Linguistics and Adjacent Arts and Sciences: Part 1 9783111659916, 9783111275512

Citation preview

CURRENT TRENDS IN VOLUME •

LINGUISTICS 12

CURRENT TRENDS IN LINGUISTICS Edited by T H O M A S

A.

S E B E O K

Research Center for the Language Sciences Indiana University

V O L U M E

12

Linguistics and Adjacent Arts and Sciences •

Associate Editors: ARTHUR S . ABRAMSON, DELL HYMES, HERBERT RUBENSTEIN EDWARD STANKIEWICZ

Assistant Editor: BERNARD SPOLSKY

Assistants to the Editor: ALEXANDRA D I LUGLIO LUCIA HADD ZOERCHER

1974

MOUTON THE H A G U E • PARIS

© Copyright 1974 in The Netherlands Mouton & Co. N.V., Publishers, The Hague The research reported herein was performed pursuant to a contract with the U.S. Department of Health, Education, and Welfare, Office of Education, under the authority of Section 602, Title VI, NDEA.

LIBRARY OF CONGRESS CATALOG CARD NUMBER: 64-3663

Printed in The Netherlands by Mouton & Co., Printers, The Hague

EDITOR'S INTRODUCTION

To my mind, the greatest weaknesses of this series so far have appeared at its ragged periphery. This book was conceived and designed to alleviate such deficiencies as are said to be characteristic of scientific communities clustering about a well-defined, particular set of problems, or, to use a fashionable single term, with reassuring linguistic resonances, a paradigm. That scientific communities are selectively closed to external influence has been implied not only by T. S. Kuhn but also by A. L. Kroeber, in his Style and civilizations (1957), while others, like M. Polanyi, have conceived of academic disciplines and research areas as a sort of honeycomb structure, that is, as consisting of chains of overlapping neighborhoods extending over the entire panorama of science. Each scholar understands the self-contained paradigm of his own field and 'just enough' about those that lie adjacent to his own—but that is about the extent of his understanding. The internal consolidation of some disciplines is more self-sufficient than that of others, and there are wide variations in this respect between, say, mathematics toward one end of the scale and the biological sciences toward the other. Or, as in the history of linguistics, periods of contention for autonomy oscillate with those striving toward integration: 'In other words', Roman Jakobson insists, 'equal attention must be paid to the specifics in the structure and development of any given province of knowledge and, furthermore, to their common foundations and developmental lines as well as their mutual dependence' (1967). Our bias of the moment supports Edward Sapir's vision, expressed in 1927, that linguists, nolens volens, 'must become increasingly concerned with the many anthropological, sociological, and psychological problems which invade the field of language.... It is difficult for a modern linguist to confine himself to his traditional subject matter. Unless he is somewhat unimaginative, he cannot but share in some or all of the mutual interests which tie up linguistics with anthropology and culture history, with sociology, with psychology, with philosophy, and, more remotely, with physics and physiology.' The volumes of this series hitherto published have seemingly reinforced the tendency of our discipline toward a high degree of specialization, on

VI

EDITOR'S INTRODUCTION

the tacit assumption that linguistic knowledge, to become at once cumulative and progressive, requires a certain amount of closure. However, since a receptivity to external ideas, assimilation of knowledge from other segments of the arts and the sciences, are also necessary to prevent our linguistic community from lapsing into stale dogmatism, this set of tomes was designed as a sort of counterweight to the previous eleven. They will be followed shortly by the publication of Vol. 13, Current trends in the historiography of linguistics, which is being typeset as I am writing these lines. The entire series will then come to an end with Vol. 14, constituting a comprehensive Index to Current Trends in Linguistics, Vols. 1-13. Since the reviews of Vols. 1-5 listed in my Introductions to Vols. 6, 8, 9, and 10 have appeared, many further accounts have come to my attention, and I would like to continue below chronicling the published reactions to the series. Many of them have proved extremely instructive, either by their positive criticism, or provision of supplementary factual information, or both; a few were merely perfunctory; but none that I read were unfairly captious. My appreciation is hereby extended to all our discussants, the following among whom were not previously mentioned: Vols. 1, 2, and 3 — Anton Vorblicher, Anthropos 67.938-41 (1972). Vol. 2 — Anon., Orientalistische Literaturzeitung 68.86-94 (1973); Bruno Lewin, Oriens 21/22.587-91 (1971); James Matisoft, JASt 28.835-37 (1968-69); Roy Andrew Miller, Journal of the American Oriental Society 92.137—41 (1972); M. V. Sofronov (on China and Tibet), V. M. Alpatov (on Japan), L. R. Koncevic (on Korea), Z. V. Severnina (on Mongolia), et al., Voprosy Jazykoznanija 1970/2, pp. 110-30. Vol. 3 — M. Doherty, Zeitschrift fiir Phonetik, Sprachwissenschaft und Kommunikationsforschung 23.619-25 (1970). Vol. 5 — Michael Agar, Language in Society 1.279-83 (1972); George Cardona, Linguistics 197.40-46 (1973); G. Fussman, BEFEO 58.305-07 (1971); A.-G. Haudricourt, L'Homme 22.159-60 (1972); K. V. Zvelebil, Lingua 30.79-88 (1972); J. C. Wright, BSOAS 34.670 (1971). Vol. 6 — Robert Hetzron, Linguistics (in press). Vol. 7 — H. A. Gleason, Jr., American Anthropologist 74.1490-92 (1972); Jan Knappert, Linguistics 124.71-89 (1974); Herbert F. W. Stahlke, Language 50.195205 (1974). Vol. 8 — Karl J. Franklin, Linguistics (in press); A.-G. Haudricourt, Bulletin de la Société de Linguistique de Paris 68/2.000 (1973); Hiroshi Kuki, Oceania 44.68-71 (1973). A parochial reviewer, apparently unfamiliar with the conception of the series as a whole, has questioned 'the justification of dividing the volumes according to geographic areas', remarking that, for example, 'the absense of Chadic makes Afroasiatic incomplete', whereas, in fact, this subfamily was actually covered, rather thoroughly, in a chapter of Vol. 7, and soon afterwards reprinted in a comprehensive paperback 'spin-off', Afroasiatic: A survey, along with the rest.

EDITOR'S INTRODUCTION

VII

Furthermore, almost one third of the volumes — 3, 11, 12, and 13 — are not geopolitically organized at all, and our underlying scheme has been expounded in previous Introductions. I am grateful for the remarkable generosity of the U.S. Office of Education, which has defrayed the cost of preparing all of Vols. 4, 5, 6, 7, 11, and 12, the latter — under contract No. OEC-0-9-097736-3732 (014), with the Indiana University Foundation — in the amount of $91,038, or, a total of $265,186. Some of the contents of this book require explanatory observations. In the original scheme, an opening section was to have been devoted to the historiography of linguistics, by way of a comprehensive chapter that was to have been written in collaboration by Stankiewicz with Hymes. However, it soon became clear to these two Associate Editors that the subject was too vast for so compressed a treatment, and, instead of producing the chapter as envisaged, they presented a preliminary design for an entire book, which was later elaborated and will shortly appear as the next volume in our series. Many readers will question, and rightly so, the inclusion in this book of a chapter on 'Linguistics and Semantics', implying, in this particular context, that semantics is somehow legislated beyond the pale of linguistics proper. Such a perspective would ascribe a misconception totally alien to me, for, in fact, I believe that most semantic questions are quintessentially linguistic questions; Professors Coseriu and Geckeler also make their point of view about the matter crystal clear in their well-integrated treatment. That their piece appears, nonetheless, where it does, came about thus: originally, the Coseriu-Geckeler article was to have been paired with a second article on the state of semantics, contributed by another prominent expert, laboring, however, in a differing scholarly tradition, who certainly would have posed the Fragestellung in quite another light. The two pieces were, of course, meant to be complementary. However, this balanced design was frustrated by the inability of this admired colleague — for reasons wholly beyond his control — to submit his hardly dispensable and very much wanted disquisition. Other readers may raise an eyebrow at the seemingly unorthodox juxtaposition of anthropology and sociology in Part Seven. This state of affairs is due to the regrettable circumstance that four anthropologists who had made binding commitments to contribute chapters on various pertinent topics simply failed to meet their obligation, whereas all those with a sociological emphasis did so punctiliously. I cannot account for the defection of the anthropologists, for they share only one manifest distinctive feature, in that they are all located in the San Francisco Bay area! The Editorial Board allocated prime responsibility for the contents of Vol. 12 in the following manner: to Abramson, for Part Ten; to Hymes, for Part Seven; to Rubenstein, for Parts Six and Twelve; to Spolsky, for Part Nine; to Stankiewicz, for Part Three; and to the undersigned, for Parts One, Two, Four, Five, Eight,

VIII

EDITOR'S INTRODUCTION

Eleven, and Thirteen. Several of the sections will shortly be published, in toto or in part, in the form of separate 'spin-off' volumes, and thus become available at prices within the reach of interested individuals, notably students. I am particularly gratified at the initiative of Professors Abramson and Spolsky in this matter. Members of the Board — each according to his predilection — have contributed editorial Overviews to sections in this book of concern to them; my chapter on 'Semiotics' may also be read in that spirit. With initial help from Lucia Hadd Zoercher, the technical preparation of this mammoth enterprise was accomplished by Alexandra Di Luglio, who also contributed the Master List of Abbreviations, the Index of Languages, and the Index of Names. Penelope Hermann extracted the Index of Subjects featured in these four tomes from the far ampler matrix of Vol. 14, now also nearing completion. This book took longer to produce than any of its predecessors. The reasons — not excuses — for the inordinate delay are attributable (in that order) to the cussedness of authors, common inadequacy of translators, inaccessibility, to a degree, of the very countries where the best work in certain areas is being carried on, the gigantism of the undertaking, and the sheer immensity of the successive processes of editing, indexing, and printing. T o illustrate: one of the most important articles to grace this book was delivered into my hands, by its distant and isolated author in person, taking advantage of my opportune visit, at a time when a very large portion of the rest of the work was already set in pages. Among the obvious alternatives, I opted for inclusion, involving great expense of Mouton funds and all our labor, and further retardation. T o edit (pace President Kennedy) is to choose; let the readers judge the sagacity of the choices. Wassenaar, March 11, 1974

THOMAS A. SEBEOK

CONTENTS

EDITOR'S INTRODUCTION

V

MASTER LIST OF ABBREVIATIONS

XIII

PART ONE: LINGUISTICS AND PHILOSOPHY

Linguistics and Philosophy, by J. M. E. Moravcsik On Logic and Theoretical Linguistics, by Yehoshua Bar-Hillel, Jonathan Malino, Avishai Margalit Linguistics and Semantics, by Eugenio Coseriu and Horst Geckeler . . . Speech Acts, by L. Jonathan Cohen

3 37 103 173

PART TWO: SEMIOTICS

Semiotics: A Survey of the State of the Art, by Thomas A. Sebeok . . . Paralinguistics, by David Crystal Facial Expression and Body Movement, by Harvey B. Sarles Proxemics, by O. Michael Watson Classification and Description of Sign Languages, by William C. Stokoe, Jr. Writing and Writing Systems, by George L. Trager Speech Surrogates: Drum and Whistle Systems, by Donna Jean Umiker . . Formalized Languages: Scientific, by Sanda Golopentia-Eretescu . . . . Zoosemiotics: Ethology and the Theory of Signs, by W. John Smith . . .

211 265 297 311 345 373 497 537 561

PART THREE: LINGUISTICS AND THE VERBAL ARTS

Structural Poetics and Linguistics, by Edward Stankiewicz Linguistics and Folkloristics, by William O. Hendricks

629 661

CONTENTS

X

Folk Poetry: General Problems, by V. N. Toporov Folk Poetry: History and Typology, by K. Horalek Folk Narrative, by Dorothy Clement and Benjamin N. Colby Growth of the Theoretical Framework of Modern Poetics, by Vyacheslav V. Ivanov Theoretical Poetics in the Twentieth Century, by F. Svejkovsky . . . . Rhetoric and Stylistics, by P. Guiraud Literary Genres, by Tzvetan Todorov Metrics, by John Lotz f

683 741 809 835 863 943 957 963

PART FOUR: SPECIAL LANGUAGES

New Formal Devices for Linguistics, by Maurice Gross Artificial Languages: International (Auxiliary), by Mario A. Pei . Cosmic Language, by H. Freudenthal

.

.

.

985 999 1019

PART FIVE: LINGUISTIC ASPECTS OF TRANSLATION

Translation, by Eugene A. Nida

1045

PART SIX: LINGUISTICS AND PSYCHOLOGY

Psycholinguistics: An Overview, by Herbert Rubenstein An Historical View of Psycholinguistics, by Arthur L. Blumenthal . . . Some Aspects of Language Acquisition, by Ursula Bellugi The Interaction of Perception and Linguistic Structures: A Preliminary Investigation of Neo-Functionalism, by T. G. Bever Syntactic Factors in Memory, by Samuel Fillenbaum Semantics and Comprehension, by Herbert H. Clark Social Perception of Speech, by Moshe Anisfeld

1071 1105 1136 1159 1235 1291 1429

PART SEVEN: ANTHROPOLOGY AND SOCIOLOGY

Anthropology and Sociology: An Overview, by Dell Hymes Some New Developments in Ethnosemantics and the Theory and Practice of Lexical/Semantic Fields, by Oswald Werner, with William Hagedorn, George Roth, Emile Schepers, and Luis Uriarte Social Class, Language, and Socialisation, by Basil B. Bernstein . . . .

1445

1477 1545

CONTENTS

Ethnomethodology, by Aaron V. Cicourel Sociolinguistics, by J. B. Pride The Sociology of Language: An Interdisciplinary Social Science Approach to Language in Society, by Joshua A. Fishman

XI

1563 1607 1629

PART EIGHT: LINGUISTICS AND ECONOMICS

Linguistics and Economics, by Ferruccio Rossi-Landi

1787

PART NINE: LINGUISTICS AND EDUCATION

Linguistics and Education: An Overview, by Bernard Spolsky Linguistics and the Language Barrier to Education, by Bernard Spolsky . Linguistics and Literacy, by Sarah C. Gudschinsky Theoretical and Experimental Bases for Teaching Reading, by Richard L. Venezky Language in Early Childhood Education, by Vera P. John and Marshall I. Peller Linguistics and the Language Arts in Elementary and Secondary Education, by Rudolph C. Troike Linguistics and Second Language Pedagogy, by E. Glyn Lewis . . . .

2021 2027 2039 2057 2101 2117 2131

PART TEN: PHONETICS

Phonetics: An Overview, by Arthur S. Abramson Phonetics in the Twentieth Century, by D. B. Fry Speech Acoustics, by John M. Heinz Physiological Aspects of Articulatory Behavior, by Katherine S. Harris . . Laryngeal Research in Experimental Phonetics, by Masayuki Sawashima . The Perception of Speech, by Michael Studdert-Kennedy On Time and Timing in Speech, by Leigh Lisker A Study of Prosodie Features, by Philip Lieberman Speech Synthesis for Phonetic and Phonological Models, by Ignatius G. Mattingjy Phonetic Fieldwork, by J. C. Catford Cross-Language Phonetics, by André Malécot

2187 2201 2241 2281 2303 2349 2387 2419 2451 2489 2507

PART ELEVEN: BIO-MEDICAL APPLICATIONS

Language in a Biological Frame, by J. Bronowski Basic Problems of Neurolinguistics, by A. R. Luria

2539 2561

CONTENTS

XII

Language Behavior and Disorders Associated with Brain Damage, by Orlando L. Taylor and Joseph P. Fox Speech Pathology, by Eugene T. McDonald Language and Psychiatry, by Harley C. Shands

2595 2641 2657

PART TWELVE: COMPUTER APPLICATIONS

Computer Applications: An Overview, by Herbert Rubenstein . . . Information Handling, by David G. Hays Language Analysis, by Paul L. Garvin Computer Recognition of Speech, by Wayne A. Lea Mechanical Translation, by A. Hood Roberts and Michael Zarechnak .

.

2711 2719 2741 2765 2825

PART THIRTEEN: LINGUISTICS AS A PILOT SCIENCE

Linguistics as a Pilot Science, by Solomon Marcus Specialty Trends in the Language Sciences, by Paul L. Garvin

2871 2889

BIOGRAPHICAL

2911

NOTES

INDEXES

Index of Topics Index of Names Index of Languages

2931

MASTER LIST OF ABBREVIATIONS

1. JOURNALS AND BOOKS AASF Acta Oto-Laryng. Acta Psychologica Acta Physiologica Scandinavia Acta Radiologica AD AEH African Music AfrS AGI AJA AJExpPsych AJPsych AJSoc AL ALH AmA American Annals of the Deaf American Behavioral Scientist American Journal of Psychotherapy American Philosophical Quarterly Amer. Naturalist Amer. Zoologist AMNH-B AmPsych Analysis

Suomalaisen Tiedeakatemian Toimituksia/Annates Academiae Scientiarum Fennicae, Series B. Helsinki. Acta Oto-Laryngologiea. Stockholm. Acta Psychologica. European journal of psychology. Amsterdam, North Holland Publishing Co. Acta Physiologica Scandinavia. Scandinavian Physiological Society. Karolinska Instituted Stockholm. Acta Radiologica. 2 Series. 1. = Diagnosis; 2.= Therapy, Physics, Biology. Radiological Societies of Denmark, Finland, Norway and Sweden. Stockholm. American Documentation. Washington, D.C. Acta Ethnographica Academiae Scientiarum Hungaricae. Budapest. African Music. Johannesburg. African Studies. Johannesburg. Archivio Glottologico Italiano. Florence. American Journal of Archeology. New York. American Journal of Experimental Psychology = JExPsych. The American Journal of Psychology. Austin, Texas. American Journal of Sociology. Chicago. Acta Linguistica Hafniensia. International Journal of Structural Linguistics. Copenhagen. Acta Linguistica Academiae Scientiarium Hungaricae. Budapest. American Anthropologist. Menasha, Wisconsin. American Annals of the Deaf. Conference of Executives of American Schools for the Deaf; American Instructors of the Deaf. Washington, D.C. American Behavioral Scientist. Beverly Hills, Calif., Sage Publications. American Journal of Psychotherapy (Supplement). Association for the Advancement of Psychotherapy. Jamaica, N.Y. American Philosophical Quarterly. Oxford, Basil Blackwell & Mott Ltd. American Naturalist, devoted to the advancement and correlation of the biological sciences. American Society of Naturalists. Chicago, University of Chicago Press. American Zoologist. American Society of Zoologists, c/o California Lutheran College, Thousand Oaks, Calif. American Museum of Natural History, Bulletin. New York. American Psychologist, American Phsychology Association, Washington, D.C. Analysis. Oxford, Basil Blackwell.

XIV

Anat. Record

MASTER LIST OF ABBREVIATIONS

Anatomical Record. American Association of Anatomists. Wistar Institute of Anatomy and Biology, Philadelphia. Anglia Anglia. Zeitschrift für englische Philologie. Tübingen. Animal Behaviour Animal Behaviour (Monographs). Association for the Study of Animal Behaviour and American Behaviour Society. London, Bailiiere, Tindall and Cassell Ltd. AnL Anthropological Linguistics. Bloomington, Indiana. Ann. d'Otolaryng. Anales d'Oto-Laryngologie. Société de Laryngologie des Hôpitaux de Paris. Paris, Masson et Cie. Anthropos Anthropos. Revue internationale d'ethnologie et de linguistique/Internationale Zeitschrift für Völker- und Sprachenkunde. Freiburg, Switzerland. Antiquity Antiquity, a quarterly review of archeology. Cambridge, Eng., W. Heffer & Sons Ltd. Architects'Journal The Architects' Journal. London, Architectural Press. Archives of Archives of General Psychiatry. American Medical Association. Chicago. General Psychiatry Arch. Klin. Exper. Archiv für Klinische und Experimentelle Ohren-, Nasen- und Kehlkopfheilkunde. Ohren-, Nasen- (Formerly Arch. Oh.- Nas.- u. Kehlk. Heilk.) Deutsche Gesellschaft der Hals-, u. KehlkopfNasen- Ohrenärzte. New York and West Berlin, Springer-Verlag, heilk. ArchL Archivum Linguisticum. A Review of Comparative Philology and General Linguistics. Glasgow. ArchNPE Annales Neérlandaises de Phonétique Expérimentale. Amsterdam. Arch. Archives of Otolaryngology. American Medical Association. Chicago. Otolaryngology ArchPsych Archives de Psychologie. Geneva. AS American Scientist; published in the interest of scientific research. Society of Sigma Xi, Scientific Research Society of America. New Haven, Conn. ASHA Rep. American Speech and Hearing Reports. Washington, D.C. Auk Auk. A journal of ornithology. The American Ornithologists' Union. Anchorage, Ky. AUPELF La Revue de VAssociation des Universités partiellement ou entièrement de langue française. Université de Montréal. Australasian J. of Australasian Journal of Philosophy. Australasian Association of Philosophy. Philosophy University of Sydney Philosophy Department. Sydney. BAE-R Bureau of American Ethnology, Annual Report. Washington, D.C. Baessler-Archiv Baessler-Archiv. Beiträge zur Völkerkunde. Museum für Völkerkunde, BerlinDahlem. Berlin, Verlag Dietrich Reimer. BEFEO Bulletin de l'École Française d'Extrême-Orient. Saigon. Behaviour Behaviour (Monographs); an international journal of comparative ethology. Leiden, E. J. Brill. BCLC Bulletin du Cercle Linguistique de Copenhague. Copenhagen. BibPhon Bibliotheca Phonetica. Basel and New York, S. Karger. BICC Thesaurus', Bolétin del Instituto Caro y Cueva. Bogatà. Biological Reviews Biological Reviews of the Cambridge Philosophical Society. New York and London, Cambridge University Press. Biologisches Biologisches Zentralblatt. Leipzig, VEB Georg Thieme. Zentralblatt BLI Beiträge zur Linguistik und Informationsverarbeitung. Munich. Brain Brain. Journal of neurology. London, Macmillan (Journals) Ltd. Brain Research Brain Research. International interdisciplinary journal devoted to fundamental research in the brain sciences. Amsterdam, Elsevier Publishing Co. British Journal of The British Journal of Aesthetics. British Society of Aesthetics. London, Thames & Aesthetics Hudson Ltd. BrJEdPsych British Journal of Educational Psychology. London, Methuen & Co. BrJPsych British Journal of Psychology. British Psychological Society. New York and

MASTER LIST OF ABBREVIATIONS

BrJSocCPsych BRMI BS BSOAS BSL BT Bulletin of Mathematical Biophysics Bull. Mus. Comp. Zool. CACM CanadJPsych

XV

London, Cambridge University Press. British Journal of Social and Clinical Psychology. British Psychological Society. London and New York, Cambridge University Press. Behavior Research Methods and Instrumentation. University of Wisconsin Regional Primate Research Center. Psychonomic Journals, Inc. Madison, Wise. Behavioral Science. Ann Arbor, Mich. Bulletin of the School of Oriental and African Studies, University of London. London. Bulletin de la Société de Linguistique de Paris. Paris. The Bible Translator. Periodical for the Assistance of Bible Translators. London. Bulletin of Mathematical Biophysics. Mental Health Research Institute, University of Michigan. Ann Arbor, Mich.

Museum of Comparative Zoology at Harvard University, Bulletin. Cambridge, Mass. Communications of the Association for Computing Machinery. New York. Canadian Journal of Psychology!Revue Canadienne de Psychologie. Canadian Psychological Association. Toronto, University of Toronto Press. CanadPsych Canadian Psychologist!Psychologie Canadienne. Canadian Psychological Association. Ottawa, Ontario. CAnthr Current Anthropology. A world journal of the sciences of man. Chicago. CFS Cahiers Ferdinand de Saussure. Geneva. CDev Child Development. Society for Research in Child Development. Washington, D.C. CEd Childhood Education. A journal for teachers, administrators, church-school workers, librarians, pediatricians. Association for Childhood Education. Washington, D.C. CHum Computers and the Humanities. New York, New York. CJL Canadian Journal of Linguistics!Revue Canadienne de linguistique. Toronto. Cleft Palate Cleft Palate Journal. American Cleft Palate Association. Gainesville, Fla. CLex Cahiers de Lexicologie. Besançon. CogPsych = JCogPsych. CollE College English. Chicago. Comp. Stud. Soc. Comparative Studies in Society and History. & Hist. Condor Condor. Cooper Ornithological Society, Inc. New Mexico State University Department of Biology. Las Cruces, N.M. Cortex Cortex. Journal devoted to study of the nervous system and behavior. Varese, Tipografica Varese. CSHVB Computer Studies in the Humanities and Verbal Behavior. The Hague, Mouton. CTL Current Trends in Linguistics, vols. 1-12. The Hague, Mouton. Deutsche Viertel- Deutsche Vierteljahrschrift für Literaturwissenschaft und Geistesgeschichte. jahrschrift Stuttgart, J. B. Metzlersche Verlagsbuchhandlung. DevPsych Developmental Psychology. American Psychological Association. Washington, D.C. Die Erde Die Erde. Zeitschrift der Gesellschaft für Erdkunde zu Berlin. Berlin, Walter de Gruyter & Co. Diogenes Diogenes. An international review of philosophy and humanistic studies. International Council for Philosophy and Humanistic Studies. Montreal, Mario Casalini Ltd. Disorders of British Journal of Disorders of Communication (incorporating Speech Pathology Communication and Therapy). Edinburgh, E & S. Livingstone Ltd. DJbVk Deutsches Jahrbuch für Volkskunde. Berlin. EE Elementary English. A magazine of the language arts. National Council of Teachers of English. Champaign, 111. EdR Educational Research. Hove, Sussex. King, Thorne & State Ltd.

XVI

ELEC Electroenceph. Clin. Neuropsysiol. Esprit Estetika ETAT Ethnographia Ethnology Ethnos Études camerounaises Evolution

MASTER LIST OF ABBREVIATIONS

English Language Education Council Bulletin. Tokyo. Electroencephalography and Clinical Neurophysiology. International Federation of Societies for Electroencephalography and Clinical Neurophysiology. Amsterdam, Elsevier Publishing Co. Esprit. Paris. Estetika. Prague, Czechoslovak Academy of Sciences. Eesti NSV Teaduste Akadeemia Toimetised, Ühiskonnateaduste seeria / Izvestija Akademii Nauk Estonskoj SSR, Serija obscestvennych nauk. Tallinn. Ethnographia. Hungarian Ethnographical Society. Budapest. Ethnology. International Journal of Cultural and Social Anthropology. Pittsburgh. Ethnos. Etnografiska Museet. Stockholm. Études Camerounaises. Douala.

Evolution. International journal of the Society for the Study of Evolution. University of Kansas, Lawrence, Kans. Experimental Experimental Brain Research!Experimentation Cérébral/Experimentelle HirnBrain Research forschung. New York and Berlin, Springer Verlag. Eye, Ear, Nose The Eye, Ear, Nose and Throat Monthly. Chicago, Professional Press, Inc. and Throat Monthly Folklore Fellows Communications. Helsinki. FFC Foundations of Language. International journal of language and philosophy. FL Dordrecht, The Netherlands. Folia Linguistica. Acta Societatis Linguisticae Europaeae. The Hague, Mouton. FLing Le Français Moderne. Paris. FM Folia Phoniatrica. Journal international de Phoniatrie. Basel and New York, FPhon Karger. FR French Review. American Association of Teachers of French. Eastern Michigan University. Ypsilanti, Mich. Genetic Psychol- Genetic Psychology Monographs', child behavior, animal behavior and comparaogy Monographs tive psychology. Provincetown, Mass., Journal Press. Germanistische Linguistik. Marburg/Lahn. GLing Globus. Hildburghausen, Brunswick ( = Petermarins Geographische Mitteilungen). Globus Grundlagenstudien Grundlagenstudien aus Kybernetik und Geisteswissenschaft (Supplement). Quickaus Kybernetik born, Verlag Schnelle. und Geisteswissenschaft Helicon Helicon. Leipzig-Budapest. Hispania Hispania. A journal devoted to the interest of the teaching of Spanish and Portuguese. Appleton, Wise. HO Human Organization. Society for Applied Anthropology. Lexington, Ky. Homme L'Homme. Revue française d'anthropologie. Paris & The Hague, Mouton. HSPh Harvard Studies in Classical Philology. Cambridge, Mass. Human Relations Human Relations. A journal of studies towards the integration of the social sciences. New York, Plenum Publishing Corp. Ibis Ibis. British Ornithologists' Union. London, Academic Press. IBM J. IBM Journal of Research and Development. International Business Machines Corp., Armonk, N.Y. Information and Control. New York and London. I&C IEEE Trans. IEEE Transactions. Audio and Electroacoustics. Institute of Electrical and Audio Electronics Engineers, Inc. New York. IF Indogermanische Forschungen. Zeitschrift für Indogermanistik und allgemeine Sprachwissenschaft. Berlin. IJAL International Journal of American Linguistics (Memoir). Baltimore. International Journal of Slavic Linguistics and Poetics. The Hague, Mouton IJSLP Indian Linguistics. Journal of the Linguistic Society of India. Poona. IL

MASTER LIST OF ABBREVIATIONS

Insectes Sociaux International Social Science Journal IRAL

XVII

Insectes Sociaux. Paris, Masson & Cie. International Social Science Journal!Revue internationale des sciences sociales. New York and Paris, UNESCO Publications.

International Review of Applied Linguistics in Language Teaching/Internationale Zeitschrift für angewandte Linguistik in der Spracherziehung. Heidelberg. ISR Information Storage and Retrieval. Oxford. IUPAL Indiana University Publications in Anthropology, Folklore and Linguistics. Bloomington, Ind. JAbPsych Journal of Abnormal Psychology. American Psychologists Association. Washington, D.C. JAbSocPsych Journal of Abnormal and Social Psychology. Washington, D.C. JAcm Journal of the Association for Computing Machinery. Baltimore. JAcS Journal of the Acoustical Society of America. Lancaster, Pa. and New York. A = Abstract; L = Letter. JAF Journal of American Folklore. Philadelphia. J. Anat. Journal of Anatomy. Anatomical Society of Great Britain and Ireland. New York and London, Cambridge University Press. JanL Janua Linguarum. Series maior, minor, practica, and critica. The Hague, Mouton. JAOS Journal of the American Oriental Society. New Haven, Conn. Jap. J. Physiol. Japanese Journal of Physiology. Japan Publications Trading Co., Ltd. Tokyo. J. Appl. Physiol. Journal of Applied Physiology. American Physiological Society. Bethesda, Md. JASt The Journal of Asian Studies. New York. JAudEngSoc Journal of the Audio Engineers Society. London. JAudRes Journal of Auditory Research. For all workers seriously interested in the scientific study of hearing. C.W. Shilling Auditory Research Center, Inc. Groton, Conn. JC Journal of Communication. National Society for the Study of Communications. Jacksonville, Fla. JCogPsych Journal of Cognitive Psychology. Washington, D.C. J. Comp. Physiol. Journal of Comparative and Physiological Psychology. American Psychological Psychol. Association. Washington, D.C. J. de Psych. Journal de Psychologie Normale et Pathologique. Boulogne-Sur-Seine, PUF. JEA The Journal of Egyptian Archaeology. London. JEdPsych Journal of Educational Psychology. American Psychological Society. Washington. D.C. JEL Journal of English Linguistics. Bellingham, Wash. JExCPsych Journal of Experimental Child Psychology. New York, Academic Press. JExPsych Journal of Experimental Psychology. American Psychological Society. Washington, D.C. JExSocPsych Journal of Experimental Social Psychology. New York, Academic Press. J. Folklore Inst. Journal of the Folklore Institute. The Hague, Mouton. J. Franklin Inst. Journal of the Franklin Institute. Philadelphia. JGenPsych. Journal of Genetic Psychology. Child behavior, animal behavior and comparative psychology. Provincetown, Mass., Journal Press. JL Journal of Linguistics. London. J. Laryng. Journal of Laryngology and Otology. London, Headley Bros. J. Neurophysiol. Journal of Neurophysiology. American Physiological Society. Bethesda, Md. J. Ornithologie Journal für Ornithologie Deutsche Ornithologen-Gesellschaft. Berlin, R. Friedländer & Sohn. Journal of Journal of Counselling Psychology. For psychologists and personnel workers Counselling concerned with the counselling of clients, students, and employees. American Psychology Psychological Association. Washington, D.C. Journal of Journal of Mammology. American Society of Mammologists. Lawrence, Kans., Mammology Allen Press. Journal of Journal of Philosophy. Journal of Philosophy, Inc. New York, Columbia UniPhilosophy versity.

XVIII

JPSocPsych JPsych J. Psychiatr. Res. J. Psychos. Res. JPTA JR J. Radio Res. Labs. JRAI JSAm JSHD JSHR JSOc JSocI JSocPsych J. Theoretical Biology JVLVB J. Zool.

MASTER LIST OF ABBREVIATIONS

Journal of Personality and Social Psychology. American Psychological Association. Lancaster, Pa. Formerly JAbSocPsych. = J. de Psych. Journal of Psychiatric Research. Elmsford, N.Y., and Oxford, Pergamon Press (Journals Dept.). Journal of Psychosomatic Research. Elmsford, N.Y. and Oxford, Pergamon Press. Journal de Physique Théorétique et Appliquée. Paris. Journal of Reading. Formerly: Journal of Developmental Reading. International Reading Association. Newark, Del. Journal of the Radio Research Laboratories. Tokyo. Journal of the Royal Anthropological Institute of Great Britain and Ireland. London. Journal de la Société des Américanistes. Paris. Journal of Speech and Hearing Disorders. Wachington, D.C. Journal of Speech and Hearing Research. Washington, D.C. Journal de la Société des Océanistes. Paris. Journal of Social Issues. Society for the Psychological Study of Social Issues. Ann Arbor, Mich. Journal of Social Psychology. Provincetown, Mass., The Journal Press. Journal of Theoretical Biology. London, Academic Press.

Journal of Verbal Learning and Verbal Behavior. New York. Journal of Zoology. Formerly: Proceedings of the Zoological Society of London. London, Academic Press. Kenyon Review Kenyon Review. Kenyon College, Gambier, Ohio. Keystone Folklore Keystone Folklore Quarterly. Pennsylvania Folklore Society, Lycoming College, Quarterly Williamsport, Penna. Kongo-Overzee Kongo-Overzee. Ghent/Antwerp. Kursbuch Kursbuch. Frankfurt (Main), Suhrkamp Verlag. KZ Zeitschrift für vergleichende Sprachforschung auf dem Gebiete der indogermanischen Sprachen. Begründet von A. Kuhn. Göttingen. Langages Langages. Paris. Langue Française Langue Française. Paris. L'Anthropologie L'Anthropologie. Paris, Masson & Cie. L'Arc L'Arc. Aix-en-Provence. Laryngoscope Laryngoscope. A monthly journal on diseases of the ear, nose, and throat. American Laryngological, Rhinological and Otological Society. Collinsville, 111., Laryngoscope Co. LBerichte Linguistische Berichte. Braunschweig. LCC-P Linguistic Circle of Canberra, Publications. Series A, Occasional Papers; Series B; Monographs; Series C, Books. Canberra. LeSt Lingua e stile. Quaderni dell'Istituto di Glottologia dell Università degli Studi di Bologna. Bologna. Lg Language. Journal of the Linguistic Society of America. Baltimore. LgSciences Language Sciences. Indiana University Research Center for the Language Sciences. Bloomington, Ind. Lingua Lingua. International Review of General Linguistics/Revue internationale de linguistique générale. Amsterdam. Linguistic Inquiry Linguistic Inquiry. Toronto. Linguistic The Linguistic Reporter. Center for Applied Linguistics, Arlington, Va. Reporter Linguistics Linguistics. An international review. The Hague, Mouton. Linguistique La Linguistique. Revue internationale de linguistique générale. Paris. Listener Listener and BBC Television Review. London, British Broadcasting Corp. LL Language Learning. Ann Arbor, Mich.

MASTER LIST OF ABBREVIATIONS

L&LL Logos L&S Man Mathematical Biosciences META Mind MIT-QPR

XIX

Language and Language Learning. London, Oxford University Press. Logos. Naples, Libraría Scientifica Editrice. Language and Speech. Teddington, Middlesex. Man. A Record of Anthropological Science. London. Mathematical Biosciences. An international journal. New York, American Elsevier. META. (Formerly: Journal des Traducteurs.) Université de Montreal. Mind. A quarterly review of psychology and philosophy. Oxford, Basil Blackwell. Quarterly Progress Report, Research Laboratory of Electronics, Massachusetts Institute of Technology. Cambridge, Mass. ML Modern Languages. Journal of the Modern Language Association. London. MLJ Modern Language Journal. Ann Arbor, Mich. Monist The Monist. La Salle, 111., Open Court Publishing Co. MPhon Le Maitre Phonétique. Organe de l'Association Phonétique Internationale. London. MS Mediaeval Studies. Toronto. Mschr. Monatsschrift für Ohrenheilkunde und Laryngo-Rhinologie. österreichische OtoOhrenheilk. Laryngologische Gesellschaft und Wiener Gesellschaft der Hals-, Nasen-, OhrenÄrzte. Vienna, Verlag Urban & Schwartzenberg GmbH. MSFOu Mémoires de la Société Finno-ougrienne. Helsinki. MSLL Monograph Series on Languages and Linguistics, Georgetown University. Washington, D.C. MSOS Mitteilungen des Seminars für Orientalische Sprachen. Berlin. MT Mechanical Translation. Cambridge, Mass. Nature Nature. A weekly journal of science. London, Macmillan (Journals) Ltd. NDok Nachrichten für Dokumentation. Zweimonatsschrift für Forschung und Praxis der Dokumentation. Fachausschuss Bibliotheks-, Buch- und Zeitschriftenwesen im Deutschen Normenausschuss. Deutsche Gesellschaft für Dokumentation. Frankfurt (Main). Neuropsychologia Neuropsychologia. An international journal. Elmsford, N.Y., and Oxford, Pergamon Press. Nigeria Nigeria. Ibadan. Nous Nous. A quarterly journal of philosophy. Detroit, Mich., Wayne State University Press. Oceania Oceania. A journal devoted to the study of the native peoples of Australia, New Guinea and the Islands of the Pacific Ocean. Sydney. OL Oceanic Linguistics, Special Publication. Pacific and Asian Linguistics Institute, University of Hawaii. Honolulu. Oxford Slavonic Papers. London. OS1P La Palabra y el Hombre. Departamento Editorial de la Universidad Veracruzana. Palabra y el Veracruz. Hombre Proceedings of the American Philosophical Society. Philadelphia. PAPhilosS Beiträge zur Geschichte der deutschen Sprache und Literatur. Tübingen. PBB(T) Prague Bulletin of Mathematical Linguistics. Prague. PBML = Globus. Petermanns Mitteilungen Philosophical Forum. Boston University Department of Philosophy. Boston, Philosophical Mass. Forum Philosophical Magazine. A journal of theoretical, experimental, and applied Philosophical physics. London, Taylor & Francis Ltd. Magazine The Philosophical Quarterly. University of St. Andrews for Scots Philosophical Philosophical Club. Dundee, Scotland, The University. Quarterly Philosophical Review. Faculty of the Sage School of Philosophy. Ithaca, N.Y., Philosophical Cornell University. Review Philosophical Studies. Minneapolis, University of Minnesota Press. Philosophical Studies

XX

Philosophy and Phenomenological Research Philosophy of Science Phonetica

MASTER LIST OF ABBREVIATIONS

Philosophy and Phenomenological Research. International Phenomenological Society, State University of New York at Buffalo. Buffalo, N.Y. Philosophy of Science. Philosophy of Science Association. St. Louis, Mo.

Phonetica. Internationale Zeitschrift für Phonetik/International Journal of Phonetics. Basel & New York. Phylon Phylon. Atlanta, Ga. Physical Review. American Physical Society. New York, American Institute of Physical Review Physics, Inc. Proceedings of the Eighth International Congress of Linguists/Actes du Huitième PICL8 Congrès International des Linguistes, Oslo 5-9 August, 1958. General editor: Eva Sivertsen. Oslo, Oslo University Press, 1960. PICL9 Proceedings of the Ninth International Congress of Linguists, Cambridge, Mass., August 27-31,1962. Ed. by Horace G. Lunt. Janua Linguarum Series Maior 12. The Hague, Mouton, 1964. PICL 10 Actes du Xe Congrès International des Linguistes, Bucarest, 28 août-2 septembre 1967. Redacteur en chef: A. Graurs. Editions de l'Académie de la République Socialiste de Roumanie, Bucarest. 4 vols., 1969-1970. Proceedings of the Fourth International Congress of Phonetic Sciences, held at the PICPS 4 University of Helsinki, 4-9 September, 1961. Edited by Antti Sovijärvi and Pento Aalto. The Hague, Mouton, 1962. Proceedings of the Fifth International Congress of Phonetic Sciences, held at the PICPS 5 University of Münster, 16-22 August, 1964. Edited by Eberhard Zwirner and Wolfgang Bethge. Basel and New York, Karger, 1965. PICPS 6 Proceedings of the Sixth International Congress of Phonetic Sciences, held in Prague, September 7-13, 1967. Prague, Academia. Plains Plains Anthropologist; a medium for the anthropological interpretation of the Anthropologist plains area in the United States (Plains Conference). Museum of Anthropology, University of Kansas. Lawrence, Kans. PMLA Publications of the Modern Language Association of America. New York. PMS Perceptual and Motor Skills. Missoula, Mont. Poetica Poetica. Zeitschrift für Sprach- und Literaturwissenschaft. Munich. POLA Project on Linguistic Analysis. Columbus, Ohio. P&P Perception & Psychophysics. Austin, Tex., Psychonomic Journals, Inc. Practica OtoPractica Oto-Rhino-Laryngologica. International review of otolaryngology. Rhino-LarynSupplement: Bibliotheca Oto-Rhino-Laryngologica. Basel, S. Karger. gologica Proceedings of Proceedings of the Royal Institution of Great Britain. Barking, Essex, Elsevier the Royal Publishing Co., Ltd. Institution Proc. Nat. Acad. Proceedings of the National Academy of Science. Washington, D.C. Sci. Proc. Zool. Soc. Proceedings of the Zoological Society of London. Now: Journal of Zoology. London London, Academic Press. PsychB Psychological Bulletin. Evaluative review of research, literature and articles on research methodology in psychology. American Psychological Society. Washington, D.C. Psychiatry Psychiatry. Journal for the study of interpersonal processes. William Alanson White Psychiatric Foundation, Inc. Washington, D.C. PsychM Psychological Monographs. American Psychological Society. Washington, D.C. Psychologie und Schweitzerische Zeitschrift für Psychologie und Ihre Anwendungen. Bern, Hans Huber. Ihre Anwendungen Psychology Today Psychology Today. Los Angeles, Calif., CRM, Inc.

MASTER LIST OF ABBREVIATIONS

Psychonomic Science PsychRev QJEP QJSp Quarterly Rev. Biol. Ratio RCAFL-P Recent Advances in Biological Psychiatry REL RES1 Review of Metaphysics RevPhonA Revue d'esthétique RF RH RJb RLing RLR RRQ RSSPh RT SA Science Science Progress Semiotica SG SIL SJA SL SMC

XXI

Psychonomic Science. Austin, Tex., Psychomonic Journals, Inc. Psychological Review. American Psychological Society. Washington, D.C. Quarterly Journal of Experimental Psychology. Experimental Psychology Society. London and New York, Academic Press. The Quarterly Journal of Speech. Columbia, Missouri. Quarterly Review of Biology. Stony Brook Foundation, Inc. Stony Brook, N.Y., State University of New York. Ratio. Oxford, Basil Blackwell. Indiana University Research Center in Anthropology, Folklore and Linguistics, Publications. Bloomington, Ind. Recent Advances in Biological Psychiatry. Formerly: Biological Psychiatry. Society of Biological Psychiatry. New York, Plenum Publishing Corp. Revue des Études Latines. Paris. Revue des Études Slaves. Paris. Review of Metaphysics. A philosophical quarterly. Philosophy Education Society, Ltd. Haverford, Pa., Haverford College. Revue de Phonétique Appliquée. Möns. Revue d'esthétique. Centre National de la Recherche Scientifique. Paris, Klincksieck. Romanische Forschungen. Vierteljahrschrift für romanische Sprachen und Literaturen. Frankfurt a.M. Romanica Helvetica. Bern. Romanistisches Jahrbuch. Hamburg. Revue Roumaine de Linguistique. Bucharest. Revue de Linguistique Romane. Lyons & Paris. Reading Research Quarterly. International Reading Association. Newark, Del. Revue roumaine des sciences sociales. Série de philosophie et logique. Bucharest, Editions de l'Académie de la République Socialiste de Roumanie. Reading Teacher. International Reading Association. Newark, Del. Scientific American. New York, Scientific American, Inc. Science. American Association for the Advancement of Science. Washington, D.C. Science Progress. A quarterly review of current developments in science. Oxford, Blackwell Scientific Publications, Inc. Semiotica. Journal of the International Association for Semiotic Studies. The Hague, Mouton. Studium Generale. Berlin, Göttingen, and Heidelberg. Studies in Linguistics. Buffalo, N.Y. Southwestern Journal of Anthropology. Albuquerque, N.M. Studia Linguistica. Revue de linguistique générale et comparée. Lund. Smithsonian Miscellaneous Contributions. Smithsonian Institution. Washington, D.C. Statistical Methods in Linguistics. Stockholm. Social Forces. Chapel Hill, N.C. Sociology and Social Research. An international journal. Los Angeles, Calif.

SMIL SocF Sociology and Social Research Sociometry Sociometry. New York, Moreno. SocScil Social Science Information!Information sur les Sciences Sociales. International Social Science Council. Paris. Southern Folklore Southern Folklore Quarterly. University of Florida, Gainesville, Fla. Quarterly SovEtn Sovetskaja Etnografija. Moscow-Leningrad. SPhon Studia Phonologica. Kyoto. SpMon Speech Monographs. Columbia, Missouri. SRAZ Studie Romanica et Anglica Zagrabiensia. Zagreb.

