28ff., 35ff., 43ff.] ) envisaged three different metalanguages dealing with three different dimensions of semioses involving signs in the object-code. The metalanguages differ in what their expressions refer to. While the pragmatic metalanguage refers to the interpreters and the semantic metalan­ guage refers to the meanings of the signs in the object-code, a metalanguage whose descriptive terms all refer only to the sign forms of the object-code is called a syntactic metalanguage. It is worth noting that each metalanguage has itself all properties of a sign system. Thus the syntactic metalanguage not only has a syntactic dimension insofar as it contains

Syntactics

1047

signs of various classes with various serial relations holding among them, it also has a semantic dimension since its signs are interpreted to represent meaning and refer to something, viz. to the signs of the object-code, and it has a pragmatic dimension insofar as it can be used by special types of interpreters, viz. by syntacticians. This being the case, it is the semantic dimension of the syntactic metalanguage that has to do with the syntactic dimension of the object-code. And it is the semantic metalanguage of the syn­ tactic metalanguage that deals with the relations between signs of the syntactic metalan­ guage and signs of the object-code. Viewed from this perspective, the text of this article can be understood to be written in the pragmatic metalanguage of the syntactic metalan­ guages devised by semioticians of various traditions to deal with object-codes. What has been said about the syntactic metalanguage so far takes care of the terms "forms of a language" in Carnap's program and "formal aspects of signs" in Morris's version thereof. Understood in this way, syntactics 1 becomes a subdiscipline of syntac­ tics2. There are, however, two problems left; one is the task of setting up a "formal the­ ory" within the syntactic metalanguage, it will be treated below; the other is the problem of characterizing the formal aspects of a sign that is not a complex sign. The latter problem was approached by Hjelsmlev, who, like Carnap (1928:11), insisted "that a totality does not consist of things but of relationships, and that not substance but only its internal and external relationships have scientific existence" (Hjelmslev 1943 :22 [ = 1963:23]). This approach had been successful in the theory of human speech sounds. Reflecting on Trubetzkoy's (1929) analysis of vowel systems, Buhler (1931) had distinguished the acoustic sign from the concrete sound event by saying that the former was connected with the latter as form is connected with unformed matter. In 1934, Buhler described the complex relation between the properties of sign matter and sign form with reference to the areas of two overlapping geometrical figures (1934:28, 42ff.). In order to recog­ nize the sign form in the sign matter produced in semiosis, the interpreter must on the one hand concentrate on the relevant properties of the sign matter and abstract from the irrelevant ones (principle of abstractive relevance), and on the other hand complement properties of the sign form not manifested in the sign matter (principle of apperceptive complementation of the sign ,:•Gestalt). The distinction between sign matter and sign form can be exemplified by the way different languages pattern the sign matter which Berliners produce in order to name their city. It becomes [bE:'ltn] in English, [�cr'li:n] in German, [�a: K 'li�n] in Danish, and [bEA.uAinu] in Japanese. Hjelmslev, who uses this example (1943 :52 [ = 1963 :56 ]), de­ scribes it by saying that one and the same sign matter ("purport") is modeled into dif­ ferent sign substance through the sign forms supplied by the different languages. The sign forms comprise the language-specific invariants patterning sign matter. According to Hjelmslev, these invariants are independent of the medium involved; thus a given sign form can be realized by sign matter in various media, as when the speech chain of German [�Er'li:n] is transformed into the written word Berlin. From this he concludes (1943:91f. [ = 1963 :103f.]): " 'Substance' cannot itself be a definiens for a language. [ . . . ]Here is a graphic 'substance' which is addressed exclusively to the eye and which need not be transposed into a phonetic 'substance' in order to be grasped or under­ stood." What is true for language also holds for all other sign systems. In phonology, the difference between substance and form is captured terminologically by the distinction

104 8

Syntactics

between phonetic and phonemic entities, studied by phonetics and phonemics, respec­ tively. This distinction was generalized in American *structuralism by isolating the suf­ fixes -etic and -emic (cf. Bloomfield 1933 ; Pike 1954) and using them in the description of nonlinguistic sign systems. Thus a description in terms of behavioremes considers all and only those properties of a given behavior that are relevant for it to represent meaning to interpreters of the kind in question, while a description including other aspects of behavior would be called behavioretic. In conclusion, it should be noted that syntactics 1 studies sign forms and disregards sign substance and sign matter; it includes phonemics and excludes phonetics. This delimitation, which relies on the differences between concrete and abstract, should not be confused with a distinction introduced by Peirce (4.537) and developed further by Reichenbach (1947:4, 21, 284, 336), which is based on the difference between individual and general. For practical purposes, sign forms must be reproducible since we want to use them on more than one occasion. The individual sign form is called a token. Thus, in the two sentences "Whatever happens, Berlin will remain Berlin" and "Berlin is situated in Germany," we have the same word Berlin, but appearing in three different tokens; and in giving the explanation, a fourth token of the word has been produced. This can be described by stating that the four sign forms are tokens of the same sign '�type. Thus the common formulation "the same sign occurs in different places" amounts to saying "(sign (form)) tokens of the same (sign (form)) type occur in different places." The independence of the type/token distinction from the form/sub­ stance/ matter distinction is demonstrated by the fact that one can also distinguish tokens from types in sign substance and in sign matter. Relations Between Signs

The statement that in a complex sign form like "Berlin will remain Berlin," two com­ ponent sign forms are tokens of the same type is a statement about a relation between these sign forms and thus falls into syntactics2 > the study of the relations between signs. Traditionally, there are said to be two kinds of relations between signs (cf. Morris 1938 :6f., 18 ff. [ = 1971 a:2 l f., 32ff.]) - relations between signs occurring in a given com­ plex sign and relations between signs in a sign system (�·code). Relations of the first kind are usually exemplified by syntagmatic relations, relations of the second kind by paradig­ matic relations (cf. Kruszewski 1884; Saussure 1916; Hjelmslev 1943). The difference can best be demonstrated with respect to the process of sign production. In producing a complex sign, the sender will make successive selections from the inventory of a sign system mastered by him and, according to certain �-rules, combine the elements chosen into an appropriate *structure. Let us take a natural *language like English as the sign system and an *utterance of the sentence "The child sleeps" as an example. Then, "if child is the topic of the message, the speaker selects one among the extant more or less similar nouns like child, kid, youngster, tot, all of them equivalent in a certain respect, and then, to comment on this topic, he may select one of the semantically cognate verbs sleeps, dozes, nods, naps. Both chosen words combine in the speech chain" Qakobson 1960 :358). The set of elements that provides the basis of selection in each step of the sign production is called a paradigm, and the result of the combination of the elements selected is called a syntagm.

Syntactics

1 049

Paradigms need not consist of semantically equivalent signs, as in Jakobson's exam­ ple, but they can consist of signs having the same distribution in syntax, like sleeps, lies, stands, signs belonging to the same lexeme, like sleep, sleeping, sleeps, slept, signs containing the same root, like sleep, sleeper, sleepy, signs having the same phonemes in certain positions, like sleep, sweep, steep, or even signs having the same subphonemic properties, like German ich, nicht, Licht, where !xi is pronounced differently from ach, Nacht, lacht. Semantic, syntactic, inflectional, derivational, phonemic, and phonetic paradigms all have the same structure. Each of them is a class of elements equivalent to one another in a certain respect. Paradigms that fulfil additional conditions, such as that of mutual substitutability of all their elements in specified types of contexts salva bene/or­ matione, i.e., with well-formedness being preserved in the interchange, are called syntactic categories (cf. Bar-Hillel 1964). While paradigms are constituted by relations of (partial) equivalence, syntagms are constituted by relations of contiguity. Contiguity can be conceived of as neighborhood in space or time, as a restriction determining the distribution of one constituent with respect to another, or as a dependency relation like agreement in number between a and child or agreement in person and number between child and sleeps in the above example. Since syntagms are complex signs produced by some interpreter, syntagmatic relations are part of the surface structure of those signs. A syntagmatic relation is to be distin­ guished from a deep-structure relation and from a syntactic rule determining either of them (cf. Posner 1 982 : 1 29-59). Although originally defined as a term for linear confi­ gurations ("serial relations ;" cf. Saussure 1 9 1 6 : 1 7 1 and Carnap 1934 : 1 ) of signs (i.e., *texts), the term "syntagmatic relation" is also applied to signs combined in more than one dimension, as occurring in '"visual art, *music, *theater, and *film. Syntagmatic and paradigmatic relations are conceptually distinct, but they can occur together as in paradigms of syntagms. "The child sleeps," "The youngster dozes, " "The kid nods," "The tot naps" constitute a paradigm o f elements equivalent with respect to the syntagmatic relations holding within each of them; another case in point is the inflectional paradigm exemplified above. The disciplines studying paradigmatic and syntagmatic relations between signs are called paradigmatics and syntagmatics, respectively. They are part of syntactics2 • Syntactics must not be confused with syntagmatics and with syntax as defined in lin­ guistics. As is obvious from the examples discussed, paradigmatic and syntagmatic rela­ tions can be found on all levels of language. This fact has been exploited by structural lin­ guists to use these relations in the definition of the levels of language and of the disci­ plines studying them (cf. Bloomfield 1933; Harris 1 947 ; Hjelmslev 1943 ) : phonetics studying the physical properties of linguistic sound matter, phonemics studying the rela­ tions between phonemes - i.e., the smallest sound forms used to distinguish the signs of a language - morphology studying the relations between morphemes - i.e., the smallest combinations of phonemes representing meaning to the language user - and their combinations into words (derivational morphology) and word forms (inflectional morphology), syntax studying the relations between phrases - i.e., combinations of word forms within and into *sentences - and lexicology studying the relations between paradigms of word forms having the same meaning. Of these disciplines, phonetics does not deal with formal aspects of signs and is therefore excluded from syntactics 1 ; phonemics deals with formal aspects of signs, but not with combinations of signs and is therefore excluded from syntactics3 ; inflectional and derivational morphology and