XXII

STL-QPSR Strumenti Critici Studia Philosophica STZ Synthese TAPA TCLC TCLP Terre et la Vie Theoria TIL TIPL TITL TLL TLP TPhS TraLiLi Trans. Phil. Soc. Tuatara UCPE Uganda Journal VFPA VJa VR Wilson Bull. Word WPP WW WZUL YCS Zchr. Laryng. Rhin. Otol. ZDA ZDPh Zeitschrift für Kolonialsprachen ZEthn ZFSL Zoologica

MASTER LIST OF ABBREVIATIONS

Speech Transmission Laboratory, Quarterly Progress and Status Report. Stockholm. Strumenti Critici. Turin. Studia Philosophica. Jahrbuch der Schweizerischen Philosophischen Gesellschaft. Basel, Verlag für Recht und Gesellschaft AG. Sprache im technischen Zeitalter. Stuttgart. Synthese. An international journal for epistemology, methodology and philosophy of science. Dordrecht, Reidel. Transactions and Proceedings of the American Philological Association. Lancaster, Pa. Travaux du Cercle Linguistique de Copenhague. Copenhagen. Travaux du Cercle Linguistique de Prague. Prague. La Terre et la Vie. Revue d'écologie appliquée. Société Nationale de Protection de la Nature et Acclimatation de France. Paris. Theoria. A Swedish journal of philosophy. Lund, C.W.K. Gleerups Forlag. Travaux de l'Institut de Linguistique. Faculté des Lettres de l'Université de Paris. Paris. Travaux de l'Institut de Phonétique de Lund. Lund, Gleerup. Tijdschrift van het Instituut voor Toegepaste Linguistiek. Leuven. Travaux de Linguistique et de Littérature, publiés par le Centre de Philologie et de Littératures romanes de l'Université de Strasbourg. Strasbourg. Travaux linguistiques de Prague. Prague, Editions de l'Académie Tchécoslovaque des Sciences, 1966. Transactions of the Philological Society. Oxford. = TLL. = TPhS Tuatara (Biological Society). Tuatara Publishing Committee, Box 196, Wellington, N.Z. University of California Publications in Education. Berkeley and Los Angeles. Uganda Journal. Uganda Society. Kampala. Viking Fund Publications in Antropology. Wenner-Gren Foundation for Anthropological Research, Inc. New York. Voprosy Jazykoznanija. Moscow. Vox Romanica. Annales helvetici explorandis Unguis romanicis destinati. Collegii Romanici Helvetiorum. Bern, Francke. Wilson Bulletin. A quarterly magazine of ornithology. Wilson Ornithological Society. Lakewood, Ohio. Word. Journal of the Linguistic Circle of New York. New York. Working Papers in Phonetics. Ohio State University. Wirkendes Wort. Deutsches Sprachschaffen in Lehre und Leben. Düsseldorf, Pädagogischer Verlag Schwann. Wissenschaftliche Zeitschrift der Karl-Marx Universität. Gesellschafts- und Sprachwissenschaftliche Reihe. Leipzig. Yale Classical Studies. New Haven, Connecticut. Zeitschrift für Laryngologie, Rhinologie, Otologie und ihre Grenzgebiete. Stuttgart, George Thieme. Zeitschrift für Deutsches Altertum und deutsche Literatur. Wiesbaden. Zeitschrift für deutsche Philologie. Berlin. Zeitschrift für Kolonialsprachen. Berlin. Zeitschrift für Ethnologie. Organ der Deutschen Gesellschaft für Volkerkunde. Brunswick. Zeitschrift für französische Sprache und Literatur. Wiesbaden. Zoologica. Scientific contributions of the New York Zoological Society. Bronx, N.Y., Zoological Park.

MASTER LIST OF ABBREVIATIONS

ZPhon ZPsych ZRPh ZS1 ZSlPh

XXIII

Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung. Berlin. Zeitschrift für Psychologie. Incorporating: Zeitschrift für Angewandte Psychologie. Leipzig. Zeitschrift für romanische Philologie. Tübingen. Zeitschrift für Slawistik. Berlin. Zeitschrift für Slavische Philologie. Heidelberg, Carl Winter Universitätsverlag.

2. OTHER ABBREVIATIONS C.A.L. ERIC IEEE IRE PEGS S.I.L.

Center for Applied Linguistics. 1611 North Kent Street, Arlington, Va. Educational Resources Information Center. London. Institute of Electrical and Electronic Engineers. 345 East 47th Street. New York, N.Y. Institute of Radio Engineers. Program for Exchange of Generative Studies. Summer Institute of Linguistics. Head Office: Santa Ana, Calif.

PART ONE

LINGUISTICS AND PHILOSOPHY

LINGUISTICS A N D PHILOSOPHY

J. M. E. MORAVCSIK

Throughout history language has been an important source of philosophic puzzlement. To many outsiders it seems that only twentieth century American and English philosophy has the study of language as its main preoccupation, and in view of this they describe this movement as linguistic, or analytic philosophy. These labels tend to become attached to the philosophies of Russell, Carnap, Austin, and Wittgenstein: the philosophers whose influence dominates twentieth century Englishspeaking philosophy. A glance at the history of philosophy, however, reveals several previous periods in which the study of language is one of the main preoccupations of the practitioners, beginning with Socrates. In order to understand a philosopher one has to know what problems he is trying to solve. In order to understand what philosophers have to say about language one has to realize that there is a large variety of problems raised by philosophers concerning language, and that solutions to some of these are irrelevant to solutions of others. Thus philosophical questions and theses about language arise on different levels, and in the contexts of diverse problems. We shall distinguish five of these, though undoubtedly other, equally adequate classifications could also be found. a) The Socratic Approach. Socrates instituted the philosophic practice of asking for the clarification of certain key terms. These terms are used very frequently, and may be part of the vocabulary in terms of which we characterize persons, e.g. 'courage' or 'wisdom'; or these may be part of the vocabulary needed to characterize scientific activities, e.g. 'knowledge' or 'truth'. Among the typical Socratic dialogues are the Meno, the Laches, and the Euthyphro. Ostensibly Socrates does not ask for the meanings of key terms. He asks questions of the form: 'what is X?' where for 'X' we can substitute one of the terms mentioned above. The ensuing discussions, however, make it clear that in order to answer this type of question to Socrates' satisfaction the meaning of the term 'X' will have to be clarified. Socrates' method has four underlying assumptions. These are worth spelling out since they constitute one of the first, rudimentary, theories of language in Western thought, i) The meanings of terms lie below the surface. Socrates believes that even the meanings of commonly used expressions cannot be determined by merely examining our associations with these, or conducting empirical examinations of the entities which these expressions denote (or are true of). According to Socrates we have to examine

4

J. M. E. MORAVCSIK

the roles that these expressions play in sound reasoning, and their places in sentences expressing conceptual truths, in order to grasp their meaning, ii) Socrates subscribes to the no-ownership theory of meaning. In Socrates' time as well as in our time there are people who think that a speaker is sole master over his own use of language and over the meanings that he assigns to his use of expressions. According to this view it is somehow up to the speaker to decide what meanings the expressions that he employs should have. One consequence of this view is expressed by the fashionable slogan that the study of language is purely descriptive. Socrates is opposed to such a conception. Like most good linguists and philosophers, he conceives of his task neither as purely descriptive nor as a series of normative reform proposals. He is interested in finding out how members of a linguistic community think we ought to talk. According to the Socratic conception a language has a life of its own, and the meanings of its expressions belong to it, and do not depend on conscious decisions made by members of the linguistic community. Pains or feelings too are not a matter of decisions on our part, but these are in a sense private; the subject has privileged access to these. Socrates would deny that the subject has privileged access to the meanings of the expressions that he uses. Introspection is not a reliable guide to the discovery of meanings; a careful examination of how reasoning is expressed in language is more likely to lead to good results, iii) Socrates believes that it is very difficult to find out what we mean by commonly used expressions, and that often we literally do not know what we mean by an expression 'e' even though we use it frequently and with success as far as communication is concerned. Thus the clarification of what someone meant by 'e' in a certain context requires great intellectual effort in the crucial cases, and the methods of clarification involve tests for consistency of beliefs, validity of accepted inferences, and the determination of what the subject really takes to be true. Yet in spite of ii) and iii) Socrates also believed iv) that finding out what we mean by a certain expression leads to self knowledge. This seems paradoxical only if we think in terms of selfindulgent and overly romantic modern conceptions of the self according to which the self is something elusive, unique to each person, and connected with feelings and other phenomena to which each person has privileged private access. We might think of Socratic self knowledge as analogous to the kind of self knowledge that involves finding out what the nature of my respiratory system or digestive system — or, in this case, thinking system — is, and then the mystery as well as the air of paradox disappears. The four theses do not require each other, and indeed in the subsequent history of philosophy sometimes one and sometimes another of the four was rejected. Although this rudimentary theory does not say anything specific about a given natural language, it gives a rationale for a certain type of detailed semantic investigation of a given natural language. This Socratic activity of unearthing initially surprizing semantic facts about a given natural language continues to be one of the key functions of philosophers. In

LINGUISTICS AND PHILOSOPHY

5

our time its chief practitioner, or artist, one might say, was John L. Austin. His study of English deals with subtle phenomena and with complexities of sense at a level rarely if ever penetrated previously by philosopher or linguist. Austin thought that this activity has salutary effects on philosophers, but it should be clear that his results are also of great interest to linguists since it provides them with really interesting data that an adequate semantic theory of English has to account for. b) The Metaphysical Concern with Language. Metaphysics can be roughly characterized as the attempt to understand some of the essential structure of reality, and some philosophers believe that this essential structure is mirrored by or is reflected in the basic structure of language. This belief leads to attempts to uncover what are claimed to be fundamental structures of language, and this investigation leads often to the construction of ideal languages. What is ideal from such a metaphysical point of view may not be ideal from several other points of view, e.g. from the point of view of ease of communication, or flexibility. Thus the notion of a fundamental structure of language, as developed e.g. by the early Wittgenstein, has to be interpreted from the point of view of the metaphysical claims which it is said to assume, or from which it is said to be derivable. Under this heading we might also consider a concern with language that derives from certain anti-metaphysical attitudes. Some philosophers, e.g. modern positivists hold that there are no non-trivial eternal verities for a special branch of philosophy such as metaphysics to establish. They want to explain those propositions that seem to be both true and non-empirical as having their source in the rules of language. c) Language and Logic. Still another group of philosophers is interested in examining the extent to which the laws of logic are expressed in the rules for various natural languages. In other words, the question can be raised to what extent a natural language is a suitable vehicle for rigorous deductive reasoning. People interested in these questions will often talk about the logical structure of a language. This is, roughly speaking, a study of those features of language that are relevant to exhibiting deductive inferential relations. Needless to say, a philosopher may be engaged in more than one of these enterprises. Nevertheless, for obvious reasons, it is important to keep these investigations and the corresponding senses of 'structure' etc. distinct. d) Philosophy of Linguistics. An integral part of philosophy is the concern with the methodology, evidential basis, and conceptual framework of the various sciences. Thus, for example, what was known earlier as natural philosophy gave way to the philosophy of science, and this in turn split into the philosophy of the a priori sciences, the philosophy of the physical sciences and the philosophy of the social sciences. In some cases a specific science occupies the attention of philosophers; thus we have subjects like the philosophy of mathematics and the philosophy of history. The philosophy of linguistics should be conceived as analogous to these. It examines the methodology, evidential basis, and conceptual framework of linguistics. What is it for linguistics to become a science? What, if anything, is unique

6

J. M. E. MORAVCSIK

to linguistics? Questions of this sort occupy philosophers working in this field, and it should be clear that the questions raised about language in this context are of a different sort and are on a different level from the ones raised in connection with enterprise a). e) The Language of Philosophy and Linguistics. Certain terms plays key roles both in philosophy and linguistics. Such terms are 'true', 'meaning', 'refer', 'understand'. These terms are also parts of ordinary non-technical English. Thus the analysis of these terms raises distinctive questions. Do these terms have the same meaning in philosophy and in linguistics? If not, why not? How are these technical meanings related to the ordinary non-technical meanings of these expressions? Analogous problems arise also with regard to the vocabulary of other sciences. We can raise the following question in connection with all of them: to what extent is it necessary to explain ultimately the technical meanings of expressions in terms of the non-technical meanings of everyday expressions? These five enterprises are logically independent of each other. Each has its distinctive problems and methodology, and each is interested in a different set of features of language. All of these deal somehow with the 'structure of language', but they mean by this different things, and unless one knows the problem and the nature of the enterprise under consideration one loses sight of what this 'structure' is supposed to be — as the practitioners themselves lose sight of this occasionally. At the same time the results of the different enterprises can be relevant to each other. Enterprises a) and c) yield data to be accounted for by a science that is studied by enterprise d). Some of the analysis under e) may resemble that conducted under a). At the present the greatest need seems to be the clarification of d), and thus this will be the focus of this paper. From time to time, however, matters falling under a), c), and e) will also become relevant, but these will be treated only in so far as they are required for the clarification of d). The decision to focus on d) does not involve a value judgment to the effect that this is the most important, or fundamental, or that this is what everybody should be doing right now, etc. Once we distinguish the five different enterprises, there is no point in raising questions of relative importance and merit. And given the characterizations outlined above it should be obvious that the decision to concentrate on d) does not imply the view that linguists should not be interested in the other topics; on the contrary, without interest in and acquaintance with what goes on under a), c), and e) linguists are likely to be missing important sources of possible data and clarification. One might add in conclusion that if this essay is to be of interest both to linguists and philosophers, then it is easier to tell philosophers something about linguistics under the heading d) than under any of the other headings.

LINGUISTICS AND PHILOSOPHY

7

I Current linguistic theory and the philosophy of language have a striking common feature. Neither field suffers from a dearth of discovered facts to be explained, and both fields lack an over-all non-vacuous conceptual framework that can account for all of the observed facts. One conceptual problem concerns the relation of syntax to semantics within transformational grammar. Another one arises when one tries to fit transformational grammar to the kind of semantic theory that philosophers have worked on in recent times. Finally, there are problems concerning the desired non-vacuousness of both semantics and syntax. These circumstances suggest the following agenda for this paper. First we shall trace historically the relation between syntax and semantics in transformational grammar, and draw some morals from this survey. We shall proceed by reviewing both the strong and the weak points of philosophical semantics in such a way as to make its applicability to the work of linguists clear. We shall also invoke some notions developed in linguistic theory and show how these help to clarify certain issues in philosophical semantics. We shall also consider some salient differences between syntax and semantics, and conclude by attempting to spell out some of the requirements that linguistic explanations have to meet before these can be regarded as scientific explanations in any serious sense, and we shall relate these requirements to certain recent developments in transformational grammar. It may come as a shock to some people that the starting date for my historical review is not 1957, but 1954-55. At that time an exchange appeared between BarHillel and Chomsky (Bar-Hillel 1954, Chomsky 1955) concerning the relation between formal syntax and semantics on the one hand, and empirical linguistic theory on the other. A careful examination shows that the positions taken and the assumptions made in this exchange had a far-reaching influence on subsequent linguistic theory. Bar-Hillel's main claim is that linguists ought to incorporate into their work on syntax and semantics the approach and some of the results of logicians like Carnap. From our point of view the most crucial part of Bar-Hillel's argument is that in which he divides semantics into 'two separate theories with two different sets of concepts' (Bar-Hillel 1954:236). These are the theory of meaning and the theory of reference. The former deals with concepts such as logical truth and synonymy, and the latter with the notions of truth and denotation. Bar-Hillel argues that though issues in the theory of reference are of no interest to the linguist except for 'methodological questions', the theory of meaning should be of great interest to the linguist since it raises questions to which the linguist should feel compelled to try to find answers, e.g. questions of synonymy and logical consequence. In order to appreciate Chomsky's reply one must keep in mind that by the time the exchange took place, an extensive philosophical literature was available, mostly by Quine, Goodman, and M. White (see Chomsky 1955, first footnote for refer-

8

J. M. E. MORAVCSIK

ences and acknowledgement), that criticized Carnap's theory of meaning. These criticisms included the claims that when applied to natural languages formal semantics furnished us only with vacuous explanations, and that with regard to natural languages a sharp distinction between matters of semantic rules and definitions on the one hand and widely shared empirical beliefs held by the linguistic community in question on the other presents an untenable dualism. In his reply Chomsky accepts Bar-Hillel's distinction between the theory of meaning and the theory of reference. He also accepts the claim that the latter is of no significant interest to the linguist. ' . . . Both Tarski and Quine have done important work in the theory of reference. This is the branch where real progress has been made; but it is also the branch that has little interest for linguists' (Chomsky 1955:41). With regard to the theory of meaning Chomsky adopts with slight variations some of the criticisms developed by Quine and others, and thus argues that formal semantics takes precisely those notions (e.g. 'synonymy', 'formal consequence') as primitive that the linguist is concerned to explain, and that the practice of listing meaning postulates or pairs of synonyms offers no help to the linguist who wants to know on what basis he should regard a pair of expressions in a natural language as synonymous, and wants to explain the relation between alleged facts of meaning and matters of empirical belief widely shared in a linguistic community (Chomsky 1955:37-9). Chomsky also claims (39-40, 44) that well known facts about conditionals cast doubt on how well we can explain the roles of logical connectives in a natural language with the help of symbolic logic, and that synonymy accounts cannot cope with belief-contexts, but for our purposes these arguments are less relevant. The main outline of what took place is clear enough. Bar-Hillel argues that semantics equals theory of meaning plus theory of reference, and that (with slight qualification) only the former is relevant to linguistics. Quine and others argue that the formal theory of meaning cannot offer non-vacuous and empirically relevant explanations of linguistic facts. Chomsky puts these arguments together and concludes that philosophical, or logical, syntax and semantics have little if anything to offer to the linguist except in so far as they set standards of rigor for a future adequate linguistic theory to be developed (45). One of the significant consequences of this exchange was that Chomsky continued to think of semantics till around 1964-65 as that part of understanding language that we know least about. It has been claimed from time to time that Chomsky's position was due to the tradition of distrust of semantics running deep in American linguistics. The development traced above, however, shows that Chomsky's doubts had much deeper clearly articulated philosophical justification. Another important consequence of this exchange is the assumed sharp distinction between theory of meaning and theory of reference and the alleged irrelevance of the latter to linguistics. Only this background might make intelligible why Fodor and Katz, as they turned to the construction of a semantic theory, insist that matters of reference are of no concern to them. The few flimsy arguments actually offered

LINGUISTICS AND PHILOSOPHY

9

in their work certainly do not explain why they adopted this view as basic to their outlook on semantics. Thus, roughly phrased, by 1965 the theory of reference is still 'out', but the theory of meaning is 'in' as far as linguistic theory is concerned. Since in Aspects Chomsky seems to approve — at least in principle — of the Fodor, Katz, and Postal approach to semantics, one wonders what factors changed his mind in the ten years between 1955 and 1965? There is no written evidence on this matter, and one can only conjecture that perhaps Chomsky and other transformalists have become overly impressed by the possibility of a parallel between syntactic and semantic structure. In any case, the shift took place without any explicit attempt to answer the objections to the theory of meaning that Chomsky himself embraced earlier. As far as the new theory has been made clear, the objections in terms of vacuousness, etc., apply just as much to notions like 'semantic marker' and 'reading' as they apply to notions like 'meaning' and 'analytic'. The next development in this area brings us up to the present. For in some recent work (Bach 1968, Lakoff and Ross n.d., Ross 1970) we see the adoption not only of the theory of meaning, but also of the theory of reference and denotation as expressed in first-order predicate calculus under the usual interpretation. This shift, like the earlier one, takes place without any sign of awareness of the theoretical considerations and implications that such an adoption carries with itself. In particular, there is no discussion in the recent literature of the theoretical issues concerning the explanatory power of the theory of meaning, of the alleged irrelevance of the theory of reference, and of the well known (by philosophers) shortcomings of the first-order predicate calculus to handle adequately a natural language with the richness of e.g. English. This development and the current situation calls for the following comments. The assumption shared in the early papers by Bar-Hillel and Chomsky that became so influential in the subsequent literature, namely, that the theory of reference or denotation is irrelevant to linguistics, is unsound. This can be seen by a review of the arguments that Bar-Hillel uses in its support. The arguments are that it is not the business of the linguist to find out which pairs of expressions in English happen to be co-extensive (have the same denotation) such as e.g. 'morning star' and 'evening star' or 'man' and 'featherless biped', and that it is not the business of the linguist to find out which English sentences happen to express something that is true; thus e.g. whether 'all cats have tails' expresses something true or not is of concern to the zoologist, but not the linguist (Bar-Hillel 1954:236). Though the claims made here are true, the irrelevance of the theories of truth and denotation does not follow. For although it is not the business of the linguist to find out whether what 'all cats have tails' expresses is true or not, it is the business of the linguist to find out under what conditions this sentence, or any other sentence in English, expresses something true. In short, it is an essential part of understanding English that we should know under what conditions a given sentence would express something true; whether a given sentence in fact expresses something true is, of

10

J. M. E. MORAVCSIK

course, not something the knowledge of which would be part of linguistic competence. Thus a theory of truth is a part of any adequate semantics for a natural language. The same considerations apply to matters of denotation. It is the business of the linguist to attempt to delineate those expressions in a natural language that are denoting expressions (can have denotation) from those that have other roles, and within the class of denoting expressions he should attempt to discover subclasses in terms of the different types of entities (e.g. concrete things, events, properties, etc.) that different types of denoting expressions might denote. E.g. it is not of interest to the linguist to note whether two given verbs happen to be co-extensive. It is, however, of interest to the linguist to develop a theory that explains what sorts of entities verbs denote; e.g. events, relations between substances, properties of substances, etc. (Davidson 1969 and Martin 1969). It is also the business of the linguist to decide how to handle from a referential point of view what functions syntactically as abstract singular terms. Countless other examples could be added. To say, however, that a theory of truth and denotation is part of an adequate semantic theory for a natural language like English is very different from saying that the scheme associated with first-order predicate logic is adequate to capture the semantics of a language like English. Philosophers have discussed over the past decades a whole range of phenomena that defy this type of simple analysis. Since this list is rarely presented in its totality and since linguists seem to be unaware of it, a brief review in this paper might not be inappropriate. The key features of first-order predicate calculus that are responsible for the break-down of the analysis are the following: the ways of conjoining elements to produce new predicate expressions are very limited, the denotation range assigned to various expressions are fixed for all contexts, and terms with the same denotation range are substitutable. The following items resist this sort of analysis. a) Sentences involving modal notions such as necessity and possibility. A sentence of the type: 'necessarily all A's are B's' will not retain its truth-value regardless of which term co-extensive with 'A' is substituted. b) Sentences involving intentional notions such as knowledge, belief, desire, etc. Here again the trouble is substitutability, for different descriptions of a belief will not retain truth-value under unrestricted substitution of co-extensive terms. c) Sentences involving intentional verbs such as seeking, hunting, trying, etc. The difficulty with these terms is that on the one hand they do not indicate simple properties of agents, and on the other hand they cannot be characterized as relations between agents and other concrete entities. d) Sentences containing indexical or token-reflexive terms such as 'here', 'there', T , 'you', 'now', etc. Though these terms retain their meaning, i.e. they are not systematically ambiguous, they denote different entities in different contexts and under different circumstances; thus one cannot be sure that simple deductive inferences turning on repeated occurrences of these terms will be valid. e) Sentences containing second-order terms such as 'is a number', 'is a shape'.

LINGUISTICS AND PHILOSOPHY

11

These terms do not characterize spatio-temporal particulars but rather properties that these particulars may have. Attempts to reduce such sentences to those that are only about particulars have been unsuccessful. f) Sentences containing adverbs and similar constructions. In a sentence like 'John read the book rapidly' we want to construe the predicate as a complex of parts, and yet we cannot derive it from 'John read the book' and 'John rapidly'. g) Ambiguity. This notion cannot be analyzed simply in terms of difference in denotation. h) Translation. One cannot translate adequately from one language to another solely on the basis of co-extensiveness. It will be left to the reader to ask himself how much of English is left as analyzable once we removed all sentences falling under a) to g). Once we abandon the naive hope that the semantics of English can be captured by the structure of first-order predicate logic, we can consider more recent developments in the theory of reference. Recent work, e.g. Hintikka 1962, Kripke 1965, and Montague 1970a, has extended the theory of reference and helped in reducing the list presented above. This work involves, however, extending the universe of discourse to cover not only actual but also possible individuals, and not only the actual but also possible worlds, and thus presents us with a new framework. Within this new framework we can define some of the notions, such as intension, that were previously part of the theory of meaning as belonging now to the theory of reference. Thus the focus of discussion in philosophical semantics has shifted in such a way that the distinction between theory of reference and theory of meaning as drawn earlier by Quine, Bar-Hillel, and Chomsky is no longer crucial. In considering recent work in philosophical semantics two main questions arise: to what extent does this work help to reduce the list of 'unmanageables' cited above, and to what extent does this work help to answer Quine's criticisms raised against the empirical applicability of the theory of meaning? In order to deal with these questions adequately Quine's criticisms have to be divided into three categories. One of these categories is that of logical criticisms. These criticisms center around the notions of identity and substitutivity. Quine insists that the entities that we admit into our explanatory framework should have clear criteria of identity, and that if something is true of an entity, it should be true of it regardless of the mode of reference applied on any one occasion to that entity. Quine claims that these conditions are not met by the explanations offered in the theory of meaning. Another set of criticisms concern the alleged vacuousness of the concepts of the theory of meaning. The third category is made up of Quine's epistemological criticisms. These include the questions of how one would know that two sentences in a language are synonymous, and of how one would establish the synonymity of two sub-sentential expressions, as well as the challenge to produce a sentence of English other than a truth of logic such that it is true by virtue of the meanings of its

12

J. M. E. MORAVCSIK

constituents and thus not subject to revision on the basis of possible future empirical evidence. This last issue received a fair amount of attention in the recent philosophical literature, and a series of more or less successive articles (Putnam 1962a, Moravcsik 1965, and Harman 1967) tends to confirm Quine's scepticism as more and more allegedly analytic propositions are shown to be disconfirmable on the basis of possible future empirical evidence. Verdict should be postponed, however, until philosophers and linguists do more work on the semantics of verbs and terms standing for relations. For the majority of the examples discussed by philosophers involve nouns and adjectives. From facts like the one that there is no strict synonym for 'bachelor' in any of its senses in English it does not follow that we might not find more promising candidates when we turn to a thorough analysis of verbs. Whether one feels that Quine's logical objections have been answered depends on the extent to which the framework of possible entities and possible worlds and functions relating these is acceptable. This topic is the subject of lively controversy among philosophers and logicians today. The three categories of criticism are logically independent. Thus even if the logical objections were to be answered that would leave the epistemological criticisms as well as the charge of vacuousness alive. Again, the charge of vacuousness does not depend on the extent to which the search for a pair of strict synonyms in English has been successful or unsuccessful. In order to consider the charge of vacuousness in a clearer light, we should restate some of the salient features of the theory of meaning that are at times ill understood by friend and foe alike. The most profound statement of the problems of meaning as well as the most penetrating solutions these problems were presented by Frege (consult the topically organized bibliography for some references).1 For Frege the fundamental underlying fact to be explained by a theory of meaning is: 'that mankind has a common store of thought which is transmitted from one generation to another' (Frege 1892 (1952): 59). In others words, the key fact to be explained is that in language thoughts are expressed, thoughts that are about reality and that are true or false. For a large and fundamental part of our language the following applies: a competent speakerhearer can grasp the thought expressed by a given sentence, without knowing who said it, when, where, and why. This is what makes the development of science, mathematics, and other intellectual endeavours possible. 'Language speaks for itself, which is to say that a large and fundamental part of our language is such that its meaning is totally independent of considerations of extra-linguistic context. On this point there is a striking similarity between Frege and Chomsky (see the latter's Cartesian linguistics). The problem for the theory of meaning becomes: how is this transmission of thought possible? It requires that sentences should be expressing 1

I would like to acknowledge my indebtedness to Mr. Fred Goldstein of Stanford University for many useful suggestions concerning the material that follows. He also contributed most of the items and classification for the logic section of the topical bibliography.

LINGUISTICS AND PHILOSOPHY

13

what is true or false, and this in turn requires that parts of sentences should refer to or denote parts of reality. (E.g. 'is a table' has as its denotation range the class of all tables). A name denotes its bearer, a predicate denotes a class of elements, i.e. those to which it truly applies. But now the question arises: what connects sentence and truth or falsity, and what connects a term with its denotation range? If the connections were arbitrary, then one could not explain the projective character of language, i.e. that we know how to go on applying predicates, and we know how to combine denoting elements in order to form new complexes. But again, the relation between term and denotation range might not be a direct one; or else in order to understand a term one would have to be acquainted with its entire range of denotation. This is, of course, impossible, since most denotation ranges are spatio-temporally unbounded. Thus the relation between term and denotation range must be mediated by something, something the grasp of which enables us to say what falls and what would fall in the denotation range of a given term. This element is the sense or meaning of the term. The senses of expressions contained in a sentence make up the thought expressed by the sentence. In certain contexts, such as the context of belief, an expression denotes indirectly, and what it denotes is its usual sense. The senses and denotations of complexes are always functions of the senses and denotations of the elements. This account also explains how identity statements of certain sorts can be both true and informative, since e.g. in the sentence 'the man in the White House in 1970 is the president of the U.S.A. in 1970' we have two denoting expressions that have different senses but denote the same thing. In order to appreciate this proposal we must invoke the distinction f a m i l i a r in recent linguistic theory but unfortunately not used in philosophical semantics between competence and performance. For Frege's theory is clearly a competence theory. The ability linked to grasping the sense of an expression, namely the ability to tell what does and does not fall within a denotation range is not the actual perceptual ability or recognitional skill. Furthermore, the notion is not brought in so as to explain what in fact we do on some particular occasion when we decide that entity x falls within the denotation range of expression 'e'. Much of the philosophical literature on semantics suffers from the confusion between competence and performance. Some philosophers tried to link the notion of meaning to that of intention, or to the use of an expression, or to the kinds of acts that one can perform by uttering a certain expression. At no point in this literature is it made clear whether these philosophers talk about what in fact takes place, or whether they are specifying a part of linguistic competence. The same lack of clarity affects the first two chapters of Quine 1960. The reason this paper is restricted to the discussion of Frege's theory is that it is the only philosophical theory of meaning that is clearly a competence model and a plausible one at that. Another distinction that one should borrow from syntactic theory and apply to semantics is between weak generative capacity and strong generative capacity. As the result of the ap-

14

J. M. E. MORAVCSIK

plication of this distinction we can grade semantic theories in terms of the extent to which they can account not only for semantic units but also for their internal structure. E.g. both Quine (1960:168ff.) and Frege can account for sentences in which someone is said to be entertaining a certain belief. But whereas on Frege's view the semantics of these sentences reveal the internal structure of the elements that make up the belief, on Quine's view each belief is in effect a different attribute ascribed to the believer. The question of psychological reality must be viewed in the light of the above. Frege certainly thinks that the notion of sense must have psychological reality, else the process whereby a competent speaker-hearer can relate expressions and their ranges of denotation becomes unintelligible. But the psychological reality in question is not the same status as that held by elements that are parts of the causal chain that makes up an average performance. Is the notion of sense vacuous? If it is so, then this is the sense of 'vacuous' that applies to something programmatic. Frege is giving the most abstract and general characterization of an element in the theory of language the existence of which he thought to have demonstrated. To make this characterization more specific is a matter of further research, both conceptual and empirical. Thus Frege's notion is not vacuous in the sense in which 'dormative power' in 'opium puts people to sleep on account of its dormitive power' is vacuous. The latter is construed as a completed explanation; the former as a general characterization or a program for a series of explanations. The related questions of indeterminacy and admissible evidence have been discussed in a recent exchange between Quine and Chomsky (Chomsky 1968b, Quine 1968). This exchange, however, concerns the notion of a rule of language in general, and Chomsky's argument that there is no more indeterminacy in linguistics than there is in any of the other sciences does not cover the specific issue of Quine's position with regard to exclusively semantic phenomena. In short, the extent of the force of Quine's arguments depend on how well we can distinguish between the nature of syntax and that of semantics. (This was urged already in Moravcsik 1967). It is thought at times that a Fregean theory of senses is incompatible with various types of lack of effectiveness discovered in semantics. This, however, is not the case. In the context of this paper it might be useful to review and distinguish the different types of inexactitude that have been discovered in languages like English. a) Terms with border-line cases of application. There are many examples for this to be found among color terms. The range of application of a color term might be such that for a large number of cases one does not know whether to apply one color term or another term naming the neighboring color. Thus in these cases, the problem is not that language does not provide a description, but that one does not know which member of a disjunction provided by language one should select. b) Vagueness. A vague term is one whose criteria of application seem unclear. This makes vague terms obviously different from terms with border-line cases since

LINGUISTICS AND PHILOSOPHY

15

for some of these e.g. colors, the criteria of application are very clear indeed. A typically vague term is 'experience'; one does not know just what this term is supposed to cover; and it makes no sense to talk in this connection about clear cases and border-line cases; what two concepts would the border-line cases fall between? c) Terms admitting only of disjunctive definitions. Wittgenstein explored some of these terms (see topical bibliography) and described the relevant phenomenon as 'family resemblance'. An example would be the word 'game'. Again, this phenomenon does not show a limitation on linguistic competence; rather, it indicates the need for complex operations in assigning sense to expressions of this type. d) Openness. By far the most interesting phenomenon of this type was discovered by Waismann (1951:120ff.) and discussed also by Austin (1961:35-6). Openness is a characteristic of those denoting terms in connection with which we find that though there is a class of entities to which the term applies and a class to which it does not apply, there are certain possible cases such that when asked if term't' applies to these — or any other term — we do not know what to say. Thus, for example, an entity might seem to fulfil the conditions required for it to be an eagle, and then it explodes, or starts reciting Homer, or does some similar 'outrageous thing' as the late Professor Austin would have said. The problem here is not one of vagueness, ambiguity, or border-line cases. In none of those cases does linguistic competence come to a halt. We have descriptions available; what we do not know is which of these to apply. But in the cases now under consideration, linguistic competence does come to a halt. There is no way of describing these possible phenomena without introducing new legislation over the semantics of the language. The semantic rules of English simply do not cover these cases. The material reviewed i nthis sketch is covered in a large part of the literature by the oft repeated and quite unilluminating remark that the semantics of natural languages is fuzzy or inexact. What we need is an exact theory of inexactitude that distinguishes the different phenomena, shows which types of expressions are affected by the different phenomena, and tries to give an account of these. For example, the phenomenon of openness might indicate that a 'principle of least effort' operates within the semantics of natural languages. The language at any one stage contains semantic rules only to cover a range of phenomena that has some likelihood of occurring. Unlike artificial and formal languages, natural languages may have semantics that do not function in such a way as to cover all logical possibilities. In any case, neither openness nor any of the other phenomena of inexactitude are incompatible with a Fregean theory of meaning. It is a part of the task of such a theory, however, to find a way to build these phenomena into a formal model of competence. This brief outline of suggestions toward an exact theory of inexactitude launches us in the task of presenting a list of problems that still await solutions in current work on semantics. Within the context of this review we must restrict ourselves to a brief mention of some of the more immediate problems.

16

J. M. E. MORAVCSIK

One of these is the problem of intensional contexts, in particular those involving belief, knowledge, hope, etc. We face the task of determining what should count as criteria for sameness of object of belief, etc., and thus of substitutability. T o say on the one hand that everything that is logically equivalent is substitutable in such contexts is to draw the line too widely; on the other hand, to say that no two distinct expressions can be substituted in such contexts is to draw too restrictive a line. A complete answer to this question has not yet been given, though interesting suggestions have been made recently (see the topical bibliography under Epistemic Terms). Another nest of problems is represented by the variety of semantic links to be found in a language like English. Four different ways of joining terms seem to yield different semantic configurations. 1) Molecular conjunction. E.g. the molecule 'this building is large and beautiful' can be built up from the atomic units of 'this building is large' and 'this building is beautiful'. From the logical point of view this is the most fundamental link. 2) Adverbial conjunction, e.g. 'John reads rapidly'. The difficulty is that though one cannot regard this as the conjunction of 'John reads' and 'John rapidly', an adequate analysis cannot treat it simply as a new non-composite predicate, since we have to account for the fact that from 'John reads rapidly' it follows that John reads. 3) Modifying conjunction, e.g. 'the Notre Dame is a tall building'. The peculiarity of this link is that though the two conjuncts can be separated without yielding nonsense, one of these in isolation is ambiguous, or even if unambiguous it requires interpretation with reference to the other conjunct. Thus the above statement is equivalent to 'the Notre Dame is a building' and 'the Notre Dame is tall', but the latter is ambiguous; is the Notre Dame tall as such, or is it tall for a building? (For dramatic effect, compare this case with the by now notorious example: 'Flora is a large flea'.) One does not want to say that terms like 'tall' or 'large' are systematically ambiguous; yet somehow the criteria of application will vary depending on the sense of the other term that the term in question quasi-modifies. 4) Non-detachable conjunctions. E.g. from 'Leo is an alleged dictator' one cannot infer either 'Leo is a dictator' or 'Leo is alleged'. It is not easy to see why a certain compound represents one link rather than another. E.g. 'this is a spurious dialogue' decomposes into 'this is a dialogue' and 'this is spurious', even though 'spurious', 'alleged', 'supposed', and 'fake' seem to be semantically similar. In terms of these four links we can form semantically significant categories. One cannot simply take expressions in English one by one, and make their ways of joining a matter of so many individual meaning postulates without m i s s i n g important generalizations. Thus the task that lies ahead is either to classify expressions in English according to these links and work out the representation of the links in formal semantics, or to reduce links 2), 3), and 4) to 1) by elaborate paraphrases of the items to be linked.

LINGUISTICS AND PHILOSOPHY

17

Another item on this list is the adequate semantics of mass-terms. For terms like 'water', 'iron', etc., function differently from those like 'table', 'animal', or 'number', and the difference is not solely a matter of syntax. Mass terms, as Quine noted (1960:91ff.) can stand both in subject or in predicate position, and as subjects seem to have a quite different referential role than that which count nouns have in that position. It is also interesting to note that from the semantic point of view a noun like 'water' and an adjective like 'red' seem to behave in a similar fashion. Both 'water' and 'red' can be construed as denoting a huge, spatiotemporally scattered object, parts of which are all the red entities and all of the waters respectively. One cannot construe the denotation ranges of count nouns and many adjectives in this fashion. Still another problem on this list of outstanding difficulties is an adequate treatment of the subject-predicate relation. In this connection again, one sees discussion proceeding in linguistics without any signs of awareness of the relevant philosophical debates that have taken place concerning this topic over the past decades — indeed, over the past centuries. The relevance and importance of these grammatical relations for semantic interpretation and for the stating of selectional rules is recognized by Chomsky (1965). In the recent literature, however, we find statements to the contrary, without careful supporting arguments. Thus in Bach's work the significance of the subject-predicate relation is belittled (Bach 1968) and in a recent paper (Lakoff Mid Ross n.d. : 2) we find the statement: 'subject of and object of are not directly relevant to semantic interpretation'. On the grammatical side, the notions must be crucial since on whichever level these are defined, we will also have to define at least some of the more important selection restrictions. On the semantic side, the issue is very complex, and no conclusions should be reached on this matter without a careful consideration of the literature by philosophers such as Frege, Russell, Strawson, and Quine (see topical bibliography). The key problems in this area are: an adequate account of predication, the distinction between singular and general terms, the significance or lack of significance of the notion of singular reference and identification, and plausible ways in which one can relate the structure of quantificational logical notation to the structure of natural languages. Thus, e.g. for Frege subject and predicate stand to non-linguistic entities in quite different semantic relations (Frege 1892 (1952) :42ff.) and issues between Russell and Strawson (Russell 1905, Strawson 1950) turn on the semantic significance of the notion of a subject, with a resulting difference in assignments of truthvalue to sentences of a certain type, and again the issues between Quine and Strawson (Quine 1968, Strawson 1968) on the intelligibility of the relevance of quantificational structure to natural language rests on the distinction between singular and general terms, which for Quine is based on the distinction between subject and predicate, and the 'distinct grammatical roles' (with respect to subject and predicate positions) that singular and general terms play in sentences (Quine 1960:96). These debates have not been resolved, but without understanding these, linguists

18

J. M. E. MORAVCSIK

are unlikely to arrive at sound positions with regard to the importance of these grammatical notions for syntactic organization and with regard to the underlying importance for semantic interpretation. As an illustration of the difficulties that careless analysis might encounter, let us consider some of the things said about this matter by Bach. In the course of an argument that, if accepted, would support the thesis that the subject-predicate distinction is not crucial for the semantics of English, Bach claims that an expression like 'the anthropologist' in the context of sentences of subject-predicate form is synonymous with an expression like 'the one who was an anthropologist' (Bach 1968 :92ff.). This claim is rather peculiar, and the source of the peculiarity can be gleaned from some of the recent philosophical literature on semantics. As Gilbert Ryle put it, a term like 'one' is substantive-hungry. Nothing is simply one; if someone describes an entity as one, we want to know immediately one what? One man, one number, one table? This feature of the expression can be seen when we observe that 'the one who' typically requires previous reference, and it points back to the previous referent. This is why one would not start a discourse with 'the one who . . . ' . In the typical case we have somewhere in discourse a complex referring expression like 'the women sitting in our living room', and later we might find a phrase like 'the one who was an anthropologist' embedded in a sentence. In this way we can show that 'the anthropologist' and 'the one who was an anthropologist' are not synonymous. If they were synonymous, then they could be substituted for each other in any context. But there is at least one context in which they cannot be substituted for each other, namely in the beginning of a discourse. Another way to show that the two expressions in question are not synonymous is to consider them in the context 'the same . . . ' . It is an important feature of some nouns and noun phrases — among these, prominently, count nouns — that understanding them involves knowing what it is to count as 'the same so-and-so'. Thus it is part of the understanding of 'anthropologist' that one knows what counts as 'the saiue anthropologist'. This feature is not correlated to pluralizability. For though both 'one' and 'anthropologist' admit of pluralization, only the latter provides criteria to decide under what conditions something is the same item, i.e. anthropologist. One does not know under what conditions something described as 'the one' simpliciter remains 'the same one' until the expression is supplemented like 'the one dog', 'the one star', etc. Bach's claim seems to be an instance of not considering a wide enough range of phenomena, in particular, regarding the question of what can begin discourse, and the question of sameness and individuation. These two questions are, however, quite vital in the discussion of the significance or lack of significance of the subject-predicate distinction. The last item on our list of urgent current problems is the need for a thorough examination of ways in which the semantics of various types of expressions is context-sensitive, in the sense in which 'context' means 'non-linguistic context'. A well-known class of expressions with context-sensitive semantics is the class of ex-

LINGUISTICS AND PHILOSOPHY

19

pressions that are or include indexicals. Recently Donellan (1966) has pointed out that singular terms in subject position that do not contain indexicals may also be prone to drastic context-sensitivity, and similar points were made earlier in Hintikka's study (1962) of belief-contexts. E.g. in a given context a man could point to an individual and say: 'the murderer is a red headed man', and be able to have made a successful identifying reference for a specified audience, even though the man in question is not a murderer but is only believed to be one. The question arises: to what extent is this type of context-sensitivity restricted to singular as opposed to general terms? Or to terms standing in subject as opposed to predicate position? To sum up, the items briefly described here have to be dealt with. Also, Quine's charge of vacuousness has to be answered in more than a mere programmatic way, and the conceptual foundations of the new theories of denotation have to be investigated. We shall have to reconsider also the question of how sharply one can separate semantic information from empirical information about the world widely shared in a linguistic community. It seems, however, that the best hope of finding an adequate semantic theory to go with transformational grammar lies in further developments of what has been characterized here as an essentially Fregean theory.

II

Since most of this part of the paper is concerned with the explanatory power and empirical content of linguistic hypotheses, it might be well to remind ourselves of the basic outlines of what and how grammatical theories are supposed to explain. Let us consider a language L. Within L we can distinguish the class of minimal meaning-bearing units; let us call this class K. We can then form another class K* by considering all possible combinations of the elements of K. At this stage we consider an abstract device that divides K* into two classes; the well formed formulae of L and those that are the not well formed. On the assumption that understanding what is or is not grammatical is a rule-following process, we assume that the device accomplishes this task by generating the well formed formulae according to a set of rules. This set of rules is the syntax of L. The more we find out about the formal properties of the syntax of L, the more we also know about the degree of complexity that we have to assign to the device that understands L. By showing that the grammar of L has certain formal properties we might be able to argue that no device or mind with a complex structure less than T, can understand L. This is the sense in which the study of generative grammar might give us limiting proofs about the structure of the mind, and this is why there is formal as well as substantive scientific interest in constructing a grammar that not only includes all of English but includes only a language like English. The languages whose grammars we are interested in in this context are the natural

20

J. M. E. MORAVCSIK

languages, and it is assumed that their grammars will have at least partly the same fundamental structure. The general justification for this assumption is that — as Chomsky has rightly insisted — language is a biological phenomenon, and thus there is no more reason to suppose that languages should be fundamentally different than there is to suppose that the bone structure or digestive system should show fundamental variations within the species. One difficulty encountered in the discussions of this topic is the specification of what is meant by a natural language. The term 'natural' admits of several contrasts. A brief exploration of these will help us to see why some delineations of the concept of a natural language are more adequate than others. 'Natural' contrasts, among other things, with 'non-natural', or 'artificial'. Accordingly, a natural language is one that originated under natural circumstances while an artificial one is generated by conscious invention and construction. This distinction is, however, trivial. For the mere fact that a language is invented and artificially constructed does not show whether that language could not also have come into being the natural way. Some artificial languages may be such that nobody could have learned them as his first language, and that could not have developed naturally; but this need not be true of all artificially produced languages. Far more significant is the contrast between a natural language in the sense of a language that can be the first language of a human being or some entity sufficiently similar to a human, as contrasted with a non-natural language that cannot serve as the first language to be acquired. In this sense 'natural' means 'initially acquirable'. From the point of view of a linguistic theory that views language as a biological phenomenon this is clearly the fundamental class of languages that we are interested in. If someone asks why we should give acquisition such a privileged position, the answer is that this follows from our conception of language as a biological phenomenon. We describe as human anatomy that anatomical structure that people are naturally born with and develop under normal circumstances. The same considerations will apply to other biological phenomena. One could also define 'natural language' as covering all of those languages that can be used by humans for certain essential tasks, such as describing empirical phenomena, formulating scientific laws, reasoning, etc. A non-natural language will be in this conception a language that cannot be used for these purposes. It is a purely empirical task to determine what the relationship is between the naturally acquirable languages and the natural languages as defined in this paragraph; most likely the latter would include the former class but would not be co-extensive with it. The distinction between natural and non-natural is at times confused with the distinction between formal and informal languages. A formal language is constructed in order to express as clearly and explicitly as possible a theory, e.g. biology, mathematics, etc. An informal language is one within which any number of different and mutually incompatible theories can be expressed. As a limiting case of informality in this sense, one could take the characteristic of universality;

LINGUISTICS AND PHILOSOPHY

21

i.e. a language is universal in its semantics in this sense if and only if it can express anything that is not logically inconsistent. It is important to see that there is no necessary correlation between the senses of 'natural' as defined above, and the notion of informality as introduced here. Rather, these distinctions enable us to state a number of interesting questions in a reasonably clear way; e.g. can everything be expressed in some natural language; can everything be expressed in all natural languages; does talking the same natural language require a certain sharing of a theory about the nature of the external world, etc.? At the present stage we are far from being able to answer any of these questions, but adequate answers to these, when they become available, would shed a great deal of light on the nature of language. In recent times the notion of intuition has figured prominently in linguistic discussions. The limits of the justification for introducing this notion into linguistic theory, however, have not been understood, nor have we so far succeeded in building this notion into a non-vacuous theory of mind. Psychologists rightly mistrust the notion of intuition, and thus it becomes that much more important to explain the justification for introducing this notion into a linguistic theory. The key concept is that of intuitions of grammaticality. The reason for regarding this notion as legitimate is that it has a corresponding part in the formal analogue of the competence model. As we saw, in explaining the device that accepts a language, we assigned to it the role of dividing K* into two sub-classes. The device accepts the well formed formulae and rejects the ill formed ones. Corresponding to this notion of acceptance and refusal we postulate the notion of intuitions of grammaticality. In general, one is justified in introducing the notion of an intuition only if one can find corresponding to the introduced entity an analogous part in the formal competence model. Viewed in this manner, it is best to regard intuitions of grammaticality as theoretical constructs, to be distinguished from beliefs about intuitions of grammaticality, with the latter as possibly evidence for the former. Another important difference between these two notions is that only instances of the former are immune to revision unless the language changes. Intuitions of grammaticality are self-warranting beliefs, while beliefs about these intuitions are fallible, like all other empirical beliefs. Under these circumstances a lot of work will have to be done in order to develop reliable testing methods for intuitions of grammaticality. That these intuitions can be reliably established was, understandably, a programmatic assumption in Chomsky's early work. The practice of people simply asking themselves in a class-room what is grammatical, etc., was meant merely as a crude temporary substitute until more objective and reliable methods were developed. Unfortunately, current practice by linguists seems to indicate that the original methodological considerations and justifiable ways of postulating intuition have been lost sight of, and the crude temporary methods of introspection etc. tend to be taken as final. Predictably, this has resulted in strange claims concerning grammaticality. The following is a sample of recent judgments concerning what is un-

22

J. M. E. MORAVCSIK

grammatical. 1) Tom doubted that Ann could swim, but nobody believed him (Ross 1970). 2) As for myself, I promise you that Tom will be there (Ross 1970). 3) John praised Mary, and then she insulted him (Lakoff 1969). There is no doubt that all three of these sentences are strange. But strangeness has many sources, and — as has been argued elsewhere (Moravcsik 1970) — there are strong methodological considerations suggesting that we should draw the line around what is grammatical very liberally. This amounts to saying that we should look for sources of strangeness elsewhere, and set up as precisely as we can criteria of what counts as grammatical. E.g. with regard to declaratives, if we can state the conditions under which what a sentences expresses is true, then the sentence is grammatical. With regard to example 1), we should note that doubts can be entertained privately or publicly. Thus there is no reason to suppose that Tom could not doubt something, that this doubt would not be noticed one way or another by other members of the linguistic community, and that the doubt could not be not shared by anyone. Thus the sentence is grammatical, and its strangeness — if any — arises from the fact that it is not indicated how others came to know about Tom's doubt. The strangeness of 2) is likewise not a question of grammaticality. The distinction between a reasonable promise and an unreasonable promise is not the same as the distinction between a grammatical sentence and an ungrammatical one. Under normal interpretation 2) expresses an unreasonable promise; but so does the proverbial sentence that indicates that the lover promises the stars for his beloved; and though the promises of lovers are often unreasonable, it will hardly do to regard them as ungrammatical. The strangeness of 3) turns on the fact that most of the time we do not regard praise as insult. But if someone were to regard it that way, he would simply show himself to be odd with regard to his relation to his fellow humans, rather than showing himself to be violating English. Through the centuries English had to accommodate itself to describing a variety of very strange social and psychological attitudes; it is doubtful that psychological abnormality could outstretch grammaticality. The difficulty is that apart from claims and counterclaims about introspective episodes there are no ways at present to settle issues of this type. Those of us who regard sentences 1) — 3) as grammatical will be regarded as deaf but long on imagination, while we describe our opponents as having perhaps sensitive ears but being definitely short on imagination. Inasmuch as we want to reformulate the structure of linguistic theory so that it should have the form of a scientific explanation, this exchange of metaphoric praise and insult is a scandal in current linguistic practice, and its replacement with more reliable tests is imperative. This discussion of the role of intuition in the treatment of syntactic phenomena leads us to the comparison of syntax and semantics, and the consideration of possible deep differences between the two. There are at least three crucial differences between syntax and semantics. One of these is their different roles with respect to non-linguistic context-sensitivity. We

LINGUISTICS AND PHILOSOPHY

23

saw above, that in order to interpret correctly the referring roles of singular terms under certain circumstances, the non-linguistic context has to be taken into account. The question arises: can we find something analogous in the case of syntax? Are there sentences such that the interpretation of the syntactic structure of one of their tokens depends on matters of extra-linguistic context? If the answer to this question is negative, as the author of this paper suspects it is, then we have found an important difference between syntax and semantics. The second difference arises with regard to the role of intuition. In the case of syntax, there are well founded arguments on the basis of which we postulate intuitions both concerning grammaticality and concerning the internal syntactic structure of that which is grammatical. These intuitions play a key role in syntactic theory since these are the data which are to be accounted for. Do intuitions play an analogous role in semantic theory? And if so, what are the relevant intuitions? Although this question has not been treated explicitly in the literature, the philosophic debates involving Quine and others, are relevant to this topic. For example, on the basis of recent and current discussions of synonymy in natural languages it seems most implausible to suppose that claims of synonymy or paraphrase should be construed as accounting for phenomena that are intuitive in the same way as the syntactic phenomena are. Again, there seems to be no need for a semantic intuition of well-formedness in addition to grammatical well-formedness. There are two candidates for the status of semantic intuition. One of these is an intuition of validity as applied to deductive arguments. Since such arguments are represented in natural language, these intuitions need to be captured by semantics. The other candidate is an intuition that entities of certain sorts fall under the denotation range of a given predicate. In any case, it seems that with respect to the role of intuitions semantics and syntax reveal important differences. The third point of comparison involves the distinction between matters of linguistic fact and matters of widely shared beliefs about the world. Quine claims that this dictinction is difficult to maintain with regard to such phenomena as synonymy and meaning-inclusion. Can one present an analogous argument concerning matters of syntax? Although there exists no argument to this effect in the current literature, some of the material presented in a recent paper by Lakoff (1969:115-6) could be used for this purpose. Lakoff argues that grammaticality is to a large extent based on presuppositions about the world shared by members of a linguistic community. Although Lakoff does not think that this should lead one to deny the distinction between linguistic and extra-linguistic facts, someone might try to draw such a conclusion from Lakoffs evidence. This evidence, however, is limited to matters involving features like Animate and Human. It is fairly obvious that questions of what is and what is not capable of performing tasks that humans perform, and what is or is not animate and thus a possible object for use, worship, etc., involve empirical beliefs about the world. Consequently any deviance from grammatical rules based on Animate and Human allow in principle two interpretations.

24

J. M. E. MORAVCSIK

Either a mistake has been made, or a new hypothesis concerning certain entities in the world has been formulated, perhaps implicitly. It is worth noting, however, that these features affect a relatively small portion of syntax. Lakoff speculates (1969:116) that not much of interest would be left in 'presuppositionless syntax'. This is false, however, unless one can show that the distinctions between concrete and non-concrete particulars, between count + and - , between proper name and general term, as well as between categories like noun, verb, etc., are also based on empirical assumptions about the world. But what change in our empirical beliefs could affect the above mentioned distinctions? Thus as long as these are independent of our beliefs about the external world, that part of grammar that makes statements of identity, individuation, and predication possible remains non-linguistic context-free. Thus with regard to this point too, semantics and syntax differ. These differences should be brought to bear upon points that cannot be discussed within the confines of this paper, e.g. comparison of the explanatory power of syntactic and semantic categories, the claim that semantic transformations are formal objects on par with grammatical transformations, etc. The observation that at present a detailed comparison between syntactic and semantic categories is not possible leads us to an important lacuna in current transformational theory. This lacuna is related to the question of the explanatory power of explanations within transformational grammar. Presumably the syntactic aspect of the capacity of a speaker-hearer to understand a sentence is to be explained in terms of his ability to recognize what is or is not grammatical, and his ability to assign a structural description to the sentence. The elements of a structural description will include the syntactic categories. This means that until we arrive at an adequate analysis of the syntactic categories, our explanation of the ability to understand a sentence is no better than the explanation offered by Moliére's doctor of the capacity of opium to put people to sleep by reference to its dormitive power. With regard to an adequate analysis of syntactic categories we must distinguish between the demand for a language-independent definition and the demand for universality. An analogy should make the difference clear. Suppose that we have to give a different account of the syntactic categories for each natural language. This would amount to saying that we can give only 'local' accounts of these categories and that no 'global' account is available. An analogous case would be if we could define only phrases like 'gold in Siberia' or 'gold in Canada', but not gold in general. This would be a scandal for science. It is obvious, then, that in order to have a viable concept of a syntactic category in linguistic explanations, we need a global account. But just as the availability of a global account of gold does not imply that there is gold to be found everywhere, so the availability of global accounts of the syntactic categories would not guarantee universality. The latter is an additional and stronger hypothesis. It is required for our conception of language as a biological phenomenon, as we saw above, but it is not required for the mere respectability of explanations in linguistics as scientific hypotheses. The current difficulty is that apart from

LINGUISTICS AND PHILOSOPHY

25

some of Chomsky's earlier programmatic comments, and a few useful suggestions in Aspects, one cannot find either constructive work on these matters within the current literature nor even any sign of an awareness of how crucial this problem is for linguistics as a science. Until this problem is tackled, linguistic accounts will have to function as if they were full-blown scientific explanations, and the danger of ad hoc, or vacuous, or near-vacuous explanations seeping into the literature remains considerable. Let us consider in this connection two examples from the recent literature. In his recent papers Lakoff claimed that quantifiers must be treated as a kind of adjective. It turns out, however, that the quantifiers that are fundamental for quantification theory ('some', 'all', 'every') — in contrast to other quantifiers like 'few' and 'numerous' — are utterly unlike adjectives. They carry no relevant selectional restrictions, they do not admit molecular conjunction, etc. So what are we to say? Why call them adjectives? Why not regard them as 'sui generis'? In view of the fact that there is no general definition available for Adjective, one does not know what counts as a conclusive argument one way or another; thus Lakoffs claim remains outside the sphere of claims with specific empirical content that could be confirmed or disconfirmed. Another relevant example is the so-called 'performative analysis' proposed by John R. Ross for declarative sentences (Ross 1970). How is one to assess the claim that in the deep structures of all declarative sentences there are performative elements that get deleted in most cases? To what extent does such a proposal have empirical content? It is curious that Ross makes this general claim, since all of the difficulties that he seeks to solve arise in connection with sentences containing indexicals. Philosophers have recognized for a long time that there is an important distinction between indexical and non-indexical sentences. Why should this distinction not be applied to Ross's work? Why is it that Ross's discussion takes no cognizance of recent attempts to work out the semantics of indexicals? The practice of postulating all kinds of elements in deep structure that have no 'traces' in the sentence itself at all creates more problems than it solves. For it casts doubt on the whole enterprise as a legitimate, empirical, and non-vacuous way of explaining a part of linguistic competence. If Ross were right, then the view of language that we attributed to Frege earlier would be wrong. But there is nothing in Ross's paper that could be construed as arguments supporting the claim that contrary to Frege the fundamental part of our language is not context-free. To illustrate the danger of adopting unclear and undefined notions let us consider the recent mushrooming of the notion of presupposition in linguistic literature. The notion was first introduced in the philosophical literature on reference by Strawson (1950) and the utility of the notion with regard to these matters is still a subject of debate. This debate has not been joined by linguists; instead we find the uncritical employment of the notion in a number of contexts. It has been suggested by Lakoff ('unpublished discussion of quantifiers') that 1) 'Pedro regrets

26

J. M. E. MORAVCSIK

being a Norwegian' presupposes that Pedro is a Norwegian. In fact, there is no logical relation between the two propositions in question. Whether Pedro regrets being a Norwegian has nothing to do with his being or not being a Norwegian. It has to do, however, with his believing that he is a Norwegian. Once we see this, we realize that there is no need for a new relation called presupposition, since 1) logically entails that Pedro believes himself to be a Norwegian. Again, we are told that 2) 'Sam's murderer reads the Reader's Digest' presupposes that Sam was murdered. But brief reflexion should convince one that in this case t09, the customary notion of entailment will suffice. For 2) entails that Sam was murdered, and if we succeed under unusual circumstances to make a true statement by using 2) even though Sam was not murdered, then this is possible only in view of the fact that 'Sam's murderer' can be given different interpretations in context (see Donnellan 1966). Something having the semblance of an argument is presented occasionally in linguistic discussion for presupposition by claiming that it helps to explain negation. Thus we are told that 'John is a bachelor' presupposes 'John is an adult male' and asserts that John is not married. As support for this interpretation we are told that 'John is not a bachelor' means 'John is an adult male who is not unmarried'. But this last claim certainly cannot be regarded as a rule of English; rather, it is a statistical correlation of some sort, to be noted as a matter of performance, not of competence. 'John is not a bachelor' says that John does not satisfy the criteria for being an unmarried adult male. He may fail, of course, in a variety of ways. He may fail because he is married, or he may fail because he is still a baby. The fact that most of the time it is the former rather than the latter type of failure that underlies our ordinary negation is not a matter of law but a matter of statistical correlation, and it is not one of those facts the knowledge of which constitutes knowing English; rather, it is a fact the knowlege of which facilitates communication and other pragmatic skills. Finally, let us see if the notion of presupposition is really needed in order to explain the difference in meaning between a pair like 'accuse' and 'criticize'. Below we shall present an analysis that accounts for the differences without invoking the notion of presupposition. Consider two examples: a) 'John accuses Bill of having stolen the money', and b) 'John criticizes Bill for having stolen the money'. Both a) and b) entail 'John believes that Bill stole the money', and both entail 'John believes that Bill's having stolen the money was bad'. Needless to say, neither a) nor b) entail or presuppose 'Bill stole the money'. The difference between 'accuse' and 'criticize' can be brought out by pointing out that a) entails 'John believes that Bill's having stolen the money is morally wrong, or violates a law' while b) does not entail this. What b) entails can be perhaps expressed by 'John believes that Bill's having stolen the money was a performance that fails to meet certain standards', where the standards in question can be technical, prudential, a matter of limited ends and means, etc. To accuse is to assert a belief that someone is guilty

LINGUISTICS AND PHILOSOPHY

27

of something, to criticize is to assert the belief that someone made a mistake; guilt and mistake are different notions. This explains why it sounds natural to say: 'John accused Bill of murder' and odd to say 'John criticized Bill for murder' though the latter makes perfectly good sense in the context in which we are assessing the fulfillment of a mission by someone, and a murder can be regarded as endangering or damaging in some other way the end for which the mission was designed. Within the confines of this paper we cannot survey each use of 'presuppose'. Hopefully, enough has been shown to convince linguists that the introduction of this notion does not solve interesting questions in semantics, and that one should restrict oneself to relatively clear and well defined notions that have been developed in logic, and not introduce new relations unless absolutely necessary. Once such introductions are made, they must be accompanied by rigorous definitions, and explanations comparing the formal properties of the new notion to the already accepted ones. In conclusion we shall take up recent formulations of anti-Whorfian hypotheses and see how the methodological remarks made above bear on these. Both Ross (1970) and Bach (1968 : 121-2) seem to regard it as an advantage if deep structure turns out to be more abstract than we think, and they think that on a very abstract level of deep structure all languages can be shown to have the same universal grammatical structure. They think that on an equally abstract level the semantics of all natural languages can be shown to be universal. The use of the word 'abstract' in this context is rather misleading. The issue is not abstractness; that in itself is neither good nor bad. E.g. mathematics is very abstract, but this fact has no bearing on its value. In the context of linguistic theory the issue centers around the notion of a structure being very remote from surface structure. Thus for 'abstract' as Ross and others are using this term, we should substitute 'remote from surface structure'. The methodological remarks made above should bring out the point that there is nothing good about deep structure being very remote from surface structure. On the contrary, the more remote, the less clear it becomes that we are making empirical claims. Given the extremely powerful notion of a transformation, as defined so far, it is not in the least surprizing that if one makes deep structure sufficiently remote from surface structure, then one should achieve universality. We are operating here with a much too powerful notion, i.e. a transformation, that has not yet been sufficiently restricted, and this notion ranges over phrase-markers, which are entities containing undefined syntactic categories as their elements. Under these circumstances, the more 'abstract' (i.e. remote from surface structure) deep structure is, the more vacuous and trivial the claim of universality is. Linguists must be careful lest their zeal for generalizations of ever widening scope leads them to vacuousness and ad hoc explanations. With regard to semantics, the claim of universality was first formulated by Tarski (1936). His remarks were made, however, in the context of comparing formal languages, ones designed to express a certain theory, with informal languages, i.e. those within which we can

28

J. M. E. MORAVCSIK

express a variety of different theories. His remarks cannot be applied, without additional premisses to a comparison of the semantics of different natural languages. This latter comparison will become more fruitful at some future time when the conceptual framework of linguistic theory is more clearly defined, and its relation to empirical evidence more clearly spelled out. To sum up, this essay is designed to show the different kinds of interests that philosophers have in language, the ways in which their semantic analysis is relevant to the work of the linguists, and ways in which distinctions from linguistic theory can be applied profitably by philosophers to their own work. Finally, in the last part some methodological suggestions have been made from a philosophical point of view. The aim of these suggestions, as well as the implicit encouragement to linguists and philosophers to borrow from each other, is to contribute toward achieving the formulation of an adequate theory of language. While such a formulation is not yet within our grasp, a survey of recent work both by linguists and philosophers cannot help but impress one with the relative amount of progress made. Further empirical work, a few more conceptual reforms, coupled with awareness of the requirements of scientific methodology, should make the future work of linguists and philosophers in this area rewarding.

SELECTED A N D TOPICALLY ARRANGED BIBLIOGRAPHY FOR LINGUISTS ON LOGIC A N D THE PHILOSOPHY OF LANGUAGE

1) Analyticity and synonymy. Carnap Donnellan 1962 Harman 1967 Lyons Mates Moravcsik 1965 Putnam 1962a, 1962b Quine 1953,1960 Wang 2) Epistemic Terms and Opacity. F0llesdal Frege Hintikka 1962, 1969a Kaplan Mates Quine 1960 3) Events, Verbs, and Tense. Davidson 1967,1969

LINGUISTICS AND PHILOSOPHY

29

Martin Prior Vendler 4) Fregean Theory of Meaning, Church 1951 Dummett Frege Furth Montague, 1970a, 1970b 5) Logic and Foundations of Mathematics. Sub-topics: i. First-order logic ii. Gentzen-type systems for first-order logic iii. Intuitionistic logic and model theory iv. Second-order logic v. Higher order logic vi. Basic notions of recursive function theory and Gòdel's theorems vii. Advanced material on recursion theory viii. Advanced material on model theory ix. Mathematical intuitionism Bell and Slomson i, viii Beth i, iii, vi, ix Brouwer ix Church 1956 i, iv Godei i, ii Hintikka 1969b i-ix Kleene i, ii, iii, vi, vii Kreisel 1965 ix; 1967 ix; and Krivine i, iv, v Mendelson i, ii, vi Mostowski esp. ix Myhillix Prawitz ii Robinson viii Rogers vii Shoenfield i, vi, vii, viii Strawson 1952 i Suppes 1957,1960 i, ii Troelstra ix 6) Logical Modalities. Hintikka 1967 Hughes and Cresswell Kripke (also ix in Logic)

30

J . M. E. MORAVCSIK

Lewis Montague 1970b 7) Mathematical Foundations of Linguistic Theory. Chomsky 1963 Davis Ginsburg and Partee Hopcroft and Ullman 8) Semantics of non-declarative sentences. Harrah Opacity see Epistemic Terms 9) Semantic Theory (general). Bar-Hillel Carnap Cartwright Caton Linsky Pap Quine 1960 Rogers Synthese 1970 Tarski Wittgenstein Ziff 10) Semantics of English (specific). Alston Austin Hall Urmson 11) Singular Terms (proper names, subject-predicate distinction, etc.). Burks Donnellan 1966,1968 Geach MacKay Quine 1953,1960,1968 Russell 1903, 1905 Searle Strawson 1950,1968 Synonymy see Analyticity 12) Theory of Knowledge for Linguistics. Chomsky 1968a, 1968b, 1969 Fodor Harman 1969 Moravcsik 1967,1969

LINGUISTICS AND PHILOSOPHY

31

Morgenbesser Nagel 1969a, 1969b 13) Vagueness and Openness. Black 1937,1963 Goguen Lyons Waismann

REFERENCES

Expressing. Philosophy in America, ed. by Max Black, 1 5 - 3 4 . Ithaca, Cornell University Press. AUSTIN, JOHN L. 1 9 6 1 . Philosophical papers. Oxford, Clarendon Press. BACH, EMMON. 1 9 6 8 . Nouns and noun phrases. Universals in linguistic theory ed. by E . Bach and R. Harms, 9 1 - 1 2 2 . New York, Holt, Rinehart, Winston. BAR-HILLEL, YEHOSHUA. 1 9 5 4 . Logical syntax and semantics. Lg 3 0 . 2 3 0 - 7 . BELL, J . , and A. SLOMSON. 1 9 6 9 . Theory of models. Amsterdam, North Holland. BENACERRAF, PAUL, and HILARY PUTNAM, eds. 1 9 6 4 . Philosophy of mathematics. Englewood, Prentice-Hall. BETH, E. W. 1959. Foundations of mathematics. Amsterdam, North Holland. BLACK, MAX. 1 9 3 7 . Vagueness. Philosophy of Science 4 . 4 2 7 - 5 5 . . 1963. Reasoning with loose concepts. Dialogue 2.1-12. BROUWER, L. E. J. 1964. Intuitionism and formalism. In Benacerraf and Putnam 1964:66-77. BURKS, ARTHUR W . 1 9 5 1 . A theory of proper names. Philosophical Studies ALSTON, WILLIAM.

1965.

2.36-45.

1956. Meaning and necessity. Chicago, University of Chicago. 1962. Propositions. Analytical philosophy, ed. by R . Butler, 81-103. Oxford, Blackwell. CATON, CHARLES, ed. 1963. Philosophy and ordinary language. Urbana, Univer. of Illinois. CHOMSKY, NOAM. 1 9 5 5 . Logical syntax and semantics; their linguistic relevance. CARNAP, RUDOLF.

CARTWRIGHT, RICHARD.

Lg 31.36-45.

. 1963. Formal properties of grammars. Handbook in mathematical psychology, vol. II. ed. by P. Luce, R. Bush, and E. Galanther, 323-418. New York, Wiley. . 1965. Aspects of the theory of syntax. Cambridge, MIT Press. . 1968a. Language and mind. New York, Harcourt, Brace, World. . 1968b. Quine's empirical assumptions. Synthese 19.53-68. . 1969. Linguistics and philosophy, Comments on Harman's Reply. In Hook 1969:51-94, 152-9.

32

J. M. E. MORAVCSIK

. 1971. Deep structure, surface structure, and semantic interpretation. Studies in general and oriental linguistics presented to Shiro Hattori on the occasion of his sixtieth birthday, ed. by Roman Jakobson and Shigeo Kawamoto, 59-91. Tokyo, TEC Company. CHURCH, ALONZO. 1951. A formulation on the logic of sense and denotation. Structure, method, and meaning, ed. by Paul Henle, H. M. Kallen, and S. K. Langer, 3-24. New York, Liberal Arts. . 1956. Introduction to mathematical logic. Princeton, Princeton U.P. CROSSLEY, J . N., and M . A. DUMMETT, eds. 1 9 6 5 . Formal systems and recursive functions. Amsterdam, North Holland. DAVIDSON, DONALD. 1967. The logical form of action sentences. The logic of decision and action, ed. by N. Rescher, 81-120. Pittsburgh, University of Pittsburgh. . 1969. On events and event descriptions. Facts and existence, ed. by J. Margolis, 75-83. Oxford, Blackwell. DAVIS, MARTIN. 1 9 5 8 . Computability and unsolvability. New York, McGraw-Hill. DONNELLAN, KEITH. 1 9 6 2 . Necessity and criteria. Journal of Philosophy 5 9 . 6 4 7 58.

. 1966. Reference and definite descriptions. Philosophical Review 75.281304. . 1968. Putting Humpty Dumpty together again. Philosophical Review 77.203-215. DUMMETT, MICHAEL A. 1967. Frege. The encyclopedia of philosophy vol. 3. 225-37. New York, McMillan. FODOR, J. 1 9 6 8 . An appeal to tacit knowledge in psychological explanation. Journal of Philosophy 6 5 . 6 2 7 - 4 0 . F0LLESDAL, DAGFINN. 1 9 6 7 . Knowledge, indentity, and existence. Theoria 3 3 . - 2 7 . FREGE, GOTTLOB. 1 8 9 2 ( 1 9 5 2 ) . Philosophical writings. New York, Philosophical Library. FURTH, MONTGOMERY. 1 9 6 5 . Introduction to G . Frege, The basic laws of arithmetic. Berkeley, University of California. GEACH, PETER T . 1 9 6 2 . Reference and generality. Ithaca, Cornell University. GINSBURG, S., and B. PARTEE. 1969. A mathematical model of transformational grammars Information and Control 15.297-334. GODEL, KURT. 1 9 6 4 . Russell's mathematical logic. In Benacerraf and Putnam 1964:211-32.

J. A. 1969. The logic of inexact concepts. Synthese 19.325-73. 1 9 5 9 . Excluders. Analysis 2 0 . 1 - 7 . HARMAN, GILBERT. 1 9 6 7 . Quine on meaning and existence. Review of Metaphysics 2 1 . 1 2 4 - 5 1 , 3 4 3 - 6 7 . . 1969. Linguistic competence and empiricism. In Hook 1969:143-51. HARRAH, DAVID. 1963. Communication. Cambridge, MIT Press. HINTIKKA, JAAKKO. 1 9 6 2 . Knowledge and belief. Ithaca, Cornell University. GOGUEN,

HALL, ROLAND.

LINGUISTICS AND PHILOSOPHY

33

. 1967. Individuals, possible worlds, and epistemic logic. Nous 1.33-62. . 1969a. Models for modality. Dordrecht, D. Reidel. , ed. 1969b. The philosophy of mathematics. London, Oxford University Press. HOOK, SIDNEY, ed. 1969. Language and philosophy. New York, New York University. HOPCROFT, JOHN, and JEFFREY ULLMAN. 1969. Formal languages and their relations to automata. Reading, Addison-Wesley. HUGHES, G . E . , and M . J. CRESSWELL. 1 9 6 8 . An introduction to modal logic. London, Methuen. KAPLAN, DAVID. 1968. Quantifying in. Synthese 19.178-214. KLEENE, S . 1 9 5 2 . Introduction to metamathematics. New York, Van Nostrand. KLIBANSKY, R . 1 9 6 8 . Contemporary philosophy. Florence, La Nuova Italia. KREISEL, GEORGE. 1 9 6 5 . Mathematical logic. Lectures on modern mathematics, ed. by Saaty, vol. 3 . 9 5 - 1 9 5 . New York, Wiley. . 1967. Mathematical logic: what has it done for the philosophy of mathematics. Bertrand Russel, philosopher of the century, ed. by R. Schoenman, 201-72. London, Allen & Unwin. KREISEL, GEORGE, and J. KRIVINE. 1967. Elements of mathematical logic. Amsterdam, North Holland. KRIPKE, SAUL. 1 9 6 5 . Semantical analysis of intuitionistic logic I . In Crossley and Dummett 1 9 6 5 : 9 2 - 1 3 0 . LAKOFF, GEORGE. 1969. Presuppositions and relative grammaticality. Studies in philosophical linguistics, ed. by W. Todd, 103-16. Evanston, Great Expectations. . 1971. On generative semantics. Semantics: An interdisciplinary reader in philosophy, linguistics, anthropology and psychology, ed. by D. D. Steinberg and L. A. Jakobovits, 232-96. Cambridge, Cambridge University Press. LAKOFF, GEORGE, and J. R. Ross. n.d. Is deep structure necessary? Unpublished. LEWIS, DAVID. 1 9 6 8 . Counterpart theory and quantified modal logic. Journal of Philosophy 6 5 . 1 1 3 - 2 6 . LINSKY, LEONARD, ed. 1952. Semantics. Urbana, University of Illinois. LYONS, JOHN. 1968. Introduction to theoretical linguistics. Cambridge, Cambridge Univ. Press. MACKAY, ALFRED F. 1968. Mr. Donnellan and Humpty Dumpty on referring. Philosophical Review 77.197-202. MARTIN, RICHARD. 1969. Events and event-descriptions. Fact and existence, ed. by J. Margolis, 63-74. Oxford, Blackwell. MATES, BENSON. 1 9 5 0 . ( 1 9 5 2 ) . Synonymy. In Linsky 1 9 5 2 : 1 1 1 - 3 8 . MENDELSON, ELLIOT. 1964. Introduction to mathematical logic. Princeton, Van Nostrand. MONTAGUE, RICHARD. 1970a. English as a formal language. Linguaggi nella società e nella technica, 189-223. Milano, Edizione di Communità.

J. M. E. MORAVCSIK

34

. 1970b. Pragmatics and intensional logic. Synthese, q.v. J. M . E. 1965. The analytic and the nonempirical. Journal of Philosophy 62.415-30. . 1967. Linguistic theory and the philosophy of language. FL 3.209-33. . 1969. Competence, creativity, and innateness. Philosophical Forum 1.40737. . 1970. Subcategorization and abstract terms. FL 6.473-87. MORGENBESSER, SIDNEY. 1 9 6 9 . Comments on Fodor ( 1 9 6 8 ) . Journal of PhilosoMORAVCSIK,

phy 66.459-72.

A. 1 9 6 6 . Thirty years of foundational studies. Acta Philosophica Fennica, vol. 17. Oxford, Blackwell. MYHILL, JOHN. 1 9 6 8 . Formalization of intuitionism. In Klibansky 1 9 6 8 : 1 . 3 2 4 MOSTOWSKI,

41.

1969a. Linguistics and epistemology. In Hook 1 9 6 9 : 1 7 1 - 8 2 . . 1969b. The boundaries of inner sense. Journal of Philosophy 66.452-8. PAP, ARTHUR. 1958. Semantics and necessary truth. New Haven, Yale. PRAWITZ, DAG. 1965. Natural deduction. Stockholm, Almquiest & Wiksell. PRIOR, ARTHUR. 1967. Past, present, and future. Oxford, Oxford University. PUTNAM, HILARY. 1962a. The analytic and the synthetic. Minnesota Studies in the Philosophy of Science, vol. IH.358-97. Minneapolis, University of Minnesota. . 1962b. It ain't necessarily so. Journal of Philosophy 59.659-65. QUINE, WILLARD V . O . 1 9 5 3 . From a logical point of view. Cambridge, Harvard University. . 1960. Word and object. New York, Wiley. . 1968. Replies. Synthese 19.264-322. ROBINSON, ABRAHAM. 1 9 6 8 . Model theory. In Klibansky 1 9 6 8 : 1 . 6 1 - 7 3 . ROGERS, HARTLEY. 1968. Theory of recursive functions and effective computability. New York, McGraw-Hill. ROGERS, ROBERT. 1963. A survey of formal semantics. Synthese 15.17-56. Ross, JOHN R. 1970. On declarative sentences. Readings in English transformational grammar, ed. by R. A. Jacobs and P. S. Rosenbaum, 222-72. Waltham, Mass. RUSSELL, BERTRAND. 1 9 0 3 . ( 1 9 5 2 ) . Descriptions. In Linsky 1 9 5 2 : 9 5 - 1 1 0 . . 1905. (1956). On denoting. Logic and knowledge, ed. by R. Marsh, 3956. London, Allen & Unwin. SEARLE, JOHN R . 1 9 5 8 . Proper names. Mind 6 7 . 1 6 6 - 7 8 . SHOENFIELD, JOSEPH R. 1 9 6 7 . Mathematical logic. Reading, Addison-Wesley. STRAWSON, PETER F. 1950. On Referring. Mind 59.320-44. . 1952. Introduction to logical theory. London, Methuen. . 1968. Singular terms and predication. Synthese 19.97-117. SUPPES, PATRICK C. 1 9 5 7 . Introduction to logic. Princeton, Van Nostrand. . 1960. Axiomatic set theory. Princeton, Van Nostrand. NAGEL, THOMAS.

LINGUISTICS AND PHILOSOPHY

35

Synthese. 1970. Issue on the semantics of natural languages. TARSKI, ALFRED. 1 9 3 6 . ( 1 9 5 6 ) . The concept of truth in formalized languages. Logic, semantics, metamathematics, 1 5 2 - 2 7 8 , Oxford, Oxford University. . 1952. The semantic conception of truth. In Linsky 1952:13-49. TROELSTRA, A N N E S. 1969. Principles of intuitionism. Berlin, Springer Verlag. URMSON, JAMES O. 1968. Criterion of intensionality. Aristotelian Society Suppl. vol. 42.107-22. VENDLER, ZENO. 1 9 5 7 . Verbs and times. Philosophical Review 6 6 . 1 4 3 - 6 0 . WAISMANN, F. 1 9 5 1 . Verifiability. Logic and language, first ser. ed. by A. Flew, 117-44. Oxford, Blackwell. WANG, H A O . 1955. Notes on the analytic-synthetic distinction. Theoria 21.15878. WITTGENSTEIN, LUDWIG. 1 9 5 3 . Philosophical investigations. New York, McMillan. ZIFF, PAUL. 1 9 6 0 . Semantic analysis. Ithaca, Cornell University.

ON LOGIC A N D THEORETICAL L I N G U I S T I C S

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

I.

INTRODUCTION

'Language, Truth, and Logic' has a familiar ring. Russell wished he had written the well-known book of that title; Ayer, no doubt, is grateful that he did. Our study, though more mundane, is perhaps a less devious treatment of these topics than Ayer's. It attempts to comprehend the multifarious interconnections among linguistic theory, the theory of truth, and logical theory. Linguistic and logical inquiry have had an affinity since classical times. The Middle Ages and several subsequent centuries saw a cultivation and enrichment of this relationship, though by the nineteenth and early twentieth centuries, it had chilled. As their fields gained academic independence, the linguist and the logician questioned the interest, relevance, and even comparability of the other's to his own work. Of late, a new cameraderie between the logician, especially the philosophical logician, and the linguist has emerged. This exchange of ideas has occurred at a time when their respective theories have progressed strikingly, and have achieved, if not a firm foundation, at least a clear direction. Integral to this development has been the concept of Form or Structure. Though not new in its centrality to either field, it has only recently become theoreticaly specifiable in a way which encourages the kind of comparative understanding we are seeking. It will occupy center stage in what follows. One note of explanation: conspicuously lacking in our study is a detailed discussion of the pragmatic element in language and argumentation. The lack is due not to disdain nor to a paucity of interesting work in the area (for some recent work see Bar-Hillel, ed., 1971), but to a shortage of space only. II.

SOME PRELIMINARY CONSIDERATIONS

What then are the interconnections among the structures proposed respectively by the linguist and the logician? The terms of the question demand refinement. But first we must consider briefly some claims which, if true, would seem to make further discussion pointless. The first such claim is that logic, being normative, can have no significant

38

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

overlap with linguistics which is descriptive. Ignoring the questionable clarity of the normative-descriptive dichotomy, we are ready to grant in some sense of these terms that logic is normative, and linguistics descriptive. But we must also acknowledge that, in the sense in which logic is normative, so is linguistics, and in the sense in which linguistics is descriptive, so is logic. Consider first the normative character of logic. We start with an infinite set of logical truths of ordinary language. Among them are the conditionals, whose validity when valid permits us to describe the antecedent as entailing the consequent. Our logical theory will be adequate only if it generates all and only the members of this initial set. In virtue of these truths, for example that A entails B, we talk of what sentences a person SHOULD accept, and how he SHOULD reason; that he SHOULD accept B if he accepts A or conclude B from A. A weaker conception of the norms would require only that he should not accept not-B if he accepts A. Logic then is normative insofar as it defines the ideal reasoner, insofar, that is, as it specifies through its rules the competence that such a reasoner should possess.1 Expressed in this way, the parallel normative character of linguistics is apparent. Corresponding to the set of logical truths is the set of grammatical sentences. In terms of these sentences we talk of how people SHOULD speak, we criticize and correct the ways in which they do speak, and we specify the competence of the ideal speaker-hearer. Turning now from competence to performance, we can locate the descriptiveness of both linguistics and logic. It would be rather absurd, although it is unclear to what extent, to claim to have specified the grammar of a language L, and yet acknowledge that the speakers of L never perform according to its standards. We want a grammar both to answer to the linguistic intuitions of the ideal native speakerhearer, and to count as grammatical many, perhaps most of the utterances of those who know and perform in the language. 'Strings heard from natives count as wellformed, at least provisionally' (Quine 1970:389).2 The normative grammatical theory alone need not be an adequate model of performance in the weakest sense of having its output match the performance output. It certainly will be part of a theory of this performance, however, insofar as it produces a partially realized 1

The problem of relating norms of reasoning to truths of logic is notorious. Our vague formulation 'in virtue of is meant to avoid any commitment to a specific relationship and to deny that the relationship can be viewed as itself a logical one. Whatever the relationship, however, certain distinctions must be made in formulating the norms. Thus the ideal reasoner should accept B if he accepts A) though it need not be the case that he should accept B (if he accepts A). — Among the issues on which the ultimate resolution of the entire question rests are: (I) failure of substitutivity in knowledge and belief contexts, (II) the relation between truths of logic and empirical facts, e.g. that the output of a certain computer is such and such, and (III) the relationship of norms to what a person believes. — We are indebted to S. Morgenbesser for the need to express caution here, though he may not agree with the specific cautions expressed. 8 The behavioral expression of linguistic intuition is also an element in performance and part of the test. But in another sense, we want to contrast what we know to be right and what we do. In this sense the intuition is not part of the performance, but functions as competence.

ON LOGIC AND THEORETICAL LINGUISTICS

39

description of it, to the extent that the performance conforms to the rules of the theory in this weak sense. Chomsky has introduced a deeper sense in which linguistic theory is descriptive (his second sense of 'theory'); namely, it is descriptive of the tacit knowledge of the native. The competence of the native will then be described in terms of the competence of the ideal speaker, and the performance of the native in terms of his competence. This is a deeper sense of 'description', because it does not merely claim to generate the observations of performance but to describe an element in the process of the native's performance.3 (Contrast the laws governing falling bodies as a description of the 'performance' of Newton's apple.) Other elements, further mechanisms, need also to be introduced to explain performance; for example, attention, motivation, and memory. An account of the application of this framework to logic is unnecessary. That the various levels of description presented here arise also in discussing the logical performance of reasoners seems plain enough. Logic, thus viewed, becomes part and parcel of a description of reasoning behavior and, in a deeper sense, of the mechanics of this behavior. Until now we have described the output of linguistic theory in terms of grammaticality only. Similarly, logical truth marked the domain of logical theory. Chomsky, of course, views grammaticality as only a small part of the adequacy conditions a linguistic theory must fulfill. In contrast to weak generative capacity, the capacity to generate all and only the grammatical strings, stands strong generative capacity, fulfillment of which assures that the strings are appropriately marked. This addition of structural description has a place in each major part of our characterization of linguistics. In these terms this means that (a) in addition to the basic norms associated with grammaticality, there are standards of labelling and bracketing to which we should conform; (b) our model must partly fit the actual performance of the natives in this endeavor; and (c) the natives will, presumably, tacitly know the right structural description of the sentences of the language, and this knowledge will figure prominently in the linguistic process. Using these terms may not be a fair way of extending the adequacy conditions of linguistics. Thus Quine objects to the idea that a native tacitly knows one of two weakly equivalent theories as opposed to the other. 4 This idea is very intimately tied to the interest in getting the strong generative capacity right. Were the notion of the right structural description given new behavioral content, at least we would know clearly whether the native's performance, other than his parsing, fits one of the two theories and not the other. Nonetheless, it is central in Chomsky's scheme 3

Talk of process (and later of mechanisms) should not be taken to suggest that from a description of the output of the theory anything specific follows about the nature of the process or mechanism. More generally, we must repeat that these preliminary considerations are meant only to head off very general objections to our concern and are not meant to be serious discussions of the often controversial topics they mention. 4 But see also Quine's comments on Geach (1970 : 391).

40

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

that we talk of one such grammar, picked out presumably by further constraints, viz. Universal Linguistic Theory, as the one internalized by the native. Does this force a divergence of logic and grammar? Janet Fodor writes: For the logician, the existence of more than one type of logical system dealing with the same subject matter is not disturbing (although he may be interested in comparing and contrasting some of their formal properties), but many linguists argue that there may be just one correct grammar for any language and that, given any two proposals, it should be possible to determine which of them is the more correct, either on the basis of further empirical data or else by reference to their relative simplicity or conformity to general universal principles concerning linguistic structures (1970: 209).

Though this argument seems to play off the logician against the linguist, it in fact only catches them in different moods. First of all, the variety of logical systems is certainly matchable by a plurality of grammars — at least at the level of weak generative capacity. If now we raise the stakes and claim a need for strong generative capacity to characterize linguistic knowledge, what is to keep us from similarly specifying greater constraints on an acceptable logical theory to characterize logical knowledge? In short, the psychologician will be as particular as the psycholinguist, and the formal (weak) logician as tolerant as the formal linguist.5 Finally, we should comment briefly on the assumption that the native tacitly knows or internalizes one grammar and/or one logical theory. This idea must face whatever strength there is in the objection implicit in Quine's notion of indeterminacy of translation. If Quine is right, there is no sense in talking about THE CORRECT linguistic or logical theory of a language L; and there is no fact of the matter about the logical or linguistic structure of a sentence. Our use of simplicity and other considerations in developing a theory is all well and good, but it does not get us closer to any reality — and in particular, no closer to the reality of the native's knowledge. This indeterminacy, if there is one, need not result in a parallel neurophysiological indeterminacy. Neurophysiological theory is one with scientific theory, towards which even Quine manages a realistic attitude. Further, it does not rule out the idea of a neurophysiological explanation of linguistic knowledge. Rather it says something about the relation between the description of mentalistic and physiological linguistic theory. • Before turning to the second claim, there is a further sense in which logic may be normative and linguistics not. This is the sense in which the language of logic might be considered an ideal language. The notion of ideal language has been closely connected with that of logical form from the time of Russell and Wittgenstein. Since we shall discuss this notion below in the section on logical form, we can be brief here: An ideal language, if it conflicts in an interesting way with the 5

It should be noted that Janet Fodor's comments were not originally intended for the purpose to which we put them. Her point is relevant to the discussion in the final section of the paper. ' For some interesting arguments related to this, see Davidson 1970a.

ON LOGIC AND THEORETICAL LINGUISTICS

41

sense of logic given above as a theory of logical truths of ordinary language7 must be a replacement of ordinary language as opposed to a characterization of its logical properties. This distinction is embodied in great part in the properties of the scheme which correlates the ideal with the ordinary language. But a new language will not qualify as a replacement merely because it is a sub-part of the ordinary language, or differs from the ordinary language in surface features, or finally contains a replacement of the non-logical, lexical elements of the ordinary language. That there are ideal replacement languages developed for the use of science may be clear. But it is also clear that everything which may appear to be such is not. In any case, there are logics which do attempt to be descriptive logics of ordinary language, and this is enough to justify further investigation. Having responded to the 'normative-descriptive' objection, we are still not quite free. A parallel and perhaps interconnected objection (images of non-Euclidean geometry(?)) might be raised in the form: Logical theory is language-independent, known a priori, and necessary, while linguistics is language-dependent, known a posteriori, and contingent. Still, we think there is a sense in which logic and linguistics share some of these characteristics. In seeing how this is so, a point of Quine's is worth noting. 'Grammar,' he says, 'is linguistic on purpose' (1970:15). Logic is not. In logic we become involved with language because of the technical difficulties in generalizing directly. We soon realize that generalities must be expressed in terms of truth and truth-bearers. Such semantic ascent is not mere quirk as would be the habit of a scientist who, instead of asserting sentences, insisted on saying that they were true. It is a necessity. On the surface, to ascend to truth is not to be language-dependent. After all, the device of a proposition is just a way of freeing ourselves from specific languages.8 If we grant they exist, then surely it is for the purpose, among others, of bearing truth. Furthermore, even if the need to talk not only of truth but also of form eliminates propositions, as ordinarily conceived, from the competition (of truth-bearers in logic), there are other entities equally language-free to take their place. But this is all beside the point, for any truth-bearer we choose, even if not sentences in specific languages, will have a clear relationship to sentences in specific languages. Further, given that in logic we must talk of truth-bearers, it remains plausible to see in this talk an extension of the domain of logic to include discussion of the relation between sentences in language to these truth-bearers. In other words, logic, however characterizable independently of language, has an 'application' to particular languages. In this process we meet the claims which invite comparison to linguistics; e.g. S in L has form F, S is a translation of S ' . . . . What now of being necessary and a priori? Granting the concepts, we will This characterization of logic in terms of logical truths of ordinary language makes irrelevant to our discussion pure mathematico-logical theories. 8 Propositions are sometimes conceived in such a way as to free us from all language. 7

42

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGAUT

likely admit a sense in which both these concepts apply in logic and not in linguistics. But, again, the contraries, being contingent and known a posteriori, apply also in logic, and present us with the field for comparison. This point may be clarified as follows: As might be expected, the application of these contrary terms is just to those sentences of logic which are 'linguistic' in character. The characterization of the form of S in L or the translation relation holding between S and S 1 , both in L, are presumably descriptions of contingencies known a posteriori. And, of course, it is just when we fix our glance on this element of logic that we begin to wonder about its relation to the various elements of a grammar. Furthermore, claims like these constitute the network which interrelates logic as formulated explicitly in terms of sentences, with logic formulated in non-explicitly linguistic terms. Finally, we must not overestimate the language-dependence of linguistics as presently conceived by many, especially those influenced by Chomsky. The idea of language universals has become a basic addition to the theoretical vocabulary of the linguist, and in the case of Chomsky, the concept of universals is integral to the notions of the descriptive and explanatory adequacy of a linguistic theory. Since we shall not return to the question of universals we shall remark about it briefly here. Semantic universals invoked in the context of the semantic component9 emerge in talk of non-linguistic entities, concepts, thoughts, readings, and in the possibly identical idea of language-neutral semantic representations. Thus, whatever moment there is to objections of such talk in logic, or to such talk independently of translation schemes between sentences or other linguistic expressions, applies to the acceptability of such talk in the context of semantic universals. More generally, care must be taken in formulating hypotheses of linguistic universals, if they are no£ to be somewhat hollow empirically. In the case of syntax, we must avoid interpreting purely grammatical categories and constructions along with the structuralist, immanently, lest there be no sense in asking about universals.10 In the case of semantics, we must avoid collapsing the claim to universality onto a claim of the intertranslatability of all languages, at least if the universal categories and constructions are transferred by translation which itself fails of objectivity. Regarding semantics (assuming Quine views predication as a fundamental construction for semantics and not just syntax), Quine writes: Someone says, let us suppose, that the subject-predicate construction occurs in all the languages he has examined. N o w of course all those languages have been translated, however forcibly, into English and vice versa. Point, then, in those languages to the translations of the English subject-predicate construction, and you establish the thesis; Or simply the grammar, if generative semantics is accepted. On this issue, see below. Quine is a superstructuralist in the sense that not only are the particular categories and constructions immanent, but so also are the notions of a category and a construction themselves.

9

10

ON LOGIC AND THEORETICAL LINGUISTICS

43

the subject-predicate construction occurs in all those languages. Or is it imposed by translation? What is the difference? Does the thesis say more than that basic English is translatable into all those languages? And what does even this latter claim amount to, pending some standard of faithfulness and objectivity of translation? To make proper sense of the hypothesis that the subject-predicate construction is a linguistic universal, we need an unequivocal behavioral criterion of subject and predicate (1970: 390).