1050

Syntactics

syntax deal with formal aspects of signs, their relations, and their combinations and are part of syntactics proper; lexicology deals with formal aspects of signs but not with their combinations and is therefore excluded from syntactics,. As it turns out, syntactics proper includes only morphology and syntax from the lin­ guistic disciplines, and it is no accident that this is exactly what linguists have traditionally called •�grammar. Thus it is justified to regard syntactics as a semiotic gene­ ralization of grammar. In many contexts, the Carnapian identification of syntactics with syntax is highly misleading. Only in sign systems which do not require a distinction between morphology and syntax is it unproblematic to equate syntactics with syntax. This is the case in sign systems such as the numerals and in most of the formal languages so far constructed in logic. In conclusion it should be noted that syntactics proper contains both syntagmatics and syntax as subdisciplines, but syntagmatics overlaps with syntax, since syntax studies not only syntagmatic but also paradigmatic relations between phrases and syntagmatics studies not only phrases but also morphemes and phonemes. Complex Signs

Analysis and synthesis. In his program for a "logical syntax of language," Carnap (1934:1) had envisioned "a formal theory of the forms of a language" to be formulated in a syntactic metalanguage. There have been various attempts to work out such a formal theory. Most of them have been guided by the idea of a calculus, i.e., an axiomatic system that has the properties of an algorithm for specifying exactly the set of all signs belonging to the object-code under investigation. The specification of a set of more or less complex objects can be given either by starting with the complex objects and introducing rules for their analysis into compo­ nents, components of components, etc. until elementary objects are reached. Or one can proceed from elementary objects and introduce rules for their use in the synthesis of more and more complex signs. The two approaches are of different value for different kinds of sign systems. For many sign systems that are in use in human or animal societies or within organisms or machines, it is by no means clear what the basic elements are. The most controversial examples include *dance, gestures, pictures, films, and •�architec­ ture. However, there are also sign systems where it is clear what the elementary signs are and hard to decide for a given complex sign whether it is part of the sign system in ques­ tion or not and what is its structure. Such cases occur in some of the richer artificial lan­ guages of logic and mathematics. For these reasons it has become customary to use the analytic approach in the study of natural sign systems that have historically grown and the synthetic approach in the study of artificial sign systems (cf. Lotman's [1969, 1971a] distinction between text-oriented and grammar-oriented *cultures, discussed in Eco 1976a:137ff. ; cf. also Pape 1980). Linguists such as Bloomfield (1926, 1933), Harris (1947), Wells (1947), Pike (1954), Hockett (1958), Hjelmslev (1943), and Prieto (1966) have developed proce­ dures for the step-by-step analysis of texts into components, components of compo­ nents, etc. The formal theory for this approach has been discussed by authors such as Hjelmslev (1943), Marcus (1967), and Harris (1970). According to Hjelmslev, syntactic theory has to provide a general calculus containing rules of partition for complex signs

Syntactics

1 05 1

in all sign systems. The application of such a calculus to a given complex sign involves a finite set of partition operations, the last of which yield basic elements of the sign system in question. The syntactic structure of that complex sign is described by describing its analysis. The basic elements of the whole sign system are obtained by analyzing complex signs belonging to the system until no new basic elements are generated (cf. Hjelmslev 1 94 3 : 27-3 1 [ = 1 963 :28-33]). In judging the value of this approach one must distinguish between the continental European tradition and the American tradition. The first relies on Hjelmslev's commuta­ tion method, which is applicable to sign systems of all kinds but does not abstract from the meaning of the signs analyzed, in the way required by Carnap and Morris. The sec­ ond abstracts from the meaning of the signs analyzed but relies on the method of pars­ ing, which is applicable to languages only. Philosophers and logicians such as *Leibniz (cf. Knobloch 1 973), Boole ( 1 8 54), •:•Frege ( 1 879, 1 893- 1 903), Schroeder ( 1 890), *Peano ( 1 894), Peirce ( 1 934), Whitehead and Russell ( 1 9 1 0- 1 3 ), Carnap ( 1 934, 1 939, 1 954 ), and Curry ( 1 963) were the first to develop step-by-step procedures for the construction of more and more complex signs out of basic elements. The formal theory for this approach has been given by authors such as Thue ( 1 9 1 4), Post (1 936), Turing ( 1 937, 1 950), Hermes ( 1 938 , 1 96 1 ), Markov (1 947), Lorenzen (1955 ), Davis ( 1 958), Trakhtenbrot ( 1 960), Chomsky and Miller ( 1 963), *Bar-Hillel ( 1 964), and Lindenmayer (cf. Lindenmayer and Rozenberg 1 976, 1 979; Rozenberg and Salomaa 1 974). According to Chomsky, syntactic theory has to specify the general form of a calculus that generates all the expressions, i.e., the simple and complex signs, of a given language, starting from a finite set of basic elements and using a finite set of rules of various types. The application of such a calculus to an initial string involves a finite set of production operations, the last of which yields an expres­ sion of the language in question. The syntactic structure of that expression is described by describing its production. The set of expressions of the language is obtained by applying the rules of the calculus to all its basic elements (cf. Chomsky 1 957: 1 8-91 ). Compared with the analytic approach, the calculi developed for the synthetic approach have reached a much higher sophistication. In addition, logicians like Carnap ( 1 934 :8), Quine ( 1 960), ''Montague ( 1 970a, 1970d), and Cresswell ( 1 973) and linguists like Chomsky ( 1 965) and Shaumyan ( 1 970) have shown that it is possible to apply the synthetic approach in the analysis of natural languages also by using the so-called "indi­ rect method" (cf. Schnelle 1 973 ). The point of this method is to introduce an artificial sign system in the metalanguage, which can be kept under tight control by means of the stipulative definitions with which it was constructed, to compare this sign system with the object-code, and to incorporate more and more features of the object-code in it so that in the end the set of signs belonging to the object-code becomes completely recon­ structed in the metalanguage. This strategy has worked well in the analysis of natural languages. If it is to be applied to the study of nonlinguistic sign systems, an additional obstacle has to be over­ come. In language, the relations between the components of a complex sign are generally thought to be based on one single serial relation, e.g., the relation "following in time" as in speech or "immediately to the right" as in European *writing systems. Complex signs governed by serial relations can be produced through application of an associative non-communicative binary operation, called concatenation. As noted above, however, there are sign systems that have either additional serial relations or equival­ ence relations or relations of more complex types governing their complex signs.

1052

Syntactics

If one wants to describe the syntactic structure of complex signs in sign systems using operations different from concatenation, one can again choose between a direct and an indirect strategy. The first strategy consists in defining appropriate operations of com­ bination and describing the complex signs directly on their basis. This strategy is applied in the logic of relations (cf., e.g., Peirce 1 897, 1934, 1 976; Roberts 1973 ), and discussed in computer graphics (cf. e.g., Faiman and Nievergelt 1969; Rosendahl 1 973; Gips 1 974; Gonzalez and Thomason 1978). In the second strategy, a system of notation is devised to represent the relevant fea­ tures of the complex signs in question in a way that makes them more amenable to lin­ guistics-type syntactic analysis. Notational systems tend to reduce multi-dimensional sign configurations to two-dimensional ones (scores) or one-dimensional ones (strings). Examples are musical notations and the phonetic transcriptions of natural lan­ guages. Theoretically, it is always possible to reduce a given n-dimensional sign com­ plex to a complex with n- 1 dimensions as long as the relations among its constituents are serial or equivalence relations (cf. Greenberg 1 957 :95f. ; Curry 1963 :50-69; Goodman 1968 : 1 27-224). In the analysis of pictures, this approach was introduced by Miller and Shaw ( 1 967, 1 96 8 ; cf. also Shaw 1969) and further developed by Dacey ( 1 970), Rosen­ feld ( 1 97 1 , 1 976, 1 979), Fu ( 1 974, 1 976, 1 982), Franck ( 1 978), and others. For a general discussion see Niemann 1974, Gonzalez and Thomason 1978, and Nag! 1 979. To sum up, the best developed branch of syntactics3 is able to describe the syntactic structure of sign systems for which two conditions are fulfilled : ( 1 ) A set of basic elements is given from which all well-formed signs of the system can be constructed by combinatory operations. (2) All combinatory operations can be reduced to or defined on the basis of one single binary operation, viz. concatenation; all complex signs of the system there­ fore are, or are reducible to, strings. Sign systems with these properties are called string codes. They include, among others, natural languages, *writing systems, clothing codes, *culinary codes, and �-traffic signs. As a technical device for the syntactic description of string codes, formal grammars have been developed which may be characterized as string production grammars. They are a special type of the so-called *generative grammars (cf. Chomsky 1957, 1965; Bar-Hillel 1964; Marcus 1968; Hermanns 1977). The rest of this section will give formal definitions for the concepts of a string and a string code, introduce some basic types of string production grammars, and explicate the concepts of the syntactic struc­ ture and the structural description of a string on this basis.

The concept of a string. Let V be a finite set. Then a string of length n (n 2::: 1) over V is a finite sequence a 1 a2 . • • an of elements from V (ai eV for i = 1 ,2, . . . ,n). A null string, A, can be defined as a string with no elements in it. We will use lower case Latin letters for elements of V and lower case Greek letters for strings. Let L(V) be the set of all non-null strings over V. There is a binary operation n in L(V) such that if a and P are members of L(V) and if a= a 1 • • • an and P = b 1 • • • b m , then a n P = a 1 • • • anb 1 • • • bm is also a member of L(V). This operation is non-commutative, i.e., for a -:la P we generally have a n p -:fa p n a, and associative, i.e., (a n p) n y = a n (p n y) = a n p n y. Such an operation as well as each result of its application is called concatena­ tion. In the following, concatenation of two strings will be represented by simple juxta-

Syntactics

1053

position of these strings. Because of non-commutativity, each concatenation result has a unique sequential order. Because of associativity, the sequential order in which concate­ nation operations are performed has no effect on the concatenation results.

String codes. An algebraic system (A, 0 ) composed of a set A and a binary operation ° is called a semigroup if O is associative in A. Clearly (L(V), n) is a semigroup. Since n is represented by juxtaposition, we will use L(V) to refer to it. If (A, 0 ) is a semigroup which contains an element e such that aO e = e0 a = a for all a E A, then (A, 0 ) is called a monoid and e is called an identity of A. If a semigroup contains an identity, this identity is unique. Let L'(V) = L(V) U {A}; then, clearly, L'(V) is a monoid with A as its identity. To exemplify an abstract system which fulfils the required conditions and is not a sign system, let us consider the set N of natural numbers 1,2,3, . . . under the operation of addition. The elements np n2, and n3 in N have the property (n1 + n2 ) + n3 = n1 + (n2 + n3 ) , i.e., addition is associative and (N, +) is a semigroup. However, (N, + ) is no monoid, since it contains no identity. Yet we may add O as an identity to N and thus obtain M = {O, 1,2, . . . }. Then (M, + ) is a monoid with identity 0, since we still have (n1 + n2 ) + n3 = n1 + (n2 + n3 ) for all np n2 , n3 E M and moreover O + n = n + 0 = n for any n E M. A semigroup L(V) or monoid L' (V) constructed from a finite vocabulary V is called the free semigroup or free monoid over V, respectively. How is a structure like a free monoid L' (V) related to a natural language? Let V be thought of as the set of phonemes ( or of morphemes or of words) in a given language. Then L' (V) would contain, among other strings, all the sentences in the language under discussion. Thus that language can be defined as a subset of L'(V). This idea applies to all string codes. A string code is a set of signs that can be characterized as a subset of a free monoid. If one wants to characterize a given string code L syntactically, the task is to set up a finite vocabulary V, to define the free monoid L'(V) over it, and to distinguish the strings of L from the rest of L'(V). In language, this task corresponds to the language user's capacity to distinguish well-formed sentences from strings not belonging to his language. Analagous considerations apply to all other string codes. Production grammars. A production grammar is a set of rules on the basis of which the strings belonging to a certain subset of a free monoid can be constructed. If the grammar produces all and only the strings of a given string code L, then the grammar is called observationally adequate for L (cf. Chomsky 1964 :30ff. for other levels of adequacy). As an example for a string code L we take the standard system of signs used for the representation of natural numbers, the Hindu-Arabic numerals LH . Let VM be the set of the ten Hindu-Arabic digits 0, 1, 2, . . . , 9. Evidently, LH is infinitely large; however, it is only a subset of L' (VM) since while it includes 10, 100, 1000, . . . , 20, 200, 2000, . . . , etc. it does not include 0, 00, 000, . . . , 01, 001, 0001, . . . , 02, 002, 0002, . . . , etc. To be observationally adequate for LH , a grammar must construct the infinitely large set of Hindu-Arabic numerals on the basis of VM and exclude strings like the ones mentioned.

1 054

Syntactics

This is accomplished through a set of stipulations such as the following:

(i) 1, 2, . . . , 9 belong to Lw (ii) If a belongs to LH , then ao, a l , . . . ,a9 belong to LH . (iii) Elements of LH can be constructed in no other way than by employing (i) and (ii).

The stipulations (i) through (iii) are said to be the rules RH of a grammar @H for the string code LH. In RH, (iii) amounts to saying that LH is the smallest string code over VM which can be formed so as to satisfy rules (i) and (ii). Using (i) and (ii), one can con­ struct all and only the strings that belong to LH, i.e., that are Hindu-Arabic numerals. Rules can be given in various forms. An alternative formulation for (i) through (iii) would be to introduce S as a starting symbol and stipulate: (ii')

(i')

01

1

9

In applying these rules one would first write down the starting symbol S and then replace it either by any one digit except 0, according to (i'), or by S followed by any one digit, according to (ii'). In order to construct the complex sign representing the number of the year 1983, the n.iles (i') and (ii') would have to be applied in the following way: (l') (2') (3') (4') (5')

S SJ S83 S983 1 983

by rewriting by rewriting by rewriting by rewriting

S in (1') as S in (2') as S in (3') as S in (4') as

SJ according to (ii') S8 according to (ii') S9 according to (ii') 1 according to (i')

The ordered set of formulas (l ') through (5') is called a derivation of 1 983. Strings like 1 983 that do not allow the further application of a rule are called terminal strings. A further alternative in the formulation of rewrite rules for the Hindu-Arabic nu­ merals is obtained when we introduce M and N as special names for the classes of signs occurring in brackets in (i') and (ii'): s.... N, N-+ 1, N-+ 2, . . . , N.... 9, (i") s.... SM, M-+ 0, M-+ 1, . . . , M.... 9. (ii") Here M and N are called non-terminal symbols since in derivations made on the basis of (i") and (ii" ) they occur only in non-terminal strings. Thus a grammar consisting of rules (i") and (ii") contains not only a terminal vocabulary Vy = TH > but also a non-ter­ minal vocabulary VNT with S, M, and N as its members. In order to make precise the terms we have used in describing this example, we introduce the following definitions: A production grammar is a 4-tuple @ = (VNT> Vy, S, R) where VNT and Vy are two disjoint non-void finite sets, the non-terminal and terminal vocabularies of @ ; S is a distinguished element of VNT> the axiom or starting symbol of @ ; and R is a finite set of rules of the form et -+ �("rewrite a as W') such that a and � are elements of L'(VNT U VT) and a contains at least one element from VNT·

Syntactics

1055

Let us call the system of rules (i) through (iii) @ H, the system of rules (i') and (ii') @ ' H, and the system of rules (i") and (ii") @ " H· It is easily seen that @ H, @ ' H > and @ " H have the same output, i.e., produce the same set of terminal strings, viz. LH. In general we say that two production grammars @ 1 and @2 are equivalent if they have the same output (for other levels of equivalence cf. Chomsky 1963 ). There are three main classes of production grammars which are interesting syntacti­ cally. These are, in increasing amounts of generality, regular grammars, context-free grammars, and context-sensitive grammars. Regular grammars. For obvious reasons it is necessary to distinguish two kinds of regular grammars, left regular grammars and right regular grammars. But it can easily be shown that for any given right regular grammar there exists a left regular grammar with the same output. A production grammar (VNT> VT> S, R} is right (left) regular if its rules take one of the following forms:

A- aB, (A- Ba,

c- c c- c)

where A, B, C belong to VNT and a, c to VT · Thus the grammar @ ' H is a left regular grammar, while the following is right regu­ lar : @ * = ({ S,P,Q}, { 0, 1,2, . . . , 9,.}, S,R':") where R�- contains the rules: 2 (P),

1

(i'') S -

(ii") P -

(0") Q -

s - .Q 0 1

(P),

9

9 (iii''")

0 1

(Q),

(iv:•)

P - . Q,

(�·) Q - m

9

and (i*) is short for the two rules S - 1

2

9

and s -

1 P. 2

9

The complex signs produced by @ '' can be interpreted as representing the rational numbers with terminating decimal expressions.