These last points apply in an interesting way to the universality of logic as well. Since logic has a precarious relation to language, it is not surprising that talk of the universality of logic modelled on its linguistic counterpart is precarious. In one sense, the 'linguistic' universality of logic is as straightforward as the 'linguistic' universality of physics. Thus, if someone asks whether the physics expressed in our language is true in some other language, the obvious answer is that the sentences of the other language which express the physical truths, expressed in our language, are true. Similarly, logic holds in any language which expresses the logical truths of our language. But since logic is handled in terms of forms of sentences, the question remains whether another language expresses the logical truths of our language as logical truths. This is just the question of whether the forms of sentences in our language are shared by sentences of other languages. If Quine's argument about the transferral of semantic categories and constructions holds, then a translation scheme from one language to another should be sufficient to answer this issue affirmatively. But caution is necessary in applying Quine's argument to our question. It is clear, for example, that just any translation of one sentence by another within a single scheme will not determine shared logical properties. Similarly, for Chomsky, all members of a maximal paraphrastic set will not have one deep structure. Thus we cannot simply argue that translation of S by S' imposes on S the logical properties, or logico-syntactic features, of S'. The fact that not all logical features turn up in every translation does not, of course, entail the falsity of the claims that every scheme will have some translation which transfers all properties, or that certain logical properties are transferred by every translation,11 or finally that constructions which are not transferred in some translations of some sentences will be transferred in other cases. These claims must await further exploration. A final question about the universality of logic, whether the speakers of other languages share our logic, should not be confused with any of the above questions. In a sense this question intersects oud earlier ones in the following way. If we consider the revisability of logic to be a coherent notion, then we will allow the possibility that a foreigner rejects our logic. We will continue to consider as true 11

Quine's example gained its plausibility because it is a construction which the grammar assigns to sentences at a very high level of generality.

44

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

the translation in his language of our logical truths, but will admit that he may not so view them. Alternatively, whether we accept revisability will determine our willingness to translate the foreigners's words in such a way that he consistently violates our logic, without adding that he is simply irrational. Quine's principle ot charity in these cases instructs us that other things being equal, we should maximize rationality. This principle by itself makes no commitment to the coherence of revisability. III.

REFINEMENT

Returning to our initial question of the interconnections among the structures proposed by linguists and logicians, we shall remark briefly on the main elements it involves. Representations are not to be confused with what they represent. Among the latter are meanings, truth-conditions, and structures (forms) of various kinds. Expressions of natural language presumably 'have' these entities, and in this sense may be considered representations of them. Linguistic and logical theories tend to come with their own representations, especially for the structures which they assign to the complex phrases of a language. Essential to these representations is the information familiarly given in brackets and labels. These incorporate the dual features of regulated composition and generality characteristic of the notion of form. Expressions of a language may or may not be perspicuous. Their perspicuity depends on the relation between the expression and its theoretical representation. It also sometimes depends on the further relation between the representation and the rules of the theory. We shall concentrate on logical, semantic, and syntactic structures. The first helps to account for logical truth and inference, the second, to explain how the meaning of a complex expression is composed from the meaning of its parts, and the third, to characterize grammaticality. To each type of structure is associated one or more types of rule. In seeking to interrelate these structures we are asking, with respect to each expression of a language, whether its various structures are identical, overlapping, completely disjoint. . . . We also must understand if these relations turn on the characterization of the structures themselves, or are contingently supported by available evidence. Either way, sensitivity to contingently related descriptions of these structures as well as to the theories in which they are developed is demanded. This need for sensitivity applies especially to the 'deep structures' of current linguistic theory over which controversy rages. IV. LOGICAL FORM

We shall begin by examining the concept of logical form as it applies to ordinary language. Specifically, we want to understand what is involved in talking about the logical form of a sentence of a natural language.

ON LOGIC AND THEORETICAL LINGUISTICS

45

As mentioned nearly at the outset, the goal of logic can be taken to be the characterization of an infinite set of sentences which we accept as 'logically' true. This set, being infinite, cannot be characterized by enumeration of its members. Such is even clear from so homely an example of logical truth as Either Sally is bald or Sally is not bald. Our only hope, then, is to call upon the abstraction of structure, specifically structure of a recursive sort, thereby enabling us to fix on logical truth as a suitable combination of structure and truth.12 So far we have assumed that people accept sentences as logically true. Sometimes we are content to interpret this in terms of the reasoner's intuition of logical truth as 'structural truth'. In a sense this amounts to a Chomskyan-flavored attribution of intuition about structure.13 An example might be the claim that 'x believes that p' and 'x knows that p' have the same form, and that the entailment of p by 'x knows that p' is not structural.14 But we need not begin with intuition about structure. Instead, we may take the truths we want to characterize to be necessary truths, especially those of the form: if S1 is true, then S2 is true.15 As we noted, structure will come anyway, since that is the only way we will begin to capture the infiniteness of the set. What will not come, however, is a characterization of every member of the set as a structural truth. The entailments which do not emerge from the theory as structural will be nonetheless necessary or true; but they will have to be characterized otherwise. The distinction between the necessary truths and the structurally or logically (in a narrow sense) necessary ones, as enforced by the theory, will be more or less the one with which we could have started, had we drawn on intuition of structure. While these matters will become clearer as we describe logic in more detail, we must now confront a remaining problem about which truths (which inferences) we aim to explain. Though we need not stop and wonder about our desire to capture the implication if Sam is happy and Sally is bald, then Sally is bald, or possibly even if Sam is a vampire then Sam is dead, we may well question whether we need worry over whether (1) My wife is either in the kitchen OR in the bedroom implies the speaker (husband) does not know which. (2) Jones has beautiful handwriting and his English is grammatical implies, when uttered in the context of reporting on a pupil's philosophy work, Jones is no good at philosophy. (3) She was poor BUT she was honest implies there is some contrast between poverty (or her poverty) and honesty (or her honesty). 12

Cf. p. 62 and logic section for further discussion of the motivation for introducing structure Intuitions can be more or less deep. One might claim to know which truths are structural, or when two sentences have the same structure, without having a sense of what their structure is. Each degree brings one deeper into Quine's problem of indeterminacy. 14 Davidson (1967a) uses this example, though with appropriate caution. ls Davidson's approach to logical form would eliminate appeal even to necessity. 13

46

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

(4) Smith has left off beating his wife implies Smith has been beating his wife. A s Grice has expertly shown (1961), these four examples must all be described somewhat differently if one takes into account the vehicle of the implication, whether the implication is detachable and/or cancellable, whether there is presupposition, and the relevance of context. For example, in the first case one is inclined to say: (1) It is not presupposition since the falsity of the claim of lack of knowledge does not render the original sentence truth-value-less. (2) The speaker and his saying what he did are vehicles of the implication (as opposed to what he said, or his words, simpliciter). (3) The implication is CANCELLABLE (i.e. one can deny the implication without annulling the assertion), but not DETACHABLE. (Any way of stating just what is stated, involves the implication.) (4) The implication is a matter not of the meaning of any word, but of the meaning of the word 'or' as given in logic plus some 'general principle governing the use of language', e.g. ' one should not make a weaker statement rather than a stronger one unless there is a good reason for doing so'. The third example differs significantly in this last point, for while the word 'but' is involved (corresponding to the fact that 'or' is involved), and though it can be claimed to be the vehicle of the implication, the cancellability and detachability of the implication in every context indicates that it is not the MEANING of 'but' which is at stake. A t least this is not the case in the sense of 'meaning' which determines the truth conditions of what is asserted.16 Further, the fact that the sentence was uttered (i.e. that the person said that) is not a vehicle of the implication. A s Quine puts it (1966), the implication is a matter of rhetoric. The existence of examples like ( l ) - ( 4 ) raises a basic question for logic and for semantics in general, namely, how wide should the net of the theory be extended? Though we have no hard and fast answer to this question, the following comments seem in order. A basic distinction must be made between conventional and non-conventional implications. Roughly, the conventional occurs with or without the intention of the speaker, the non-conventional only with his intention. Similarly, there is conventional and non-conventional 'following from'; the converse of the speaker's intention being what the hearer takes to follow from the utterance. Within the conventional, however, further distinctions must be made. Thus the description of example (1) provides a case of conventional implication, yet one dependent not solely on the meaning of the words, but on this meaning plus a rule of COMMUNICATION. In the second example the implication depends again not solely on the meaning of the words but on the fact that these words WERE UTTERED IN THE CONTEXT of reporting on performance. These factors are brought out when we say that it was the speaker's SAYING what he said that did the implying, and not WHAT HE SAID. 16

This point may go counter to Grice who seems to suggest that it is part of the meaning.

ON LOGIC AND THEORETICAL LINGUISTICS

47

Now a theory whose goal is to handle the implications or 'truths' which hold in the act of communication, that is, when sentences are used, must explain these implications. Such a theory may be a part of a theory of langue; but it shares a key feature of any theory of parole absent from the standard theory, in its aim to capture n-place properties of utterances of sentences, which, if they be expressible as properties of sentences at all, will become n + i-place properties (the additional places being necessary to absorb relevant features, e.g. context).17 Logic, as it has developed so far, makes no effort to develop such a 'communication' theory. Similarly, most of the work in theoretical linguistics has dealt with language at its farthest reaches from parole. These tendencies are not without their critics both in logic and in linguistics. The parallel here is especially to be expected, since what is at stake is the delineation of the competence of the speaker-hearer. Given our initial comments on its symmetry with logic, one would expect that advocacy of an extension of linguistic competence to include the communicative competence to produce and understand utterances which are appropriate to the context in which they are made would be matched by the desire to create a theory of inference so relativized. Whether one should go beyond the descriptive claim that logic has not dealt with such a theory to maintain that it should not or cannot, is unclear. The argument of Katz and Fodor (1964) to the effect that semantics should not and cannot explain the 'meaning' of sentences, when this is taken to include all that is conventionally communicated, if valid, would yield the parallel conclusion in application to our question. This is clear from the parallels between what the person means and what he intends to imply; what an expression of a language means and what the expression implies, and so on. In any case, it remains unquestioned that if logic is to attempt such a theory, it will be faced with a challenge far beyond that with which it is normally faced. Since we are attempting to characterize logical form at its present development, it seems plausible merely to ignore this possibility and therefore to eliminate from the set of inferences to be explained most of the examples like the above.18 We turn now to logical form as it emerges from the standard treatments of logic. We begin with a 'categorized' canonical notation in terms of which formation rules generate an infinite set of grammatical or well-formed formulae. The infinity is gained by recursive constructions among the formation rules. Typically, a very austere language consists of an infinite number of variables (x, y, z, x ' . . . ) , indivi17

Inductive logic may be relevant in formulating a pragmatic theory in a broader sense. For its application to metaphor, see Margalit (1970). The theories of Montague and Lewis discussed below provide the means for specifying contextual features as part of a theory of truth for sentences with indexicals. They do not, however, attempt to capture communicative competence in the sense intended here. Alternatively, their theories include pragmatics but in a narrow sense. 18 It should be noted that nothing we have said bears on: (a) whether Grice's analysis of utterer's meaning is acceptable; (b) whether Grice's attempt to base conventional meaning on utterer's meaning works.

48

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

dual constants (a, b, c, a ' . . . ) and sentence letters (p, q, r, p ' . . . ) (all gained by construction from a finite set), and a finite number19 of 1-place, 2-place... predicate letters (F, G, H, F ' . . .)• Of the key constructions, one yields atomic predications (that is, open sentence letters (Fx) or closed sentence letters (Fa)) from predicate letters and variables or individual constants, while others yield complex sentence letters (open (FXAGX) or closed (FaAGa, 3xFx)) from other sentence letters. The latter constructions generally consists of the application of a 'logical constant' (simple A, v , or complex 3x) to one or more sentences. These symbols may be taken as part of the construction itself, though one could easily imagine removing these symbols from the construction, and, instead, introducing a new construction which would treat these symbols as separate members of a category. What cannot be sacrified is the recursivity of some of these constructions. The result of the application of formation rules, as we have said, is an infinity of 'grammatical' formulae or schemata. So far, the sentences have no interpretation, and truth is not a meaningful issue. This lack is remedied, however, by assigning to the constants an interpretation in some natural language. This interpretation is assigned by a rule which specifies how the truth or satisfaction conditions of sentences, built up syntactically by the constructions associated with the constants, are themselves 'built up' out of the truth or satisfaction conditions of their parts. Syntactic recursivity thereby goes semantic, in this sense, at least.20 An example of a semantic rule using sentence letters is: The result of applying the V-construction to sentences 'p' and 'q' is true if and only if (iff) 'p' is true or 'q' is true. The rules become more complicated, however, when we talk of predicates, for then we must introduce sequences. Sequences are basically ordered n-tuples of objects to which predicates may apply. We say that a member of a sequence satisfies an open sentence if the predicates in the open sentence are true of this object. More specifically, if 'xn' is the n th variable in the list of variables and xn is the n th member of any sequence then A sequence satisfies the result of applying the predication construction to ' F and 'xn' iff Fx n . The extension to complex sentences follows clearly. Thus, A sequence satisfies the result of applying the V-construction to the result of applying the predication construction to 'F' and 'xn' and 'G' and 'x n ' if the sequence satisfies 'Fx„' or 'Gx n \ As a final example, consider the rule for existential quantification, 18

The motivation for this restriction to finitude is considered in the discussion of Davidson. This seems to make the constants semantically recursive, even if they are not part of the syntactically recursive constructions S. 20

ON LOGIC AND THEORETICAL LINGUISTICS

49

A sequence satisfies the result of applying the 3xn-construction to the result of applying the predication construction to ' F and 'xn' if and only if there is some sequence which differs from it in at most the n th place and which satisfies 'Fx n \ Finally, truth is defined as satisfaction by all sequences. Of course, the formation and semantic rules for our canonical notation are in a meta-language; in this case, English, more or less. Given interpretations for the constants, we can then introduce the notion of assigning an interpretation to the atomic elements of the theory. For the subpart of our language known as the sentential calculus this would just amount to assigning a truth value (truth or falsity) to single sentence letters. The predicate logic introduces a s s i g n m e n t s to predications from a non-empty domain of discourse over which the bound variables (and therefore the quantifiers) range. Specifically we assign individual members of the domain to free variables and individual constants, and sets of ordered n-tuples to n-place predicates. Finally, we may then define logical truth for the formulae of the language by saying that a formula is logically true if and only if it is true on every interpretation (in every domain) assigned to the atomic parts. Again referring to the sentential calculus, this definition results in the identification of logical truths with tautologies. For example, given the standard truth rules for 'V' which is interpreted as 'or' and which is interpreted as 'not', we can see that 'pV~p' is a logical truth. The constants of the standard theory which are normally interpreted as 'and' (A), 'or' (V), 'not' ( ~ ) , 'if-then' (->-), 'if and only if' 'there is an x such that' (3lx), have occasioned wide discussion in treatments of logic and epistemology, due to their central role in defining logical truth. The medievals saw them (their translations in Latin) as syncategorematic in contrast to categorematic terms. Wittgenstein deemed that they were not part of the picture in his picture theory of meaning, though, of course, they were very much in his picture of language. Many have seen them as topic neutral items. Tarski has characterized them as transformational (not linguistically) invariants. It should be clear that in logic, as outlined above, their special virtue is in their ability to create semantic recursivity as part of a theory of inference or logical truth. Beyond this, we have merely enumerated them. There is, however, a further characterization of the constants which we shall present, while leaving unexplored its connections with the traditional ones. Abstract, for a moment, from the standard constants, and think of a constant as any term whose occurrence in a schema which is true under every assignment to the nonconstant terms in the schema makes it a logical truth. Now any term may be treated as a constant once it has been interpreted, provided only that we agree to label all such sentences basically true. We may then define it to be a true logical constant if and only if, when treated as a logical constant in the sense above, any other term also so treated and substitutable for it in every schema salva-logical-veritate is

50

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

also substitutable for it in every schema salva veritate on every interpretation. The contribution to the meaning of a sentence by a term meeting this condition is completely fixed by its contribution to the logical powers of the schema, in the narrow sense of 'logical'. Any term whose meaning is given as a truth-function clearly meets this definition; for two truth-functions could not determine the same logical truths, and be defined differently. What other terms meet the definition awaits further examination. In our austere grammar, the possible types of terms are few. Some predicates would clearly determine no new logical truths, e.g. 'is a sweety'. Others, for example, 'is identical to' and 'is a thing', would. When we branch out to sentences not yet absorbed into current logical theory, we find expressions which might determine new logical truths, but are not constants. 'Quickly' and 'slowly', for example, seem to determine the same logical truths, but are not substitutable salva veritate. So far we have sketched out the syntactic and semantic elements of standard logical theory. While indicating the semantic rules, we gave an example in which the meta-language was English. The constants of our theory were interpreted via English words — 'and', 'or', 'not'. . . . This, then, raises a question whose implications will be made explicit below: D o these words maintain in logic the meaning(s) they have in ordinary language? It is already clear at the outset of an introductory logic course that the constants do not bear all of the meanings of their English counterparts. In accepting the commutativity of conjunction ('A', above) we are eliminating the interpretation 'and-then', a common interpretation of the word 'and' in ordinary language. This disambiguation occurs also when we separate the exclusive and the non-exclusive 'or'. It is instructive at this juncture to consider some comments of Hintikka (1968) to which we shall return later when exploring (in greater depth) the relation of ordinary language to logic. Hintikka views a branch of logic, not as a 'regimentation of the relevant sectors of ordinary discourse', nor as a revision of our ordinary ways with certain concepts, but as an explanatory model which '. . . may be thought of as giving us a way of using language in so far as this use is determined by one main purpose which the part of language in question is calculated to serve' (1968). Thus the model does not 'reflect what happens in ordinary language', for what happens there is also influenced by many other facts and pressures, e.g. (1) other competing purposes, (2) pragmatic pressures, (3) built-in limitations of the human mind, (4) pressures due to particular contexts. An example of this non-reflection of ordinary language is found in the negation-constant. In ordinary language, Hintikka claims, a doubly-negated expression seldom, if ever, has the same logical powers as the original unnegated statement. The difference in use of the constants in logic and in ordinary language was similarly noted by Tarski years ago in his introductory logic text, and the standard identification of 'but' with 'and' would seem to reinforce this judgment.

ON LOGIC AND THEORETICAL LINGUISTICS

51

What is instructive about Hintikka's comments is that they show how careful we must be in drawing certain distinctions. Hintikka's contrast between logic as regimentation and explanatory model is that, unlike the former, the later need not 'reflect what happens in ordinary language'. This failure to reflect is exemplified by the variance in ordinary language of two logically equivalent expressions. But if we are to accept this dichotomy at all, we will want to do so in the context of some distinctions about levels of logical and semantic theory like those noted at the outset of this section. Whether logic reflects ordinary language must then be relativized to those features of ordinary language one is talking about. Hintikka's indication along this line suggests that he has in mind as non-reflected precisely those features of ordinary language which are questionably in the domain of logic and semantics. In short, then, we must distinguish reflecting ordinary language in the sense of reflecting the logical and semantical features. This done, the distinction between regimentation and explanatory model will sometimes collapse;81 and further it seems that the constants may be taken to have their ordinary meaning (or some of their ordinary meaning) despite their failure to maintain all features of the ordinary words. A final question about the constants is which ones do we use? Obviously, we must use whatever constants are necessary to give us the inferences we want. Beyond this, the question amounts to which of the interdefinable constants we should call upon, i.e. given that ' A ' , w i l l yield equivalent results to ' A ' , ' V ' , do we use 'V' — or, given that 'V', will suffice, do we use 'A', or given that a single constant ( \ , Sheffer stroke) will suffice, do we use 'A', ' V ' . . . ? The only theoretical relevance these questions have is to the issue of perspicuity to which we will return. Our goal is to characterize logical truth for ordinary language. So far we have succeeded in characterizing it for our canonical notation only. The obvious move now is to correlate sentences of ordinary language with the well-formed formulae of our canonical notation, so that the logical truth of the latter will be passed on to the former. Let us call R a representation of S if and only if R is a well-formed formula of canonical notation, to which S is correlated. How, then, do we establish such correlations? A minimum is required to establish one segment of the correlations. This It is worth noting, further, that (a) 'regimentation' is not generally used in Hintikka's sense of 'reflecting all features of ordinary language'; (b) the issue of logical theory as a paraphrase into a subset of ordinary language is not dependent on regimentation in this sense; (c) Hintikka views an explanatory model as a theory in contrast to a mere report of observation. But this will depend on what features of the language are considered. One could try to maintain a theory that made semantics the observational content and the model an abstraction f r o m this. This would make the constants theoretical terms even relative to semantics; and would make translation into ordinary language as an issue of observation and theory. But this idea may not make sense, for to specify adequately the truth conditions of the sentence with these constants w e may have to define them in terms of ordinary words. This would still give the constants ordinary meaning though perhaps not their ordinary meaning. 21

52

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

is so because our canonical notation was itself explained in ordinary language. Once we assign lexical items to the grammatical categories of our canonical notation we have provided a 'way out' and a 'way in' for a whole range of sentences. As Quine notes (for a lexically interpreted canonical notation): The artificial notation of logic is itself explained, of course, in ordinary language. The explanations amount to the implicit specification of simple mechanical operations, whereby any sentence in logical notation can be directly expanded if not into quite ordinary language, at least into semi-ordinary language. Parentheses and variables may survive such expansion, for they do not always go over into ordinary language by easy routine. Commonly also the result of such mechanical expansion will display an extraordinary cumbersomeness of phrasing to and an extraordinary monotony of reiterated elements; but all the vocabulary and grammatical constructions will be ordinary. Hence to paraphrase a sentence of ordinary language into logical symbols is mutually to paraphrase it into a special part still of ordinary or semi-ordinary language (1960 : 159). The simplest examples are from the sentential calculus. Here paraphrase or correlation amounts to the trivial task of replacing sentences by sentence letters. This done, we have no trouble in correlating 'Sally is silly or Sally is not silly' with ' p V ~ p \ Subtlety enters when we begin to assign structure within a sentence. But here, too, our use of an ordinary meta-language assures us that if we assign 'is silly' to the category of one-place predicates and 'Sally' to the category of individual constants, 'Sally is silly' will correlate with 'Fa'. Similarly, 'there exists something x such that it is silly' correlates with '3x (Fx)'- These last two examples illustrate how we have 'swallowed' in this mechanical procedure some operations which syntactically are transformations. Thus we have the trivial rearrangement of 'is silly' and 'Sally'. These rearrangements may be indicated in the semantic rules by which we correlate the canonical notation to ordinary language.22 Once we have assigned lexical elements of ordinary language to categories, we can reformulate the earlier definition of logical truth in terms of substitution. Roughly, a sentence will be logically true if it remains true under every lexical substitution of the non-logical parts. Assuming a rich lexicon, this definition will yield identical results to the earlier one. We have seen that implicit within our canonical notation was a correlation scheme for an infinity of sentences of ordinary language. This means we have characterized logical truth for this infinity and, in so doing, have ascribed a logical structure to its members. For we have done what amounts to categorizing the parts of the sentences, and specifying the method by which these parts are to be combined, in order to yield the logical character of the sentences. What is to be said for the remainder of ordinary language? The scheme which correlates these remaining sentences with representations in canonical notation can be viewed as proceeding indirectly, first correlating these sentences with already 81

The important question of transformations and correlation schemes will be discussed in the final section on linguistics. The general issue of to what extent correlation can be mechanized is beyond the scope of this paper.

ON LOGIC AND THEORETICAL LINGUISTICS

53

correlated sentences of ordinary language. The question is what the conditions are which this scheme must satisfy. Quine, in distinguishing these correlations from the earlier mechanical ones, warns us not to invoke synonymy. S' might indeed naturally enough be spoken of as synonymous with the sentence S" of semi-ordinary language into which S' mechanically expands according to the general explanations of logical symbols; but there is no call to think of S' as synonymous with S (a further sentence paraphrased as S'). Its relation to S is just that the particular business that the speaker was, on that occasion, trying to get on with, with the help of S among other things, can be managed well enough to suit him by using S' instead of S (1960:160).

Formulated in this way, Quine suggests what is even more clearly evidenced in his next paragraph, namely, that his model is that of individuals on given occasions using language for specific purposes. This stance seems not to avoid calling upon synonymy. We may reject laying down synonymy as an adequacy condition while maintaining the stance adumbrated above of language as something conventionally given. Logical form is evidenced in sentences, not just in uses of sentences for specific purposes. What, then, will replace synonymy or 'preservation of purpose' as our conditions? On one level of generality, we may say that the correlated sentences must be such that on the basis of this correlation we can determine the entailments of the correlated sentences via the theory, and such that no unacceptable inferences are counted acceptable. But this, of course, is just a restatement of the adequacy condition of our entire theory. Having once stated it at the outset and assuming we had the acceptable inferences in mind, nothing need be laid down about further correlations. Rather, we are 'free' to correlate at will, providing only that we are not forced to accept the unacceptable.88 What is implicit here is the fact that the correlation does not proceed by pairing sentences as wholes, but rather by matching parts of the correlated sentences. Basic unstructured elements are thereby assigned to categories, whereas structured phrases are assigned structure and their parts categorized. This process will be essential if we are to use our subset of ordinary language and ultimately our canonical notation in defining logical truth for the remaining sentences. Examples of how the correlation process need have no unique restraints are not far off. Consider the age-old problem of whether expressions like 'some 0' 'no 0' should be viewed as names. Given our logic, we know what happens to sentences of certain forms. Quine gives the following example (1969b: 330): (1) Copernicus was Polish and wrote Latin

is entailed by (2) Copernicus was Polish and Copernicus wrote Latin. If then, instead of correlating !S

Though we do not appeal directly to synonymy, it cannot be claimed that we have avoided the indeterminacy which Quine believes lurks behind this notion. For all the same problems arise in staking out the set of inferences which we appeal to as our test.

54

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

(3) Some astronomer was Polish and some astronomer wrote Latin

with our quantificational structure, we pair it with a structure equivalent to (2), we shall be forced to accept the unacceptable inference from (3) to (4) Some astronomer was Polish and wrote Latin.

Similar difficulty arises if we assign the wrong structure to structured phrases of a sentence, even if we get straight the basic categories of the parts. Thus 'Susie is a sorrowful lass' will soon create problems for a logic that ascribes to 'sorrowful lass' the structure of a disjunction in place of a conjunction. Our PRACTICE in correlating sentences is not so unguided. In order to cover a wide range of inferences as simply as possible, we tend to uncover structure wherever ordinary language suggests it is residing. This is not to say that we always follow the suggestions of ordinary language as to where the structure resides or what structure resides where. Such practice would lead us to count a deposed mayor among the mayors as we count a lovely man among the men. In brief, our principle is: other things being equal, maximize PERSPICUITY. Thus, capturing 'is unmarried and a man' as 'bachelor' would leave us at a loss to explain the former's entailment of 'unmarried', as well as to correlate sentences with only 'unmarried' or 'man'. In the end we would need to introduce 'unmarried' and 'man' into the theory separately anyway. Our principle suggests that we strive for this at the outset. Reversing the circumstance, we might ascribe to 'bachelor' the structure of 'is unmarried' and 'is a man'. This way we would not exclude any inferences. But the theory then would have a step avoidable by directly considering 'bachelor' a one-place predicate, and thereby allowing some sentences with 'bachelor' to be automatically correlated with the canonical notation. In both cases we are permitted, of course, to add the rule: 'bachelor if and only if is unmarried and a man'. But this further rule would take on a role of an additional principle. This last example illustrates the point, already noted, that synonymous sentences need not have the same logical form.84 We are now in a position to introduce specific concepts of perspicuity and representation. (1) R is a representation in theory T of the logical form of S if and only if R is a well-formed formula (WFF) of T, to which S is correlated. (2) S is perspicuous in T if and only if there is an R such that R is correlated with S by abbreviation, and/or rearrangement,25 and/or replacement of terms by symbols, and such that the semantic rule for R yields a sentence in the meta-language identical with R except for the constants. 84

The perspicuity principle accords with those who may want to start off with intuitions of structure though here it functions as a heuristic only. Whether other things are equal can be very controversial and may depend on fundamental matters of approach. Thus the 'principle' is very rough. !S This is a bit too loose. A more complete specification will have to distinguish acceptable from unacceptable rearrangements.

ON LOGIC AND THEORETICAL LINGUISTICS

55

(2a) S is perspicuous if and only if there is a T such that S is perspicuous in T. (3) R is a fundamental representation of S in T if and only if R is a representation of S in T and [S is perspicuous in T or R is a representation of some perspicuous S' in T].28 (3a) R is a fundamental representation of S if and only if there is a T such that R is a representation of S in T and [S is perspicuous in T or (it is not the case that there is a T such that S is perspicuous in T and R is a representation of some perspicuous S' in T)]. Thus, if S is perspicuous in T, it is perspicuous. If it is not perspicuous in T, nothing follows about its perspicuity. If S has a fundamental representation in T and S is not perspicuous, nothing follows about whether this representation is fundamental. If S has a fundamental representation in T and S is perspicuous, then the representation is a fundamental representation of S. It can be seen now that the interdefinability of the logical constants will simply affect whether a sentence is perspicuous in a theory 'P and Q' will turn out not to be perspicuous in a theory containing only 'not' and 'or'. But given that there is a theory with 'and', the sentence is perspicuous. With these definitions we can talk objectively, not just pragmatically, of a sentence's being perspicuous. If, in addition to the fundamental R of S in T, we define an additional representation which applies to sentences as they appear, perspicuity will become equivalent to a relation between these two representations. It is in this light that we understand the familiar cry that grammar misleads about logic. Grammar (i.e. the second representation) misleads if its representation is not identical with its logical representation. Grammar may also mislead about when two sentences have the same logical representation. In such a case, however, only one of the two sentences need fail to be perspicuous. That we can be misled about whether sentences have the same logical form should engender caution in our talk of 'the logical form of sentences of kind K'. Unless we carefully restrict the class we mention, or choose it according to logical criteria, there is no guarantee that all its members will have the same logical form. Grammar may mislead about other things besides logic. Below we will see its relation to syntax and semantics. Another topic on which grammar may be misleading but which is different from the three areas mentioned is verification. We might claim that 'John is popular' misleads, because it suggests that we can verify someone's being popular in the same way we can verify his being black or white. But it is clear that we are no longer talking about the logical entailments of 'John is popular'. At best we might claim that this notion of verification is tied to the logical powers of the sentence, but not in a way that allows a distinction between structural 26

This latter clause may always be satisfied, thereby collapsing fundamental representations to mere representations. It is included only because of the uncertainty of whether it is trivially satisfied. An interesting test case which we have no space to discuss is modal logic where the modal operator notation is explained in terms of possible worlds and quantification (generality).

56

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

and non-structural inferences. In Gilbert Ryle's classic "Systematically misleading expressions" (1967) much of the misleadingness concerns non-formal 'selectional' features of language. With the technical details behind us, we are free to inquire more broadly about the nature of logical theory and its implications for 'philosophical analysis' generally. Russell's theory of descriptions, one of the major pieces of logical analysis produced in the twentieth century, is instructive in this regard. More than any other piece of contemporary analysis, it has aroused varying estimates of the adequacy conditions which Russell intended it to meet. Thus to Strawson Russell appears to claim for the theory of descriptions that it gives an exact account of the working of one class of definite singular terms, viz. singular descriptions ( 1 9 6 9 : 106).

In contrast, Strawson finds that . . . Quine does not claim that the sentences which replace those containing definite singular terms have the same meaning as the latter. N o r presumably would he claim that they serve exactly the same purpose. . . . H e would claim that in some weaker sense the sentences containing singular terms could be replaced by the sentences in canonical notation (107).

And yet, Strawson notwithstanding, Quine clearly views his elimination of singular descriptions as of a piece with Russell's. Briefly, to expand Strawson's characterization, Quine feels that, for himself as for Russell, theory construction lays down goals and then sets out to meet them. If the theory meets problems along the way, it does its best to avoid them. What is problematic in cases of singular description is the 'inconvenience of truth-value gaps and the paradoxes of talking of what does not exist'. Quine, following Russell, points the way to solve these problems via the theory of descriptions. Of interest here is not the historical question of what Russell really intended, nor the parallel question of how to interpret Wittgenstein's Tractatus remarks (5.5563a) which suggest that for him, in contrast to Russell, ordinary language just as we know it is in perfect logical order. 27 Rather, it is the dichotomy — Logic as Replacement vs. Logic as Descriptive Theory — which draws attention. Already in Hintikka's remarks mentioned earlier, we found a distinction between logic as an ideal revision and as an Explanatory Model (identical with Regimentation in certain cases). Now we have traced it through Quine and Strawson. Finally, we may note Davidson's remark t h a t ' . . . the task of a theory of meaning (for which read theory of logical form), as I conceive it, is not to change, improve or reform a language, but to describe and understand it' (1967b: 314). 87 Russell may well have been unclear, and Pears' distinction (1968) between correct and comprehensive analyses seems to be of little help. Black (1964) seems to support Quine's view of Russell while making early Wittgenstein a Strawsonian of sorts.

ON LOGIC AND THEORETICAL LINGUISTICS

57

What are we to make of this proposed dichotomy? Let us be a bit indirect and argue for the irrelevance of certain considerations. So far we have suggested that, given a theory of logical truth, some entailments which turn up in our original adequacy set will not be logically valid conditionals. Such entailments will thus be labelled non-structural entailments, and their explanation will turn on what we have to say about individual theoretically unstructured terms occurring in the relevant sentences. What we have is parallel to the old distinction between logical truth and analytical truth, where the latter, viewed restrictively, consists of truths which can become logical truths by substitution of synonymous terms. Now it should be clear that replacement vs. description at least applies to what we can do with terms. The triple sins of ordinary language, vagueness, ambiguity, and contradictoriness, successfully tempt numerous TERMS of our language. Often we are out to battle against these sins, and, in systematic theory construction, the avoidance of such sin is cardinal. Further, the goal of much philosophical analysis centers around replacement programs for terms (cf. R. Carnap, Nelson Goodman). Quine's example of the ordered pair as paradigm of philosophical analysis is just such a term replacement program. But, since this kind of replacement program is delineated only AFTER the theory of LOGICAL entailments is complete, it should not be viewed as directly relevant to it. More specifically, there is nothing problematic about having a DESCRIPTIVE theory of the logical form of S1 . . . S n alongside a REPLACEMENT program for the category member terms of S1 . . . Sn. Such replacement, as Quine notes, need not claim synonymy and may allow all 'don't cares' to vary. To a theory of logical form this is all of no moment.28 Furthermore, to return for a moment to terms, it is not even clear exactly when we should be careful not to mislead about replacement vs. description. Suppose we give an ambiguous term of ordinary language more than one route into our theory, that is, we correlate it with more than one term. Though the term itself no longer appears, the theory remains descriptive of this property of the term.29 Ambiguity need not penetrate its self-description. On the other hand, if we explicitly change the clear extension of a term or its selectional features by our uncorrelated scheme, then we will have replaced the term, in contrast to describing it. Quine's explication as elimination counters, as he notes, the descriptive goals of Oxford. In between these two extremes is the status of vagueness. If our adequacy condition requires descriptions of CLEAR CASES, then elimination of vagueness is no offense against a descriptive theory. If, however, we are to describe sentences 28

Quine's comments (1969b) on Davidson's "On saying that" (1968) indicate clearly how Quine's desire to replace such concepts as the modal ones is compatible with handling their logical form descriptively. 29 This makes explicit what may have been unclear in the use of 'replacement', i.e. it is replacement versus description of various properties of terms, as determined by the theory and the correlation scheme, and not the inconsequential absence or modified use of the terms themselves in the theory. Perhaps revision vs. description would be clearer. Talk of Ideal Language is also subject to this dual interpretation. For further remarks on ambiguity see p. 68.

58

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

in all their vagueness, elimination of vagueness will just be replacement.80 Anyway, terms are not forms, and a replacement program for non-logical terms need not conflict with a descriptive theory of logical form. Neither, of course, does the fact that in non-perspicuous cases correlation amounts to transformation. For what is transformed is not the logical form of the sentence, but some other form which has no positive bearing on it. To require of the theory of logical form that it make all sentences perspicuous is just a confusion. The straightening-out of structural ambiguity is also compatible with a descriptive theory of logical form, just as we saw the straightening out of term ambiguity to be compatible with a descriptive term theory. In fact, we really have no choice in generating structural ambiguities. Consider the familiar 'everyone loves someone'. If we theorize about this sentence, as we normally do, by correlating the first word with a universal quantifier, the second with a two-place predicate,31 and the third with an existential quantifier, then our rules FORCE US to have two sentences with which to correlate it, barring, that is, further restriction. But what now of vagueness? The twin cases familiar to philosophers of logic and susceptible to the same treatment, according to some, are the vagueness of the truth-condition of the indicative conditional 'if-then', and the similar vagueness of the truth-conditions of sentences in which a 'presupposition' (especially of singular terms) fails to hold. Of course, the uncertain relevance of vagueness in the issue of replacement or description of terms applies here also. If our theory is to describe clear cases only, then how we fill in the 'truth-value' gaps need not concern us. But if our adequacy condition is not so specified, we will be led to agree with Strawson in saying that standard logic and Russell's theory are not descriptive.32 One additional point may bear on this question. We might try to distinguish the goal of simply preserving or describing vagueness from the goal of precisely describing a clear but complex situation. Seeing the latter goal as applicable here would amount to claiming that a pair of terms, apparently contradictory, are only contrary. In 'between' there would be a further term which applies precisely to describe the failure of either contrary to hold. Thus instead of merely talking about uncertain truth, which is like uncertain baldness, there would be a clear stage between truth and falsity, and this stage should be described here. At times the advocates of three-valued logic, which is suggested to handle both of the above problems, talk in this way. If a third value, intuitively clear, were forthcoming, their 59

An extensive effort in constructing a logic of vagueness is now going on mainly under the title of 'fuzzy sets', first introduced by Zadeh (1969). Fuzzy sets are collections of objects with vague ('fuzzy') boundaries, that is, sets without precise criteria of membership (see Brown 1971). The fuzzy sets can be viewed in terms of a multivalued logic with a continuum of truth values in the interval [0,1]. Thus, for example, the truth assignment (V) in fuzzy logic of ctA/? will be min (V(ct), V(/?)), while that of aV/S will be max (V(a), V(tf)). (See Lee and Chang 1971.) 81 This may be untenable. See discussion of Davidson below. ** Actually Strawson's willingness to have truth-value gaps itself derives from his interest in describing other features of ordinary language associated with singular terms.

ON LOGIC AND THEORETICAL LINGUISTICS

59

arguments might constitute ground for insisting on broadening the adequacy conditions, so that standard logic would be a clear replacement and not a description. Until now we have mentioned only clear and uncertain irrelevancies to the issue of Description vs. Replacement. Implicit in our discussion all along, however, has been the presumption that logical theory, in one of its guises at least, seeks to DESCRIBE the logical features of ordinary language, and that standard logic is a candidate for part of this task. Our comments about vagueness nonetheless raise a key point: that it is pointless to contrast replacement and description without a firm grasp of the purpose(s) for which we construct our theory. Quine comments: In passing, I would touch on a second and lesser point of Geach's, when he deplores my policy of eliminating singular terms other than descriptions. He is right in so far as one's purpose is an analysis of E n g l i s h ; . . . . For that matter, in so far as one's purpose is analysis of English, there is something to be said also for truth gaps rather than falsity in the cases where singular terms lack designation. On both points, my deviant course is defensible only in so far as one's objective is a medium having certain advantages over English (1969 : 331).

Of course, talk of purpose is helpful only if we specify our purposes clearly; 'analysis of English' hardly so qualifies. Nonetheless, the point remains that only relative to some clearly specified purpose, description vs. replacement makes sense. Our uncertainty about truth gaps and vagueness generally was really due to an unclarity about purpose. Still, we should not be misled. Just any talk about the purposes which one theory serves and another does not, does not make one theory descriptive of something ignored by the other. To be 'positivist', only purposes identical with some empirical adequacy condition will do this. Alternatively, purposes which determine ACCEPTANCE would not have the force of purposes which determine CONFIRMATION. Quine's epistemic vagueness on this point is deliberate, for he will have no truck with the set of terms on which the distinction rests. Yet, the distinction does come to the fore when indeterminacy of translation and related notions is relevant. For here, our purpose may allow a differentation of theories where there is according to Quine no fact of the matter. In such cases, then, even for Quine, not every distinction of purpose yields replacement vs. description.88 Pending a further clarification of our purposes, it seems standard logic can be viewed as descriptive. There is a further argument, however, which, if sound, would refute this. Indeed, the argument would seem to entail the impossibility of providing a descriptive logical theory of ordinary language. Thus Dummett writes: Now I do not think we can tell . . . whether Frege himself was right in thinking — as I believe he did — that natural language could not have for it absolutely precise, " It should be noted that even if standard logic or the elimination of singular terms, etc., is vindicated as description, philosophers and logicians in pursuing philosophical interests may seek replacement. Essentially this is revisionary metaphysics. The correlation to ordinary language does not preserve the inferences of correlated sentences and properties of ordinary terms do not recur.

60

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGAUT

determinate, unambiguous rules — that is, as far as I understand him, his belief was that the methods of expression of generality, in particular, and other things in natural language, were simply incoherent. That is not to say o n e could not describe as true or false sentences of natural languages, but one could give no complete account of the rules determining truth or falsity and no coherent such account (Staal 1969 : 277).

Standard logic whose treatment of generality is coherent, would no longer be descriptive, for the correlation of incoherent sentences with coherent ones would violate our adequacy conditions. More importantly, no revision in this logic which maintained the same correlation scheme would do, since ultimately no logic can be built on an incoherence. In any case, as Dummett notes, whether Frege is correct must await further investigation. Assuming that standard logic is a descriptive theory, we must finally ask to what extent this theory meets our initial adequacy conditions, that is, how large a part of our set of inferences or truths it covers. In a sense the question is narrower, for we are not really interested in how extensively the theory characterizes non-structural truths with the help of further assumptions. We assume that for terms which have already found their way into the theory, the addition of meaning postulates34 or the elevation to constant status is relatively straightforward.35 Rather the issue centers around the extent to which our correlation scheme extends throughout the language. Are there sentences which simply are not correlated? The answer, not surprisingly, is yes; and it is with these sentences that the open problems in the logical theory of ordinary language lie. Two main approaches are exploited to remedy this lack. The first continues to use the structure outlined above in new combinations, while possibly increasing the range of the variables to include new sorts of entities. The second adds grammatical and semantic categories, or possibly further devices as well, and by thus enriching the structural framework incorporates more sentences. We shall encounter both tacks below. Davidson, in attempting to treat prepositional modification, uses the former. Montague and Lewis choose the latter in dealing with similar problems. Perhaps the archetypal rivalry between these approaches is that of Quine and Rescher. Quine exploits the austere standard logic to the fullest. Rescher (1968) looks at every turn to the possibility of new structure.38 The reductionist character of the Quinean approach should not be confused, however, with the issue of description vs. replacement, though, for Quine, reduction is easier since one is freer to replace. 34

How the addition of meaning postulates to a theory allows more inferences to be covered is straightforward. Much about meaning postulates, of course, is not. ®5 Since treating a term as a constant in a theory does not mean that it is a logical constant in the sense introduced earlier, the 'elevation to constant status' does not equal enrichment of logical structure in a serious sense. " Compare Rescher and Quine in attitudes towards tense logic.

ON LOGIC AND THEORETICAL LINGUISTICS

61

Two general conclusions emerge from our study of logical form. 1. Logical form is presumed to be a feature of ordinary language presented in a theory descriptive of this language. 2. A key element in characterizing logical form is the concept of truth. 37

V.

PHILOSOPHICAL SEMANTICS Meaning is truth; truth, meaning. That is all ye know on Earth and all ye need to know. "Ode on a piece of Polish crystal" — Davidsen, obscure Scandinavian poet

From logic, we turn to Semantics. Our route will lead us through the complexities and subtleties of work by Davidson, David Lewis, and Richard Montague. We begin with Davidson because (1) the theory with which he works is closest to standard logic as outlined above, (2) his adequacy conditions are enough different from those of logic to yield illuminating comparison, and (3) later discussion will find numerous points of contact with the issues raised here. Davidson's initial formulation of his challenge in "Truth and meaning" goes like this: 'A satisfactory theory of meaning must give an account of how the meanings of sentences depend upon the meanings of words' (1967b: 304). Variant formulations call for an account of how the 'meaning of every sentence depends on its structure' (1969:160) as well as an effective method which will 'give the meanings of all independently meaningful expressions on the basis of an analysis of their structure' (1968:177). In considering possible ways of meeting this challenge, Davidson rejects the claim that a solution is automatically provided by a recursive theory of syntax plus a dictionary: . . recursive syntax with a dictionary is not necessarily recursive semantics . . . . ' This claim, as Davidson notes, is convincing only in the context of a clear specification of the adequacy conditions of a semantic theory. When presenting the example of belief-sentences as having an unproblematic syntax and a problematic semantics, Davidson has in mind his adequacy condition which we shall specify shortly. One might wonder whether this claim is warranted on a different conception of semantics. We shall take up this point below in considering other proposed semantic programs (Katz-Fodor) at which time we shall also discuss the warrant for Davidson's conception of semantics itself. What, then, is Davidson's proposal? Essentially it amounts to this: a theory of meaning for a language consists in a Tarskian truth definition for that language. Let us unpack this a bit and see what emerges. Roughly, a truth definition for a language L is a finitely axiomatized theory whose 37

We have ignored throughout provability as a way of characterizing logical truth. Such characterizations are, of course, equivalent to those we have mentioned and do not really eliminate the fundamental importance of the concept of truth in logic.

62

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

theorems include all sentences gotten from the schema x is true when 'x' is replaced by a standard name of a sentence of L 3 8 and 'p' is replaced by a sentence correlated with that sentence. Davidson specifies the notion of a correlation in a number of ways. Keeping in mind that the language of the truth-definition is a meta-language of and richer than the language L, a requirement needed to avoid paradox, we can see the question of correlation as: how should we correlate the sentences of L with sentences of the meta-language so that the relevant bi-conditionals are specified? Initially, Davidson takes the case where L is part of the meta-language. Correlation is, then, a matter of identity, for some sentences at least. Beyond this, 'translation' and 'guaranteed equivalence' are employed, but finally the condition for correlation is weakened to the truth of the truth-functional bi-conditional. 'p' is replaced by a sentence true if and only if the sentence named is. 39 Any way it is characterized, the correlation scheme is a set of empirical, contingent truths. Therefore the theory is an empirical one. Earlier we noted the relevance of correlation to the contingent aspects of logic. Such a theory of truth, Tarski has shown, must employ recursive devices for compounding a finite number of truth relevant elements of sentences, at least if the theory is to encompass an infinity. This framework explains why finite axiomatization will provide an explanation of how the meaning of a sentence depends on its structure. Actually it explains how the truth conditions of a sentence depend on its structure.40 Care in requiring finite axiomatization has indicated, no doubt, that not every 'theory of truth' is a finite theory. One could claim to have a theory of truth if one took 'it is true that p-p' as an axiom schema, but then one's axioms would be infinite in number (Fodor 1970b: 299). Alternatively, one could employ substitutional quantification to avoid the problem of quotation and define truth in a single axiom. Here provability is sacrificed. The failure of some truth-theories to be adequate for semantics in no way vitiates the search for one that is. The motivation for a theory which assigns semantic structure is independent of the demand for a theory of truth. On exactly what this demand depends is unclear. Davidson (1965) ties it to learnability,41 and to the 38 Ultimately, the standard name can be in terms of the categories and constructions out of which the sentence is built. Perspicuous sentences will have structural descriptive names consisting only of semantically relevant constructions. 39 A further condition prohibits 'the appearance of a semantic term in the statement of the truth conditions of a sentence unless it already contains a semantic term' (Davidson 1970b : 179). ,0 It also explains that it is the proof that is of interest for structure and not the biconditional alone. 41 Davidson's argument for recursive semantic structure on the grounds of learnability (1965) may not be as strong as first appears. The catch is that there are other notions of learnability, sensitive to infinite accomplishment by finite means, which do not require this; for example, the sense in which we would be said to have learned an infinite vocabulary when we have mastered in a finite time the ability to use an infinite dictionary. Thus Davidson must switch to an explicit argument that natural language is learnable in his sense. Though this seems by and large the case, it might break down in key spots. The stimulus for these points is due to S. Morgenbesser.