1 056

Syntactics

If a grammar @ is right or left regular, then its output L(@) is called a right regular or left regular string code. Similary, if a string code is generated by a grammar of any type (*context-free, context-sensitive, etc.), it is referred to as a string code of that type (context-free string code, context-sensitive string code, etc.).

Context-free grammars. Context-free gramJl}ars are a less restrictive kind of production grammar. A production grammar (I/NT> VT> S, R} is context-free if each of its rules takes the form A-,. a, where A belongs to VNT and a to :E(VNT U VT). A context-free grammar is exemplified by @ 11 8 . However, L8 , the output of @ 11 8 , can also be produced by @ ' 8, and since @ ' 8 is regular, L8 is not only context-free but also regular. An example of a string code that is context-free but not regular is the set L0 of Eng­ lish sentences connected by the logical connectives . . . and---, . . . or---, it is false that . . ., and if. . . then---. If a and � are sentences of English, then a and �,aor �, it is false that a, and if a, then � are also sentences of English. These considerations can be embodied in a context-free grammar @0 as follows: Let the non-terminal vocabulary be composed of the symbols S, E, N, I, Tand let the terminal vocabulary be composed of and , or, it, is, false, that, if, then, and CJ (where CJ would have to be expanded into an English sen­ tence in a larger grammar containing the present fragment). Let the rules be (i0 ) S-,. SES, (ii0 ) S-,. JSTS, (iii0 ) S-,. NS, (iv0 ) S-,. CJ, (v0 ) £-,. and, (vi0 ) £-,. or, (vii0 ) N-,. it is false that, (viii0 ) J-,. if, (ix0 ) T-,. then.

@0 will produce strings like if CJ then CJ which become normal English sentences when every occurrence of CJ is replaced by an English sentence. However, @0 will not produce such strings for all English sentences containing the logical connectives specified. Strings such as CJCJCJ and CJ, which have the syntactic structure of listing expressions like (2a) Ann played the piano, Peter beat the drums, Mary plucked the bass, and Paul blew the horn.

do not belong to the output of @0• Another problem is the fact that, when interpreted in the usual way, some of the strings produced by @ will be ambiguous. For example, it will be left open whether CJ and CJ or CJ should be read as (CJ and CJ)or CJ or as CJ and (CJ or CJ). Depending on what sentences are substituted for CJ, these two readings can have rather different consequen­ ces. The two readings are mirrored in two different derivational histories for CJ and CJ or CJ : (1. 1 ) (1.2) (1.3) (1 .4) (1.5) (1.6) (1.7) (1 .8)

s

SES SESES CJE S E S CJ and S E S CJ and CJ E S CJ and CJ or S CJ and CJ or CJ

by by by by by by by

(i0 ) (i0 ) (iv0 ) (v0 ) (iv0 ) (v0 ) (iv0 )

(2. 1 ) (2.2) (2.3) (2.4) (2.5) (2.6) (2.7) (2.8)

s

SES SESES CJ E S E S CJ and S E S CJ and CJ E S CJ and CJ or S CJ and CJ or CJ

by (i0 ) by (i0 ) by (iv0 ) by (v0 ) by (iv0 ) by (v0 ) by (iv0 )

Syntactics

1 0 57

The two derivations ( 1 . 1) through (1 .8) and (2. 1 ) through (2.8) use the same rules, but apply them differently: In S E S of (1 .2), the first occurrence of S is rewritten as S E S according to rule (i0 ) . In S E S of (2.2) the second occurrence of S is rewritten as S E S according to rule (i0 ) . However, this difference cannot be reflected by the formal properties of the derivations as defined so far. Logicians solve the problem of *ambiguity by introducing parentheses into the ter­ minal vocabulary and restating the first three rules above as follows: s- ( SES), s­ (ISTS), S- (NS). The resulting grammar @ '0 will allow the following derivations: S, ( SES), ((SES)ES), etc. and S, ( SES), ( SE( SES)), etc. The parentheses can be dispensed with if we employ what is often called Polish notation (cf. Bar-Hillel 1 964). A grammar avoiding ambiguity by using the Polish nota­ tion can be obtained if we restate the rules of @0 as follows : s- KSS, s- ESS, s- ISTS, s- NS, s- a, K- and, E- or, N- it is false that, I- if, T- then. The resulting Grammar @ " 0 will allow the following derivations: S, KSS, KSESS, etc. and S, ESS, EKSSS, etc. Here the terminal strings are and a or crcr and or and aaa, respec­ tively. As is seen from the form of the rules, each of the grammars @0, @ '0, and @ " 0 is con­ text-free but not regular. Strings like the ones produced by @ '0 and @ " 0 fulfil the requirements of logic since they have only one syntactic structure each, but they do not correspond to sentences of ordinary English since they either contain additional ele­ ments or are ordered in a different way. Strings like the ones produced by @0 do not have these disadvantages, but they are ambiguous. Neither grammar accounts for listing expressions like (2a). While context-free grammars are capable of generating such expressions, they are not able to generate them in such a way that their structural related­ ness with expressions like (2b) is indicated. (2b) Ann played the piano, and Peter beat the drums, and Mary plucked the bass, and Paul blew the horn. This could be done through a grammar deriving expressions such as (2a) from expres­ sions such as (2b) by dropping unnecessary connectives like the first two tokens of and in (2b). (For discussion of the question whether a natural language like English can be regarded as a regular or context-free string code, cf. Chomsky and Miller 1 963:297-300; Bar-Hillel 1 964:94-98, 1 1 4f. ; Hiz 1 968; Brainerd 1971 :1 76-81 , 186-95; Pullum and Gazdar 1 982).

Context-sensitive grammars. A production grammar (VNT> VT> S, R) is context-sensitive if its rules take the form a1Aa2- a2coa2 where A belongs to VN> co to l:(VNT U VT), and a 1 and a2 to l:'(VNT U VT)- As an example, consider the grammar @ *0, which differs from @0 in that its non-terminal vocabulary contains an additional element Q and in ° that the rule (i ) s- SES is replaced by the three rules s- QEQ, QE- Q QE, and Q - S, where QE- Q QE is context-sensitive. @ *0 easily produces sentences of the form aaEa, crcrcrfu, etc. Context-sensitive rules are sometimes presented in the form A- col a 1 _ a2 , read "rewrite A as co in the context a 1 _a2 . ' ' On certain occasions this is preferable to the form a 1 Aa2 - a 1 coa2 because the a;'s may also contain A's, which renders the actual con­ text that the writer had in mind unclear. For example, the rule QE- Q QE can be read as a 1 = Q and a2 = A or as a 1 = A and a2 = E. If we employ the above convention, the

10 5 8

Syntactics

former intent is rendered E--+ Q E I Q_and the latter Q --+ Q Q /_ E. In case either ai is empty, the corresponding side of the bar is left empty. (For further details, including alternative terminology, more explicit definitions, and proofs, cf. Chomsky 1963; Gross and Lemin 1967; Brainerd 1971; Wall 1 972 ; Corcoran, Frank, and Maloney 1974.) In special fields such as computer programming, string production grammars are used that hold an intermediate position between context-free and context-sensitive grammars. Such grammars are usually obtained on the basis of context-free rules by imposing special restrictions on the process of derivation. The most well-known are matrix grammars, controlled grammars, ordered grammars, scattered grammars, pro­ grammed grammars, and valence grammars (for discussion cf. Salomaa 1973). While all these grammars can be regarded as special types of context-sensitive gram­ mars, there are also string production grammars that do not fit in the hierarchy of gram­ mars discussed above. The most important are configurational grammars (cf. Gladkij 1963, 1968), Lindenmayer systems (cf. Lindenmayer and Rozenberg 1976, 1979; Herman and Rozenberg 1975 ; Rozenberg and Salomaa 1974), and contextual gram­ mars (cf. Paun 1 982). Lindenmayer systems and contextual grammars are both charac­ terized by the fact that they do not have a non-terminal vocabulary. Besides, Lindenmayer systems require that every element of a given string be processed in each step of a derivation. Grammars with such properties have been successfully used in the description of cell structures and their development (cf. Claus, Ehrig, and Rozenberg 1979). Today, progress of syntactic research has transcended the structure of string pro­ duction grammars even in linguistics. Among the new features introduced are the fol­ lowing: 1.

2.

3.

Ordering of rules: Production grammars impose no restrictions on the order in which the rules are applied outside of their intrinsic order, i.e., the order imposed by the availa­ bility of non-terminal elements in the string µpon which the rules are to act. Transformational rules : These rules are far more complex than production rules because their application depends upon the history of the derivation of the strings they act upon rather than just the constituents in these strings, as in the case of production rules (see the discussion above of ambiguity in strings produced by ® 0) . Rule schemata : These allow for the application of an infinite set of rules. One of the fundamental restrictions of production grammars is that the rule set is finite. With the introduction of rule schemata this finiteness is lost.

Syntactic structures and structural descriptions. Production grammars are not only a tool for the study of the way in which signs of various classes are combined to form complex signs, they can also be used to describe the syntactic structures of complex signs. This is possible since the structure of a complex sign produced by a production grammar can be related to the derivational history of this sign (see above). This idea has already been hinted at in the discussion of context-free grammars (see above). There brackets were introduced to disambiguate strings that can be structured in more than one way, i.e., that are syntactically ambiguous. By labeling such brackets with the appropriate elements of the non-terminal vocabulary each string in the output of the grammar can be given a description of its syntactic structure which is complete with respect to this grammar. (For discussion of this point cf. Reichenbach 1947:164-167, Hermes 1 965, Wang 1969, Schnelle 1970, and Posner 1971).

Syntactics

1059

Let us consider once again our example of a syntactically ambiguous string: (3) cr and a or a

Instead of introducing brackets by including them in the rewrite rules as in @'0, we can construct labeled bracketings in the course of derivations by using the following brack­ eting convention: If a rule A ---Hl is applied to a string � = X 1AX2 , write the result as :x 1 [a]AX2 > where [a]A is a bracketing of a labeled by A. (For other versions of the bracketing convention cf. Montague 1970b, Brainerd 1971:212ff., and Hermanns 1977:160ff.) In order to obtain the two structural descriptions indicated for (3), we proceed as fol­ lows: (1 .1' ) (1.2') (1 . 3') ( 1 .4' )

(1 .5' ) (1 .6') (1 .7') ( 1 . 8') (2. 1') (2.2 ') (2.3')

s

[S

E [ [S [[ (o] s E [[ (o] 5 [ and ] E [[ [and] ( crJs E [[ (oJs [and]E [ [ (oJs [and]E

[ s [ s

s

E E E

E E E E E

s s s

Js Js Js o ( Js Js [a]s Js [ or]E [o]s Js [ or]E

s

s s s

E E E [ [ [o] s E [ [

s s s s s s

Js Js Js Js Js Js [oJs sJs

s s s s s

(2.4') [ (o] s (2. 5') [ [o] 5 [ and] (2. 6') [ [cr] 5 [an d]E (2.7') ( [oJs [and]E [ [o] 5 [ or]E (2. 8') [ [o] 5 [an d]E [ [o] s [ or]E [o]s

J sJs JsJs J sJs J sJs J sJs J sJs

by (i°} by (i°} by (iv°) by (v°) by (iv°) by (v°) by (iv°) by (i°} by (i°} by (iv° ) by (v°) by (iv°) by (v°} by (iv°)

Each line in the above derivations contains a string within labeled bracketing indicating the syntactic structure of the string. The difference in structure between (1.8') and (2.8') is due to the structural difference between (1.3') and (2.3'), which is caused by the appli­ cation of rule (i0 ) to different element tokens in the string [ SES] 5 in (1.2') and (2.2'), respectively. While the unstructured strings (1.8 ) and (2.8) are identical, the structured strings (1.8') and (2.8') are different: (1.8') contains cr and a as a constituent, (2.8') does not. (2.8') contains cr or cr as a constituent, (1.8') does not. We are now in a position to give more exact definitions for some of the terms used in describing the ambiguous string (3) and its structure. If for a given context-free grammar @ we apply the bracketing convention, then any of the bracketings assigned to a terminal string a E L(@) is a description of the syntactic structure (or in short, a structural description) of a in @. Each partial string of a included within a labeled pair

1 060

Syntactics

of brackets is a constituent of a. The non-terminal symbol occurring as a label of the brackets indicates the syntactic class or category which this partial string belongs to. These definitions explicate the central terms used in the characterization of syntactics3 given by Morris ( 1 938 :14 [ = 1 971a :28f.] and 1 946 :354f. [ = 1 971a:367]). Even for a simple string like (3), the bracketings assigned to it by the grammar are quite complex. Thus there is a certain amount of possibility for error due to inaccurate transcription. This is why some syntacticians prefer to think of structural descriptions in terms of parsing trees (or P-markers) rather than in terms of labeled bracketings. The two methods are interchangeable. Therefore, we also define a parsing tree. Let ® = 0fNT> VT> S, R} be a context-free grammar. Then each derivation S, a t > a2 , • • • , an has associated with it a unique parsing tree which is constructed as follows: S is the root. If a 1 = a 1 a2 • • • an where aiEVNT U VT for i = 1 ,2, . . . ,n, then the parsing tree has nodes 1 1 , 1 2, . . . , 1 n labeled respectively a p a2, • • • , a0, which are connected to the root by lines as in Figure 1 . If a2 is obtained from a 1 by application of a rule ai - b1 b2 • • • bk where aieVNT and bieVNT U VT for 1 � j � k, then the nodes l i l , l i2, . . . , l ik are connected to the node labeled ai by lines and are labeled bt > b2, • • • , bk respectively as in Figure 2. If this process is continued until the terminal string is reached, we ultimately obtain a unique labeled tree corresponding to the derivation in question. This tree is the parsing tree which ® assigns to the terminal string of that derivation.

Figure 1.

Figure 2.

Figures 3 and 4 give the structural descriptions of (2) in the form of parsing trees corresponding to derivations ( 1 . 1 ') through ( 1 .8') and (2.1 ') through (2.8'), respectively:

Syntax

1 06 1

s

0

and

0

or

0

Figure 3.

A r \" . I

S -�� E

. 0

.

and

.

0

.

or

.

0

Figure 4. (For further details about froms of structural descriptions for strings cf. Brainerd 1 97 1 :220-33 and Wall 1 972 :2 14-53.) R.Pos.

Syntax

Syntax is generally agreed to be concerned with the ways in which words and mor­ phemes can be combined into larger units, especially into �·sentences, though the various linguistic schools differ considerably in their positions regarding the nature of syntactic *structure, the relationship of syntax to other domains of *language structure, and the range of phenomena that are regarded as ''syntactic.

Syntactic Structure

The conception of syntactic structure that figured in the works of the Greek and Roman grammarians was that of dependency among words. The core of the analysis of any sentence was a specification of the ways in which some of its words were dependent on others, e.g., in Filius Marci recitat carmina ma/a 'The son of M:1rcus recites bad poems',

1062

Syntax

Marci was described as modifying filius, filius as the subject of recital (not filius Marci as the subject of recital carmina mala- the relations were among individual words, not among larger units), mala as modifying carmina, and carmina (not carmina mala) as the object of recital. Restrictions on what a word could be combined with were formulated in terms of the relations in which it could stand to other words, e.g., obedio required a direct object in the dative case. The psychologist Wilhelm Wundt (1880:53-58) proposed an alternative conception of syntactic structure that, during the first half of the twentieth century, came to supplant dependency as the most widely accepted basis for syntactic structure. For Wundt, the basic structure of a sentence consisted not in the dependency relations that held among its words but in the way in which it could be divided into parts. Thus, for Wundt the structure of the above Latin example would be that it consisted of the parts filius Marci and recital carmina mala, which consisted in turn of filius and of Marci and of recital and carmina mala, with the latter item (or constituent, to use current terminology) consisting of carmina and mala. Wundt's conception of syntactic structure was adopted with ak­ knowledgment by Bloomfield (1914, 1916) and remained central to his thinking on syn­ tax even after he rejected Wundt's psychological theory in favor of behaviorism. It is lar­ gely through Bloomfield's influence that since the 1930s constituent structure has been accepted almost unanimously by American linguists as the key structural notion in syn­ tax. According to the constituent structure conception of syntactic structure, a sentence is an object of the sort known in �·graph theory as a tree: a set of items (the nodes of the tree, here the constituents of the sentence) with a relationship of "is a part of" (repre­ sentable by the connections in a diagram such as the following):

NP

--------

��

N

I

filius

Det

�-i I

NP N

-----� � I

--------VP V

recitat

NP

I canruna N

I

A

mala

MarcThe diagram here represents in fact an ordered labeled tree, in which sister constituents stand in a determinate ordering relationship (indicated by the left-to-right arrangement in the diagram), and each constituent is assigned to a syntactic category (indicated by the labels on the non-terminal nodes). Conceptions of syntactic structure in which both dependency and constituent struc­ ture are combined are found in the works of Jespersen (1937), of certain American de­ scriptivist linguists such as Harris (1946), and in relational ':grammar (Perlmutter 1979; Johnson and Postal 1980), the offshoot of transformational grammar in which syntactic structures are regarded as based on grammatical relations such as "is subject of." In these conceptions of syntactic structure, grammatical relations and dependency rela­ tions are taken as holding between syntactic constituents, rather than just between indi­ vidual words, as in the pure dependency structure of traditional grammarians and such

1 063

Syntax

modern linguists as Tesniere ( 1 959) and Mel'cuk (1 979). In Jespersen's approach to syn­ tactic structure, sentences are analyzed in terms of a constituent structure in which gram­ matical relations, dependency relations, coreference relations, and understood elements are indicated and elements may stand in multiple relations, as in the sample analysis given here, in which the formula indicates that a good man is both subject of is and object of find (asterisks connect non-adjacent "related" items ; the numerals relate to Jes­ persen's system of "ranks": a "secondary" modifies a "primary"). SAMPLE ANALYSIS (of A good man is hard to find)

Dependency snucture (the labeled arrows indicate different kinds of dependency)

Subj.

n. I

man

A good

A good man is hard to find S( 2 1) (O''") V P(2 p V*)

Jespersen 1937

Pred.

is

s

/

NP

S

VP

�

I

/::,.