ON LOGIC AND THEORETICAL LINGUISTICS

63

ability to produce and understand an infinity of sentences after mastering a finite vocabulary and a finite number of rules. Fodor seems to rely on only the second of these (1970b: 300). Further, the idea that truth conditions in some sense are an integral part of semantics would remain even if no finite axiomatization were possible. It is just the possibility of such a theory, however, which allows Davidson to proclaim his theory to be a semantic theory. How then does Davidson propose to meet his adequacy condition? Until now he has exploited precisely the standard logic which we outlined above. It will be recalled that in defining logical truth we used rules of the form' "pAq" is true if and only if "p" is true and "q" is true'. When we went internal we saw the need to call on satisfaction (and with it, sequences) as our semantical notion. Either way, the rules match Davidson's in style: on the left, we name a (closed or open) sentence, and on the right, we use a sentence suitably correlated. The logical constants, whose semantic recursive function was duly noted, serve here in exactly the same role, though new importance is attached to the satisfaction clauses for the non-structured elements. Certainly there must be only a finite number of these clauses.42 All this considered, proof of the required sort of bi-conditionals will be forthcoming. What this means is that we have a truth definition in English for our canonical notation. But we also have a correlation scheme for a sub-part of English to the canonical notation. Together these constitute a truth definition in English for the sub-part of English: the canonical notation merely fades away. 43 In short, for at least that part of natural English covered by standard logic, we already have the required theory. This consequence is not at all surprising, for Tarski tailored his truth definition to fit the language of standard logic. More generally, Davidson has reiterated our points about the relation between a theory of truth and a theory of inference (logic) for a language (1969:160). If then Davidson is correct in claiming that semantics amounts to a truth definition, and a truth definition is the core of logic, then logical form and semantic structure are interchangeable. Yet we should not be misled by the intimate connection between Davidson's theory and logic into thinking that there are no significant differences between them. The major difference is the knowledge we need to call on in framing the adequacy conditions of the theory. From the perspective of the adequacy condition, logic and truth definitions are at right angles. Logic tells us to characterize the logical truths which are presumed known. In so doing, we establish a correlation scheme and therefore do enough to handle the goal of a truth definition. The truth definition, however, only asks of us that we generate a bi-conditional for each senthe competence associated with conversational implicature (see pp. 46-47). But this must await development. 42 Davidson calls non-structured elements 'primitives'. This is not the old sense of 'primitive' as absolutely undefinable, or even undefined in the theory. 43 Actually the canonical notation fades away only in the instances of the theory. It is as essential to the theory over-all as is the form which it graphically captures.

64

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

tence of the language. It takes a correlation scheme as the goal. This generating of the bi-conditional turns out to be sufficient to result in a theory of inference. But clearly we ask less, and ask it more simply, when we start with a call for a truth definition. Further benefits accrue to one who seeks truth definition directly. Specifically, the new angle admits of a more determinate distinction between a theory of logical form and a theory of term analysis. In particular, it vitiates the need to call on the distinction between perspicuous and non-perspicuous sentences. This is clear from the following considerations. Perspicuity in logic essentially depended on two relations: the relation between S and the representation of S in canonical notation on the one hand, and the relation between the representation of S and the representation of its truth condition in the meta-language. If we eliminate the canonical notation in which the representation of S is formulated and continue to assume that the truth definition for English is in English, then a sentence will be perspicuous in the theory if and only if the output of the satisfaction rules and definition of truth is a homophonic bi-conditional, i.e. the sentence named on the left is used on the right. Correspondingly, a sentence will not be perspicuous in the theory if and only if the bi-conditional is heterophonic. (Transformations which preserve truth might even everything up in the end.) But now consider the question of whether to correlate (1) 'unmarried man' with (2) 'unmarried and a man', or with (3) 'bachelor'. As in logic, the theory must be able to handle 'unmarried' and 'man' individually. Logic would lose inferences otherwise, and the truth definition would lose bi-conditionals with the sentences containing only one of the words. Thus it is simpler to correlate (1) with (2) than with (3). But, and here is the crucial point, once we make one correlation, the theory demands no more. Any other correlations are gratuitous. In logic, further correlations complicated the theory, but they still contributed to the success of the theory in meeting the adequacy condition. The distinction, then, between the two correlations could not be made solely in terms of the adequacy condition and simplicity.44 The same argument applies in the reverse. (3) must correlate with itself and not (2) for simplicity's sake. When this correlation is made, we are finished. The over-all result is that the theory itself demands that we adopt the 'Perspicuity Principle' which we formulated as independent. A closely related point is that in a truth definition, unlike in logic, we cannot correlate sentences in a way requiring the correlation of a basic element with just any other element of its category. This is so because the initial correlation must at least be a true truth-functional bi-conditional. Of course, if the correlation is a true bi-conditional, it does not mean that we have finished. The element-element correlation may unleash proofs of other bi-conditionals which may be false. To use Davidson's example, we may try to correlate 'Snow is white' with 'grass is green'; 44

Of course if Davidson required that 'unmarried man' relate to 'unmarried', then simplicity would be irrelevant.

ON LOGIC AND THEORETICAL LINGUISTICS

65

thereby, if successful,proving ' "snow is white" is true if and only if grass is green'. So far, so good — but how are we to avoid ' "John's pants are white" is true if and only if John's pants are green'? Again the fact that correlation proceeds by parts takes its toll. A wrong correlation will make mistakes in a bi-conditional as surely as it ruins inference. This point also explains why the referential character of Davidson's theory of meaning does not counter our intuitions that NOT just any true bi-conditional of the right form GIVES THE MEANING of a sentence. Davidson does not give up the distinction between bi-conditionals that do and do not 'give the meaning' of a sentence; he merely gets it to take a referential shape. Only directly provable bi-conditionals 'give the meaning' of a sentence. 45 The close identification of standard logic and theory of reference with semantics on Davidson's model is in sharp contrast to the skepticism of many linguists about the relevance of the logician's constructed language to any theory of natural language. This skepticism was clearly formulated in Chomsky's reply (1955) to BarHillel (1954). These arguments were further extended in Katz and Fodor (1964). Chomsky in his reply to Bar-Hillel made two claims. In order to understand the first, we shall sketch briefly the background of the controversy. Bar-Hillel argued that linguists should learn from logicians such as Tarski, Quine, and Carnap to do semantics rigorously. He adopted Carnap's position in Meaning and necessity (1964: § 45) that the theory of meaning (intension), unlike the theory of reference (extension), is relevant to linguistics proper. Grasping the meaning of an expression should be distinguished from knowing its application. The first activity is explained by the theory of meaning, according to Carnap, whereas the second activity, application, has to do with theory of reference. For the first only knowledge of the language is required, but for the second, knowledge of the world is involved as well. In reference to this distinction, Chomsky argued that Tarski and Quine, two out of the three mentioned by Bar-Hillel, deal explicitly with the theory of reference (mainly Tarski); thus their studies are not directly relevant to linguistics since, ex hypothesi, theory of reference is irrelevant to linguistics proper. Bar-Hillel had therefore made two incompatible assertions. The one on which Davidson's theory rests is the claim that the theories of Tarski and Quine are highly relevant to linguistics proper, and not, as Bar-Hillel had intimated, to its methodology only. Standard arguments for the irrelevance of the theory of reference do, in fact, seem to be based on a partial misconstrual of this theory.46 One can regard the theory 45

But of course it still does not follow that the distinction in referential terms is exactly as we would have it; alternatively, in the terminology of what follows, having characterized some distinction between knowledge of the world and knowledge of language does not entail having adequately characterized either. This is just the basic question of whether a pure theory of reference can really be viewed as a theory of meaning. We have only been concerned to point out fallacious arguments against this reduction. More subtle arguments may well exist, however, but their exploration is beyond this work. For similar remarks see Hintikka 1969. 48 This misconstrual is found also in Katz's mentalistic semantics.

66

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

of reference as a theory of notions such as truth, extension, designation, and the like. But, as noted above, one must also distinguish theorems of the theory from just any true statements including true bi-conditionals which employ concepts of the theory. Not every statement that consists of concepts of the theory is a theorem of the theory. A sentence such as: (1) ' "All cats have tails" is true' is not a theorem of the theory of reference. In contrast the following most likely is: (2) A sequence S satisfies the universal sentence 'for all x, if x is a cat, then x has a tail' if and only if, for all sequences S which differ from S at most in an assignment to x, if S satisfies the sentential function 'x is a cat' then it statisfies 'x has a tail'. Statements like (1), which are not theorems of theory, do seem to require knowledge of the world in order to be known. But no such claim follows regarding statements like (2), which are theorems. Though the distinction between knowledge of the world and of language can be made in terms of Davidson's theory, no use should be made of this distinction as an adequacy condition of the theory. Thus, as noted above, Davidson need only require that the theory generate true bi-conditionals. To add that the theory must generate only truths constitutive of knowledge of the language would be a fundamental addition to the adequacy condition of the theory. It might be questioned, however, whether without such a prior restriction, Davidson is not committed to the claim that in CONSTRUCTING the theory, one must know a great deal about the world. The theory is mistaken only if it generates falsehoods at some point. But to be sure that a constructor of the theory could tell when a mistake was made would then seem to require that he know the truth value erf every material bi-conditional. This argument assumes that to check the theory we must know of true theorems that they are true and of false ones that they are false. But less is required. For, since we may say that we know the theory is true if and only if we know of every theorem that it is true, we may specify the principle: accept only theories you know to be true. (Other epistemic notions can be substituted.) Checking, then, does not require knowing of a false theorem that it is false. Not knowing that it is true is enough. This may eliminate certain otherwise acceptable theories since, though not false, we may not know them to be true, but we need not worry about acepting the false ones. 47 Davidson's theory does not include any special apparatus for distinguishing theory of meaning from theory of reference. Some distinctions which are thought to depend on this distinction, can, we have shown, be formulated simply in terms of the theory of reference.48 But the remaining concepts such as analyticity, though not integral 47

Compare Davidson (1967b) on omniscience. Our claim that a distinction can be made between knowledge of the language and of the world in terms of the theory in no way implies that there is a unique way of determining this relation, and therefore that the relation has any meaning, except as relative to the theory. It also 48

ON LOGIC AND THEORETICAL LINGUISTICS

67

to Davidson's theory, can be added via meaning postulates. We shall see below, in contrast, that much of the recent work in semantics makes the concept of analyticity integral to the theory without specifying any non-logical truths as analytic. The linguists did not reject the theory of reference, only because of its putative commitment to knowledge of the world. They looked askance, as well, at the constructed languages employed in this theory. We have already commented on the relationship of constructed and natural languages in our discussion of description vs. replacement. Several further points are appropriately noted here. Montague (1970a), Scott (1970a), and Davidson (1970b) claim that there is no theoretical difference between constructed languages and natural languages. The only difference is in the degree of explicitness of the formation rules. But we can formulate the rules governing natural languages and make them as explicit as the prescriptive rules of constructed languages. So the difference is not significant. A full grammar of a natural language will be counted as a grammar of a 'formal language', once its rules are explicit, complete, and formal, up to the standards of the logicians. There is no empirical evidence, not to say a priori reason, why this cannot be achieved. In principle, then, there is no theoretical difference between constructed languages and natural languages. Obviously, not all the constructed languages share the theoretical properties of natural languages. Natural languages are subjected to empirical constraints from which constructed languages may be free. Consider Davidson's criterion of learnability. It is not clear that languages such as generalized first order theories which differ from ordinary first order theories in having a non-countable number of predicates, functions, and individual constants, satisfy the criterion of learnability, since the number of semantic primitives involved is infinite. Still these languages have mathematical interest. An example of a different kind is ordinary first order predicate theory with identity but without singular terms. It may very well be the case that, because of empirical constraints, all natural languages contain singular terms. These differences are theoretical differences, but they do not affect Scott and Montague's basic claims; for Scott and Montague have not claimed that ALL the features of natural languages are preserved in ALL constructed languages. Tarski claims that there is a theoretical distinction between the two types of languages. He says, A characteristic feature of colloquial language (in contrast to various scientific languages) is its universality. It would not be in harmony with the spirit of this language if in another language a word occurred which could not be translated into it. It could be claimed that if we can speak meaningfully about anything at all, we can also speak about it in colloquial language (1956 : 164).

But Tarski proved that languages (of sufficient strength) cannot include their own leaves completely open the usefulness of the distinction between conceptual analysis and knowledge of the world, even relative to a theory. See Margalit (1970) for use of inductive logic in this respect.

68

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

truth definitions. Thus for any such language, there is something meaningful it cannot express. Davidson in "Truth and meaning" (1967b: 314) has questioned how seriously we should consider the putative universality of natural language. On the one hand, he suggests that natural language just may not be universal, though in his concluding remark he asks whether the very concept of universality may not itself be suspect. These seem to be fair questions and do, we think, justify the semanticist in moving forward despite the antinomies. In contrast, Davidson's remark that 'most of the problems of general philosophical interest arise within the fragment of the relevant natural language that may be conceived as containing very little set theory' is at best irrelevant. A further problem which Tarski mentions and Davidson discusses briefly (1967b: 316) is ambiguity. Davidson feels that ambiguous terms present no difficulty, certainly for a homophonic truth-definition. In that case, as he puts it, 'a truthdefinition will not tell us any lies'. If telling a lie requires that a theorem of the theory be false simpliciter, then such a truth-definition will tell none; for even if a term is ambiguous, at least two readings of a theorem containing it will be true. Yet such sentences may also be false on other readings. If we follow Davidson's comment, then we are forced at the very least, to revise Convention T in terms of not lying, that is false simpliciter or true on a reading. This is certainly to take notice of ambiguity in a way Davidson ignores. A parallel point for definitions of logical truth is made by Strawson (1957). Alternatively we might be able to handle ambiguity in a purely referential setting, without these changes. Either way, ambiguity cannot be ignored. Referential semantics partially vindicated, we return to the details of Davidson's theory. The structural framework in which we are good at doing truth definitions is the standard formalism of first order predicate logic plus identity. This means that our semantics is only as advanced as this logic. As we noted in discussing this framework, there are numerous sentences which resist, at first blush, the tight jacket of so austere a theory. Often, however, resistance melts away at the suggestion of new combinations of old structures. Davidson hopes to seduce all the sentences of the language by exploiting, fully, attractive combinations of dull elements. This 'conservative'49 though rather appealing approach emerges in his handling of action sentences. In a by now familiar paper, "The logical form of action sentences" (1967a), Davidson proposes a way to handle action sentences, designed specifically to explain the 'variable polyadicity' arising from the unlimitedly many prepositional phrases they may contain. Let us review for a minute the problem. From a logical point of view, we want an account of the entailments among (1) Jones buttered the toast in the bathroom at midnight. (2) Jones buttered the toast in the bathroom. (3) Jones buttered the toast. 48

So dubbed by Clarke.

ON LOGIC AND THEORETICAL LINGUISTICS

69

Clearly we must find some general rule, since there are an infinity of such examples. We might try to treat 'buttering' in (1) as a four-place predicate, in (2) as a threeplace predicate, etc., and correspondingly formulate a rule relating predicates of greater and lesser degree.50 But such a tack would create more wonder than it explains, even leaving aside the wonder of how an infinity of such predicates is generated. The rule would obviously pick up the four-place buttering-predicate, and relate it to the buttering-predicates of other numbers of places. But why should it relate these buttering-predicates? According to the theory they share nothing but some of their letters. What requires that they bear any syntactic or semantic relation to each other? Nothing, unless we add some semantic rules of generality equal to our 'inference' rule to explain the relationship. But this move would be just another way of introducing iterative structure into the sentences resulting in a repeated element 'buttered' of a fixed number of places. In addition to a general rule, then, we must construct a theory which attributes structure to these sentences, and which picks out a recurrent element in the related sentences. This demand is equally clear from the perspective of a truth-definition. Davidson's solution, after rejecting plausible but unacceptable alternatives, is to ascribe to these action sentences a form not obvious from their surface grammar. In light of our discussions of perspicuity, this solution amounts to associating with each action sentence a different sentence of ordinary language, which itself can be straightforwardly handled by logic and truth definition alike. In Davidson's words: The basic idea is that verbs of action . . . should be construed as containing a place for singular terms or variables, that they do not appear to. For example, we would normally suppose that "Shem kicked Shaun" consisted in two names and a two place predicate. I suggest, though, that we think of "kicked" as a three place predicate, and that the sentence be given in this form: 3 x) (kicked (Shem, Shaun, x ) ) . If we try for an English sentence that directly represents this form, we run into difficulties. "There is an event x such that x is a kicking of Shem by Shaun" is about the best I can do . . . (1967a: 92).

Thus the basic shift from the obvious occurs in the reclassifying of apparently n-place predicates as n+l-place predicates. Once this reclassification is accopplished, and with it the explicit specification of the range of some variables to include events, the remainder of the problem is solved by construing the prepositions as predicates attached to event names and descriptions, predicates i.e. which contain places for names of events and places, times.... Symbolically, (l)-(3) emerge as (1)' (3x) (Buttered (Jones, the toast, x) and In (the bathroom, x) and At (midnight, x)) 50

This approach should not be confused with one that makes 'buttering' a part of a 3, 4, . . . , n-place phrase. Here there is syntactic structure, and a recurrent part. Of course, a semantics would still be necessary, but its form would be clear cut. For example, the fixed elements would already be given some semantics. In the above case, we have no idea what to do with the semantics.

70

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

(2)' (3'x) (Buttered (Jones, the toast, x) and In (the bathroom, x)) (3)' (3x) (Buttered (Jones, the toast, x)) 'Buttered' is now constant three-place predicate, 'In', 'at' . . . are themselves predicates (of constant place) of events, and inference is handled by the standard equipment familiar to truth definitions.51 To sharpen the image of Davidson's theory, we might briefly compare it to one version of Reichenbach's suggestions (discussed in Parsons 1970). Reichenbach also correlates action sentences with other sentences of the language, thereby making the former non-perspicuous. Like Davidson, he specifies a range of variables to include previously unmentioned values. In Davidson's case, however, the values, events, presumably are not individuated intensionally. Reichenbach, however, explicitly draws on properties which are so individuated. Thus, corresponding to (1)' Reichenbach would suggest (1)" (3f) (f(Jones, the toast) and Buttered (f) and In (the bathroom, f) and At (midnight, f)) to be read: 'There is a relation f52 . . . such that Jones bears a relation of f-ing to the toast, and f-ing is a Buttering-relation . . . ' Clearly the technique here is fundamentally similar to Davidson's. Now, however, another difference connected to the value range becomes visible. Davidson makes the standard predicates take on a new place; Reichenbach achieves a similar effect by 'nominalizing' the predicate, and keeping it n-place while adding an additional conjunct. This approach is possible because in naming properties one can absorb the idea corresponding to being an event into the having of a property. How adequate is Davidson's proposal? Jerry Fodor (1970b) has argued that indeed it is not adequate at all. He presents examples to illustrate the deficiencies and then tries to zero in on the general ills creating the trouble. We are concerned with his assessment of the latter. Consider, he suggests, (4) John spoke clearly, and (4)' John spoke, clearly. 111

A nice use to which Davidson's theory can be put is in explaining why (1) and (2), though paraphrases and apparently built out of singular terms, which are coreferential when used by a single speaker can be one true and the other false for the same speaker. (Example is from Lewis 1970.) (1) 'I am talking in hexameter trochaic.' (2) 'In hexameter trochaic am I talking.' According to Davidson (1) and (2) must be analyzed with a variable ranging over an event of talking. But then, the references of the singular terms in the logical translation of (1) and (2) WILL differ for different talkings of the same speaker. 52 This use of second order logic need not be confused with a version which treats predicates as simultaneously names of properties. In such a system, , 3f (fx)' is read: there is a property f such that x is f. In our interpretation this is read: there is a property f such that x has f, where 'f is replaced by an abstract singular term (not a grammatical predicate). This latter reading could then be taken as the standard way of reading all predication or as the way of reading only certain style predication.

ON LOGIC AND THEORETICAL LINGUISTICS

71

clearly an ambiguous sentence. Wherein lies the difference? The most natural answer, he thinks, is in a difference in structure. In one, 'clearly' modifies 'John spoke', a sentence. In the other 'clearly' modifies 'spoke', a verb. Thus there are two distinct classes of adverbial phrases, 'sentence modifiers' and 'constituent modifiers'.5* Fodor contends, however, that Davidson cannot possibly handle the constituent modification version of (4). (As Fodor notes, Davidson does not claim to handle any adverbs — Fodor thinks this is irrelevant, and for purposes of our reply to him, we agree.54) That is, the parallel to Davidson's analysis of (1), namely (5): (5) (3x) (spoke (John, x) and clearly (x)) while being a 'natural way of paraphrasing', the sentence modification in (4) is a 'non-starter' as an analysis of the constituent modification in (4)'. This is true of all such cases. The general form F simply cannot handle constituent modification. F: (3x) ((V (pi, p2, . . p n , x)) and (Dx)) where 'pi', . . . , 'pm' (m < n) range over persons 'pm+ ¡', . .., 'p n ' range over persons, events, things, and 'x' ranges over events. T o put it succinctly, instantiations of (Dx) in a Davidsonian paraphrase are most naturally read as having the syntactic analysis (adverb (nominalized sentence)). But in (7) (our (4) interpreted constituency) w e have a modifier not o n a sentence, but o n a non-sentential phrase (1970b : 308).

This argument is peculiar, to say the least. Firstly, it is odd because a Davidsontype solution would be to treat adverbial modification by way of adjectives, thus making the second conjunct (adjective nominalized sentence). Though we write 'clearly (x)' it is just for heuristic purposes, just as 'spoke' retains its verb status. But both of these terms are given new form. Secondly, the argument is questionable because Fodor seems to assume that if a nominalized sentence is used, then the paraphrase is a paraphrase of sentence modification. But why should this assumption be so? At least it requires argument. In fact, however, it seems clear that Davidson's analysis is at best an analysis of the constituent modification version. This is clear if we consider 58

Fodor seems not to recognize that (4)' is ambiguous. But if 'it is clear' is as opaque as it seems to be, we need to give it a de re and a de dicto reading as well as the constituent modification. Now either the constituent modification is structurally different from the structures of (4) on both of these readings, or it is the same as the de re structurally, in which case 'clearly' is lexically ambiguous. Fodor seems to want the first alternative, since he denies that 'clearly' is ambiguous on the grounds that it appears in the unambiguous readings. (4) and (4)' seem to justify Montague's distinction between adverbs and ad-formulas. A grammatical argument for using these structures is: 'John spoke slowly' is formally like 'John spoke clearly'. But if the latter is a lexical variant of 'it is clear of John that he spoke', we should have 'it is slow of John that he spoke' which seems objectionable. 54 Note that Quine (1970 : 31) talks of how Davidson might do it.

72

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

(6) John spoke unclearly. Now Davidson has a ready paraphrase of this. Prefix 'un' to the 'clearly' of (5). Call it 51. Just as it should, (6) would then entail (7) John spoke. If Fodor were correct, and in addition (6) is unambiguous, how account for the plausibility of (5') as an analysis of anything? Alternatively, if we stretch things55 and give (6) the benefit of a second reading corresponding to (8) It is unclear that John spoke, (5') still could not represent this, since (8) does not entail 'John spoke'.59 The failure of Fodor's argument to show Davidson incapable of handling constituent modification is repeated in a second group of examples. Here, however, Fodor's error is more illuminating. One of the examples in which Fodor finds ambiguity due to constituent vs. sentence modification is (Fodor's numbering): (27) John aimed (his gun) at the target. Bracketed as in (28) At the target (John aimed his gun) he believes we have an unambiguous sentence modification, while bracketed as in (29) John ((aimed) at the target) his gun, we have unadulterated constituent modification. Fodor again claims Davidson to be unable to handle (29). (Quite peculiarly, he says that his main point is only that Davidson has no theory to handle both readings. Yet he also claims that being incomplete is not a serious deficiency of the theory. Rather, what is crucial is the general problem, namely constituent modification.) The view that (27) is ambiguous due to sentence vs. constituent modification is discussed at some length in Aspects. There Chomsky vacillates between taking sentence modification as it is, or as a verb PHRASE (as opposed to verb) modification. His point, however, seems much the same as Fodor's, and many of his examples, among them, 'John decided on the train', are close parallels to (27). Even if we agree that (27) is structurally ambiguous in much the same way as (4) and (4)' are, we should, we think, acknowledge a lexical ambiguity here, not found in (4)'. (27) is most naturally disambiguated by replacing 'at' with 'toward' or 55

The reason that this reading of (6) does not seem acceptable is similar to the unacceptability of 'p, but I don't believe it' in contrast to 'I don't believe that p \ 56 A reason for construing Davidson as intending to capture (4)' as constituent modification is that his analysis here is in contrast to that of 'On saying that' (1968), which is clearly sentence modification.

ON LOGIC AND THEORETICAL LINGUISTICS

73

'in the direction of', unlike the 'clearly' case where 'clear' was prominent in the most obvious paraphrase for each reading. More importantly, the phrase 'at the target' can itself be viewed as a 'directional' or 'locational' phrase. Surely here, this cannot be accounted for by structure. Rather we need to appeal to the readings of 'at' as well as the subcategorization of the following noun phrase. Thus 'at midnight' would fail to be ambiguous due to the selectional restriction on 'at'. Finally, without appealing to ambiguity, even if Fodor is correct, how could he explain why some prepositions can be verb- and sentence-modifiers with some verbs, and not with others?87 ' The recognition of lexical ambiguity may kindle the hope that all one needs in disambiguating (27) is such ambiguity. But hopes fade on recognition of the need to explain the unambiguous passive versions of sentences like (27). For example: (9) The boat was decided on by John. (10) The target was aimed at by John. 58 Without a significant difference of structure, it is not clear how to explain the variation which passivization introduces.68 Granted then a need for structural diversity in (28)-(31), and granted as well an appeal to lexical ambiguity, we may yet wonder whether Fodor has hit on the 'right structure'. Several points suggest that he may not have. Namely, (1) How is one to account for the ambiguity of the phrase 'aiming at the target' when there is no sentence modification — there is no sentence — and when the verb phrase is identical to the verb. Relatedly, how explain the ambiguity of 'John's aiming at the target' or 'John's aiming is at the target' in the face of the unambiguous 'John's speaking is clear'? (2) Why do we think of 'deciding on', 'aiming at' . . . as units by themselves, ignoring the remainder of the prepositional phrase? (3) Why is the preposing of the prepositional phrase so unconvincing a way of disambiguation, while the questions 'What did John aim at?' and 'What happened at the target?' are not ambiguous? 60 These questions suggest an alternative to Fodor's theory, one which makes the prepositional phrase ambiguity fundamentally dissimilar to the ambiguity of (4). Briefly it goes as follows: All prepositions can occur as independent elements which 57

These points, though crudely put, could, we think, be refined. (27) has no unambiguous passive, since the passive there is not related to the use of a preposition. 59 One should not confuse this 'structural' argument with the question of why (50) 'John aimed at something' is more or less unambiguous. Here the explanation is selectional. Thus one might accept as ambiguous: 'There is an x such that John aimed at x.' The problem is in adding 'thing', for then 'at' is locational, and the noun phrase names a location, but locations may not fit our ordinary notion of 'thing'. Thus (50) fits only the non-locational reading. 60 A possible further objection is how to explain why 'laughed at the circus' is ambiguous, and why on one reading 'the circus* fails to allow substitutivity salva veritate. It is not clear that Fodor's distinction explains this. 68

74

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

combine with noun phrases to form modifying prepositional phrases. Some prepositional phrases, of course, will fail to 'sit right' with certain verbs, e.g. 'John aimed until the target'. But until one provides a theory to account for the oddness of these sentences, if one does at all, they may simply be accounted false. ai Now, some phrases which APPEAR to consist of verbs plus prepositional phrases will turn out instead to consist of verbs plus noun phrases. Though apparently complex, these verbs will be unstructured. Thus we will find among the primitive predicates: 'deciding on', 'aiming at', 'working on', 'laughing at' . . . (cf. German aufbauen). Whether we ascribe this structure depends on whether we need to do so in order to account for the facts, e.g. passivization. Further, what the relation is between the unstructured predicates and the words which seem to be their parts will be left to the rest of the semantic theory, just as the oddness above was. Such practice, no doubt, means missing entailments similar to the entailments missed by not structuring 'single word' transitive verbs to relate them to their intransitive counterparts; but, as often noted, inferences are but a prima facie guide to structure. So long as we have only a finite number of such predicates, the theory will work. Davidson's approach can easily be adapted to this theory; 'on' will be a predicate giving us (11) John decided on the boat. (12) 3x (Decided (John, x) A on (the boat, x)) Additionally, since 'decide on' will be a primitive predicate, we will also find (13) 3lx (Decided on (John, the boat, x)) Of course, these structures need not account for all the ambiguities in such sentences. 82 Alternatively, one might try a 'structural' theory which would have the same results as this theory. Thus, corresponding to '(aim at) the target' in which 'aim at' is an unstructured lexical item we would have (aim (at)) the target. 'At' would be a preposition which could attach to a noun phrase to make a prepositional phrase which then modifies verbs, as in the earlier theory, or it could attach to a verb to make a new verb which has an additional place. Again, structurally all prepositions M

Having such a theory is compatible with Davidson, though it is not part of his adequacy condition any more than it is a characterization of analyticity, metaphor, etc. Pace Fodor, 'John left the train at the station' seems to be lexically ambiguous which is further justified by 'left at* not being able to stand alone, and by the total lexical non-ambiguity of 'John left his friend at the station'. After all, if John drives to the station and leaves off his friend there, both John and his friend part at the station. Similarly 'John will do the job in a minute' is lexically ambiguous. Above we claimed that 'at' and other prepositions be counted ambiguous. Its paraphrases will presumably give semantic variations of some non-structured predicates, 'aims at', 'aims toward' . . . Further, when 'at' means 'towards', we seem to be able to get two structures which are synonymous on one reading of the 'complex' structure. This will be satisfactory, since there will not be a passive of this sentence at all, although there will be of the 'simple' one.

ON LOGIC AND THEORETICAL LINGUISTICS

75

could do either, but selectional rules governing overlapping or contrasting subcategories of prepositions, nouns, and verbs, would eliminate readings. Here, ambiguity of readings will depend on structure plus lexical ambiguity of prepositions. " It is unclear whether Davidson could handle this theory because, as drawn, the distinction depends on allowing prepositions to attach to verbs, and this process might not be adaptable. This is not because the VERB is modified, but because the PREPOSITION would be functioning alone. But if instead one treated the difference as one of closed or (partially) open prepositional phrases attaching to the verb, Davidson might be able to handle it. For example, 'At (the target, x)' could be added to a phrase specifying the verbal element corresponding to 'x', or 'At (y, x)' could similarly be attached following which 'y' would be specified. These sentences would have different but logically equivalent structures, though lexical ambiguity would keep the meaning separate. When the lexical item is the same in meaning, the sentences, therefore, would be structurally different, but semantically equivalent. The passive rule would only function for one of the sentences and, consequently, would explain why it is unambiguous. It should be clear that either the 'structural' or 'non-structural' theory will explain the questions raised earlier. The upshot then appears to be that Fodor has provided no counter-examples to Davidson.64 Furthermore, if there are objections of a general sort to Davidson's program, constituent modification seems not to be one. Fodor, of course, is right in noting that if standard logic does not expand to handling constituent modification directly, i.e. homophonically, there is no guarantee we can paraphrase away all our worries. But this is no argument that we cannot. In the same vein, Fodor's comments about standard truth definitions prove little. While it may be true that the STANDARD WAY of formulating these truth definitions cannot allow the constituent modification to be added directly,65 the spirit of these truth definitions seems to be preserved in formulations which do allow this. In particular, though all constituents may not be the kind of thing that have truth con" Partially analogous is the way semantic assumptions about possible worlds affect whether de re and de dicto structures vary in meaning. M Standard logic (even without Davidson's analysis of action sentences) can also handle Fodor's points about negation, i.e. it can distinguish the basic structures: (1) ~ ;(3x) (x = J A Lxh)] (2) (3x) [Lxh A ~ (x = J)] (3) (3x) [LJx A ~ (x = h)] where 'J' = John, 'L' = left, and 'h' = his house. (35) It is not the case that John left his house, (36) It was not John who left his house, and (37) It was not his house that John left can then be translated respectively as (l)-(3), in which case what Fodor calls presuppositions turn out to be entailments, OR they may all be translated as (1), leaving (2) and (3) to indicate the presuppositions of (36) and 37). Either way, we do not 'lose precisely the structure we need to keep these distinctions [of presuppositions] clear' (Fodor 1970b: 312). 65 For more on this cf. pp. 79-80.

76

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

ditions, it is not at all clear that truth theoretic semantics need require that all the single parts have truth conditions. Predicates and not just open sentences can be assigned relevant interpretations. Even more appropriately, different truth theories can count as non-basic what standard theories count as basic — verbs, for example — and therefore need not give them truth conditions directly. Though Fodor's critique of Davidson seems of little moment, recent work on adverbial and adjectival modification does raise fundamental questions (Clarke 1970, Parsons 1970, Montague 1970a, Lewis 1970). Below is a sketch of various types of modifications in terms of inference patterns: A. Adjectives I. In predicative position: x is A (predicative) II. In attributive position: x is an AN a. analyzable as predicative via conjunction (predicative): x is A and x is an N b. not analyzable as predicative via conjunction (non-predicative): 1. standard non-predicative x is an AN -/->• x is N 2. non-standard non-predicative x is an AN -/->- x is an N Further subdivisions are possible (Clarke 1970). B. Adverbs I. Standard: x O's A-ly — > x 's II. Non-standard: x ®'s A-ly —>• O's C. Prepositional Phrases I. Standard: x ®'s prepositional phrase — » x x O's Logic provides us with a way to treat predicative adjectives. Davidson has further explicitly tried to handle action sentences with standard prepositional phrase modification. On the surface his approach lends itself to handling (at least) standard adverbs; this is what Fodor presented and what Quine has suggested. But this plausibility is superficial. For if events are individuated as Davidson wants, odd results are forthcoming; e.g. if John's driving crosscountry is identical to his traveling crosscountry (suppose he does it non-stop), then if he drives fast he must travel fast. But in the jet age the latter will most likely be false even if the former is true. More generally, Davidson's analysis works for adverbs only if they are treated as non-predicative adjectives. Thus, as yet untreated, even if Davidson's prepositional theory is correct, are adverbs, non-predicative adjectives, and non-standard prepositional phrases.86 These This does not exhaust Davidson's problem: (a) e.g. there are opaque verbs which would question his analysis of 'laugh at' and 'aims at' even in the first clause; (b) Clarke notes that 'John was drunk on Tuesday in the pub' . . . would by analogy have a different analysis from standard logic; (c) Parsons has noted the question of iterated adverbs.

ON LOGIC AND THEORETICAL LINGUISTICS

77

all share a feature which creates difficulties if the standard approach is used: apparent failure of substitutivity of co-referential terms.67 It is this feature that Fodor might have pointed to in order to question Davidson's ability to handle all sentences. Of course, the failure of the standard way to bring these expressions into the theory does not mean that Davidson will not find some way; witness his "On saying that" (1968). 68 It does raise the question, however, of whether we cannot find direct ways of handling these types of sentences through an addition of structural elements to the theory, and through extension of its apparatus. More clearly, extension may be possible through expanding the syntax and semantics of the theory and by relativizing truth to possible worlds. Wallace (1970) considered such relativization of the truth predicate in discussing how to deal with modal sentences in a truth theory. Any way other than relativization seems to fail. His conclusion, however, is that such an extension fails to meet the adequacy conditions of the theory, because of the 'expressive' incompleteness of the object language. It is difficult to evaluate the seriousness of this incompleteness. 69 Further, since much of the key work in formal truth semantics uses just this relativization of the truth predicate to possible worlds, we shall sidestep Wallace's arguments and consider these theories directly. After a general sketch of the form of these theories we shall return to adverbs and explain the alternatives for them which these theories offer.70 Carnap, as we saw, stressed the difference between the theory of meaning and the theory of reference. He also showed how to connect the two. His idea was to treat intensions as functions, the domain of which consists of state descriptions (possible worlds) and whose range consists of extensions. Thus a name will pick the thing named (if there is one) in each possible world, a one-place predicate, the appropriate set in each possible world, and an n-place predicate, a set of ordered n-tuples in each possible world. There is a natural way to extend the method to more and more constituents. Montague, Scott, and Lewis expanded Carnap's idea to include some more factors which are relevant for determining extensions. Among the factors we shall find, along with possible world coordinates, are speaker, audience, time of the utterance, place of the utterance, and the set in which the utterance occurs. (Montague mentions fewer factors.) Intensions, in the wide sense (Carnap's intensions 17

The failure is particularly obvious in the non-standard cases but can be made apparent in the others as well. 68 This counters Parsons' claim (1970 : 326) that we must quantify over intensions if we are to use extensional logic. 69 This difficulty does not apply to Wallace's other arguments in which he interestingly shows the effect of the structures a truth definition lays down. 70 In what follows, we consider a possible world semantics in only a narrow range of applications. We do not mean to suggest that the theory offers a wholly satisfactory solution to these problems nor to any others, for example, the failure of substitutivity in contexts of prepositional attitude. In fact, serious problems may well lurk here, when the failures of substitutivity are of logically equivalent expressions.

78

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

are intensions in the narrow sense), are functions which map indices, that is, ordered sets of factors (points of references) onto extensions. The syntax, at least in part, is characterized by an Ajdukiewicz-style categorial grammar, not by a phrase structure grammar familiar from the base component of Chomskyan grammars.71 In Bar-Hillel, Gaifman, and Shamir (1960) it has been proved that phrase structure grammars are equivalent to categorial grammars. Moreover, it has been proved that for each phrase structure grammar there exists a restricted uni-directional categorial grammar weakly equivalent to it. Lewis makes a further claim, however. He argues that by having three basic categories S (sentence), C (common noun), N (noun), and a uni-directional rule of cancellation, it is possible to have a grammar which is a notational variant of a phrase structure grammar. Furthermore, there are other triples of basic categories which can do the same job. In addition to the basic categories the grammar can generate an infinite number of derived categories, derived in the sense that if a \ . . . a n (n ^ 1) are basic or derived categories, then a/a x . . . a n is a derived category. In Lewis's grammar, an adjective is in the category C/C, while an adverb modifying an intransitive verb is in the category S / N / / S / N . The advantage in having a categorial base as an input to semantic theory was pointed out by Cohen and Margalit (1970:479). They suggest that the syntactic rule of cancellation can supply direction to the application of the projection rules. The rule of cancellation that Lewis uses is the following: let c / c j . . . cn be a derived category which is concatenated successively (with) cx . . . cn, then we shall get C as the category of the whole string; in symbols: c/c± . .., c„, c 1; . . . , c n ->-c. In terms of semantic interpretation we have the direction for applying the projection rules so that if 0O is the intension of the derived category c/c1 . . . c n , then the result of concatenating c/c1 . . . cn whose intension is 0 1; c 2 whose intension is 02 afld so on, until cn whose intension is 0n, is the compositional intension: 0o (0i . . . 0n) of c. As an illustration of how the theory might be applied, consider 'John drives slowly', a sentence with adverbial modification. John drives slowly S S/N S/N/ /S/N slowly

N John S/N drives

71 Lewis's categorial grammar differs from that of Ajdukiewicz and Bar-Hillel in having a third category — common noun.

ON LOGIC AND THEORETICAL LINGUISTICS

79

Let / / f / / denote the intension of the structure whose surface representation is T. (1) / / J o h n / / is an n-intension; that is, a function from indices to things. (2) //John drives slowly// is an S-intension; that is, a function from indices to truth values. (3) //drives// is an S/N intension; that is, a function from N-intensions to S-intentions; that is, a function from functions from indices to things to functions from indices to truth values. (4) //slowly// is an S / N / / S / N intension; that is, a function from S/N intensions to S/N intensions; that is, a function from functions from N-intensions to S-intensions to functions from N-intensions to S-intensions; that is, a function from functions from functions from indices to things to functions from indices to truth values to functions from functions from indices to things to functions from indices to truth values. Thus, the intension of 'slowly' is that function 0 such that, if 0X is an S/N-intension, 0 2 is an N-intension, 0 3 is (0(0i)(02)) and i is an index, then 03 (i) = truth, if 0X expresses an activity-kind K72 and 0 2 (i) performs some activity of kind K slowly at the world and time given by the appropriate coordinates of i; falsity otherwise. Having sketched out the elements of the theories of Montague and Lewis, including an example of its semantic account of adverbs, to which our discussion of prepositions is easily assimilated, it may be useful to comment briefly on the overall relationship between these semantics and that of Davidson. The initial most conspicuous feature of the theories of Montague and Lewis is the widened list of syntactic and semantic categories, which enables the deep representations in their theory to mirror more closely the familiar grammar of natural languages. Such an extension of categories results in an increased homophonicity of translation into the theory as is clear from the fact that, in contrast to Davidson, verbs, adverbs, and prepositions are no longer misleading in their form. One might wonder whether Davidson could also extend the categories of his theory. As presently formulated in terms of satisfaction rules, there would, it seems, be a sheer grammatical limit to the possibility of such extension. Structure, after all, is indicated in such a theory by the interposition of clauses about satisfaction between the constituents of a complex expression. These clauses, then, must either be connected by sentential connectives or must be modified by one of the constituents of the complex expression. Simple rules of grammar will then determine the possibility of such modification. A separate category of adjectives and the distinction between 72

An S/N-intension 0, expresses an activity kind K iff, whenever 02 is an N-intension, 0 3 is 01 (02), and i is an index, then 03(i) = truth, if 0 2 ® performs some action of kind K at the world and time given by the appropriate coordinates of i; falsity otherwise. Here and throughout this section we are indebted to David Lewis for kind assistance.

80

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

adverbs and ad-formulas 73 would thereby seem to be ruled out. It is plausible, however, to view all this as an artificial limitation due to the way truth theories are formulated. We might reformulate the truth theory in a way that allows for the expansion of categories. At this point, however, the fundamental difference between Lewis and Montague on the one hand and Davidson on the other emerges. Davidson's theory, so reformulated, would treat the intensions of elements of basic categories as functions from indices to extensions which indices, though referring to speaker, t i m e , . . w o u l d not include possible worlds. This would result, if only the categories of Lewis and Montague are used, in an inability to handle failures of substitutivity of co-extensional terms, and would permit the theory to cover only those cases which Davidson previously handled heterophonically. The key, then, to Lewis and Montague's success is the increased power of a theory which makes use of possible worlds. The use of a possible world semantics also has its effect on what inferences are directly covered by semantics. Thus, on Montague's theory, 'if x is an unmarried man, then x is unmarried' is not logically true. This differs from Davidson and the reason is clear. Davidson's prevalent mode of giving structure is via the logical constants; therefore all structure yields inferences. Here it does not. On the other hand, in Montague 'alleged communist' and 'unmarried communist' have the same logical form. 74 Montague's theory could, however, characterize such sentences as 'if x is an unmarried man, then x is unmarried' as logically true, by adding new categories.75 Or one could specify the inferences as valid in some sense, other than logical; e.g. analytically. This could be achieved either by adding meaning postulates for specific words and calling a sentence analytically true if it is logically true under replacements, or by adding postulates such that a sentence is analytically true if it is deducible from a postulate, e.g. every d £ is a Finally, one might simply increase the set of logical constants. A brief comment about ontology: if the functions of Lewis and Montague's theory are viewed as meanings, as Lewis and Montague seem to view them, then meanings do seem to have an integral role in the theory of meaning. The role that they do have, namely, as required in the meta-linguistic formulation of the theory, should not, however, be confused with the service they perform, e.g. in Reichenbach's conception, as references of object-language expressions.

73

Ad-formulas themselves can easily be added to the truth theory, though, as Wallace has shown, in discussing homophonic modal logic, the subsequent addition of replacement restrictions will yield false or non-sensical results. 74 Anyway Davidson cannot complain that these inferences must be logical, given the flexibility he notes in 1967a. 75 Montague calls this 'changing the definition of logical truth' which it is if 'logically true' is defined relative to specific categories. Note also that the new categories are characterized via logical constants.

81

ON LOGIC AND THEORETICAL LINGUISTICS

What seems to pose a serious problem for referential semantics in the style of Davidson or Lewis and Montague is the treatment of non-declarative sentences. All the referential semantic theories share the assumption that truth and falsity are properties of declarative sentences (or tokens of such sentences). If a theory of meaning turns out to be only an extended theory of truth, then how are we to treat sentences such as interrogatives or imperatives which prima facie fail to be truth bearers? To this question Lewis replies that: (i) The distinction between declaratives and non-declaratives is a syntactic distinction with regard to surface structures only and has no bearing on semantics. (ii) There is always the possibility of reducing all non-declarative sentences to sentences with performatives which constitute a subset of declarative sentences. Thus the sentence (13) Don't be late! can be reduced to (14) I command you not to be late which is a declarative sentence. In order to grasp the novelty of Lewis' approach, one should compare it to its natural alternative (as Lewis does himself). A relatively established alternative suggests that every sentence in a language should be analyzed into two basic components, dubbed 'sentence radical' and 'mood'. The first component presents the descriptive content of the sentence, whereas the mood 'determines' whether the state of affairs described by the sentence radical is the case, may be the case, ought to be the case, etc. Truth and falsity are properties of sentence radicals (or what Frege calls the 'supposition' and Hare the 'phrastic' of the sentence). In categorial grammar the analysis of an interrogative such as (15) Does John drive? in terms of sentence radical and mood is Sentence

Sentence Radical

Mood

drive

John

Sentence

(16)

kill

Caesar

Sentence

(18)

Cassius kill Caesar

83

ON LOGIC AND THEORETICAL LINGUISTICS

The semantic analysis is done in pairs of moods and sentence radicals. Sentence radicals will get truth values as extensions, and a sentence as a whole will get its truth value derivatively, in terms of the embedded sentence radical. For the sentence radical we shall map indices on extensions or intensions as required, but for moods we shall have arbitrary objects. Obviously the interesting things to say about moods are left to pragmatics (Stenius 1969). The above analysis can be applied to more complicated cases such as: (16) Did someone kill Caesar? Following Lewis we shall use the abstraction operation in analyzing (16), so we get (17) Someone is something x such that x killed Caesar. But what happens if we move from yes-no questions to wh-interrogatives? For some types of wh-interrogatives the analysis in terms of sentence radicals and moods is (18) Who killed Caesar — Brutus or Cassius? But once we have wh-interrogatives of the type (19) Who killed Caesar?, the method described collapses. The sentence radical is an open sentence (i.e. a sentence with free variables), and it is pointless to say that the sentential function (20) x kills Caesar is true (or false). In Lewis's method the sentence (19) is paraphrased into (21) I ask you who killed Caesar. S

N

S/N

I

S/N//N(S/N) ask - who S/N//N

|

kill (past)

Caesar

84

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

What about the truth value of such a sentence? Well, 'I ask you who killed Caesar' is true in a world, at a time, for a speaker, and for a hearer if and only if the speaker asks the hearer in that world at that time who killed Caesar. Presumably, if 'asks' is taken very literally, then when Marc Anthony 'asks' his rhetorical questions, they are just false sentences since he does not ask them. How many moods are there? Wittgenstein (1963: §23), who characterizes moods as moves in a 'language game', says that the 'number is countless'. But to argue that there is an infinite number of moods in syntactic-semantic theory is to run the risk of having an infinite number of primitives. Lewis says (1970:165) that the number is finite; although we do not know the upper limit, we know that it is not too high. In any case, Lewis discusses the possibility of mood reduction, so that we shall have only those moods to which can be ascribed truth value. Aquist presents an interesting example of such a reduction of interrogatives. He characterizes normal circumstances for asking questions in the following way: let the interrogative be 'Is the philosophy of language still alive?' (i) The questioner does not KNOW an answer to the question. (ii) The questioner COMMANDS an appropriate audience to remove his ignorance (in relevant respects i.e. as to the state of the art). According to this characterization an interrogative is equivalent to an imperative sentence which includes in addition to the imperative operator 'let it be the case' an epistemic 'I know that', and so the translation is 'Let it be the case that I know that the philosophy of language is still alive or know that it is not alive'. Aquist proposes a semantic system for the reduction program of interrogatives. The system is a mixture of imperative logic and Hintikka-type epistemic logic. The advantage of this system for our purpose is that in such a system we have the definitions 'true in a model' and 'false in a model' and so we can apply truth and falsity to interrogatives, although indirectly.