Ai A Pres V AP be 1

�NP

V

�

Det

I a

A 1

I

hard

hard

NP

A

N

�

to find

V

VP

AP

S

�

I I I I good man is A

I

VP

I AV

hard to

Dct �S 1

_,

i� =-i_ I --

Surface structure s

Deep snucture

Transformational grammar (Chomsky 1965)

Comp!.

man NP �VP � I A Det N Pres V AP �e

mL

ll

g1

The most obvious distinguishing trait of transformational grammar is that a sentence is analyzed in terms of a multiplicity of syntactic structures rather than a single structure. In the version of transformational grammar presented in Chomsky 1 965, a grammar of a language contains a set of base ,:•rules that specify what deep structures are possible in that language, and a set of transformations that convert the deep structures by successive steps into the corresponding surface structures (In the example given, the steps include operations of relative clause reduction - in which the embedded S that could be realized in full as who is good is reduced to its predicate element good -, of adjective preposing - in which that adjective is moved to a position in front of the noun that it modifies, of "tough movement" - in which the object NP of find is made into a derived subject of the main clause and the residue of the embedded clause is placed after the predicate element of the main clause -, and of agreement and "affix hopping" - which yield a third-person singular present tense form to be. The deep structure, the surface struc­ ture, and all the intermediate steps in the syntactic derivation are labeled trees, indeed

1 064

Syntax

ordered labeled trees, although subsequently to Chomsky 1 965 some dispute has arisen as to whether any underlying order of constituents need be assumed.

Goals of Syntax

Many (though far from all) transformational grammarians endorse the proposition that a grammar of a language must be a *generative grammar, that is, that it must specify which strings of words are sentences of that language and which ones are not. (Not­ withstanding a popular tendency to freely interchange the terms transformational and generative, a grammar can be transformational without being generative, or vice versa.) The notion of "generative grammar" presupposes that language can be identified (or at least, associated) with a set of sentences. Exactly what set of sentences, though, has always been a matter of controversy. Transformational grammarians generally do not hold that whether something is a sentence of a given language can be decided in any straightforward fashion (such as by having an informant make a simple judgment as to the "normalness" of the putative sentence). Since the appearance of Chomsky 1 965, they have generally distinguished between "grammaticality" and "acceptability" : an informant's negative judgment about a sentence may show it to be "unacceptable" but does not show it to be ' 'ungrammatical," and to determine whether an unacceptable sentence is ungrammatical one must in practice determine why it is unacceptable and decide whether the source of its unacceptability is within or outside the realm of gram­ mar. Accordingly, while there has been fairly general agreement as to what is responsible for the unacceptability of various sentences (e.g., it is generally agreed that the "unacceptability" of "John frightened sincerity" is the result of the violation of the selectional restriction that the object of frighten denote a being who is conscious at the time in question), there has been much vacillation as to whether the factors involved are matters of grammar, and thus as to whether the sentences in question are treated as "grammatical." The significance of the question of which factors lie "within grammar" is denied in works such as Lakoff 1 97 1 and McCawley 1 976, 1 980, in which the notion of a language as a set of sentences is rejected, as a concomitant of the rejection of the notion of the grammatically or even acceptability of sentences in isolation from *con­ text. A "grammar" is conceived of in those works as a system of rules sensitive to linguistic and extralinguistic context that specify the correspondence between *meaning and surface *form. What relationship different linguists have seen between syntactic structure and meaning has correlated closely with their views of meaning: whether they take sentence meanings to be objects of the same formal nature as syntactic structures (e.g., labeled trees, as in McCawley 1 972), whether they regard them as well-defined objects of some other structural type (as in Katz and Fodor 1 963 and Katz 1 972), whether they regard only certain aspects of meaning as open to formal analysis (as, apparently, in most of Chomsky's work), or whether they hold meanings to be basically amorphous (as in '�Wittgenstein 1 953 and Sampson 1 979). In Chomsky 1 965, the meanings of sentences were held to be derivable from though not identical with their deep structures. The many linguists who subscribed to that position in the mid 1 960s have almost all rejected it in favor of either the position that a distinction between syntax and linguistic *semantics cannot be drawn (McCawley 1 97 3 ; Lakoff 1 97 1 ), or the position that there are details of meaning that correlate in a systematic way with surface structure but not with any

Syntax

1065

"deeper" level of syntactic structure (Chomsky 1972; Jackendoff 1 972), or the position of *Montague grammar, to be described below. Chomsky (1980) adopts an "enriched" conception of surface structure (e.g., surface structures contain various phonologically null but semantically significant elements, such as "traces" of constituents that have been moved) that allows all semantic interpretation to be based on surface structure. A tight integration between syntax and semantics is assumed in Montague grammar (Montague 1974; Partee 1976; Dowty 1979), in which to every syntactic rule (i.e., rule for constructing expressions of a given syntactic category out of constituents that enter into it) there is to correspond a semantic rule which constructs a semantic representation for the derived item (i.e., a formula of intensional logic that expresses its contribution to the *truth conditions of sentences in which it occurs) out of the semantic representa­ tions of the items from which it is constructed.

Some Important Syntactic Notions Combinatoric Rules A full description of a language must include information both about gross combinatorics (i.e., the overall shapes that expressions may have, e.g., that in English an "adjective phrase" consists of an adjective, optionally followed by a prepositional phrase) and about petty combinatorics (i.e., the possible choices of lexical items in particular syntactic constructions, e.g., throw but not sleep or destroy can be the V in the combination V the boxes through the window). In the Aspects version of transformational grammar, both gross and petty combinatorics were treated as primarily matters of deep structure: the base rules, which specified what deep structures were possible, included both rules of gross combinatorics (the phrase structure rules) and a *lexicon, whose entries contained specifications of petty combinatorics (both strict subcategorization features, specifying what gross deep structure combinations the given item could be used in, e.g., that throw can be used in the configuration [-NP PP], and selectional features, giving restrictions on the items in the frame in which the given item was used, e.g., that the NP with which throw is combined must have the property "physical object," and the PP the property "direction" or "destination"). Whatever combinatoric regularities were observable in surface structure were held merely to reflect regularities of deep structure combinatorics that happened not to be destroyed by the transformations that the language happened to have. For example, that articles always precede their nouns in surface structure was held to reflect merely that they precede them in deep structure and that English happens to have no transformations that alter the order of article and noun. Most transformational grammarians have abandoned this view in favor of one of the following : (1) that transformations are restricted in such a way that, with limited classes of exceptions, surface structures exhibit the same combinatoric patterns as do deep structures (this is the "structure preserving" conception of transformation of Emonds 1976), (2) that languages have rules of gross combinatorics both for deep structure and for surface structure (as in Ross 1967 and Perlmutter 197 1, in which output constraints, i.e., rules of surface combinatorics, are proposed), or (3) that gross syntactic combina­ torics is basically a matter of surface structure (McCawley 1980). In addition, the Aspects position on selectional restrictions, which were held to be restrictions on the lexical item that formed the "head" of the given constituent (e.g., the

1066

Syntax

noun that formed the head of the NP with which throw was combined was to have the feature [ + concrete]), has largely been rejected in favor of the position that they are restrictions on the denotation of the constituent (e.g., that the NP with which throw is combined, regardless of its lexical composition, must denote a physical object). For arguments in favor of this position, see McCawley 1968 (1 32-35) and Jackendoff 1972 (18-21). This view is close to the position that selectional violations constitute "category mistakes" in the sense of Ryle 1949, except that it allows for the possibility that some lexical items impose selectional restrictions that are narrower than the require­ ment that the predication be meaningful, such as the restriction that ill (unlike sick) is normally predicated only of human beings. Syntactic Categories Words, even words that appear to have the same meaning, can differ from one another in the ways in which they combine with other words: This container is sufficiently/enough large. This container is large enough/sufficiently. My brother fears/ afraids snakes. My brother is afraid/fear of snakes. Fred is so/such stupid that he's sure to fail. Fred is such/ so an idiot that he's sure to fail.

Linguists have generally dealt with this fact by setting up syntactic categories and formu­ lating in terms of them their rules for the ways in which linguistic elements can combine, e.g., fear but not afraid belongs to a class of items that can be inflected for tense and can appear at the beginning of a "verb phrase." Since traditional grammar did not recognize syntactic units other than words and clauses, the syntactic categories of traditional grammar were word classes. The numerous attempts by traditional grammarians to define their word classes (or parts of speech) rigorously have suffered from an unwillingness by their authors to be systematic in applying their criteria for category distinctions, and from a prejudice in favor of having the number of word classes be small (eight seems to be the favorite number). Familiar definitions of word classes are sometimes in terms of what is denoted by words of the given class (e.g., that a noun denotes persons, places, and/or things), sometimes in terms of the grammatical relations that they stand in (e.g., that an adjective modifies a noun), and sometimes in terms of anaphoric and other relations (e.g., that a pronoun "stands for" a noun). It is only in such more modern approaches as categorial grammar (discussed below) that a serious attempt has been made to apply some of these criteria for category distinctions systematically, e.g., to distinguish modifiers systematically with regard to what sort of things they modify. In approaches to syntax based on constituent structure, it is necessary to set up syn­ tactic categories in such a way that constituents larger than a word are also assigned to categories. Harris (1 946) provided a criterion for category membership that proved influential in subsequent years - namely, that two items (whether single words or larger expressions) belong to the same category if and only if substituting either for the other does not affect the grammaticality of the item in which the substitution is made. This criterion has the problems that (1) its implications depend on how broad or narrow a conception of grammaticality one takes (thus, whether dog and log belong to the same

Syntax

1067

category will depend on whether "John frightened the log" counts as grammatical), and (2) it imposes no structure on the system of categories: items that are not always mutually substitutable simply belong to two separate categories rather than to distinct subclasses of a more inclusive category. In early transformational grammar, Harris' conception of category was widely en­ dorsed but was applied only haphazardly in setting up category distinctions. The first major break with Harris' notion of category came with Chomsky 1965, where strict subcategorization and selection were separated out from syntactic category per se as independent dimensions of syntactic behavior. Whereas in earlier transformational grammar, kick, sleep, and elapse had been treated as simply belonging to three distinct syntactic categories, in Chomsky 1965 they were treated as all belonging to the same category "V" but having different selectional and strict subcategorization features in their ,;dictionary entries. In addition, differences in the distribution of words that could be attributed to differences in whether the word allows or requires application of a par­ ticular transformation were not regarded as category differences, e.g., want and desire were assigned to the same category despite the difference that want but not desire requires the for of an immediately following object complement to be deleted ("I want (for) them to leave," but "I want very much for them to leave." "I desire for them to leave"). It was soon realized by Lakoff (1970) that many traditional word-class distinc­ tions could be reinterpreted as distinctions in applicability of transformations. For exam­ ple, adjectives and verbs could be taken as filling the same deep structure positions but differing in that adjectives but not verbs required insertion of the copula be. Lakoff, like virtually all transformational grammarians, assumed that syntactic combinatoric rules related primarily or exclusively to deep structure and that syntactic categories could not change in the course of a derivation. These assumptions led different transformational grammarians to profoundly different conclusions : while they provided Lakoff with grounds for rejecting category distinctions such as that between verb and adjective, in that they can be dispensed with in deep structure, the same assumptions led Chomsky (1970) to reject many arguments for "abstract" deep structures, in part because the ana­ lyses gave up category distinctions that he maintained were needed in surface structure. In the conception of syntactic category currently accepted by Chomsky, the so­ called "X-bar syntax" (Chomsky 1970; Jackendoff 1977; Koster 1978), most syntactic categories consist of two components: the word-class of the head of the constituent, and a count of the depth to which the head is embedded in the constituent (frequently expressed by multiple overlines, whence the name "X-bar syntax." For example, afraid of snakes is an A and very afraid of snakes an X; for typographical convenience, numerical superscripts are often used in place of overlines). In addition, a number of categories not based on word-classes have figured in X-bar syntax, notably COMP ( = complementizer) and S (though see Koster 1978 for arguments that the head of a clause is its verb and thus that S = vn for some n), and items (chiefly "grammatical mor­ phemes" of various kinds) that belong to no syntactic category at all are admitted. The conception of syntactic category adopted in Montague grammar is a variant of 'fAjdukiewicz's (1935) categorial grammar, which is an elaboration of ideas developed by �·Frege, ''·Husserl, and Lesniewski. For Ajdukiewicz there were two "basic" catego­ ries s 'sentence' and n 'name' and an infinite number of derived categories XIY 'com­ bines with Y to yield X'. For example, VP was interpreted as sin ('combines with name to yield sentence') and sentential adverb as sis; this approach is further developed in

1 068

Syntax

*Bar-Hillel 1953. Roughly speaking, the syntax of Montague grammar is to pure cate­ gorial grammar as transformational grammar is to pure phrase structure grammar, in that it supplements Ajdukiewicz's basic principles of category formation in several ways : ( 1 ) syntactic rules need not merely concatenate the operands but can perform other operations such as inserting one operand into a given position in the other, (2) in addition to X/Y combining with Y to yield X, it may also combine with other things, as in Montague's rule that combines a NP (a category whose name means "combines with VP to yield S") with an S by substituting the NP for a pronoun in the S, and (3) it is pos­ sible for there to be syntactic distinctions between syntactic categories that both "com­ bine with Y to yield X." In the latter case, Montague-grammarians have adopted the makeshift of using multiple slashes to distinguish the categories : X/Y versus XI IY. Ajdukiewicz's and Montague's conceptions of syntactic category distinguish systemati­ cally among different kinds of modifiers (the many things traditionally classed as adverbs are distinguished on the basis of the category of what they modify) but provides no basis for setting up the traditional word classes "noun," "verb," and "adjective, " since combinatorically distinct words (e.g., sleep, hit, and put) must be assigned to different categories. Coordination and Subordination One particularly important way in which syntactically complex sentences can be formed is through the embedding of clauses, that is, by constructing from a clause an expression that may play a role in a superordinate clause such as that of subject ("That this book will be reprinted is unlikely"), adverb ("Nero fiddled while Rome burned"), or adjec­ tive ("The books that you ordered have arrived"). Expressions involving non-finite forms of verbs can usually be argued to be reduced or altered forms of embedded clauses. For example, in "I want to leave," to leave is the surface manifestation of an underlying embedded clause / leave, and in "Bill seems to me to admire Alice," the underlying subject of seem is "Bill admires Alice" (as argued in Jespersen 1 9 1 3, 1937). Another way in which sentences or parts of sentences can be combined is through conjoining, as in "Warsaw is in Poland, Budapest is in Hungary, and Sofia is in Bulga­ ria," or "The cat is either on the mat or under the bed." Some noteworthy characteris­ tics of conjoined structures are : ( 1 ) the number of conjuncts is unlimited except where the particular conjunction (e.g., but) requires that there be only two conjuncts ; (2) con­ joined clauses differing in only one item can often be replaced by a single clause with conjoined parts ("Either Tom got away or he was arrested" - "Tom either got away or was arrested" ) ; and (3) any syntactic process affecting a conjoined structure must affect all the conjuncts, as in "The theory which Schwartz lectured on and Blau argued was wrong is absurd," in which the relative pronoun stands for items in both conjuncts (the object of on and the subject of was wrong), and "I've never eaten pizza and drunk cham­ pagne at the same meal," in which the have causes both conjuncts of its complement to be put in the past participle form (this is the coordinate structure constraint of Ross 1 967a). There are a variety of cases in which elements of a subordinate clause may be moved into a superordinate clause. In many languages relative and interrogative expressions appear at the beginning of the relative or interrogative clause even if they play a syntactic role in a lower clause ("What did he say he wanted to buy?"; "This is the book

Syntax

10 6 9

whose author Ted insisted he would never speak to"). Movement of relative and inter­ rogative expressions in this case is unbounded: there is no limit in principle as to the degree to which their surface position is superordinate to the clause to which they are grammatically related. Many grammatical phenomena that are describable in terms of movement transformations are bounded, that is, elements are moved either within a single clause or into an immediately superordinate clause, as in the phenomenon of rais­ ing, in which the subject of an embedded clause assumes the role of the subject or object of the clause in which it is embedded ("Fred believes Susan to be a genius"). The involve­ ment of a movement transformation in a syntactic construction can often be diagnosed through the use of embedded clauses containing idioms, whose pieces must be units in underlying structure but can be separated from one another by movement transforma­ tions (e.g., "Whose leg did he say he wanted to pull?"; "The cat seems to have gotten out of the bag"). Relations Among Linguistic Units

Grammatical relations. Relations such as "subject" and "direct object" have figured in virtually all approaches to syntax, though in quite different ways. While they have fig­ ured centrally in the syntactic analyses of traditional grammar and of such linguists as Jespersen, in the more orthodox varieties of transformational grammar grammatical relations have played only a peripheral role. In Aspects, Chomsky took grammatical rela­ tions to be definable in terms of syntactic configurations, e.g., the subject of a clause is the NP that is directly dominated by the S-node in question. Aside from vague sugges­ tions that grammatical relations play some role in semantics (possibly serving as the basis for cross-linguistic identification of the different adjuncts of corresponding predi­ cates), no grammatical role was provided for grammatical relations to play. Still, trans­ formational grammarians frequently spoke informally in terms of grammatical rela­ tions, e.g., they would speak of the passive transformation as turning a certain NP into a derived subject. This informal practice has been developed into a formal descriptive device in rela­ tional grammar, in which "direct object becomes subject" is a formal statement of a pas­ sive transformation rather than an informal paraphrase of a transformation whose formal statement is in terms of the notions of linear order and syntactic category. Rela­ tional grammar is particularly concerned with formulating universal principles by which the side effects of changes in grammatical relation can be predicted. For example, Postal and Perlmutter's "relational annihilation law" states that when an item assumes a gram­ matical relation in a given clause, the previous holder of that relation loses it, as when the promotion of a direct object to subject by passivization causes the previous subject to lose its status as subject and become what relational grammarians call a chomeur. The independence of grammatical relations from morphological cases is argued by Shibatani (1 977), who shows that some grammatical phenomena in Japanese (e.g., the possibility of serving as antecedent of a reflexive pronoun) operate in terms of subject­ hood, while others (such as the conversion of quantifiers into adverbs, as in the Japanese analog to The boys are all here) operate in terms of nominative case, as demonstrated by examples involving dative subjects (for which only the former sort of phenomena are allowed) and examples involving nominative non-subjects (where only the latter sort occur).