VI.

LINGUISTIC THEORY

Until now, we have concentrated our attention on the interrelationship between logical- and 'formal'-semantic structure. Waiting in the wings, throughout, has been the grammatical theory of Chomsky and his 'followers'. We conclude our paper by turning to this theory and exploring some of its features which are relevant to previous discussion. Reversing the normal procedure, we begin with the semantic theory developed and maintained by Katz, Fodor, and others (hereafter CT, 'the classical theory'). As we have grown to expect, a key aspect of the adequacy condition of this semantic theory is that it explains how the meaning of complex expressions is determined by the meaning of their parts. Additional conditions emerge, however, as CT requires from any acceptable theory an account of analyticity, entailment, paraphrase,

ON LOGIC AND THEORETICAL LINGUISTICS

85

semantic anomaly, ambiguity, and other semantic properties and relations. CT envisions the test of such a theory to be a straightforward matching of the theory's output e.g. 'S is ambiguous' with the judgments of fluent speakers of the language. However problematic this form of testing may be, we want to know as well the comparable test of the 'compositional' adequacy of the theory. One might claim that no additional test is required since any theory which met the above conditions would automatically meet this one. Still, the question of a direct test is fruitful, and we shall pursue it briefly. The most plausible suggestion, based on our earlier remarks and those of Katz, is that a theory is compositionally adequate if it generates, for each sentence of the language, sentence(s) of the form'S has reading R' (where 'S' is replaced by a name of a sentence and 'R' by the name of a reading; the basis of the generation would be a finite set of sentences about readings (senses) of lexical items (constituents); the means, semantic rules. Readings of sentences are taken to be thoughts; those of constituents, ideas. Thus the entities to which linguistic expressions are tied are language independent. Without sacrificing language independence (if this is in fact achievable) one might modify the above condition slightly by requiring the provability of sentences like 'S reads that R' where 'R' is no longer replaced by the name of a reading but by a linguistic expression which is used. 78 In any case, we should distinguish the reading of a sentence from its special representation in the theory, the latter being the expression which replaces the 'R\ These representations, unlike the readings and other entities which are represented, are not language independent. They need not, of course, be expressions of any natural language, but if they are not, this will have little theoretical interest since, as we saw with the canonical notation of logic, eventually the representations must be tied to a natural language. We might as well think of them, therefore, as expressions of a familiar language. CT frequently talks, however, of the vocabulary of these representations as 'theoretical' and 'universal'. It maintains as well that 'the vocabulary offers a language independent means of representing the common conceptual system underlying communication in natural languages' (Katz 1967: 129). But language independence here is different from what we encountered in the case of readings. Its significance is rather that the semantic representation be such that 'it is possible to formulate the fact that translationally equivalent sentences express the same conceptual content by giving each sentence in a translationally equivalent n-tuple the same semantic representation . . . ' (129). The existence of a unique semantic representation for each such set is at the heart of Katz's proposed mechanical tests for paraphrase, analyticity, etc. Essentially it hinges on the assumption that there is a finite vocabulary of unanalyzable terms (atomic markers) 76

These variations are discussed in Davidson 1967b: 305ff. The new form might run into grammatical difficulty in the rules for constituents. See earlier remarks on the grammatical limitations of Davidson's truth theory.

86

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

into which all concepts (or any lexical item in any language) can be analyzed, the analyses presumably making use of whatever syntactic or semantic structures exist in natural language.77 That the atomic markers must be undefinable seems clear if we are to guarantee that all paraphrases have a single semantic representation (whether it is possible to determine paraphrases mechanically without this latter requirement is unclear, though if it is, it would seem to require the notion of all possible definitions of a term; for if we could define a marker by others, then the resulting semantic representation would also be a legitimate representation). The strength of this 'undefinability' assumption should not be underestimated. It imparts to the goal of analysis an 'ultimacy' not found in the requirements of normal reduction programs which the project otherwise resembles. It further leads to the conclusion that all the atomic markers have isomorphic (hence, atomic) translations in every natural language — at least, if all natural languages are expressively equivalent. We shall not pursue any further the issue of analysis of terms and the semantic properties which depend on it, for these have been conspicuously irrelevant to our foregoing discussion. Furthermore, the discussion of the lexicon in the literature of linguistics has been weak, never confronting the numerous issues about definition and analysis posed by philosophers from Plato on. Instead we shall concentrate on the compositional aspect of semantics. To this end, we shall assume that the sentences to be proved in the theory make use as much as possible of the very lexical items mentioned.78 One final comment before pushing on. We remarked that semantic representations must be in natural language, preferably the language for which the theory is being given. It is clear, however, that sentences of the form 'S means that P' in which 'P' is replaced by the sentence named by the name replacing 'S' could not be used to explain to anyone the meaning of S. Equally, one could not be said to learn the meaning of S from such sentences. Quine, noting the parallel point for the homophonic 'constant clauses' of a Tarskian truth theory suggests that the purpose of such a theory is simply misconstrued, if taken as an explanation of the constants. Following Tarski, he sees the purpose of the theory 'not as explaining negation, conjunction, and quantification, which would be untenable, but as contributing to a definition . . . of truth' (Quine 1970:40). Quine's distinction in no way requires that a person could be said to 77

The use of syntactic structure in the lexicon is blurred in some of Katz's formulations but see Bierwisch 1969. 78 Ignoring the lexicon will also save our reviewing the obvious defects in CTs definition of properties like analyticity, etc., on which cf. Linsky 1971. Note that Bierwisch 1969 contains some replies to Linsky's objections. It is also worth noting that Linsky frees Katz from Quine's strictures on analyticity because Katz appeals to a list of antonymous n-tuples, the correct discovery of which is 'irrelevant to the question of the truth or adequacy of the theory discovered* (1970 : 442). But this seems to miss the strength of Quine's argument, if it be accepted, which seems to call into question the meaningfulness of any objective description of the list, and this makes impossible any discovery of it so characterized.

ON LOGIC AND THEORETICAL LINGUISTICS

87

understand the definition of truth without understanding the constants; nor that one could see exactly how the truth or meaning of a sentence depends on its structure without understanding those constants which give it its structure. It's just that knowing the constants is not the same vis-à-vis the homophonic constant clauses as it is vis-à-vis the truth theory as a whole. There is, however, a further distinction entailed by Quine's. For a definition of truth will be successful only if it shows how to give or present finitely the truthconditions of an infinity of sentences. The homophonic biconditionals then must be construed as presenting the truth-conditions of sentences. Thus emerges a distinction, applicable to individual sentences of the theory including the constant clauses, between explaining, and giving or presenting the meaning (or truth or satisfaction) conditions of an expression. It is a far more difficult and controversial question whether these distinctions constitute a reply to objections brought against the idea that in learning language one 'internalizes' a homophonic theory of meaning. This is too broad an issue to enter into here. We return now to the question of composition. What kind of rules will we need to carry through proofs of the required sentences? Most generally, we shall require rules which specify in some form or other the meaning of complex expressions as a function of the meaning of non-complex expressions. As we have sketched the theory, its interest lies in the contrast of these rules to the satisfaction rules familiar to us from Davidson. The difference lies in the use of a non-extensional 'means that' in place of the extensional 'is satisfied iff'; alternatively, in the individuation of intensions in contrast to extensions. What Katz's theory would yield if completed would be an account of how the meaning of the whole depended on the meaning of the parts and not on the extension of the parts. Of course, it should be much easier to satisfy an intensional theory than an extensional one, since failure of substitutivity of co-extensional terms would in no way raise doubts about the functionalism of the theory — i.e. would in no way impede carrying out the proofs. The theory so understood is much more similar to the possible world semantics of Lewis, Montague, Hintikka, etc., than to Davidson's semantics, yet it diverges from all of these theories in its total neglect of truth. Thus we must confront a question postponed earlier, of how integral is the notion of truth, in whatever form, to the goals of semantics. In a sense, the issue has a fundamental character about it which makes it difficult to bring to bear further considerations. Two points are, however, in order. Katz agrees that semantic theory should provide an account of entailment. As Linsky (1970:447) has noted, however, in the absence of an account of truth, Katz has no reason to call analytic sentences true. More generally, it emerges as accidental that entailment defined within a CT-type semantic theory should be identical (if it succeeds in being so) with entailment normally characterized in terms of truth. This oddity is apparent in Fodor's suggestion (1970b: 315) that perhaps we

88

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

should distinguish the theory of meaning from the theory of truth or logical form while at the same time looking to the theory of meaning for the logical consequences of a sentence. At the very least, then, the requirement that semantic theory deal with truth would put its account of inference on more solid footing. Secondly, it seems a fair question to ask of a theory of meaning that it explain what meaning is, and it is hard to imagine any beginning in this direction without calling on the concepts of truth, extension, etc.79 One may argue that truth in possible worlds offers a minimum of clarification; but it would seem to explain, at least, why meaning ties language to the world.80 In this familiar slogan — semantics deals with language in relation to the world — we seem to reach the foundation stone. In the remaining discussion, then, we shall assume CT semantic theory to be some form of a truth theory, most plausibly one like those of Lewis and Montague. 81 To this could be added the program of analysis discussed above, thereby making applicable to the entire theory the standard definitions of logical truth, analyticity, etc. If our arguments are correct, such a theory is what CT-theorists really want, despite their disclaimers.82 What now of syntax? So far our comments have been restricted to the brief remark of Davidson's that recursive semantics does not mean recursive syntax plus a dictionary. In this comment Davidson had in mind the idea that semantically problematic sentences like those of 'prepositional attitude' have a fairly straightforward syntax. How then could semantic and syntactic structure be easily identified? We noted earlier that this comment might depend on Davidson's conception of semantics as Tarskian truth definition; and if Davidson's suggestion (1968) for handling propositional attitudes along these lines be accepted, it does seem that their syntactic and semantic structure will diverge. A further motive for viewing this divergence as due to Davidson's strong extensional requirements is the greater similarity to ordinary grammar seemingly enabled in Lewis and Montague by the weakening of this extensionalism. Still, one can wonder whether there is any a 79

It has gone unmentioned so far, but perhaps is worth noting, that truth theories and talk of meaning as truth conditions in no way is a positivist idea of meaning as verification. The latter, unlike truth, is an epistemic notion. Further evidence of this distinction is that a positivist could not be satisfied with a homophonic theory, in the terminology used above; positivists want to explain meaning. 80 Some seem to argue that a semantic theory must be given as a standard (and not a possible world) truth theory on the grounds that one gets more insights from the standard theory. (This would be a demand for the standard theory as opposed to the proclamation of p. 62). But getting more insights cannot justify a demand for the standard theory unless these insights are those needed to satisfy the adequacy conditions. Whether these insights are needed remains an open question on which see p. 76. Quine's indeterminacy question would seem to be a relevant consideration. 81 Again we must emphasize the cautionary remarks of fn. 70. 82 Katz and Fodor have presented their theory as 'in between' those of Oxford and the formalists. We have pushed them over to the formalists. Of course, if no coherent formal theory (see objections to logic as descriptive) is possible, neither is the CT program so interpreted. But it would not really be possible on their interpretation either, since they too want to 'formalize' ordinary language.

ON LOGIC AND THEORETICAL LINGUISTICS

89

priori reason, founded on the conceptions of syntax and semantics themselves, to interrelate syntactic and semantic structure. We shall argue that there is not. A key feature, perhaps the key feature, of Chomsky's contribution to grammar has been the introduction of deep grammatical structure into detailed linguistic description. This idea immediately bridges an initial gap in the attempt to view syntax and semantics together, namely, that in semantics one readily acknowledges the distance between appearance and reality. (Recall the critical role of perspicuity in logical theory.) This acknowledgment was not always present, and Russell at one time criticized Leibniz for ignoring grammar (surface grammar!). But throughout most of the development of analytic philosophy in the twentieth century, the distinction between apparent and real form has been central. Chomsky, of course, refers to this distinction in a footnote in Aspects, and there is little reason to doubt its influence. If then, syntax has its superficial form, logicians and philosophical semanticists can no longer blithely contrast logic with grammar and must, if they choose to redefine the contrast, locate the surface-deep distinction in grammar in their parallel logical scheme. The obvious proposal is that there is an a priori connection between the deep grammatical and semantic structures. The best way to check this would seem to be to review the types of arguments which justify agreement that a sentence S has form F. Logical and semantic arguments for the assignment of structure are fairly uncomplicated. In the case of inference, we test a proposal by seeing whether, relative to the rules of inference defined for the proposed structure, a sentence would be assigned a mistaken consequence relation. Davidson and truth theorists generally test their theories on the acceptability of the biconditional which are proved. Finally, if we consider semantic theory according to CT, the properties of anomaly, paraphrase, and ambiguity function as do the biconditionals and inference to directly test a proposal. In all of these cases, we match a property assigned via the deep semantic structure with one assumed known for the sentence. No rules intervening between the deep structures and sentences are called on, though there exists the assumption that such rules can be specified. In contrast, arguments in transformational linguistics juggle the dual tests of the grammatical transformation and the grammaticality of the generated sentence. There need be no matching of the grammaticality of the sentence with some property which the theory assigns on the basis of a proposed deep structure. But what remains true, in any case, and this is what is relevant to our question, is the lack of any a priori reason to see, as connected in a specific way, the structure that makes for grammaticality and the structure that makes for inference, truth, ambiguity, etc. The separateness of these properties is rendered unclear by the use of the latter in arguments for the former. A typical argument might run: S has a different deep structure from S* because when transformation T is applied to both on the supposition of same structure, it yields S' and S*' such that the semantic relations between

90

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGAUT

S and S' are different from those between S* and S*' (or S* and S*' are falsely assigned a semantic property P). But what kind of structure is this an argument for? If syntactic, then it requires at least the assumption that the deep syntactic structure is used in determining the semantic properties of a structure and the entailed assumption that transformations preserve these relations. But this has no a priori plausibility. We do not mean to suggest that most arguments are of this form. More often they show the interconnection among assumptions of structural description, transformation, and grammaticality. Thus, though Chomsky uses a 'semantic' argument for 'expect' and 'persuade', he is careful to point out that in the transformational explanation of the acceptability of such phrases as 'his eagerness to please', 'he is eager to please us', 'for us to please him is easy', 'it is easy for us to please him', 'he is an easy fellow to please', etc., but not the corresponding forms 'his easiness to please', 'he is easy to please us', 'for us to please him is eager', 'it is eager for us to please him', 'he is an eager fellow to please', and so on . . . one can hardly seriously maintain that these (corresponding) forms are merely semantically excluded as nonsense (1962 : 549). Furthermore, we are not claiming the illegitimacy of semantic arguments for deep syntactic structure in the context of developing an empirical claim about the interrelatedness of grammar and semantics, anymore than we object to the semantic fruitfulness of transformations as an empirical argument for a transformational syntax. Our only aim has been to bring out sharply the empirical nature of these relationships. Though we have found no reason for seeing any necessary connection between syntactic and semantic structure, there is a very general consideration, familiar as an oft-repeated goal of linguistics, which points to SOME strong connection between these forms: the need ultimately to explain the verbal behavior and linguistic capacity of those who know a language in terms of a connection between sound and meaning. The high level of generality of this consideration warrants only a vague claim that it is most likely that any acceptable linguistic theory will integrate in an empirically significant way the structure relevant to an explanation of phonology, syntax, and semantics. Calling any theory integrated88 which presents such a relation, we shall explore theory sketches of several presently proposed integrated theories in the context of our foregoing remarks. Before the details, one cautionary remark: the truth theorists (philosophical logicians) seem to differ from the linguists in their attitude toward syntax. The former suggest that semantics, not syntax, is the hard, significant stuff. Montague (1970a) advocates working on syntax and semantics at the same time, but his point is that there are many adequate grammars, and one should develop one relevant to semantics. (His remark is, he feels, important since the 'relevant syntax' probably is not the simplest.) " Obviously, following Katz and Postal.

ON LOGIC AND THEORETICAL LINGUISTICS

91

The contrast between this attitude and Chomsky's is clear. While Montague makes it sound as though there is more 'picking' than 'discovering' in syntax, Chomsky talks of a uniquely correct syntactic theory, the discovery of which is immensely significant. Syntax, too, is of great interest. These contrasting attitudes towards syntax, of course, have their impact on any discussion of the interrelationship between syntax and semantics. Without committing ourselves to any uniqueness claim for grammar, or the correlative claim of objectivity for deep structure, we shall assume that syntax is a theoretically interesting area, bound by significant enough constraints to require talk of good and bad theories. Thus our comments about integrated theories refer to an integration of theories which themselves significantly satisfy syntactic and semantic adequacy conditions. Integrated theories all have their pineal glands. What interrelate at this point in the theories, we have assumed, are semantic and deep syntactic structures. Nothing we have claimed so far about syntax and semantics, however, rests on reference to deep syntactic structure in contrast to any other non-terminal structure represented in the generation of a sentence. But the specific proposals which we shall consider do talk explicity about deep structure. To avoid prejudice we must take care in characterizing deep structure. Fair to all would seem to be: The deep structure of a sentence S is that structure represented by the output of the phrase structure component of the grammar prior to the application of any transformations; that is, represented by the first tree or phrase-marker in the derivation of the sentence. Equivalent to trees are ordered sets of labels and brackets. Formatives may be considered present or absent. We shall consider them absent. (Occurrence of identical formatives within a sentence must, of course, be indicated.) The generative derivation of a sentence contains representations of numerous structures, specified in more or less detail. The terminal sequence of labels and brackets represents the surface structure. Subtract some of the labels and brackets of constituents — ultimate to immediate — and variations emerge. A particularly important variation is the SHALLOW STRUCTURE represented by the bracketing minus the labelling of the terminal representation. The interest in non-terminal representations derives from linguistic explanations afforded by their failure to be identical with terminal ones. The degree to which surface or shallow representations resemble deep ones is a measure of the syntactic perspicuity of the sentence.85 Having ignored the presence of specific formatives, we may compare in one sweep some possible relationships between terminal and non-terminal representations of sentences. 84

We are not able to pursue this topic further. A deeper discussion must, however, take account of Chomsky's comments about kernel sentences vs. underlying strings. These comments are also relevant to the issue of logical formulae as semantic representations.

92

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

Most interesting are the folowing configurations: 1. a single sentence S f single terminal bracketing 1 with a { multiple terminal labelling J e.g. Flying planes can be dangerous 2. a group of sentences i single terminal bracketing 1 S i . . . S n with a | single terminal labelling j e.g. Sj I persuaded John to leave S 2 1 expected John to leave

Î Î

single shallow structure multiple surface structure multiple deep structure single shallow structure single surface structure multiple deep structure

multiple shallow structure 3. a group of sentences Jmultiple terminal bracketingl and a { multiple surface structure S i . . . Sri with a { multiple terminal labelling J single deep structure e.g. Si They headed the tub S 2 The tub was headed by them

Although formatives are omitted from the definition of structure, the above summary indicates the need still to distinguish the possible relationships between terminal and non-terminal representations in a single sentence case (one set of formatives) from a group case (several sets of formatives). Thus, though having different shallow structures entails having different surface structures, a single sentence with multiple surface and shallow structures unlike a group must have different deep structures (except in trivial cases). Conversely, while the presence of multiple deep structures among a group of sentences does not entail that there are multiple surface or shallow structures, a single sentence with more than one deep structure would seem necessarily to have different surface structures. All of this is compatible with the various integrated theories. What the theories do differ over, as far as this can be reconstructed from their often vague and imprecise formulations, is the extent to which semantic and deep syntactic representations are integrated. Alternatively, they disagree over the amount of semantic information encoded in the deep structure. Given our decision to restrict attention to the compositional aspect of semantics, this amounts to a disagreement over how much the deep structure determines the semantic opposition rules. Finally, and again equivalently, they propose differing views about the degree of overlap among transformations connecting deep and surface syntactic representations and the rules connecting deep and surface semantic representations. In the integrated theory of Katz and Postal the deep structure completely determines the applicability of the semantic projection rules. This does not entail, for them, an identification of syntactic and semantic representations, however, for the latter includes the representation contributed by the analysis of formatives into semantic markers. Furthermore, there may be syntactic information in the deep structure which is irrelevant to semantics. Having chosen to ignore analyses into semantic markers, we may then say that the syntactic representation contains all

ON LOGIC AND THEORETICAL LINGUISTICS

93

the relevant semantic information. Each bit of significant semantic structure is defined by a projection rule which applies uniquely given the presence of certain syntactic structures. 'So there is one projection rule for attribution, one for the subject-predicate relation, one for the verb-object relation and so on' (Katz 1967 : 128). The use of deep syntax to define the semantic notions of logical subject and object well exemplifies this theory. Bierwisch (1969), while 'logicizing' Katz's formulation of the lexicon, specifies in some detail the interconnection between syntax and semantics. In a more popular article, he summarizes it as follows: The semantic representation of a sentence as a whole is derived from the syntactic deep structure by certain universal operations that combine the meanings of the lexical elements of a deep structure according to the relevant syntactic relations. Several proposals have been made for the formulation of the procedure. I will follow here the version suggested in Bierwisch (1969). This proposal relies on the fact that all noun phrases, except for predicate nominals in sentences like Alexander is an unlucky fellow, are referential constituents, i.e. are constituents of the sentence which refer to objects rather than describing them, and are therefore marked with a reference index, say an arbitrary natural number. Two noun phrases with identical indices refer to the same (set of) object(s), those with different indices refer to different objects. Thus in When they^ came in, Paul2 asked the studentsj the pronoun they is coreferential with the students. Given these referential indices, the semantic interpretation of a deep structure is then derived by two operations. The first of these interrelates the semantic components by means of appropriate arguments. Thus an argument X s of a verb or an adjective or a predicate noun is replaced by X ; if i is the referential index of the subject noun phrase. An argument X^ is replaced by Xj if j is the referential index of the direct object. And so on for all syntactically indexed arguments in the lexical elements. The second operation connects the meanings of the individual words by logical constants, mainly (but not only) by 'and'. Thus we would get (23b) as the semantic interpretation of (23a) if we abbreviate the specific semantic components of dog as DOG: (23) (a) [[the boy] NP* [kills [the dog] NP*] VP]S (b) HUMAN X t and MALE X1 and not ADULT Xx and X t CAUSE (X2 CHANGE TO (not ALIVE X 2 )) and ANIMATE X2 and DOG X2 I have oversimplified the matter in several respects. I have ignored, in particular, the fairly complex problems involved in quantifiers, like all, many, some, in definite and indefinite determiners, etc. The meaning of these elements must be incorporated in the semantic representation of sentences essentially in the form of certain operators which further specify the referential arguments X;, Xj, etc. It should be clear, however, that in principle the meaning of a sentence can be derived in a definite form on the basis of the meaning of its words and its syntactic deep structure, and that this derived meaning represents the crucial properties of its cognitive content in a plausible way. If appropriately refined the semantic structures arrived at in this way will presumably turn out to be nothing but a suitably adapted realization of the principles of formal logic. Hence the logical rules of transformation and deduction apply to semantic representations, explaining how we are able to carry out logical operations in natural language (Bierwisch 1970b : 178-9).

94

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

Bierwisch seems to construe his theory as identical to Katz and Postal's at least to the extent in which it integrates syntax and semantics. But it is unclear that this is so, for two reasons. The first is that the introduction of variables and eventually quantifiers would seem to introduce new structural forms for which projection rules must be specified. If so, deep syntax will not indicate all relevant semantic structure, straightforwardly.88 The second reason is that the introduction of the logical constants seems not to be mechanically determined by information in the syntactic representation, 'and' may be the most frequently used constant, but 'if-then' also has been shown to have a role. In short, the semantic composition rules, seem no longer to be determined completely by information in the deep structure. This creates an empirical gap between the theories. Bierwisch's general point about semantic and logical form 86 is, of course, in keeping with our earlier discussion. Our conclusion that semantic and logical form are interchangeable entails that representations of one are representations of the other. It also justifies our discussion of integration as the device for exploring the relation of logical form to syntax. 87 Lastly, we mention generative semantics. All prejudice falls by the wayside in this theory; integration is complete. Deep structure is claimed identical to semantico-logical form, and one set of transformations leads to its representation. Lakoff, a main proponent of generative semantics (along with Ross, McCawley, and others), presents an example where there is a single form and then writes: This is a trivial example of a case where there is a correspondence between grammatical structure and logical structure. It does, however, raise an interesting question. Is this an accidental case? Or is there some necessary connection between the grammatical structures of these sentences and the corresponding logical structures? Intuitively, one would guess that the connection was not accidental. If this is true, one would like such a fact to be represented in a theory of linguistic structure. Not all theories of linguistic structure guarantee that such a correspondence is not accidental. For example, the theory given in Chomsky's Syntactic structures leaves open the question as to whether such correspondences are accidental. The reason is that, in that theory, the sentences of English are to be generated by rules that do not take into account the meaning of the sentences. Any rules relating English sentences to their logical forms would be independent of the rules assigning those sentences grammatical structures, though the rules assigning logical form might or might not depend on the grammatical structures assigned by rules of grammar. To the extent to which a theory of grammar assigns grammatical form independently of meaning, to that extent that theory will be making the claim that correspondence between grammatical form and logical form is accidental (1970a: 152-3). 85

'Straightforwardly' here is crucial, for we defined perspicuous logical representations partly in terms of how the semantic rule applies. 8 * We say 'general point' since Bierwisch seems to suggest also that standard logic, possibly including second order logic, may be adequate. Such a suggestion is premature. 87 Janet Fodor, in her article "Formal logic and formal linguistics" (1970), assumes some general form of integration and then raises the question of whether, given that logic and semantics cover the same facts, they represent them identically. We have begun in a different place, but the issues covered are the same.

ON LOGIC AND THEORETICAL LINGUISTICS

95

It is not clear how to understand Lakoff's contrast between being accidental and being necessary or guaranteed. He explicitly claims that this theory is EMPIRICALLY better than less integrated theories and so does not seem to be disagreeing with our earlier remarks in this vein. In any case, generative semantics does seem to claim that the best grammatical theory will turn out to have identical syntactic and semantic representations. Generative semanticists seem to differ from Katz in their assumptions about the nature of semantic markers and their place in a full semantic representation. This means that on the surface, at least, the claim of identity for syntactic and semantic representation includes more than the Katz-Postal claim of integration. Katz has argued (1970) that indeed this is only an appearance. Our policy to ignore lexical analysis again frees us from discussing this. We cannot avoid, however, Katz's general claim that generative semantics is a mere notational variant of his theory. If in addition to abstracting from lexical analysis we sidestep the fact that generative semantics would seem to allow no syntactically relevant but semantically irrelevant information in the deep representation, the two theories do seem to make equivalent claims. And yet, an odd situation arises. For Katz maintains this identity even when considering as an hypothesis that semantic representations are formulas of the applied predicate calculus. The problem is that if semantic representations do take on this form, Katz's theory comes to resemble that of Bierwisch. But then the empirical distinction which we noted between Katz and Bierwisch should apply to Katz and generative semantics. Katz as well as Bierwisch seem not to see the distinction between their own theories. But surely it is an empirical issue, whether one needs separate operations to derive an additional semantic representation from a syntactic one. The crux is that both Katz and Bierwisch think that this 'additional' representation MUST appear in any adequate theory. But they have not shown this, and so the difference remains empirical. The need for extra operations and representations in a Bierwisch-type theory should not be confused with the incorrect claim that generative semantics in contrast to this type of theory dispenses with projection rules. Postal (1970) seems to have made this mistake, possibly through a confusion of representation and reading. In any case, if projection rules are taken to be semantic composition rules, no semantic theory can do without them. To eliminate them would be to eliminate satisfaction rules (the truth-definition) from logic; this leaves only syntax. The fluidity in current syntactic and semantic theory means that generative semantics like the other theories must wait on a great deal more evidence for helpful evaluation. At present it is being tested through the use of familiar logical systems to provide the semantic representations of the theory. Bach, McCawley, and Harman have explored the use of variables and quantifiers in pronominalization,88 Harman 88

Bach and McCawley's theories would clash with the Katz-Fodor claim that 'two chairs are in the room' and 'there are at least two things in the room and each is a chair* have no grammatical relation.

96

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

has discussed propositional attitudes, and Lakoff has considered a wide variety of topics in current logical theory. Of course, generative semantics has no special tie to prevalent logical systems; some other logical system might prove a more adequate semantics;89 it could even turn out that the current logic is very adequate but for a less integrated theory; more generally, particular semantic representations are not tied to any specific assumptions about the over-all form of a linguistic theory, though they may fit more or less well into different theories. In addition to investigations of the applicability of already developed logical theories to syntax, work is proceeding on the use of logical techniques to solve outstanding semantico-logical problems in the context of linguistic theory. Presupposition mentioned briefly in discussing logic is a good example of such a problem. Roughly three types of presupposition should be handled in a complete theory. (1) Presuppositions of ordinary individual lexical items. (2) Presuppositions of a category of lexical items. (3) Presuppositions of logical terms or constants. These three types parallel in an obvious way types of entailments. The first are entailments generated by individual meaning postulates, the second, those generated by generalized meaning postulates, and the third, logical entailments. Lakoff (1970b) combines syntactic and semantic information in his discussion of an example of the first type, the presuppositions of 'realize'. He suggests that both (a) Sam realizes that Irv is a martian and (b) Sam doesn't realize that Irv is a martian presuppose (c) Irv is a martian. Problems are not far off, however, for the relation between (a) and (c) seems to be identical to that between (d) Sam knows that Irv is a martian and (c). Since the latter is generally treated as an entailment we might try construing the former also as an entailment, reserving presupposition for the relation between (a) and (c). Yet the failure of this move is all too clear. If (c) is false then according to modus tollens and the assumption of entailment so is (a). But ac88

Our earlier discussion of action sentences points to one possible source of inadequacy in standard logic and the cautionary remarks of fns. 46 and 71 should make clear that we are far from an adequate and complete logico-semantic theory.

ON LOGIC AND THEORETICAL LINGUISTICS

97

cording to the common conception of presupposition, if (c) is false, (a) is neither true nor false. This problem is just an application of the general difficulty of defining presupposition in terms of the ordinary material conditional. We cannot discuss these problems now, nor need we; for our brief remarks illustrate an open problem in semantics which must be solved while developing an over-all linguistic theory. A more explicit example of semantic facts about presupposition worked out in the context of syntactic theory is that presented by Paul and Carol Kiparsky. In distinguishing factive from non-factive predicates, they consider the second type of presupposition. They specify syntactic criteria for the distinction — only factive predicates can take as object the noun 'fact' with a gerund while only non-factive predicates admit accusative and infinitive — and then indicate some correlative semantic facts. The most important is that what follows a factive predicate is presupposed to be true. According to our earlier remarks on the form of semantic versus syntactic arguments (p. 89), the detailed working out of transformations plays a more crucial part in syntax than in semantics. This is evidenced by the lack of interest on the part of logicians and philosophical semanticists generally in specifying the correlation scheme required for the application of their theories. We noted the central role of correlation schemes in semantic and logical theory, but also conspicuously ignored any of its detailed working out. Closely related is the absence of interest in and discussion of semantic generalizations. While one finds frequent comments on the simplicity of a certain axiomatization of a branch of logic or the relationship between this and the variety of rules used for deduction, rarely is there consideration of how some semantic generalization can or cannot be formulated in a particular system. Finally, psycho-semantics has been almost totally unexplored. Reason for this general neglect lies in the ability to test a semantic theory independently of specifying and examining the relevant transformations. Nonetheless, significant work on a general integrated linguistic theory will depend on investigations in these areas, and thus on a broadening of the aims of semantico-logical inquiry. REFERENCES References mentioned in the text are listed below. For a full bibliography the reader should consult Barbara Partee, Sharon Sabsay, and John Soper, Bibliog.: Logic and Language, Indiana University Linguistics Club, Bloomington, June, 1971.

L. 1965. A new approach to the logical theory of interrogatives. Uppsala, Sweden. BACH, E. 1968. Nouns and noun phrases. Universals in linguistic theory, ed. by E. Bach and R. T. Harms, 91-122. New York, Holt.

AQUIST,

98

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGAL1T

E., and R. T. HARMS, eds. 1968. Universals in linguistic theory. New York, Holt. BAR-HILLEL, Y. 1954. Logical syntax and semantics. Lg 30.230-7. . 1964. Language and information. Jerusalem, Jerusalem Academic Press, and Reading, Mass., Addison-Wesley. , ed. 1971. Pragmatics of natural languages. Dordrecht, Holland, Reidel. BAR-HILLEL, Y., C. GAIFMAN, and E. SHAMIR. 1960. On categorial and phrase structure grammars. The Bulletin of the Research Council of Israel 9F.1-16. (Reprinted Bar-Hillel 1964:99-115.) BIERWISCH, M . 1 9 6 9 . Certain problems of semantic representation. F L 5 . 1 5 3 - 8 4 . . 1970a. On classifying semantic features. Semantics, ed. by D. D. Steinberg and L. A. Jakobovitz, 410-35. Cambridge, Cambridge University Press. . 1970b. Semantics. New horizons in linguistics, ed. by J. Lyons, 166-84. Middlesex, England, Penguin Books. BLACK, M. 1964. A companion to Wittgenstein's Tractatus. Cambridge, Cambridge University Press. BROWN, J . G. 1971. A note on fuzzy sets. I&C 18. CARNAP, R. 1964. Meaning and necessity. A study in semantics and modal logic. Chicago, University of Chicago Press. CHOMSKY, NOAM. 1955. Logical syntax and semantics. Lg 31.36-45. . 1962. Explanatory models in linguistics. Logic, methodology, and philosophy of science, ed. by E. Nagel, P. Suppes, and A. Tarski, 528-50. Stanford University Press. . 1965. Aspects of the theory of syntax. Cambridge, M.I.T. Press. CLARKE, R. 1970. Concerning the logic of predicate modifiers. Nous 4.311-36. COHEN, L. J . , and A . MARGALIT. 1 9 7 0 . The role of inductive reasoning in the interpretation of metaphor. Synthese 2 1 . 4 6 9 - 8 7 . DAVIDSON, DONALD. 1965. Theories of meaning and learnable languages. Logic, methodology, and philosophy of science, Proceedings of the 1964 International conference, ed. by Y. Bar-Hillel, 387-94. Amsterdam, North-Holland. . 1967a. The logical form of action sentences. The logic of decision and action, ed. by N. Rescher, 81-95. Pittsburgh, University of Pittsburgh Press. . 1967b. Truth and meaning. Synthese 17.304-23. . 1968. On saying that. Synthese 19.130-46. (Reprinted in Davidson and Hintikka 1969:159-74.) . 1969. On events and event descriptions. Fact and existence. Proceedings of the University of Western Ontario's Philosophy Colloquium, ed. by J. Margolies, 74-84. Oxford, Blackwell. . 1970a. Mental events. Experience and theory, ed. by L. Foster, 79-101. Amherst, University of Massachusetts Press. . 1970b. Semantics for natural languages. Linguaggi nella società e nella tecnica, 177-88. Milano, Edizioni di Communità. BACH,

ON LOGIC AND THEORETICAL LINGUISTICS

99

and J . HINTIKKA, eds. 1969. Words and objections, essays on the work of W. V. Quine. Dordrecht, Holland, Reidel. FILLMORE, C. J. 1965. Entailment rules in a semantic theory. Project on Linguistic Analysis, Report 1 0 . 6 0 - 8 2 . Columbus, Ohio. . 1966. Deictic categories in the semantics of 'come'. FL 2.219-27. . 1968. The case for case. Universals in linguistic theory, ed. by E. Bach and R. T. Harms, 1-88. New York, Holt. . 1970. Subjects, speakers, and roles. Working Papers in Linguistics 4.3163. FODOR, J. A. 1970a. Three reasons for not deriving 'kill' from 'cause to die'. Linguistic Inquiry 1.421-38. . 1970b. Troubles about actions. Synthese 21.298-319. FODOR, J . A . , and J . KATZ, eds. 1964. The structure of language; Readings in the philosophy of language. Englewood Cliffs, New Jersey, Prentice-Hall. FODOR, JANET D . 1970. Formal linguistics and formal logic. New horizons in linguistics, ed. by J . Lyons, 1 9 8 - 2 1 4 . Middlesex, England, Penguin. FREGE, G. 1952. Translations from the philosophical writings of Gottlob Frege. Translated by P. Geach and M. Black. Oxford, Blackwell. GRICE, H . P. 1961. The causal theory of perception. Proceedings of the Aristotelian Society. Supplementary Volume 3 5 . 1 2 1 - 6 8 . HARMAN, G. 1969. An introduction to 'translation and meaning'. Words and objections, ed. by D. Davidson and J. Hintikka, 14—26. Dordrecht, Holland, Reidel. . 1970. Deep structure as logical form. Synthese 21.275-97. HINTIKKA, J . 1967. A program and a set of concepts for philosophical logic. Monist 5 1 . 6 9 - 9 2 . . 1968. Epistemic logic and the methods of philosophical analysis. Australasian Journal of Philosophy, 37-51. (Reprinted 1969b: 3-19.) . 1969a. Semantics for propositional attitudes. Philosophical logic, ed. by J. S. Davis, D. J. Hockney, and W. K. Wilson, 2 1 ^ 5 . Dordrecht, Holland, Reidel. . 1969b. Models for modalities. Dordrecht, Holland, Reidel. KATZ, J . J . 1966. The philosophy of language. New York, Harper and Row. . 1967. Recent issues in semantic theory. FL 3.124-44. . 1970. Interpretative semantics vs. generative semantics. FL 6.220-59. . 1972. Semantic theory. New York, Harper and Row. KATZ, J. J., and J. A . FODOR. 1964. The structure of a semantic theory. The structure of language: Readings in the philosophy of language, ed. by J. Fodor and J. J. Katz, 4 7 9 - 5 1 . Englewood Cliffs, New Jersey, Prentice-Hall. LAKOFF, G. 1970a. Linguistics and natural logic. Synthese 22.151-271. . 1970b. On generative semantics. Semantics, ed. by D. D. Steinberg and L. A. Jakobovitz, 232-96. Cambridge, Cambridge University PressDAVIDSON, DONALD,

100

YEHOSHUA BAR-HILLEL, JONATHAN MALINO, AVISHAI MARGALIT

LEE, R . ,

and

C.

LEWIS, DAVID.

T.

1971. Some properties of fuzzy logic. I&C. General semantics. Synthese 2 2 . 1 8 - 6 7 . 1970. Analytic/synthetic and semantic theory. Synthese

CHANG.

1970.

LINSKY, LEONARD.

21.439-48. P. 1969. Logic and grammar. Monist 53.195-203. LYONS, J . 1970. New horizons in linguistics. Middlesex, England, Penguin. MCCAWLEY, J. D. 1968a. Concerning the base component of a transformational grammar. FL 4.342-69. . 1968b. The role of semantics in grammar. Universals in linguistic theory, ed. by E. Bach and R. T. Harms, 125-69. New York, Holt. MARGALIT, A. 1970. The cognitive status of metaphors. Ph.D. Dissertation, The Hebrew University. MONTAGUE, R . 1 9 6 8 . Pragmatics. Contemporary philosophy, ed. by R . KLIbansky, 1 0 2 - 2 2 . Florence, La Nuova Italia Editrici. . 1970a. English as a formal language. Linguaggi nella società e nella tecnica, 189-223. Milano, Edizioni di Communità. . 1970b. Universal grammar. Theoria 36/3.373-98. PARSONS, T. 1970. Some problems concerning the logic of grammatical modifiers. Synthese 21.320-34. PEARS, D. F. 1968. Bertrand Russell and the British tradition in philosophy. London, Collins: The Fontana Library. QUINE, W. V. O. 1960. Word and object. Cambridge, M.I.T. Press. . 1966. Elementary logic. Cambridge, Harvard University Press. . 1969. Replies. Words and objections, ed. by D. Davidson and J. Hintikka, 292-352. Dordrecht, Holland, Reidel. . 1970a. Methodological reflections on current linguistic theory. Synthese 21.386-98. . 1970b. Philosophy of logic. Englewood Cliffs, New Jersey, Prentice-Hall. REICHENBACH, HANS. 1947. Elements of symbolic logic. New York, MacMillan. RESCHER, N. 1968. Topics in philosophical logic. Dordrecht, Holland, Reidel. RYLE, GILBERT. 1 9 6 7 . Systematically misleading expressions. The linguistic turn: Recent essays in philosophical method, ed. by Richard Rorty, 8 5 - 1 0 0 . Chicago, University of Chicago Press. SCOTT, D. 1970a. The problem of giving precise semantics for formal languages. Linguaggi nella società e nella tecnica, 225-54. Milano, Edizioni di Communità. . 1970b. Advice on modal logic. Philosophical problems in logic, ed. by K. Lambert, 143-73. Dordrecht, Holland, Reidel. STAAL, J . F., ed. 1 9 6 9 . Formal logic and natural languages (a symposium). FL

LORENZEN,

5.256-84. STALNAKER, ROBERT

C.

1970.

Pragmatics. Synthese

22.273-89.

ON LOGIC AND THEORETICAL LINGUISTICS STEINBERG, D., and L . JAKOBOVITS, eds.

1970.

101

Semantics: A n interdisciplinary

reader. Cambridge, Cambridge University Press. STENIUS, ERIK. 1969. Mood and language-games. Philosophical logic, ed. by J. W. Davis, D. J. Hockney, and W. K. Wilson, 251-71. Dordrecht, Holland, Reidel. STRAWSON, P. F. 1957. Propositions, concepts, and logical truths. The Philosophical Quarterly 7.15-25. . 1969. Singular terms and predication. Words and objections, ed. by D. Davidson and J. Hintikka, 97-117. Dordrecht, Holland, Reidel. . 1970. Meaning and truth. Inaugural lecture. Oxford, Oxford University Press. TARSKI, A . 1956. The concept of truth in formalized languages. Logic, semantics, meta-mathematics, ed. by A. Tarski, 152-278. Oxford, Oxford University Press. WALLACE, JOHN.

1970.

O n the frame of reference. Synthese 22.117-50.

WITTGENSTEIN, LEO. 1962. Tractatus. Translated by B. Pears and B. E. McGuinness. Oxford, Blackwell. ZADEH, L.H. 1965. Fuzzy sets. I&C 18.

LINGUISTICS A N D SEMANTICS

EUGENIO COSERIU and HORST GECKELER

LINGUISTIC, ESPECIALLY FUNCTIONAL, SEMANTICS*

0. At least three different tendencies can be distinguished in the use of the term semantics: 0.1 Linguistic semantics, which has to do with the scientific study of linguistic meanings, in principle is concerned with all kinds of linguistic meaning, including grammatical meaning; traditionally, however, it is first and foremost the study of lexical meaning which has been associated with semantics. The survey given in this article will refer exclusively to this type of semantics. 0.2 The semantics of logicians, as it is defined by R. Carnap (1942) and, to some extent, by Ch. W. Morris (1938), affects only one of three possible aspects within the range of questions raised by semiotics: 'If we abstract from the user of the language and analyze only the expressions and their designata, we are in the field of semantics.' The other two are: 'If in an investigation explicit reference is m a d e . . . to the user of a language, then we assign it to the field of pragmatics' 'And if, finally, we abstract from the designata also and analyze only the relations between the expressions, we are in (logical) syntax.'1 0.3 General semantics, a trend initiated by A. Korzybski (1933), which has had since 1943 as its most important mouthpiece the journal ETC.: A Review of General Semantics,2 'is a technique for correcting certain abuses of language such as the uncritical use of ill-defined abstractions' (Ullmann 1962:10); it proposes to fight against alleged detrimental consequences of the 'power of language over thought' in human society. P. Guiraud (1962) characterizes this approach as 'une psycho-socio-logique du signe'. This trend, however, has rightly been severely criticized from various sides (e.g. Black 1949 : 221-46; Schlauch 1943 :130-2; and Coseriu 1958 : 113 and 1962 : esp. 237-8 fn. 6). * The sections on E. Coseriu's structural semantics (esp. 4.2.2.) were written by H. Geckeler alone, who therefore assumes complete responsibility for the statements contained therein. 1 Carnap 1942 :9. Cf. Morris 1938 :21: 'Semantics deals with the relation of signs to their designata and so to the objects which they may or do denote.' In later works Morris modified his position; cf. his definition of meaning: 'those conditions which are such that whatever fulfills them is a denotatum will be called a signification of the sign' (1955 :17). 2 Its editor is S. I. Hayakawa, of whose works, cf. his 1949 and 1 9 5 4 : 1 9 - 3 7 .

104

EUGENIO COSERIU AND HORST GECKELER

0.4 Among the comprehensive summaries of semantics,3 none treats all three directions with equal completeness. The works by Ullmann, Regnéll, Guiraud, and George, however, give at least some information about 'the semantics of logicians' and on 'general semantics'. As already indicated above, the following discussions will deal exclusively with 'linguistic (lexical) semantics'.

1.0 Before briefly illuminating a quite specific situation of linguistic semantics as it existed before the various types of a structural semantics came into being, we must first of all make a few preliminary observations on terminology and on the delimitation of the object we are considering. 1.1 During the course of the development of linguistics since the end of the nineteenth century, the terms semasiology and semantics both came to be used to designate that linguistic discipline which is concerned with the study of linguistic, especially lexical meanings. Nowadays, this competition has been resolved in favor of the term semantics (Read 1948, Ullmann 1951). The term semasiology was introduced into linguistics before 1829 by K. Reisig (1839) and became current in Germany more than elsewhere, but did also to some extent spread from there to Anglo-American territory.4 This usage can be followed from G. Stern (1931) to H. Kronasser (1952) and finally to K. Baldinger (1957; cf. Zvegincev (1957)), whose survey seems to stand at the end of this typically German philological-linguistic tradition. The term semantics,5 which has served since the 1950s in international linguistic terminology as the undisputed designation for this discipline, was to originate in France. M. Bréal first used this technical term (1883) and it was immediately taken over by A. Darmesteter (1887), but the real breakthrough for this terminological innovation came as a consequence of Bréal's well-known monograph of 1897: Essai de sémantique (Science des significations). While Bréal defined semantics very generally as 'la science des significations', as opposed to phonetics ('la science des sons') (1897: 8, fn. 1 — see 3rd ed.), Darmesteter, entirely in the spirit of the time, conceived of semantics as 'la science des changements de signification dans les mots' (1887:88, fn. 1 — see 7th éd.). In works on linguistics written in German, Bedeutungslehre has appeared as a terminological constant at least since the time of K. Reisig throughout the entire 3

The best syntheses have been written by S. Ullmann 1963, with more bibliographical material, and 1962. With strongly traditional orientation: see H. Kronasser 1952 and K. Baldinger 1957. Among the shorter introductory works may be mentioned H. Regnéll 1958 and P. Guiraud 1962; disappointing from a linguistic point of view is F. H. George 1964. 4 Cf. Read 1948 : 82, and E. P. Hamp 1966. Cf. also S. Ullmann 1946, who here still employs semasiology for that which he consistently designates as semantics in his later publications. 5 Cf. the titles of the works by S. Ullmann, H. Regnéll, P. Guiraud, and, further, the different approaches to a 'structural semanticsetc.

LINGUISTICS AND SEMANTICS

105

nineteenth century, and in the twentieth century as well from H. Sperber (1923) up to E. Gamillscheg (1951) and H. Kronasser.6 Now there was no lack of attempts to differentiate the terms semantics and semasiology.7 But they remained only attempts, for at times both terms were used as synonyms according to the authors' preferences, or else only one of the two terms was used for the entire discipline within a particular philologicallinguistic tradition. J. R. Firth (1957a: 27, fn. 2), for example, made the following suggestions for differentiating the terms: 'Taking advantage of what Coleridge called the "desynonymizing" process, I would use the term "semasiology" for the historical study of changes of meaning.8 Another suggestion is that phonetics and semantics be regarded as branches of general linguistics, the corresponding fields in special grammar being phonology and semasiology.' The terminological parallelism implicit in this quotation, i.e. phonetics is to phonology as semantics is to semasiology, or, phonetics is to semantics as phonology is to semasiology, was followed to some extent (Hamp 1966:48, 54), but without any lasting effect. Since the relationship between phonetics and phonology had not been uniformly defined before the papers of the Prague school appeared,9 no uniform delimitations for semantics and semasiology were forthcoming either. A distinction between semantics and semasiology based on such considerations will not be made here. We call the entire discipline of the science of lexical meanings semantics, which can be either of descriptive-synchronic (analytic) or of historical-diachronic orientation. By semasiology we mean only a sub-discipline with a very limited range of application: semasiology takes the word qua signifiant as a point of departure and investigates the contents (meanings) associated with it in their multiplicity and their change (polysemy and change of meaning), while onomasiology10 proceeds from the contents (signifiés) or concepts — in practice even, in part, from the objects of extralinguistic reality itself — and studies the various signifiants (designations11) which can designate the content in question (in diachronic p&rspective=Bezeichnungswandet). " 1952. The formation parallel to German Bedeutungslehre exists in Dutch, Norwegian, Hungarian, and Finnish linguistic terminology (cf. Ullmann 1963 :4). 7 On the specific relationship between semasiology and onomasiology cf. below. 8 By 'semantics', on the other hand, he means the descriptive, synchronic study of meaning. 0 Before the Prague phonologists had clearly differentiated the terms phonetics and phonology, phonology was used at times as a synonym for phonetics, at times for the designation of a specific section of phonetics: this is the case especially in Anglo-American tradition, where phonology appears for 'historical phonetics'. Cf. also F. de Saussure's idiosyncratic terminology (adopted only by M. Grammont) (1964 :55-6): 'phonologie' in Saussure's terminology corresponds to the present conception of phonetics; his 'phonétique' corresponds to our historical phonetics. On these questions, see N . S. Trubetzkoy 1967 :12, J. Marouzeau 1 9 6 1 : 1 7 6 , F. Lázaro Carreter 1962 : 192-3, and M. Pei 1966 : 207. 10 The term onomasiology has come into general use in linguistics since A. Zauner's publication (1902). B. Quadri gives an excellent survey of the onomasiological line of research (1952). Cf. also R. Hallig and W. v. Wartburg 1963 and K. Baldinger 1964. On the theoretical bases of onomasiology, see K. Heger 1964 and 1969. 11 In 4.2.2. we will give a different definition of the term designation (Bezeichnung).

106

EUGENIO COSERIU AND HORST GECKELER

The traditional study of meaning, whether it was called semantics or semasiology, had, in practice, changes in meaning, i.e. our historical-diachronic semantics, as its focal point. By comparison, there are far fewer efforts towards a descriptivesynchronic semantics in linguistic tradition. A few such studies will be discussed in connection with the precursors of structural semantics in 3.2. — Among the achievements in the realm of descriptive semantics must be counted lexicographical practice insofar as it concerns dictionaries ordered according to concepts.12 This point will not be further treated here. We also find a descriptive approach in onomasiology, which has been developed — in opposition to semasiology in the narrower sense of the term — as another kind of sub-discipline within the field of semantics. Onomasiology (cf. fn. 10 above) will not be discussed here either.13 1.2.0 In the following discussion we will restrict ourselves to a brief panorama of a specific situation which had important historical consequences for linguistic study. Before the appearance of the various forms of a structural semantics, linguistic semantics within North American linguistics was in an extremely peculiar situation, which can be explained chiefly by means of the following factors: 1.2.1 North American structuralism in the wake of L. Bloomfield and particularly of his followers was shown to have been hostile toward meaning.14 In his striving to bring to linguistic analysis the rigor of the methods of the natural sciences, and guided by mechanistic-behavioristic premises,15 L. Bloomfield encountered a difficult obstacle, namely linguistic meaning: In order to give a scientifically accurate definition of meaning for every form of a language, we should have to have a scientifically accurate knowledge of everything in the speakers' world. The actual extent of human knowledge is very small, compared to t h i s . . . The statement of meanings is therefore the weak point in language-study, and will remain so until human knowledge advances very far beyond its present state. In practice, we define the meaning of a linguistic form, wherever we can, in terms of some other science. (Bloomfield 1965 : 139-40; cf. 1943, 1960.)

According to this, the investigation of lexical meaning, and therefore semantics, would lie outside of the actual domain of linguistics. Under the weighty influence of Bloomfield's Language (first published New York 1933), 'semeophobia' (A. Reichling) crystallized among the North American structuralists, and with it came the tendency to eliminate lexical meaning from 1J On this topic cf. K. Baldinger 1952 and 1960, F. de Tollenaere 1960, and A. Rey 1965. J. Casares' Diccionario ideoldgico de la lengua espanola (1942) can probably be considered the best conceptual dictionary hitherto realized. 13 E. Coseriu makes critical statements concerning the theoretical basis of onomasiology, 1964 :162 and 1968a : 4. 14 However, a line which starts with E. Sapir and continues via B. L. Whorf to the ethnolinguists must be clearly separated from Bloomfield's line. 15 A fundamental discussion of the principles of the mechanistic approach in linguistics can be found in E. Coseriu 1954.

LINGUISTICS AND SEMANTICS

107

linguistic analysis, as for example in the work of B. Bloch and G. L. Trager (1942: 6, 68) and, most consistently, in the work of Z. S. Harris (1963), who believed he had found a reliable basis for the formal description of language in distribution,1S which could dispense with any reference to meaning: However, this differentiation of life and rife on the basis of meaning is only the linguist's and the layman's shortcut to a distributional differentiation. In principle, meaning need be involved only to the extent of determining what is repetition. If we know that life and rife are not entirely repetitions of each other, we will then discover that they differ in distribution (and hence in 'meaning'). It may be presumed that any two morphemes A and B having different meanings also differ somewhere in distribution: there are some environments in which one occurs and the other does not (Harris 1963 : 7, fn. 4).

Since the beginning of the 1950s, voices were increasing among the very structuralists in the USA to rethink the problem of meaning within the framework of linguistic analysis. It was the contribution of Ch. C. Fries (1954; cf. Fowler 1965) to have checked, according to the texts, the hostility towards meaning ascribed to Bloomfield, which in the meantime had become almost a myth, and to have brought back a proper perspective. Bloomfield's 'efforts to achieve statements in physical rather than "mentalistic" terms do not lead to the conclusion that he "ignores meaning" or that "he takes no account of meaning'" (Fries 1954:59). 'With Bloomfield, no serious study of human language can or does ignore "meaning"' (p. 60). Basically, North American structuralism had, according to Fries, arrived at its 'repudiation of meaning' through 'inferences drawn from a somewhat superficial reading of (Bloomfield's) discussions of mentalism and mechanism' (p. 58). However, one gets the impression that Fries was primarily attacking expressions with which others had characterized Bloomfield's procedure, since he could not do away with the fact that Bloomfield simply excluded the investigation of lexical meaning from the field of linguistics. To be sure, Bloomfield acknowledged the importance of the content-level for the study of language (one need only refer to his terminological framework in Language, 1965:264), but he saw no way for it to be methodically studied according to a rigorous linguistic approach. In Bloomfield's linguistic system, 'meaning' always shows up as a point of reference but he does not make 'meaning' an object of linguistic investigation. Because of methodological scruples which originated in a falsely understood, rigorous scientific ideal, he sacrificed the investigation of lexical meaning. M. Schlauch (1946) pointed to the fact that the hostile attitude with respect to 'meaning' and 'mentalism' corresponded to an early phase of behavior16

Cf. also Harris 1954. A comprehensive discussion on the theme "The importance of distribution versus other criteria in linguistic analysis" with two reports by P. Diderichsen and H. SpangHanssen, as well as numerous contributions to the discussion by leading linguists can be found in the Proceedings of the Eighth International Congress of Linguists, 156-213 (Oslo 1958). Cf. further on this topic: H. Frei 1954 and P. Naert 1961.

108

EUGENIO COSERIU AND HORST GECKELER

ism which Bloomfield took over without later taking into account the differentiated further development of this theory. Already in E. Haugen's 'Presidential Address' to the Meeting of the Linguistic Society in Chicago (on Dec. 29, 1950),17 criticism of the attitude of many North American structuralists to the question of meaning comes through: 'The minimizing of meaning as a factor in linguistic description was at first a healthy reaction against the misuse of meaning in establishing linguistic categories, but has now become almost a fetish with some linguists. It is curious to see how those who eliminate meaning have brought it back under the covert guise of distribution' (1951:219). In 1951 an article by E. A. Nida appeared, with the title (unusual for the situation then current in North American structuralism): "A system for the description of semantic elements" (Nida 1951). This study suggests, for the first time, a comprehensive and coherent terminology for the description of meaning.18 Although the well-known North American introductory linguistics handbooks by Ch. F. Hockett (1958), H. A. Gleason, Jr. (1961), A. A. HiU (1958), and R. A. Hall, Jr. (1964), still give only minimal space to semantics,19 this development slowly reversed during the fifties, so that concepts such as 'meaning' and 'mentalist', which were then regarded as dirty words, have once again become respectable in the most widespread North American scholarly circles. We owe the definite victory over this attitude of hostility towards meaning, which had completely paralyzed investigation in the area of lexical content for many years, to the onset of generative linguistics 20 against the already undermined position of doctrinaire Bloomfieldianism. 21 1.2.2 A second important factor is the equation, often found in North American linguistics, of 'meaning' and 'thing-meant' (cf. A. Gardiner (1951: 29-33)), i.e. a reducing of the linguistic content to extralinguistic reality, whether it be as thing (cf. for instance Bloomfield's example salt— 'sodium chloride (NaCl)' (1965:139), or whether it be as situation (Ch. W. Morris, on the other hand, arrives at a distinction between thing and meaning). Thus, this erroneous view also underlies Bloomfield's definition of meaning: We have defined the meaning of a linguistic form as the situation in which the speaker utters it and the response which it calls forth in the hearer (Ibid). 17

Published under the title, "Directions in modern linguistics", see Haugen 1951. Basing his distinctions on L. Bloomfield and on E. A. Nida, J. H. Greenberg provides, somewhat later (1954), a likewise coherent contribution to the terminology of semantics (1963 : esp. 7-8). 19 The most comprehensive discussion of meaning known to us in the North American linguistic publications of this time can be found in Pike 1954-60: ch. 16 and particularly in E. A. Nida 1964:30-119. 20 N. Chomsky's estimation of the role of 'meaning' and 'intuition' for linguistic analysis has undergone basic changes from his first publications up to the present time. 21 H. Geckeler gives a somewhat more extensive survey of the position of semantics in North American linguistics (1971a: chapter I). 18

LINGUISTICS AND SEMANTICS

109

By uttering a linguistic form, a speaker prompts his hearers to respond to a situation; this situation and the responses to it, are the linguistic meaning of the form (Bloomfield 1965:158).

In practice, however, linguistic investigation has never fully traversed the path of the determination of meaning via the sum of the situations, so that the question must arise as to whether or not it can be done. D. Abercrombie (1965 :116) denies the possibility of such a procedure and, in connection with this, speaks of a 'pseudo-procedure'. For additional illustration of this conception of meaning, corresponding statements by B. Bloch and G. L. Trager, and A. A. Hill may be cited as examples (cf. also Nida 1951:4 and Harris 1963:190): The MEANING of a linguistic form (a word, a part of a word, or a combination of words) is the feature common to all the situations in which it is used. Meaning is thus a matter of the practical world around us — a matter of social and cultural relations as well as of purely objective 'reality' (Bloch and Trager 1942:6). Meaning proper is ultimately correspondence between a linguistic item and an item in the nonsymbolic world, or between a linguistic structure of many items and a similar structure in the nonsymbolic world (Hill 1958 : 410).

The relations between linguistic sign, signifié, signifiant, and extralinguistic reality become clear when one distinguishes resolutely between signification (Bedeutung) and designation (Bezeichnung), cf. 4.2.2. This essential distinction is ignored by transformational-generative linguistics too (see Coseriu 1970a), and this misconception is thus a heritage of 'taxonomic' North American structuralism which has been unconsciously carried over. The conception of meaning (lexical meaning) in North American structuralism just outlined led certain linguists to separate the investigation of this phenomenon from linguistics itself ( = 'microlinguistics') and to assign it to a new discipline called 'metalinguistics'22 or 'exolinguistics' (J. B. Carroll 1961:29). 1.2.3.0 Up to the most recent times, one found, in the United States, the beginnings of a descriptive semantics among the anthropologists (respectively ethnolinguists) and psychologists rather than among the very linguists themselves.23 1.2.3.1 The most prominent position is here occupied by the North American 'school' of anthropologists and ethnologists (or ethnolinguists). Their tradition goes back to A. L. Kroeber, and they operate to some extent with procedures similar to those of the European type of content-analysis. This school's most significant representatives may be said to be H. C. Conklin (1962), W. H. Goodenough (1956), and F. G. Lounsbury (1956, 1964) whose works have been 22

Trager 1950. In more recent times, however, we can discern an altered conception in Trager. The investigation of lexical meaning in 'semology' is again included within the framework of linguistics (1963). 2» We will return in another context to the more recent studies by M. Joos, S. M. Lamb, U. Weinreich, and the semantics of transformational-generative linguistics.

110

EUGENIO COSERIU AND HORST GECKELER

assessed e.g. by U. Weinreich as 'tide-turning papers'24 with respect to the 'concept of semantic component'. In the investigation of certain clearly delimited areas of vocabulary (especially systems of kinship relations,25 disease names, names of colors, folk taxonomies, etc.)26 in various languages (often in North American Indian languages) these scholars, partly with predominantly anthropologicalethnological interests, partly with predominantly linguistic interests, have created an apparatus for linguistic analysis that has no equal in the field of North American linguistics. We have found the most precise formulations in F. G. Lounsbury's Report to the 9th International Congress of Linguists (Cambridge, Mass., Aug. 27-31, 1962), and would like, therefore, to cite some representative definitions from it. Lounsbury regards the system of kinship terms in a given language as a 'semantic field' and states: A kinship vocabulary can be regarded as constituting a paradigm. It can be subjected to a kind of analysis similar to that given other paradigmatic sets in a language (1964: 1073). Thus, he conceives of the semantic field as a paradigm; cf. our definition of the lexical field as a lexical paradigm in 4.2.2. Lounsbury defines the paradigm as follows: We shall regard as a paradigm any set of linguistic forms wherein: (a) the meaning of every form has a feature in common with the meaning of all other forms of the set, and (b) the meaning of every form differs from that of every other form of the set by one or more additional features. The common feature will be said to be the ROOT MEANING of the paradigm. It defines the semantic field which the forms of the paradigm partition. The variable features define the SEMANTIC DIMENSIONS 27 of the paradigm (1964:1073-4). Lounsbury makes the following statement on 'componential definitions': A term belonging to a paradigm can be defined componentially in terms of its coordinates in the paradigm. The definition represents a bundle of features . . . The componential definition of a term is the expression of its significatum (1964:1074). Although we are not going to pursue this line further here, we assume that this brief outline will have shown clearly enough how close ethnolinguistic componential analysis (specifically, in the case of Lounsbury) is to the most modern structural semantics of European tradition (cf. 4.2). 24 1963a : 148. Cf. also Pike 1954-60 :111.98: 'The most significant recent attempt to treat meaning structurally lies, in my judgment, in the work of Lounsbury (1956), and Goodenough (1956). They have each, independently, attempted to illustrate semantic analysis by way of kinship systems.' 25 On the method of componential analysis in its application to the vocabulary of kinship relations cf. A. F. C. Wallace and J. Atkins (1960). M However, it must be remarked here that L. Weisgerber had concerned himself, long before the ethnolinguists, with certain of their preferred areas of research (e.g. kinship terms, names of colors). " The concept of the 'semantic dimension* has been further developed and applied in H. Geckeler (1971a: ch. VIII).

LINGUISTICS AND SEMANTICS

111

1.2.3.2 Another, though less important contribution to descriptive semantics, this time from the field of psychology in the U.S., is represented by the psychosemantics of Ch. E. Osgood and his circle.28 Here, with the help of the so-called 'semantic differential',29 meanings are supposed to be measured. What is then in fact measured is not linguistic meaning, but rather the reactions of hearers to linguistic utterances, i.e. not further differentiated reactions, which are delimited by means of prescribed oppositional pairs of adjectives. The purely linguistic value of such investigations may be assessed as relatively minor, since only the area of connotative meanings (Rosiello 1 9 6 2 : 3 8 ) is here appealed to. This, however, can become important for stylistic analysis.

2.0 The more recent and most recent developments in semantics have, respectively, been understood to be structural or have been termed structural. Three possible types of 'structural' semantics can be distinguished: 2.1.0 'Structural': understood as the structure, or better, as the configuration of associations of one sign with other signs in the vocabulary. These associations depend on similarity or on contiguity30 between signifiants, or between signifiés, or between signifiants and signifiés. Associations may, however, be based on the copresence of things in extralinguistic reality, too. 2.1.1 F. de Saussure (1964:173-5), in his treatment of the 'rapports associatifs', analyzed French enseignement as an example of a 'série associative' or 'famille associative' (we can here omit Saussure's well-known schema). 31 In this, he comes to the following important conclusions: Donc il y a tantôt communauté double du sens et de la forme, tantôt communauté de forme ou de sens seulement. Un mot quelconque peut toujours évoquer tout ce qui est susceptible de lui être associé d'une manière ou d'une autre... les termes d'une famille associative ne se présentent ni en nombre défini, ni dans un ordre déterminé... Un terme donné est comme le centre d'une constellation, le point où convergent d'autres termes coordonnés, dont la somme est indéfinie (Saussure 1964 : 174). 2.1.2 28

F. de Saussure's 'rapports associatifs' or 'séries associatives' certainly

Cf. Ch. E. Osgood, G. J. Suci, and P. H. Tannenbaum 1957. This work was preceded by a series of articles by Osgood; cf. on this topic, R. Wells 1957. On this approach see U. Weinreich 1958, and the answer to this by Osgood (1959), with rejoinder by Weinreich (1959). Cf. also the comprehensive discussion by J. B. Carroll (1959). 29 'By semantic differentiation, then, we mean the successive allocation of a concept to a point in the multidimensional semantic space by selection from among a set of given scaled semantic alternatives' (Osgood, Suci, and Tannenbaum 1957 : 26). 30 On these two terms, cf. Ullmann 1963 : 220-44. 31 E. de Bustos Tovar (1967) proposes, with respect to Saussure, an expanded schema of association which proceeds from Ogden and Richards' semiotic triangular model; this more complex schema, however, contains questionable relations.

112

EUGENIO COSERIU AND HORST GECKELER

formed the point of departure for Ch. Bally's 'champ associatif' (1940:195-7). Bally characterizes this particular type of a field as follows: Les signes qui déterminent plus particulièrement la valeur de celui qu'ils entourent dans la mémoire forment son 'champ associatif: notion toute relative, puisque tout, dans la langue, est, au moins indirectement, associé à tout Le champ associatif est un halo qui entoure le signe et dont les franges extérieures se confondent avec leur ambiance (1940:195).

Although Bally regards this field-type as belonging to the level of the languagesystem (langue) on the one hand, he accords it no complete intersubjective validity on the other hand. As an illustration of his field-type he cites the famous example boeuf: Le mot boeuf fait penser: 1) à vache, taureau, veau, cornes, ruminer, beugler, etc., 2) à labour, charrue, joug, etc., à viande, abattoir, boucherie, etc., enfin 3) il peut dégager, et dégage en français des idées de force, d'endurance, de travail patient, mais aussi de lenteur, de lourdeur, de passivité ( 1 9 4 0 : 1 9 6 ) .

2.1.3 The 'champs morpho-sémantiques', introduced by P. Guiraud (1956),32 also represent associative configurations. He defines such a morpho-semantic field as 'le complexe de relations de formes et de sens formé par un ensemble de mots' (1962:82); such fields can assume gigantic dimensions: the field 'chat', which Guiraud presents as an illustration, contains, at an early stage of the analysis, about 2,000 words, but even after a process of elimination still about 300. Thus, the word chat stands at the centre of a vast constellation which is made up of materially-founded and content-founded associations. 2.1.4 Finally, G. Matoré's 'champ notionnel'33 is likewise based on a configuration of associations.34 The most important constituent elements of the hierarchical arrangement of Matoré's 'champ notionnel' are the 'mots-témoins'35 and the 'mots-clés',36 wherein the 'mot-clé' stands at the center of the field. Matoré's delimitations are not of a linguistic, but of a sociological nature. His procedure concerns 'den Sprachgebrauch, mehr noch die Situationen und Themen des Sprechens, die Einstellung gegenüber den gemeinten Sachen, nicht eigentlich die Sprache' (Coseriu 1967a: 21). Matoré's notional fields are so extensive that they pose the problem of the situating of the words within the total vocabulary. 32

Guiraud published further examples of such fields in a series of articles, especially in BSL. Matoré's study (1953) gives the theoretical basis for his conception of the 'champ notionnel' after a practical application had already appeared (1951). 34 Cf. Matoré's schema of the 'champ notionnel d'ART et de TECHNIQUE vers 1765' (1953 :102). 35 'Le mot-témoin est le symbole matériel d'un fait spirituel important; c'est l'élément à la fois expressif et tangible qui concrétise un fait de civilisation' (1953:65-6). »« 'Nous donnerons à l'unité lexicologique exprimant une société le nom de mot-clé. Le mot-clé désignera donc non une abstraction, non une moyenne, non un objet, mais un être, un sentiment, une idée, vivants dans la mesure même où la société reconnaît en eux son idéal' (1953 :68). 33

LINGUISTICS AND SEMANTICS

113

2.1.5 Within the various subtypes of his 'champs linguistiques', the Czech linguist O. Duchâcek (1959) 37 also includes the 'champs associatifs'. These fields are based on associations either on the level of expression or on the content-level, or on associations from a combination of the two levels. 2.1.6 A critical estimation of the forms of associative configurations briefly discussed in 2.1.1-5, which are usually regarded as belonging to a structural semantics, 38 must, from the standpoint of a proper structural semantics such as we will subsequently present (4.2.2), include the following three points: 39 a) The associations, if they are not of an infinite character, then at least reach an uncontrollable extent. b) The associations are largely individual and to this extent neither predictable nor systematizable. c) The associations are not necessarily dependent on the language; they can also be based on the real context of things. 2.1.7 The chief emphasis of the investigations into associative relations is in the area of description, but noteworthy attempts have also been made in diachronics. S. Ullmann (1957), on the basis of investigative results by W. von Wartburg (1962:117-18), has 'structurally' interpreted the changes which have occurred in the sphere of Lat. coxa-femur-crus in the Romance languages. In this, he arrives at the following conclusion: 'one m i g h t . . . lay it down as a postulate of etymological research to replace each word within its associative field' ( 1 9 5 7 : 2 9 5 ; cf. Coseriu 1964:147). G. Matoré (1953), too, takes the historical dimension of language into consideration when he demonstrates the change of the 'champs notionnels' with respect to the change of the 'mots-témoins' and especially of the 'mots-clés'. 40 2.2.0 'Structural': understood as the structure of the interpretation of a signifiant. Here we are dealing with a semasiological point of view: Which signifié is (or which signifiés are) to be assigned to a particular signifiant (disambiguation)? 2.2.1 This is the procedure usually applied in lexicographical practice, even though today there is at least a tendency to typographically mark as different dictionary entries homonyms (better: homophones) which are in fact separate lexical units. 2.2.2 The semantics of transformational grammar, as it was first conceived by J. J. Katz and J. A. Fodor (1963) and then further developed especially by Katz (1966, 1967)," is based on the approach outlined in 2.2.0. The TG type of 87

The corresponding monograph with the same title appeared in Prague in I960. Cf. for example P. Guiraud (1962:74) with respect to Matoré's lexicology. 38 On the evaluation of associative configurations from the standpoint of a purely contentoriented structural semantics, see E. Coseriu 1964 :155, fn. 21; 1966 :180, 185-6, 208; 1968a : 7; and H. Geckeler 1971a: ch. III.4. Cf. also G. S. 35ur 1969. 40 Matoré's assumption that vocabulary changes in a rhythm which parallels generations must be taken with reservation. 41 For critical statements on the semantics of Katz and Fodor see U. Weinreich 1966, 1967. 38

114

EUGENIO COSERIU AND HORST GECKELER

semantics will not be further discussed here.42 Cf. 6.2 for possible points of contact or agreement between the semantics of TG and structural semantics (in the sense used here). 2.3 'Structural': understood as structure in an analytical respect, as the structuring of the content-level by means of functional lexical oppositions. The analysis of lexical meanings leads to the decomposition of the contents into smaller elements, i.e. into meaning-differentiating features. In the following sections, our survey will be concerned only with this conception of structural semantics, which we consider the real analytical semantics.48 Only on the solid basis of such a paradigmatic semantics can a well-founded combinatorial semantics be built up.

3.0 Before beginning our discussion of purely content-oriented structural semantics as it has been briefly characterized in 2.3, we should like to refer to a long and quite varied series of tentative approaches which from a historical viewpoint can be termed precursors of this proper form of structural semantics. 3.1.0 Attempts approaching an analytical semantics as we define it can be found to some extent in lexicography:44 3.1.1 Among the definitions in the larger dictionaries (the bilingual as well as the monolingual ones) from good lexicographical tradition, one can discern attempts to make use of meaning-distinguishing features, although this is not done consistently. Of course such features were not discovered on the basis of any method, but rather were found out intuitively. Sometimes there is also an attempt to determine and delimit a base-meaning for a given lexical unit, but in such cases it is often apparent that this determining of meaning is etymologically colored. Occasionally, too, restrictions concerning the ability of certain words (e.g. adjectives) to combine with other words (e.g. nouns) are formulated almost in terms of classematically-determined selection restrictions.45 3.1.2 Not to be overlooked are the contributions of the dictionaries of syno" For criticism concerning the principles of the semantics of TG, see E. Coseriu 1967b : 493-4, fn. 3; 1968a: 4-7; 1968b: 36-8; and 1970a. See also brief discussion and critical evaluation of the semantics of TG in H. Geckeler 1971a : VII.3; and H.-M. Gauger 1969. 43 H. Geckeler (1971a) provides a fairly complete bibliography and a survey of the development and the forms of this type of structural semantics. 44 The practical and theoretical value of the presently available dictionaries for modern semantics has been quite differently assessed by semanticists. While some linguists, as for example L. Hjelmslev, E. Coseriu, and A. Rey, emphasize primarily the fact that the lexicographical work which has been accomplished up to now often provides a valuable starting point for the practical realization of content-analysis, other scholars, as for example B. Pottier, U. Weinreich, and F. Rodriguez Adrados, are chiefly concerned with the weak theoretical bases of the existing dictionaries. 48 Cf. for example the famous German-Latin dictionary by Georges.

LINGUISTICS AND SEMANTICS

115

nyms, dictionaries of antonyms, and other dictionaries which contain information on synonyms and antonyms, for these two content-relations hold a solid place in content-oriented structural semantics (cf. for example the work of J. Lyons). 3.2.0 Older approaches, which one can term 'structuralistic' avant la lettre at least as concerns certain aspects of them, are to be found in linguistic tradition. 3.2.1 One may here refer first and foremost to K.W. L. Heyse (fl855), who had already had 'die Intuition eines strukturierten Wortfeldes'.46 In his work System der Sprachwissenschaft (1856:31-2) Heyse gives an analysis of the lexical field 'Schall' which amounts to a nearly perfect structural content-analysis, even though it was not so intended. He formulates differences of content in terms of distinctive features via intuitive application of the method of commutation; he discovers the hierarchical organization within his 'field' (i.e. the relation between archi-unit and unit, in more modern terminology) and already establishes the existence of contentcorrelations in the vocabulary. This analysis is thus valuable above all from the point of view of method. 3.2.2 The question of precursors and pre-stages of the lexical field is to be treated only briefly here. First of all reference can be made to the tradition of lexicography, specifically to onomastics, i.e. to the arrangement of vocabulary according to meaning-groups and/or object-groups. Outside of this tradition, E. Coseriu has already referred to 'die Intuition eines strukturierten Wortfeldes' in K.W.L. Heyse (before 1855) (cf. 3.2.1). The concept of the field is said to have been formed as early as in the work of E. Tegnér (1874) and C. Abel (1885). R. M. Meyer's 'Bedeutungssysteme' (1910) already parallel the Trier fields quite closely. The paradigmatic character of the lexical field has been intuitively discovered in the work of F. de Saussure (1916) (cf. 3.3.1). A. Stöhr (1910) and H. Werner (1919) were also forerunners. The first explicit formulation of the field-concept prior to Trier is by G. Ipsen in the year 1924. We wish to quote it primarily because of its undisputed influence on the terminology of subsequent research on the field: Ferner, die Eigenwörter stehn in einer Sprache nie allein, sondern sind eingeordnet in Bedeutungsgruppen; damit ist nicht eine etymologische Gruppe gemeint, am wenigsten um chimärische 'Wurzeln' aufgereihte Wörter, sondern solche, deren gegenständlicher Sinngehalt mit anderen Sinngehalten verknüpft ist. Diese Verknüpfung aber ist nicht als Aneinanderreihung an einem Assoziationsfaden gemeint, sondern so, dass die ganze Gruppe ein Bedeutungsfeld absteckt, das in sich gegliedert ist; wie in einem Mosaik fügt sich hier Wort an Wort, jedes anders umrissen, doch so, dass die Konturen aneinanderpassen und alle zusammen in einer Sinneinheit höherer Ordnung auf-, nicht in einer faulen Abstraktion untergehen (1924 : 225). And in fact, Trier does not exclude the possibility that his field theory was 48

E. Coseriu discovered this ante-litteram contribution to structural semantics. He has discussed and interpreted it in his contribution to the new Festschrift for R. Jakobson (1967b).

116

EUGENIO COSERIU AND HORST GECKELER

influenced by these formulations of Ipsen's (Trier 1931:11, fn. 1). Ipsen returned (1932) to his definition of the 'Bedeutungsfeld' in the Streitberg-Festschrift; however, he there modified his ideas in a direction which departs from a purely content-oriented study of language, namely in the direction of semantic and formal affinity among the members of the field. Ipsen's new concept of the field has achieved hardly any practical significance in linguistics; it merely represents a special case within content-oriented field theory. 3.2.3 L. Hjelmslev (1958:646) has drawn attention to another approach, this time from the area of comparative linguistics: A. F. Pott (1861), following the comparative method, studied the different lexical structuring of a section of the system of kinship relations in various languages, determining the relevant features by means of which the different lexemes are marked off with respect to one another. 3.2.4 Most probably many other approaches in this direction could be discovered in the history of linguistics. We shall allow the above mentioned examples to suffice. Still, it would be a worthwhile undertaking to explore linguistic tradition for such forerunners of a structural semantics. 3.2.5 In this context mention can also be made of an attempt to arrive at a more differentiated conception of word-meaning: around 1900 K.O. Erdmann distinguished the following three components in lexical meaning: 1. den begrifflichen Inhalt von grösserer oder geringerer Bestimmtheit,... 2. den Nebensinn, 3. den Gefühlswert (oder Stimmungsgehalt) (1910: 107 — emphasis ours).

This classification was taken over by a number of scholars, either in precisely this form, as for example by H. Sperber, E. Gamillscheg, K. Ammer, W. Schmidt; or in a slightly modified version, as by H. Kronasser, F. Kainz, A. Sieberer; or in a considerably changed form, as by H. Giintert, E. Otto, A. Scherer. 3.3.0 In the following section we are already dealing with approaches to problems which belong, or almost belong to the proper domain of structural semantics; these efforts were, however, generally not consistently followed through. 3.3.1 F. de Saussure's Cours de linguistique générale (which first appeared in 1916) does not provide a completed treatment of semantics; in fact, this discipline is touched upon only occasionally. In addition to his treatment of the 'rapports associatifs' (cf. 2.1.1), among which only one type ('la seule analogie des signifiés [enseignement, instruction, apprentissage, éducation, etc.]' 1964 : 174) belongs to purely content-oriented semantics, Saussure formulates, especially in his statements on 'la valeur linguistique considérée dans son aspect conceptuel', certain ideas which can be regarded as belonging to the fundamentals of a structurally understood field theory: Puisque la langue est un système dont tous les termes sont solidaires et où la valeur de l'un ne résulte que de la présence simultanée des autres... (1964: 159).

LINGUISTICS AND SEMANTICS

117

Even more clearly — because examples are supplied as illustrations (Fr. mouton — Engl, sheep/mutton and cf. below) — does the following quotation from Saussure point in the direction of the modern concept of the field, for here the paradigmatic character of the lexical field has been intuitively anticipated: Dans l'intérieur d'une même langue, tous les mots qui expriment des idées voisines se limitent réciproquement: des synonymes comme redouter, craindre, avoir peur n'ont de valeur propre que par leur opposition; si redouter n'existait pas, tout son contenu irait à ses concurrents (1964:160 — cf. 161-2). 3.3.2.0 A large number of semanticists agree that field theory represents a significant stage in the history of modern semantics. For example, S. Ullmann expresses the following estimation of the effect of Trier's field theory: 'The whole perspective has changed: a "Copernican revolution" has taken place in semantics' (1963:160). 3.3.2.1 In the following section the essential features of field theory, as developed by J. Trier (especially in 1931, 1932 a and b, 1934 a and b, 1938) and taken over and to some extent continued by L. Weisgerber (primarily in 1942, 1954, 1957, 1962 a and b, 1963, 1964) will be discussed briefly." Trier's conception of the field is based primarily on ideas from W. von Humboldt and F. de Saussure. From W. von Humboldt he took the principle of articulation (Gliederung) in language. This is probably the fundamental link between J. Trier and L. Weisgerber. From F. de Saussure he took the conception of language as a system. He was the first to consistently apply this principle of Saussure's to the study of vocabulary. Thus, he speaks of the 'Ideen der Ganzheit, der Gliederung und des Gefiiges' as the 'Leitsterne' of his work (1931 : 25). Trier conceives of the vocabulary of a synchronic stage of a language as a whole arranged according to principles of content; that is, organized in 'Wortfelder' which can stand side by side, or in a hierarchical relationship to one another. The 'Wortfeld'48 or 'sprachliches Zeichenfeld' itself represents in turn 'ein gegliedertes Ganzes, ein Gefiige'. The global meaning of the field is articulated and distributed, leaving no gaps, by means of the lexical units functioning in this field. The content of the different units is determined by mutual delimitation with respect to their neighbors in the 47

The following publications give information on the development and discussion of lexical field theory. The already quoted introductions to semantics by S. Ullmann 1962 and 1963, P. Guiraud 1962, and H. Kronasser 1952; S. Ohman 1 9 5 1 : 7 2 - 8 9 , and 1953; O. Duchâcek 1960a; N . C. W. Spence 1961; H. Gipper and H. Schwarz 1962 : lx-lxvi (Einleitung, fascicle 7 [1966]). The most recent publications which deal in detail with lexical field theory are L. Seiffert 1968a : 9-51, and, probably the most comprehensive, H. Geckeler 1971a: esp. Chapter III. 48 Trier does not make explicit distinctions between 'Wortfeld' (lexical field), 'sprachliches Zeichenfeld', 'Begriffsfeld' (conceptual field) and others. It seems that by 'Begriffsfeld' he means simply the content-side of the 'Wortfeld'. In lexicology, we prefer to speak of lexical field, or possibly of semantic field as well, provided that one uses this adjective in such a way as to refer exclusively to lexical content. The term linguistic field is more extensive, encompassing at least lexical field and syntactic field.

118

EUGENIO COSERIU AND HORST GECKELER

field ('die Einzelworte bestimmen sich durch Zahl und Lagerung im Gesamtfeld gegenseitig ihre Bedeutungen' (Trier 1931 : 7)). To illustrate, let us quote some of the most important passages for Trier's conception of the lexical field: In der Sprache ist alles Gliederung. Wie die Worte sich aus dem Feld ergliedern und darin ihr Wesen haben, so sind die Felder auch nur in der Gliederung übergeordneter Grössen und so stufenweise aufwärts bis zum Ganzen der Sprache (1934a : 188). Trier's definition of the field concept: Felder sind die zwischen den Einzelworten und dem Wortschatzganzen lebendigen Wirklichkeiten, die als Teilganze mit dem Wort das Merkmal gemeinsam haben, dass sie sich ergliedern, mit dem Wortschatz hingegen, dass sie sich ausgliedern. Die Ordnungshöhe ist dabei gleichgültig (1934b: 430). Das Wortfeld ist zeichenhaft zugeordnet einem mehr oder weniger geschlossenen Begriffskomplex, dessen innere Aufteilung sich im gegliederten Gefüge des Zeichenfeldes darstellt, in ihm für die Angehörigen einer Sprachgemeinschaft gegeben ist... Die das Wortfeld, den Wortmantel, die Wortdecke mosaikartig49 zusammensetzenden Einzelworte legen — im Sinne ihrer Zahl und Lagerung — Grenzen in den Begriffsblock50 hinein und teilen ihn auf (1931:1). Die Worte im Feld stehen in gegenseitiger Abhängigkeit voneinander.51 Vom Gefüge des Ganzen her empfängt das Einzelwort seine inhaltliche begriffliche Bestimmtheit (1931 : 2). Die Bedeutung des Einzelwortes ist abhängig von der Bedeutung seiner begrifflichen Nachbarn. Alle schliessen sich zu der Aufgabe zusammen, in den Block ungegliederten Bewusstseinsinhalts52 gliedernde Grenzen einzuziehen, ihn zu klären, ihn begrifflich fassbar zu machen (1931 : 3). After J. Trier had ceased publishing on field theory, L. Weisgerber continued these ideas in the spirit of their founder, so that nowadays we can quite correctly speak of the Trier- Weisgerber field theory as of one single conception. Weisgerber then incorporated the field theory into a comprehensive theoretical framework erected on Humboldtian principles. He distinguishes a 'static' and an 'energetic' study of languages, whereby he characterizes the former as grammatical procedure ('grammatisches Verfahren') and the latter as a fully linguistic procedure ('voll sprachwissenschaftliches Verfahren'). Static language study includes the 'formoriented' (gestaltbezogen) and the 'content-oriented' (inhaltbezogen) approach; energetic language study includes the 'efficiency-oriented' (leistungbezogen) and the 'effect-oriented' (wirkungbezogen) approach (Weisgerber 1963 : 11-18). Accor48

The comparison of the internal organization in a lexical field with a mosaic, which Trier probably took over from G. Ipsen, has been rejected in the critical literature by opponents and supporters of field theory alike. Especially noteworthy is the criticism of this kind of illustration of linguistic relations in the lexical field on the part of content-research ('Sprachinhaltsforschung') (e.g. by H. Gipper, H. Schwarz; more cautiously in L. Weisgerber). 50 What J. Trier here designates as 'Begriffsblock' seems to correspond to 'content purport' in Hjelmslev's terminology (Prolegomena to a theory of language, 1963). 61 Saussure's definition of the concept 'valeur' would seem to underlie this and the following determinations. 52 Cf. footnote 50.

LINGUISTICS AND SEMANTICS

119

ding to L. Weisgerber, these are the 'four stages in the investigation of languages'. In this, he is of the opinion that the 'leistungbezogene' approach is closest to the essence of language, which he sees in the ' "word"-ing of the world' (das Worten der Welt) (1963:36). Weisgerber defines the linguistic field as follows: Ein sprachliches Feld ist also ein Ausschnitt aus der sprachlichen Zwischenwelt, der durch die Ganzheit einer in organischer Gliederung zusammenwirkenden Gruppe von Sprachzeichen aufgebaut wird (1962a : 100).

As Weisgerber understands them, 'linguistic fields' (sprachliche Felder) include 'lexical fields' (Wortfelder) as well as 'syntactic fields', the latter representing fields of 'sentence patterns' (Satzbaupläne) (e.g. syntactic patterns for questions, commands, wishes in German) (not to be confused with Porzig's 'syntactic fields',58 coextensive with 'lexical solidarities'). Weisgerber includes the study of fields, especially of lexical fields, within the scope of content-oriented language study: Die grösste Bedeutung des Feldgedankens besteht aber darin, dass er zum methodischen Zentralbegriff der Sprachinhaltforschung und damit zum Schlüssel für das Aufdecken eines sprachlichen Weltbildes wird (1964 : 71). Dem Aufzeigen des Bestandes und der Struktur der in einer Sprache vorhandenen Wortfeder gilt die Hauptarbeit der inhaltbezogenen Wortlehre (1963 :70).

Weisgerber sees field study so emphatically within the framework of his total linguistic conception that he warns against over-estimating it, for even though the lexical field represents the most prominent form of content determination, there nevertheless exist other kinds of content determination besides it (cf. below concerning 'Sinnbezirk'). He has been occupied with the structure of lexical fields in a theoretical as well as in a practical sense. He has gone beyond Trier in distinguishing the following levels of organization and has illustrated them with more or less worked out lexical field outlines (as in, e.g., 1962a :176ff.): 'SINGLE-STRATUM FIELDS':

Arrangement in a series (Reihengliederung) : e.g. the numerical series or the grades used in evaluating achievements of pupils (example from Trier): e.g. sehr gut/gut/befriedigend/ausreichend/ mangelhaft/ungenügend. Surface arrangement (Flächengliederung): e.g. in the field of kinship terms in modern German: dominance of a single classificatory viewpoint. Arrangement in depth (Tiefengliederung) : e.g. the color cone in German: a two-level organization. Basically, this already goes beyond the type of the single-stratum field. 'FIELDS OF MORE THAN ONE STRATUM':

e.g. the linguistic articulation of dying in German. Weisgerber speaks of the 53

Cf. below fn. 59 and Duden 1959 : § 842.

120

EUGENIO COSERIU AND HORST GECKELER

fact that 'die deutsche Sprache einen dreifachen Ring um das sterben legt' (1962a: 184). Here, in the innermost circle, human dying (sterben) is contrasted with that of animals (verenden) and of plants (eingehen); in a second circle the 'cessation of life' is organized by means of objective points of view, e.g. erfrieren, verhungern; in the third and outermost circle by means of subjective points of view : e.g. hinscheiden, heimgehen, verröcheln, abkratzen, verrecken, etc.54 In criticism of Weisgerber's view of the lexical field 'cessation of life', it must be noted that he includes the entire 'historical language' (esp. 'diastratic' and 'diaphasic' differences come clearly to the fore here), instead of analyzing a 'functional language' (on these distinctions introduced by E. Coseriu, cf. 4.2.2.). Weisgerber also made reference to three different kinds of lexical fields as they can be distinguished on the basis of the particular domains which they organize: lexical fields from the domain of 'natural phenomena', lexical fields from the domain of 'material culture', and lexical fields from the domain of the 'intellect' (des Geistigen) (1964:72, 1942:30-3). Here, the lexical fields from the domain of the intellect assume an important specific role, for this part of vocabulary has 'viel mehr den Charakter des Vorstosses in die "geistige Zwischenwelt", teils in solchem Masse, dass tatsächlich die Begründung dieser "geistigen Gegenstände" nur von der Sprache aus verständlich wird' (1964:72). The transition from 'Wortfeld' to 'Sinnbezirk' ('semantic area') (in Weisgerber's sense) means, simultaneously, the changeover from a 'contentoriented' to an 'efficiency-oriented' perspective; from the 'static' to the 'energetic' point of view. Weisgerber wishes to propose the concept of the 'Sinnbezirk' as basic for efficiency-oriented word-study, paralleling the concept of the 'Wortfeld' which is primary in content-oriented word-study. In view of the fact that Trier's usage of this term was not consistent, Weisgerber wants to redefine it in a twofold way: Einmal als übergreifenden Begriff, innnerhalb dessen die Betrachtung nach Feldern, aber auch die anderen Formen des Aufzeigens der Bestimmtheit von Wortinhalten zu ihrem Recht kommen. Sodann kann man ihn eher der leistungbezogenen Forschung zuweisen (1963:104).

As early as 1956, the group of linguists named 'Sprache und Gemeinschaft' defined the concept of the 'Sinnbezirk', with the concurrence of J. Trier, as follows: Unter Sinnbezirk verstehen wir einen relativ selbständig erscheinenden Ausschnitt aus dem sprachlichen Weltbild, wobei ausser- und innerspachliche Bedingungen bei der Umgrenzung zusammenwirken. Es ist damit zu rechnen, dass wir in jedem Sinnbezirk Ausprägungen aller genannten Formen sprachlichen Zugriffs, grammatisch gesprochen also inhaltlicher Bestimmtheit, antreffen (1956-57 : 70). 64

Concerning the field 'Aufhören des Lebens', only briefly outlined by Weisgerber in 1962a : 184-5, K. Baumgärtner (1967 : esp. 190-2) has made a proposal for systematization and for a real content-analysis.