1070

Syntax

The notion of subject is so frequently confused with those of topic and agent that each of the three terms can be found defined in terms of any of the three definientia. As used here, subject is a syntactic notion, agent a semantic notion, and topic a notion of *discourse structure. The subject of a clause is the grammatically most prominent NP, in the sense of most readily serving as antecedent of pronouns elsewhere in the clause, of most readily undergoing the effects of grammatical processes in superordinate clauses, of controlling agreement, if any, etc.; see Keenan 1976a for a detailed list of syntactic and semantic properties typical of subjects. An agent NP denotes a person or entity that is spoken of as performing an action that is under his, her, or its control. A topic is a constituent that the rest of the clause is interpreted as presenting information about. The three notions are related only in that an agent, if any, is usually an underlying subject, and there is some tendency for a topic to be a surface subject. However, it is easy to find sentences in which the three notions are not coextensive. For example, in No one understands him, no one is the subject, him is the topic, and there is no agent. Semantic ''case relations. In the works of Par:iini (see Kiparsky and Staal 1969), Gruber (1976), and Fillmore (1968, 1977), a level of underlying syntactic structure is posited in which NP's are specified for semantic relations such as "agent," "patient," "expe­ riencer," and "instrument." One of the chief appeals of these approaches is that they allow related sentences in which corresponding NP's stand in different grammatical rela­ tions to be treated in terms of the same "case relation," e.g., the ball in both "The ball rolled across the lawn," and "He rolled the ball across the lawn" stand in the patient relation to roll. The various "case" approaches all involve rules specifying what syntactic or morphological means are used to express the different case relations, with Par:iini's rules being particularly ingenious: case relations can be expressed either in the morphological case of the noun or in the voice of the verb (e.g., "agent" can be expressed by means of either instrumental case or active voice), and, if the relation is expressed in the verb, the noun is put in the unmarked (nominative) case. In the underlying structures developed by Gruber (1976) and Talmy (1975), semantic case relations such as are discussed here are combined with the relation that Gruber calls "theme" and Talmy (following �·gestalt psychologists) calls "figure" the relationship of an item to a frame of reference with respect to which it is located. For Gruber, for example, the difference between "The circle contains the dot" and "The circle surrounds the dot" is that in the former the dot is theme, and in the latter the circle. Gruber's approach has been influential in the conception of semantic structure deve­ loped in Jackendoff 1972. A naphoric relations. Languages abound in devices by which one can make reference to items mentioned elsewhere in the sentence or discourse: not only personal pronouns but also processes of elision (e.g., the omission of the repeated VP in "Tom bought a car and Susan did too"), adverbs expressing the relation of the given clause to the preceding discourse (such as therefore and however), and a miscellaneous set of words that can be taken as incorporating a personal pronoun (e.g., else can be paraphrased as 'other than X', for an appropriately chosen pronoun X). Considerable work has been done on the question of where such anaphoric devices (or anaphors) can appear in relation to their antecedents, the most celebrated result (Langacker 1969; Ross 1967; Lakoff 1976) being that, subject to some qualification, the antecedent must either precede or be in a super-

Syntax

1071

ordinate clause to the anaphor (e.g., she may refer to Alice in "After she left the office, Alice went to the bank," but not in "She went to the bank after Alice left the office"). The once widely accepted transformational analysis in which all anaphors were derived from copies of their antecedents has been shown untenable in virtue of Bach-Peters sen­ tences (Bach 1970; Karttunen 1971) such as ' 'The boy that deserved it got the prize that he wanted," in which each of two pronouns is contained in the antecedent NP of the other: if the pronouns were to be derived from copies of their antecedents, an infinite underlying structure would be needed. Lakoff (1976) has argued that anaphors differ from one another with regard to whether they need to be derived from copies of their antecedents. Specifically, some types of anaphor can have as antecedents transforma­ tionally derived constituents (as in "Alice was arrested by the police but Dick wasn't," in which the antecedent of the zero VP is arrested by the police) ; those types that can have transformationally derived antecedents cannot serve as the basis of Bach-Peters sentences. Since both kinds of anaphor are subject to the restriction that the antecedent must precede or be in a superordinate clause, the latter restriction cannot be embodied in the statement of a particular transformation such as VP-deletion but must be a general condition on the relationship between anaphors and their antecedents regard­ less of how they are derived. The distinction that Lakoff draws between two types of anaphor has subsequently been retined by Hankamer and Sag (1976), who demonstrate that precisely those anaphors that are not transformationally derived allow �-pragmatic control, that is, can be used without a linguistic antecedent to refer to something present in the situation. For example, in sight of a person standing on a high window ledge one can say "Do you think he'll do it?" without any previous reference to his jumping from the ledge, but not "Do you think he will?" J.D.M.

T Tactile Communication

A way of *communication based on all constructions of a vicarious experience between two (human) individuals which make use of the mechanical sensitivity of the skin on the surface of the body, and which arise out of an original pattern built purpo­ sively by a transmitter. Most of the communication processes that we generally consider depend upon the visual or auditory ''channels to convey a �·message between two individuals : "reading" for instance, is based on the recognition of stereotypical visual signs, "spoken language" on recognizable patterns of an acoustic variation of pressure on the surface of the ear conveyed by the air. In principle, any kind of recognizable variation of the individual's surroundings, so far as they belong to a definite repertory of elements, could be used - if appropriately encoded - for transfering an experience to that individual. The sense of touch, whose characteristics vary considerably from one part of the body to another, can thus be used as a means of transmission, either to reproduce directly the sensations that the body of a transmittor receives at different points and wants to make known by another individual (isomorphic message) or, possibly, as an indicator of conventional *signals (semiological message). Touching someone's shoulder to make him turn around to look at one is an encoded signal commonly used by ,:•culture to attract his attention (possibly with a view to another, more elaborate, kind of communication). In practical terms, the skin's tactile receptors can only react to direct contact, which is to say that the spatial range of communication is limited to gestures in which the limbs come into contact with another body. As, in most cases, both individuals have generally other means of communication at their disposal, the practical use of tactile communica­ tion is rather limited. This is not, however, always true, e.g., in noisy surroundings, dar­ kness, etc., in which case the individual has to rely upon surrogate means of communica­ tion. People who, in various situations, are constrained, e.g., prisoners, lovers, etc., thus construct some kind of �·code which opens up a new field for the semiotics of contact. Any precise approach to tactile communication must first be aware of the vast variety of physical stimuli that the skin, as a sensory organ, is capable of distinguishing. Under the outer layer, the skin has a number of specific receivers (Krause, Meissner, Pacini sen­ sory cells), each of which reacts to a different type of sensory stimulus and the distribu­ tion of which over the vast surface of the body is quite different from one place to another. Physiologists have, for many years, been drawing anamorphic diagrams of the human *body according to the approximate distribution of the latter. Table 1 summar­ izes some of the communicational capacities of the various sensory channels, one of which is the tactile system proper which includes sensitivity to pressure, thermal sensitivity, sensitivity to pain, sensual sensitivity (erogenous), vibrational sensitivity, muscular reactivity. No accurate exploration of all of these fields of contact between the individual and the surrounding medium had been carried out until recent years, whether the medium be a human being, the texture of a chair, the feeling of velvet or cloth, the vibration of an object, etc. Today, however, a systematic study is being made of surro­ gate communication channels, either for functional purposes - encoding in a new way

1 073

Tactile Communication

auditory or visual signals which the individual could not directly grasp (blind or deaf people) - or for �·aesthetic purposes or erotic sensations, which has called for a much more thorough knowledge of this field. This constitutes an opening for coded or artifi­ cial messages sent by one individual to another, which implies a new science of signs transmitted in this manner. Table 1. Communication capacities of the various sensory channels

0 � 3! ..c:

0

......

.,