LINGUISTICS AND SEMANTICS

121

The decisively important 'Formen sprachlichen Zugriffs' respectively 'inhaltlicher Bestimmtheit' are represented by the 'Wortfeld' and the 'Wortstand' (cf. 3.3.3.).55 In conclusion let us cite a statement by Weisgerber on the relationship of Wortfeld/ Sinnbezirk: In dieser Stellung soll der Begriff Sinnbezirk den des Wortfeldes überbauen, sowohl systematisch, indem er die statischen Überlegungen der inhaltbezogenen Grammatik ins Energetische fortführt, wie auch in der Reichweite, indem er einer Überbelastung des Begriffes Wortfeld mit seinem Grundgedanken der Gliederung und wechselseitigen Umgrenzung durch andere Beobachtungen vorbeugt (1963 :206). Weisgerber's long-term goal is a synthesis of the four stages of language study (ganzheitliche Sprachbetrachtung).M 3.3.2.2 The number of publications having to do with field theory in a more or less relevant way is so extensive as to be almost impossible to survey. The scale of reactions ranges from the direct adoption and application of Trier's field theory to related areas (this particularly by Trier's own students) through criticism of certain features of the concept of the field and criticism of the application and material results of field-research, up to a total rejection of field considerations as a theory. We have tried to give a detailed discussion of these objections elsewhere (Geckeler 1971a: ch III. 3). Among the critics of field theory will be here mentioned only: F. Dornseiff (1938), F. Scheidweiler (1942), W. Betz (1954); also E. Oksaar (1958), whose approach is based on practical suggestions by W. Betz and theoretical affinity with E. Leisi (1953); finally, from the side of onomasiology57: U. Ricken (1961a and b) and W. Bahner (1962). A positive stance toward field theory, although with criticism of specific points, has been taken by W. von Wartburg (1937, 1962), S. öhman (1951, 1953) and also S. Ullmann. To the advocates of field theory belong, among others, those scholars concerned with the investigation of language content ('Sprachinhaltsforschung', e.g. H. Gipper, H. Schwarz (1962)) and also those interested in structural semantics (explicitly in E. Coseriu and J. Lyons). Peculiarly, criticism with regard to field theory is almost always directed against Trier alone, and only rarely against Weisgerber, although the latter is precisely the one who should be regarded as the protagonist of the wordfield idea ever since the fifties, at least. Thus, criticism lags behind the development of the field concept to some extent, for example in that it still dwells on the mosaic simile as originally applied to the organization within the lexical field. Weisgerber himself admits that 55

The most comprehensive enumeration of such 'Formen inhaltlicher Bestimmtheit' (6) is found in Weisgerber 1962b : 206ff. 56 The following articles give information about some aspects of the 'Sprachinhaltsforschung', which in Anglo-American tradition is occasionally characterized as 'Neohumboldtian' linguistics: H. Basilius 1952 and L. Seiffert 1968b. 57 On behalf of the onomasiologists, on the other hand, B. Quadri (1952 : 153-4) makes a very positive assessment of the value of Trier's field theory.

122

EUGENIO COSERIU AND HORST GECKELER

'die Feldforschung selbst gewisse Vereinfachungen, die in der ersten Ausbauzeit kaum vermeidbar waren, korrigiert [hat]' (1963:184-5). From the view of a structural semantics as we intend it here, it must be noted in criticism of the Trier-Weisgerber lexical field theory that this theory was indeed structural at the level of the interpretation of linguistic facts, but that its chief failing consisted in the lack of a linguistic method. Thus, the oppositional principle, which constitutes precisely the field-creating power, appeared only implicitly in linguistic discussions of the field. Then too, the fact that the difference between phenomena which are purely linguistically conditioned and others which are extralinguistically conditioned was not clearly recognized stood in the way of the clarification of many a controversial point. In addition, various scholars did not observe the fact that in the case of many fields (e.g. the evaluation scale for scholastic use by J. Trier, the titles of military rank, which R. M. Meyer had already investigated as terminology (1910b)), the question is not one of primary linguistic structurings, but of technical language. But it was in structural semantics (esp. by E. Coseriu) that a linguistic method was created for field theory by means of its connection with the principle of functional linguistic oppositions (cf. 4.2.2). 3.3.2.3 In the following section, we shall take a look at other kinds and conceptions of lexical field, first of all from the early years of field research. In an article in 1934, A. Jolies (1934),58 in opposition to J. Trier, proposed a field concept of his own, which he claims to have found anticipated in the work of the ancient grammarian Dionysios Thrax. Jolies' 'Bedeutungsfelder' are minimal fields, each with only two members, as for example Vater-Sohn, rechts-links, Tag-Nacht, Tod-Leben. J. Trier took this field type critically to task in that same year (1934 b) and recognized its limitations clearly. As a matter of fact, this field concept has scarcely become relevant for further progress in field research if one abstracts from L. Hjelmslev's 'petites classes fermées' (1958:652-3) of adjectives, e.g. grand : petit, beau : laid, etc., and likewise from J. Lyons' 'sense-relation' of the 'oppositeness' of meaning (cf. 4.2.1.4). Subsequent to various earlier attempts, W. Porzig, in an article (1934), developed in detail his conception of the field (he speaks of 'elementare Bedeutungsfelder' 'elementary fields of meaning'). Porzig points out that groups such as greifen-Hand, sehen-Auge, hören-Ohr, lecken-Zunge; bellen-Hund, wiehern-Pferd; jällen-Baum and others belong together by virtue of meaning. He does not conceive of them simply as 'Konsoziationen' in H. Sperber's sense; rather, for him it is a question of 'eine beziehung, die im wesen der gemeinten bedeutungen selbst gründet. Ich nenne sie deshalb wesenhafte bedeutungsbeziehungen, (1934:70). Like Jolies, Porzig presents his conception of the field in contrast to Trier's, and defends the minimal field; J. Trier in turn made known his critical stand on these 'elementary fields of meaning' and, with respect to these semantic fields and to those of Jolies, he 58

In addition, this article contains many valuable individual observations, also with regard to a structural diachronic semantics.

123

LINGUISTICS AND SEMANTICS

makes the following statement: 'In ihnen liegt nicht Teilgefüge eines grossen Gefüges vor, sondern sie sind kleine gefügte Stellen innerhalb einer Wüste des Ungefügten' (1934b: 449). In his admirable introductory work Das Wunder der Sprache, Porzig presents a revised conception of the field problem. He recognizes the justification for the existence of the type of lexical fields as developed by Trier and Weisgerber in addition to the elementary fields of meaning which he himself had described. The former he henceforth designates 'paratactic fields' ('paradigmatic fields' would be better); his own he designates 'syntactic fields'.59 The evidence that Porzig had made a truly relevant linguistic discovery, which, however, he had at first confused with Trier's lexical field, can be illustrated by the fact that various linguists concerned with modern semantics have given a firm place in their considerations to his 'wesenhaften Bedeutungsbeziehungen' (or 'elementaren Bedeutungsfeldern' or 'syntaktischen Feldern'). We encounter them in work by E. Leisi under the designation 'semantische Kongruenz' (1953:68-70, 119, fn. 3); H. Schwarz calls them 'Prädikativklammer' or 'Wertigkeitsbereich' (1959:251); P. Grebe, on the other hand, terms them 'Sinnkopplung' or 'semantisch-syntaktischer Hof' (1967:111-12). In E. Coseriu's outline of a comprehensive semantic theory60 they appear classified under the 'syntagmatischen (oder kombinatorischen) lexikalischen Strukturen', as 'lexikalische Solidaritäten'.61 Chronologically considerably later than the origin of Trier's conception of the field, certain linguists proposed others, among which we will mention here only the best-known. Ch. Bally outlined his 'champ associatif' (1940); G. Matoré his 'champ notionnel' (1953), and P. Guiraud introduced his 'champ morpho-sémantique' (1956). We have already briefly characterized these three field types in conjunction with the associative configurations (cf. 2.1.0-2.1.7). In conclusion, O. Duchácek's conception of the field62 should be mentioned here. CHAMPS LINGUISTIQUES

de mots morphologiques

d'idées

syntagmatiques (syntaxiques)

conceptuels

sémantiques

associatifs 5

* 'Wortinhalte werden also in doppelter Weise feldmässig festgelegt: durch die syntaktischen und durch die parataktischen Wortfelder' (1967 : 126). The principles of this theory are contained in 1966 and 1968a. 61 E. Coseriu gives a refined analysis of Porzig's 'wesenhafte Bedeutungsbeziehungen' in the light of his structural semantics (1967c). Esp. set forth in his monograph Le champ conceptuel de la beauté en français moderne (1960) and summarized earlier (19S9).

124

EUGENIO COSERIU AND HORST GECKELER

Its complexity is made clearly apparent in the schema which precedes (1960:20). As far as a purely content-oriented study of the lexical field is concerned, Duchácek's 'champs linguistiques d'idées', which in turn include 'champs conceptuels' and 'champs sémantiques', are all that is relevant. These are distinguished in the following manner: Les champs sémantiques diffèrent des champs conceptuels par un degré plus petit d'homogénéité, par une plus grande complexité et une plus grande étendue, unissant les mots qui concernent par exemple les travaux agricoles ou l'administration ou l'âge humain ou les degrés de parenté, etc. (1959 : 300).

We have tried to give a somewhat more detailed discussion, as well as an attempt at criticism, of Duchácek's field concept elsewhere (Geckeier 1971a: ch. III. 4). 3.3.2.4 Finally, let us refer to the danger of misinterpretations of the field concept. E. A. Nida apparently assumes that the various possibilities of the use of a word and the tree diagram of signifiants (as e.g. bachelor) customarily used in the semantics of TG represent a lexical field (1964:39-40): Such a schematic structuring can be said to define the semantic field, including the relationships between the various dictionary entries for a particular t e r m . . . The semantic field of any lexical item is always much greater than the meaning which occurs within a specific context.

Furthermore, a field of the type developed by Matoré, contrary to the opinion of its author, has little more than the name in common with the Trier-Weisgerber type.63 3.3.3 Approaches to structural semantics also come from such representatives of the theory of word-formation who see the principles of this linguistic subdiscipline not first and foremost as morphologically determined, but rather, as contentdetermined. If one overlooks such forerunners as Christian Wolff (1730) and O. Jespersen (1924), then the decisive points of view do appear first in studies by W. Porzig, and then, more explicitly, in the work of J. Kurylowicz, who refers expressly to Porzig. On the one hand, Porzig (1930-31, 1967 : 104-5, 128-35) finds that abstracta summarize a sentence by focusing the predicate, e.g. die Rose ist rot die Röte der Rose; on the other hand, he recognizes that the two expressions are synonymous with respect to designation (Bezeichnung), but not with respect to signification (Bedeutung) (cf. the rigorous distinction between Bedeutung and Bezeichnung by E. Coseriu in 4.2.2). J. Kurylowicz64 broadened Porzig's outline into a theory. Among the important results of his investigations should be mentioned in this context: 1) not only abstracta, but also other derivations com63

Cf. also Matoré's scarcely credible claim: 'N'ayant pas su s'imposer, elle [la linguistique allemande des champs] est restée inconnue des linguistes français qui ont dû la réinventer en partant du concret linguistique' (1953 : 64). 44 In a first article (1936) the Polish linguist advocates a somewhat different conception from that in his contribution to the 6th International Congress of Linguists (1949a) and in 1949b : 54-6.

LINGUISTICS AND SEMANTICS

125

prise a syntactic function; 2) certain derivations do not contain any sentence function: e.g. château ->- châtelet; 3) the various phases of the underlying transformations are contained in the end product, e.g. arc (1) -» (se servir) d'arc (2) -> (se servant) d'arc (3) (individu se servant) d'arc (4) -> archer (5). The difference between primary and secondary lexematic structures (cf. 4.2.2) has been clearly recognized here. Kurylowicz's distinction 'dérivation syntaxique' / 'dérivation lexicale' appears in Ch. Bally (1965 : §180) as 'transposition fonctionnelle' / 'transposition sémantique'. This distinction of Bally's has been taken over by other scholars, as for example by H. Marchand (1966, 1969). The perspectives in word-formation opened up by Porzig and Kurylowicz were later pursued by B. Pottier, to some extent, but are completely integrated only in the total conception of lexematic structures by E. Coseriu (cf. his word-formation types of Modifikation, Entwicklung, Komposition in 4.2.2). In conclusion, let us here mention the content-structures of the 'semantische Nische' ('number of derivations of a suffix forming a semantic group') and of the 'Nischenuberdachung' ('the coinciding of the same meaning-group of two suffixes'), which were introduced by K. Baldinger (1950 : 279). The niohe, then, constitutes a subgroup within a materially identical derivational type which is held together by common content. Corresponding to the 'bridging over of niches' is the concept 'Wortstand' developed by the 'Sprachinhaltsforschung' ('Wichtig ist, dass hier ein bedeutsames Aufbauprinzip getroffen ist, und zwar der Aufbau grosserer Sinneinheiten vornehmlich durch das Zusammenwirken verschiedener Ableitungsmittel' (Duden 1959 : §§832-3)) e.g. the various means, in word-formation, for the formation of words designating professions in German (whereby the formations in -er represent a semantic niche). 3.3.4 In connection with the more recent discussion of the problem of homonymy (and of polysemy), certain points of view became apparent which at the same time, are relevant to structural semantics and even implicitly presuppose the existence of such a semantics. It was the contribution of R. Godel to have taken up the problem anew in an important article with the characteristic title "Homonymie et identité" (1948) and to have suggested a solution based on a formal, synchronic criterion. He underscores the fact that his mode of inquiry concerns the paradigmatic axis, and differentiates homonyms (better: homophones) on the basis of their belonging to different derivational series, e.g. French polii - polir, dépolir, polissage . . . / polh - impoli, poliment, politesse, etc. French louer (louange, louangeur) is considered a homonym with louer (location, locataire, sous-louer), and, in the case of the latter unit louer, he assumes 'identity' (i.e. polysemy) for 'donner en location'/'prendre en location'.95 In summing up, Godel defines the 'signes homonymes' in the following manner: • 5 H. Frei ( 1 9 6 1 : 4 4 - 5 ) , on the other hand, points out that louer ('locare'), too, is the source of two different derivational series.

126

EUGENIO COSERIU AND HORST GECKELER

ce sont des monèmes, phonologiquement semblables, mais distingués par leur place dans les rapports associatifs. C'est dans ces conditions qu'il y a lieu de discerner les cas d'homonymie et ceux d'identité et la confrontation des séries mémorielles fournit un critère moins subjectif que le sentiment linguistique des individus (1948:14-15).

H. Frei (1961 : 43) presents a procedure which permits a distinction between polysemy and homonymy, although, to be sure, only for the substantives. It has to do with a pronominal replacement test. In order to prove that French mouton represents only one single lexical unit with two variants ('acceptions'), Frei cites the following perfectly acceptable sentence: 'On n'élève pas seulement le mouton pour en manger.' To show that in French police, on the other hand, there are two separate lexemes and not merely two variants of a system-unit, the Geneva linguist adduces contexts in which the pronominal replacement cannot be interpreted with reference to the unit police appearing in the main clause: 'On informera la police [ = 'police'] si vous ne la [ = 'policy'] signez pas; Il a des ennuis avec la police [ = 'police'] parce qu'il n'en [ = 'policy'] a pas.' In his discussion of homonymy in Fr. voler, E. Benveniste (1954) goes beyond the beginning made by Godel in that he places the viewpoint of the signifié in the center of his considerations. On the ground that they belong to two different '•classes sémantiques', he distinguishes two separate lexemes voler: voleri belongs to the semantic class of 'marcher, courir, nager, ramper, . . . '; voler2 to that of 'dérober, soustraire, . . . In principle, we here already have before us the dissolution of homonymy by means of the lexical field as it is expressly represented by W. Porzig ('Die Zugehörigkeit zu ganz verschiedenen Feldern müsste für die Scheidung genügen' - 1959: 161) and L. Weisgerber (dissolving of the 'Bedeutungs"klumpen" ' by means of assigning the different units to different lexical fields - 1962a : 209-10). Benveniste uses other criteria as well, however: the grammatical construction and the derivational series (as defined by R. Godel): voler 'fly' is intransitive, but voler 'steal' is transitive. The two derivational series have the following form: voleri : vol\, voleter, s'envoler, survoler, volée, volatile, volaille, volière voler2'.vol2, voleur. The problem, here only just touched upon, of homophony and of polysemy,68 has, among other things, in common with the focus of structural semantics that the paradigmatic axis of the language is emphasized first and foremost, and that the lexemes as linguistic units are clearly distinguished and delimited with respect to one another. 3.3.5 If we now briefly present here L. Hjelmslev's contribution of paramount importance to the founding of a structural semantics, then the question arises as to whether he is to be classified here, historically, among the forerunners, or whether M

For additional bibliography, the following articles should also be mentioned: O. Duchâiek 1962, K. Heger 1963, and especially W. A. Koch 1963.

127

LINGUISTICS AND SEMANTICS

he should not rather be included under point 4 among the very representatives of modern structural semantics; various arguments, which we will cite in the following paragraph, speak for the classification finally adopted by us. There can be no doubt that it was L. Hjelmslev who after all laid the foundation for the possibility of a structural semantics with his idea that the content-level of language can be analyzed in a way analogous to the level of expression. Hjelmslev's project, however, did not get beyond the initial stages, chiefly for two reasons: first of all, because he completely eliminates semantic substance;87 secondly, because the separation between the lexical and the grammatical sphere is not carried out. Thus pleremics of the glossematician J. Holt (1946, 1961, 1964) deals almost exclusively with grammatical content. (Cf. Alarcos Llorach 1969). Hjelmslev's plerematics (the counterpart to kenematics) is concerned with everything belonging to the content-level. By analogy to the level of expression, Hjelmslev (1959) 88 finds the following differentiations for the content-level, which is precisely what is being investigated in plerematics: central (radicals) marginal (derivatives) intense (nominal morphemes) extense (verbal morphemes) What is especially important here is the separation of the derivational elements from the grammatical elements (morphemes). Let us now return to the structural content-analysis. Proceeding from the principle that there is solidarity between expression and content (which was interpreted as isomorphism), Hjelmslev in his chief work, Omkring sprogteoriens grundlaeggelse,69 transferred the decisive idea of passing under the sign-threshold to the analysis of vocabulary as well. Considerations of linguistic economy in the creation of new signs led Hjelmslev to the insight that the unlimited number of linguistic signs is made up of a limited number of 'non-signs' called 'figurae': 'Thus, a language is so ordered that with the help of a handful of figurae and through ever new arrangements of them a legion of 67

On the form/substance problem in Hjelmslev's conception of language, see E. Coseriu 1954, especially chapter V. 68 As a critical introduction to glossematics, see Siertsema (1965). A concise presentation of this direction of modern linguistics is offered by E. Fischer-J0rgensen (1952); (our schema was taken from this article). 69 We are using the English translation by F. J. Whitfield, revised edition, 1963.

128

EUGENIO COSERIU AND HORST GECKELER

signs can be constructed' (1963 : 46). He regards 'the construction of the sign from a restricted number of figurae' as 'an essential basic feature in the structure of any language' (47). Hjelmslev's now-famous statement: 'une description structurale ne pourra s'effectuer qu'à condition de pouvoir réduire les classes ouvertes à des classes fermées', contained in his report on structural semantics for the 8th International Congress of Linguists (1958:653), is clearly anticipated already in the Prolegomena (e.g. 1963 :71). Here, the question is one of attempting to reduce the open lists of the vocabulary to closed lists according to the grammatical model; basically, he strives toward a grammaticalization of the vocabulary. While he considers, on the level of expression, already the phonemes as figurae, his content figurae correspond at least in part to what we now call distinctive content features or semes. In the Prolegomena (1963:70), Hjelmslev gives some examples to illustrate how he conceives of the first stage of the analysis. Unfortunately, the great Danish linguist, who died far too soon, neither suggested new examples nor continued his analysis in his later works. Hjelmslev cites the following examples: 1) 'ram' = 'he-sheep' 2) 'ewe' = 'she-sheep' 3) 'man' - 'he-human being' 4) 'woman' = 'she-human being' 5) 'boy' = 'he-child' 6) 'girl' = 'she-child' 7) 'stallion' = 'he-horse' 8) 'mare' = 'she-horse' The content of eight units of vocabulary is thus determined by means of combinations of six elements (content figurae). If one considers, e.g. the examples 1) and 2) and 1) and 7), one can state that 'The exchange of one and only one element for another is in both cases sufficient to entail an exchange in the other plane of the language' (1963 : 70). Hjelmslev called this procedure 'exchange test', which ultimately belongs to the commutation test. This he defines in the following manner: L'épreuve qui doit servir à montrer si le remplacement d'un élément par un autre dans le plan de l'expression de la langue peut entraîner une distinction dans le plan du contenu, ou si le remplacement d'un élément par un autre dans le plan du contenu peut entraîner une différence dans le plan de l'expression (1966:173).70 The commutation test serves first of all to identify the functional units, the invariants, within a paradigm. However, this service is not necessary for content analysis, since the units are here given as already identified, if one disregards liomophony and polysemy. On the other hand, however, the commutation test forms the instrument for the delimination of the functional level and for the analysis of lexemes into distinctive features, and as such it can be used as a criterion for 70

The linguistic term commutation was first used by Hjelmslev himself.

LINGUISTICS AND SEMANTICS

129

the distinction of the two most important directions within structural semantics (cf. 4.1 and 4.2). L. J. Prieto (1956, 1957) has formulated proposals for a content analysis based on Hjelmslev. But Prieto's "noologie" (1964) remains a disappointment since it stops at the point where structural semantics begins. There is also an important line of affiliation running from Hjelmslev to A.-J. Greimas, in whose semantics the content-substance is not eliminated, however; and finally also to E. Coseriu. 4.0 Various important developments71 have grown out of the structural approaches and assumptions discussed in Section 3. They can be differentiated on the basis of their differing methods (distribution/commutation). 4.1.0 Determination of content by means of distributional methods. In the following paragraphs we will briefly discuss some important representative figures of this methodological approach. Yet here it must be stated, however, that this orientation of semantic research did not progress beyond the experimental stages. 4.1.1. M. Joos has attempted, in his article "Semology: A linguistic theory of meaning" (1958) with the help of a 'purely linguistic test', namely the process of 'collocation',72 to define differences in meaning as differences of distribution in various contexts. Joos illustrates his procedure by means of the English lexeme code, and arrives, via the 'congruence'73 operative in the collocations among the meanings of the combinations with code, at a 'continuous 14-place CODE ring', which he interprets as follows: 'CODE has just one 'sememe' with 14 'allosemes'' (1958:64). Via collocations such as rigid code, strict code, ethical code, military code, religious code, etc., in which certain allosemes are congruent, others incongruent (and are therefore eliminated), Joos arrives at the following fourteen allosemes of code: Formalism, Codification, Law, Rule (public), Custom (public), 71

On these most recent developments, the aforementioned introductory handbooks on semantics do not yet give information. For orientation we refer the reader especially to the following studies: Tz. Todorov 1966; A. Rey 1969; L. Lerot 1967; and H. Geckeler 1971a. Individual authors and their approaches are treated briefly also in R. Barthes 1964 and Y. Ikegami (1962, 1967). 72 'In lexicography, a collocation is a word-combination which throws light on the meanings of the words involved' (1958:62); Joos' definition of 'collocation' reads: 'concurrence of morphemes which eliminates meanings (others then [w'c] surviving)' (55). 73 'Congruence: matching of surviving meanings between morphemes within a collocation' (1958:55); simplified graphic representation in the case of the collocation RIGID CODE:

130

EUGENIO COSERIU A N D HORST GECKELER

Ethics, Morals, Ritual, Crypto-Secrecy, Clique-Language, Condensation (public), Recoding (public), Language, Lexicon (1958 :55). His 'semology' is concerned with 'distributional meaning' or 'inside meaning' (Joos 1962:46). 74 4.1.2 J. Dubois, based on Z. S. Harris' conception of distribution, undertook (1964) to distinguish 'synonyms' on the basis of distribution. His statements on the French examples aigu-pointu, route-chemin-voie, and briser-casser-rompre utilizing the 'environnements (de droite et de gauche)' which represent the distribution, exhibit the danger of confusion between signification and designationJS (cf. 4.2.2). 4.1.3 S.M. Lamb's 'sememic approach' (1963) which is placed within his 'stratificationaT (cf. 1966) method, is distributional, too. Lamb distinguishes the following 'strata' of language structure: phonemic, morphemic, lexemic, sememic; and, in addition: 'two peripheral strata which relate to the structure but are outside it: the phonetic and the semantic' (1963:4). The relationship of respectively two consecutive strata to one another is characterized by means of types of 'representational relations', whereby the principle is: 'emes are represented by their alios on the next lower stratum' (1963 :3). With respect to the two strata which are particularly interesting to us in this connection Lamb states: ''the sememic stratum is above the lexemic, and . . . it is on the lexemic stratum that we find the alios of the sememes, that is, the allosemes' (17). - 'Sememes have their representations, i.e. their allosemes, on the lexemic stratum, not on the semantic' (Ibid.), whereby the semantic stratum would correspond to Hjelmslev's 'content substance.' Lamb's example, which exhibits 'diversification' as a representational relation, concerns the sememe S/konnen/which has, in English, on the lexemic stratum the two allosemes can and be able to, cf. he can go/he will be able to go. In fact this is a matter of distribution of suppletive forms. But is it justifiable to assume a particular stratum for material suppletivism? In his publications, Lamb works with a great profusion of terminology; the analyses, however, do not come up to expectation. 4.1.4 Excursus. A t this point, we would like to insert an excursive discussion of the function of context and of contextual determination of meaning. T w o things must not be confused here: for one, contextual meanings, i.e. discourse-meanings or discourse-variants (as such, not relevant to our discussion), and second, the determination of meanings on the basis of context as a method. Here, then, we are dealing with meaning which is determined and investigated in terms of context. The meaning of a word is here equated to the sum of the different contexts in which it occurs, i.e. to its linguistic distribution. Publications from the English, respectively London, school of linguistics were probably the first in which the connection of the situational with the contextual determination of meaning was made. J. R. Firth took over the term context of situation from B. Malinowski (1960:306), and has developed this concept further in a series of studies (since 74 75

A succinct resume of Joos' semology is given e.g. in Y . Ikegami 1962 : 10-11. Such a confusion is clearly present in J. Dubois 1960.

LINGUISTICS AND SEMANTICS

131

1930).78 He applies the principle of context to all levels of language, from phonetics to semantics; for him, contexts are hierarchically arranged, ascending up to the 'context of culture'. With respect to his technique, Firth writes: It can be described as a serial contextualization of our facts, context within context, each one being a function, an organ of the bigger context and all contexts finding a place in what may be called the context of culture ( 1 9 5 7 a : 32).

Thus, he then defines 'meaning' in terms of context of situation: Meaning is best regarded in this way as a complex of relations of various kinds between the component terms of a context of situation ( 1 9 6 4 : 1 1 0 ) .

These 'component terms' are, for him, made up in the following way:77 A context of situation for linguistic work brings into relation the following categories: A . The relevant features of participants: persons, personalities. (I) The verbal action of the participants. (II) The non-verbal action of the participants. B. The relevant objects and non-verbal and non-personal events. 78 C. The effect of the verbal action.

Even though the schema of the components of the situational context may seem attractive at first glance, this combination of linguistic factors with extralinguistic factors points up weighty theoretical (Abercrombie 1965) and practical79 problems. Firth's 'test of "collocability" '80 does not solve the practical difficulties either. Thus, we must agree with the rather negative judgement at which J. Lyons (1966) arrives in a critical sifting of what J. R. Firth has to offer semantics in the way of theory and method. In the more recent theoretical literature in linguistics, the problematics of context is usually treated together with that of the situation. Thus, E. A. Nida (1962: 152-5) classifies 'environment', with the help of which he would like to determine the 'meaning',81 in the following way: A. Nonlinguistic Environments 1. Objective Environments 2. Subjective Environments B. Linguistic Environments 1. Structural Environments 2. Contextual Environments 76

The most important publications of Firth in this regard are Speech (1930), The tongues of men (1937) [reprinted together as The tongues of men and Speech (London 1964)], and a series of articles collected in his Papers in linguistics 1934-1951 (1957b). 77 "Personality and language in society" (1950), reprinted 1957b : 177-89, see p. 182. 78 This category is here reproduced in the expanded form given by Firth (1962 :9). 79 Firth himself points out practical inadequacies (1962 : 9). 80 On this topic see his "Modes of meaning" (1951), reprinted 1957b : 190-215, see pp. 194 ff. 81 'Meaning is definable by environment' (1962 :152).

132

EUGENIO COSERIU AND HORST GECKELER

K. Ammer (1958:67) proceeding from E. Otto's three-part classification (1965 : 102), suggests a four-part division of the context complex: 1. Linguistic Context 2. Situational Context 3. Emotional Context 4. Cultural Context.82 E. Coseriu (1955-56) has developed the most comprehensive and most richly categorized context theory to date. Starting from distinctions suggested by Ch. Bally, K. Biihler, and W. M. Urban, he worked out a substantially expanded and refined classification which can be reproduced only schematically below. He establishes a differentiation consisting of four possible basic types of 'entornos' (= 'environments'), with further subdivisions: I. situación: (= 'el "espacio-tiempo" del discurso, en cuanto creado por el discurso mismo y ordenado con respecto a su sujeto', 1962:310) 1. situación inmediata ('creada por el hecho mismo de hablar', 311) 2. situación mediata ('creada por el contexto verbal', 311) II. región: (= 'el espacio dentro de cuyos límites un signo funciona en determinados sistemas de significación', 311) Three types can be distinguished: 1. zona (= 'la "región" en la que se conoce y se emplea corrientemente un signo', 311) 2. ámbito (= 'la "región" en la que el objeto se conoce como elemento del horizonte vital de los hablantes o de un dominio orgánico de la experiencia o de la cultura', 311) 3. ambiente {= 'una "región" establecida social o culturalmente', 312) III. contexto: (= 'toda la realidad que rodea un signo, un acto verbal o un discurso, como presencia física, como saber de los interlocutores y como actividad', 313) Once again, three types are distinguished: 1. contexto idiomàtico 2. contexto verbal ai) inmediato a.2) positivo ('aquello que efectivamente se dice', 315) bi) mediato bí) negativo ('aquello que se deja de decir', 315) 3. contexto extraverbal with the subtypes: a) contexto físico b) contexto empírico c) contexto natural d) contexto práctico u ocasional e) contexto histórico ai) particular a2) actual ft) universal fo) pretérito 81

This could well be an echo of Firth's 'context of culture*.

LINGUISTICS AND SEMANTICS

133

f) contexto cultural IV. universo de discurso: ( = 'el sistema universal de significaciones al que pertenece un discurso (o un enunciado) y que determina su validez y su sentido', 318). T. Slama-Cazacu published a study (1961:207-23) of considerable length which has, among other things, the complex 'context' as theme. She distinguishes various aspects of context, which, however, are a disappointment in comparison to the rich systematic differentiation of 'entornos' in E. Coseriu. 4.2.0 Determination of content by means of the commutation method. Here we find ourselves in the realm of true content-analysis, of paradigmatic lexematics. 4.2.1.0 In the following paragraphs, the decisive ideas of the most important representatives of this orientation will be discussed briefly, whereby mention will be made of the particular achievements as well as of the deficiencies of these investigations. It may be noted that the linguists discussed below developed their versions of structural semantics almost simultaneously and independently of one another (this is particularly the case for B. Pottier, A.-J. Greimas, and E. Coseriu). 4.2.1.1 In France, B. Pottier (primarily in 1963, 1964, 1965) worked out an adequate conceptual apparatus for the analysis of content into distinctive features. This was done for the most part in clear parallel to already existing differentiations in the domain of phonic sciences. He introduces the following terms as relevant for lexematics: Corresponding to the phème, le sème appears in lexematics, defined as 'le trait sémantique pertinent' (1963 : 8); parallel to the phémème is le sémème, defined as 'l'ensemble des traits sémantiques pertinents (ou sèmes) entrant dans la définition de la substance d'un lexème' (1963:8); corresponding to the phonème is le lexème which is understood as the lexical realization of a sememe. As a further parallel to phonology, B. Pottier introduces, for the area of vocabulary, archi-units: thus l'archisémème and its lexical realization as archilexème, also called cover-word and inclusif (1967a : 55). The classème introduced by Pottier does not, however, go back to a model in the area of phonematic analysis: 'Le classème est une caractérisation d'appartenance de sémèmes à des classes générales sémantico-fonctionnelles: animation, continuité, transitivité' (1964:125). While semes and classemes constitute the denotative meaning of a lexeme or of a lexie (as Pottier designates a lexical unit), the element virtuème introduced by him belongs to the domain of connotative meaning: Chaque lexie a ainsi un certain nombre de virtualités combinatoires, qu'on peut appeler ses virtuèmes. Ceux-ci peuvent être caractérisés par un indice, très approximatif, de probabilité (1964 : 130-l).8s He even regards the virtueme as a kind of seme: 'les sèmes variables forment le Cf. also his definition (1967b : 190).

134

EUGENIO COSERIU AND HORST GECKELER

virtu^me, et sont connotatifs ( . . . ) ' (1967a: 27). Thus, according to Pottier, a lexical unit is made up of the following four components: sémème (particularisant)

classèmes (généralisants)

virtuèmes

fonctème** constante

variante

Si l'on considère une lexie, on a les composantes sémème et classème, fondées sur un choix essentiellement paradigmatique, et les composantes fonctème et virtuèmes, déterminées surtout par la situation syntagmatique (contraintes diverses au moment du choix paradigmatique) (1964:133).

With respect to the linguistic evaluation of the virtueme, we are of the opinion that it is not a question of a fact of language, but rather of a category which is based on the knowledge of extralinguistic reality, of things (cf. e.g. 'mouette' 'blanc'). In summary and as a transition let us quote from Pottier's recent fundamentals of a theory of linguistics: Le contenu sémique d'un lexème est son sémème. Le sémème est l'ensemble des sèmes. Le sème est le trait distinctif minimal de signification, et se révèle par opposition dans un ensemble lexical. - Ce n'est donc qu'en travaillant sur de petits ensembles lexicaux qu'on peut établir les sèmes d'un sémème (1967a: 26).

B. Pottier has actually demonstrated the analysis of such a 'petit ensemble lexical' (1963:11-17), which corresponds approximately to the concept 'lexical field'. His study concerns the field 'siège' ('seat') in modern French.85 He carries out his analysis on the following five lexical units: chaise, fauteuil, tabouret, canapé, pouf. Thus, for example, he analyzes the 'lexie' chaise into the following semes: Si: 'avec dossier', S2: 'sur pied', S3: 'pour 1 personne', S4: 'pour s'asseoir'. The 'lexie' fauteuil presents the same sememe (S) as chaise plus the seme S5: 'avec bras'. The result of Pottier's analysis is schematically represented as follows: Si

S2

S3

S4

fauteuil

+ +

tabouret

—

+ + +

+

+ + + +

+ + + +

—

+

chaise

canapé 8

pouf « 84

—

+

+

S5

Sg

+

+ = Si + = S2 + = s3 + = s4

—

—

—

+ —

= S5

'Nous proposons d'appeler fonctème l'ensemble des traits grammaticaux fonctionnels d'une lexie' (1964:127) and 'fonctème ("parties du discours" et leurs implications)' (133). However, a complete field is not investigated. 86 In his 1965 article, Pottier has undertaken some minor modifications in his analysis. Thus, he no longer includes the unit pouf in this field, so that the seme s6 becomes irrelevant. 85

LINGUISTICS AND SEMANTICS

135

51 : 'avec dossier' 52 : 'sur pied' 53 : 'pour 1 personne' s4 : 'pour s'asseoir' S5 : 'avec bras' s6 : 'avec matériau rigide' The semes S2 and S4 are common to all the lexical units in question: they make up the archi-sememe of the field; their lexical realization as archi-lexeme is represented by the 'lexie' siège. We can now oppose the different lexemes in pairs so that in every case they are distinguished only by a single distinctive feature (the other semes being identical): pouf-tabouret tabouret-chaise chaise-fauteuil fauteuil-canapé

: : : :

s? Si S5 S3

sg, S4 make up the common basis, the archi-sememe.

Pottier's illustration of an analysis87 raises the question as to whether or not this is really a matter of an analysis of linguistic content or, at least in a first phase of the analysis, rather of a description of a series of functionally related objects, which is to say, of a part of extralinguistic reality. This point has been subject to criticism (e.g. Coseriu 1968a : 8-9). And in fact, Pottier starts from a description of the objects, cf. e.g. 'Prenons une suite d'objets désignés par la lexie chaise, et décrivons chacune de ces chaises de la façon la plus complète' (1963:11). On the basis of his test, he is able to eliminate the features of the description which are not relevant and thus arrives at the inventory of the pertinent features, whose combinations we have seen above. Do these features really function as semes in the language? 'Wo liegen die Grenzen zwischen sachlichen und sprachlichen Gesichtspunkten?' (Gipper 1959:276). Certain restrictive comments on Pottier's structural semantics88 notwithstanding, his important contribution remains his having introduced the idea of the classeme into modern semantics and having reconsidered the old problem of polysemy in the light of his semantics. Indeed, he prepared a linguistic apparatus which is most highly suitable for application to the content-analysis of lexical fields. 4.2.1.2 Likewise from France is A.-J. Greimas' contribution (1966) to contentanalysis, which was not available in published form until 1966, but had been worked out some years earlier in the form of lectures. From this work, which is 87

Pottier does not refer to H. Gipper's article (1959), which would have been of particular interest by way of comparison for his analysis. 88 The introduction of the category of the virtueme was no doubt conditioned by practical exigencies and aspects of machine translation.

136

EUGENIO COSERIU AND HORST GECKELER

somewhat unsurveyable in spite of its systematic external arrangement, we will discuss only a few points, since the fundamental ideas of this semantics (and to some extent the terminology, too) coincide with the ideas of Pottier and Coseriu, who, moreover, state them in a clearer form. In his book, Greimas cites as an example for an analysis of his type 'le système sémique de la spatialité' (1966:32-6) within the domain of the adjective in modern French. This analysis, in which it is not apparent in just what way the author progresses immediately to the semes, looks like this: spatialité dimensionalité

non-dimensionalité

horizontalité verticalité I (haut/bas) I 1 perspectivité latéralité (long/court) (large/étroit)

superficie (vaste/x)

volume (épais/mince)

or, in a different form of graphie representation: SÈMES LEXEMES

spatialité

dimensionalité

i haut 1 bas

+ +

+ +

i long 1 court

+ +

+ +

i large 1 étroit

+ +

+ +

f vaste 1 épais

+ +

verticalité

+ + — —

— —

horizontalité

perspectivité

—

—

—

—

—

—

+ + + +

+ + — —

latéralité

— —

+ +

— —

In the hierarchical arrangement of the semes (e.g. 'dimensionalité' is hierarchically higher than the semes 'horizontalité' and 'verticalité'), every next-higher seme represents, with respect to the next lower seme, an 'axe sémique',89 the notion which we 89

Elsewhere, Greimas speaks rather of an 'axe sémantique', which he defines in the following manner: 'ce dénominateur commun des deux termes, ce fond sur lequel se dégage l'articulation de la signification. On voit que l'axe sémantique a pour fonction de subsumer, de totaliser les articulations qui lui sont inhérentes' (1966 : 21).

LINGUISTICS AND SEMANTICS

137

have called 'dimension'. (Cf. Geckeler 1971a: oh. VIII.) The opposition between adjectives situated on the same semic axis (e.g. long/court; haut/bas) rests on their polar relationship with respect to the category 'quantité relative', which is subdivided into the two semes 'grande quantité' and 'petite quantité'.90 It is Greimas' contribution to have attempted, in his structural semantics, to progress immediately to very abstract features. One must note as a restriction, however, that his conception does not concern the entire domain of semantics and that his semantics is situated somewhere between a semantics of the langue and a semantics of the text. 4.2.1.3 In the U.S., U. Weinreich (1962, 1963a)91 and his disciple E. H. Bendix (1966) may be regarded as the exponents of content-analysis closest to European structuralism. U. Weinreich, who died all too early, has in his publications addressed himself primarily to combinatorial semantics.92 He has ultimately discussed and criticized in detail the so-called 'semantic theory' of Katz (and Fodor) (Weinreich 1966), which is first and foremost a sentence-semantics. We know of no study by Weinreich in which he explicitly treats the paradigmatic side of semantics, but over and over again one finds statements in his works which prove that he means by 'componential structure' and by 'covert semantic components' (e.g., 'generation', 'sex', 'alive' vs. 'dead') the same linguistic phenomena as the representatives of European structural content-analysis mean by semic or semantic structure or by semes or distinctive features of content. Perhaps it is also legitimate to see, in the following quotation from E. H. Bendix, a reflection of the views of his teacher Weinreich. An approach unusual for North American linguistics is formulated in the introduction (Chapter 1) of Bendix's above-mentioned monograph: Our approach is structural in the paradigmatic sense of the word. The meanings of the forms in a given language are presented as standing in opposition to one another within the system of the language and as being distinguished by discrete semantic components acting as the distinctive features (1966:1). We will look upon a minimal definition of the meaning of a form as a statement of semantic components that are sufficient to distinguish the meaning paradigmatically from the meanings of all other forms in the language (p. 2). 4.2.1.4 Another important contribution to structural semantics in our sense comes from the English school of linguistics, from J. Lyons. After Lyons had Criticism of Greimas' method and practical analyses has been made by K. Togeby (1965 :7) and M. Wandruszka (1968 :620-3) although Greimas expressly declares that the examples chosen by him are to be taken as illustrations of his theory and make no claim to material precision of analysis. Setting out from different theoretical premises, M. Bierwisch investigates (1967) among other things the content-structure of German adjectives of space. 91 A good survey of the state of semantics in Soviet linguistics is given by the same author (1963b). " Weinreich distinguishes 'linking* and 'nesting' as types of the combining of sign-contents. 'Linking may be described as that effect of a grammatical conjunction of two signs which yields a product of their designata' (1963a : 163), e.g. in yellow flower, (to) walk fast. 'Nesting' is not really defined by Weinreich in this article; he merely shows that 'nesting' appears in 'asymmetrical two-place relations' as e.g. buy flowers, under water (pp. 164-5).

138

EUGENIO COSERIU AND HORST GECKELER

first presented his theory in a monograph form (1963), he modified and refined it for the discussion of semantics in his introduction to linguistics (1968:443-70), which has since become well-known. In our presentation of Lyons' ideas about semantics we will keep for the most part to this more recent version. Lyons recognizes 'that the vocabulary of a language will contain a number of lexical systems the semantic structure of which can be described in terms of paradigmatic and synitagmatic sense-relations' (1968 : 443).M The lexical systems are made up of 'lexical items' between which the 'senserelations' obtain. Lyons determines the content of a lexical item in the following manner: . . . , the sense of a lexical item m a y b e d e f i n e d to be, n o t o n l y dependent u p o n , but identical with, the set of relations [i.e. paradigmatic sense-relations] w h i c h h o l d b e t w e e n the item in question and other items in the s a m e lexical system ( 1 9 6 8 : 4 4 3 ) . 9 4

Recently, Lyons has distinguished the following 'sense-relations': A) Synonymy, whereby he emphasizes, in opposition to S. Ullmann, that this sense-relation 'is not essential to the semantic structure of language' (p. 452). Of decisive importance, on the other hand, are the following: B) Hypony my (formed by analogy with synonymy, antonymy) is defined as 'unilateral implication' or as 'inclusion', e.g. scarlet-red, tulip-flower, whereby in each case the first lexical unit represents the 'hyponym', the second the 'superordinate term' (or 'hyperonym'). This relationship corresponds to that between unit and archi-unit in our terminology. C) Incompatibility, defined 'on the basis of the relationship of contradictoriness between sentences' (p. 458). Thus, the color-terms represent 'a set of incompatible lexical items' (Ibid.). This sense-relation must be distinguished from mere 'difference of sense'. Thus, e.g. English crimson and soft are 'different in sense, but not incompatible', while crimson and scarlet are 'similar in sense (.. .), but incompatible' (p. 459). D) 'Oppositeness' of meaning, with three subtypes: a) Complementarity, e.g. single: married, male: female, corresponds to the logical principle of 'tertium non datur'. b) Antonymy, e.g. big : small, good : bad, forms the sense-relation which obtains between the ' "opposites" par excellence' (p. 463). These are characterized by the fact that they are 'regularly gradable' (ibid.) (cf. Sapir's concept of 'grading'). c) Converseness, e.g. buy : sell, husband: wife. This sense-relation has not been 83

His 'lexical systems' correspond to our 'lexical fields'; his 'sense-relations' in the Introduction correspond to the 'meaning-relations' in Structural semantics; Lyons does not treat the 'syntagmatic sense-relations'. 04 Cf. also: 'I consider that the theory of meaning will be more solidly based if the meaning of a given linguistic unit is defined to be the set of (paradigmatic) relations that the unit in question contracts with other units of the language (in the context or contexts in which it occurs), without any attempt being made to set up "contents" for these units. This I should mark as one of the principal theoretical points that is being made in the present work' (1963 :59).

LINGUISTICS AND SEMANTICS

139

sufficiently defined by Lyons. 'Converseness' represents a content-relation in which the one member implies the other. In actual fact, however, this relation has to do with different content relationships: thus, a common basic content may be assumed for buy and sell, whereby the differentiation of the two lexemes is given only by means of the different (classematic) viewpoint of the verbal process; husband and wife on the other hand are relational polar concepts. Lyons' semantics differs from the other forms of structural semantics presented in 4.2.1 in that it has a different object of investigation, namely the kinds of 'senserelations' within the field- or class-structures (in this regard it represents probably the most extensive treatment). One of the major achievements of this interesting approach is that it has provided, with its sense-relation of 'incompatibility', which actually falls outside of the framework of the other sense-relations, a contribution to the problem of the delimiting of lexical fields. The fact that Lyons has not attacked the entire area of a structural semantics can be regarded as a failing, as can the fact that he does not arrive, within the domain he does treat, at a proper analysis of content into distinctive features on the basis of the explicit principle of functional oppositions.95 4.2.1.5 At this point the most extensive outline of a structural semantics to date ought to be presented, namely, E. Coseriu's semantics. We will discuss it rather exhaustively in 4.2.2. 4.2.1.6 After the discussion of the most important proponents of the structural analysis of lexical content,98 it must be emphasized that the analyses of the individual semanticists (thus esp. in the case of Pottier, Greimas, Coseriu), despite certain theoretical or methodological divergences, do parallel one another, and can also be traced back to one another.97 A failing that can be found in the work of almost all these linguists is that a delimitation of the object, i.e. of semantics, either is not proposed at all, or else is silently assumed to be already given. 4.2.2.0 E. Coseriu98 has proposed the most comprehensive conception of a semantics of lexematic structures to date. His structural semantics has a twofold goal: for one, he wishes, by precisely delimiting the object of his semantics, to avoid the inadequacies of the approaches of other linguists, cf. the seven preliminary distinctions in 4.2.2.1; secondly, he succeeds in tracing all the problems of a structural semantics back to a single system, at present only outlined. This system comprises the paradigmatic structures (i.e. the primary structures, such as lexical field and lexical class as well as the secondary structures, which correspond to the 95

On further problems of Lyons' semantics, cf. the review of the Introduction by H. Geckeler (1970). •6 Attempts in this direction are found also e.g. in Hattorf 1956 (the idea of the classeme, cf. also 4.2.2), and F. Rodríguez Adrados 1967 (concept of 'polarización'). 87 This has been pointed out by E. Coseriu (1967b : 492 ff.). 98 Esp. in 1966; Romanische (insbesondere französische) Semantik, course given at the University of Tiibingen in the winter term 1965-66; 1967c; and 1968a.

140

EUGENIO COSER1U AND HORST GECKELER

three content-determined processes of word-formation) and also the syntagmatic structures (i.e. the 'lexical solidarities'), cf. 4.2.2.2 and 4.2.2.3. As concerns general problematics, it must be said that the study of vocabulary has remained very far behind in comparison to the millenary grammatical tradition. This statement is true to an even greater extent with respect to a structural approach to vocabulary. In general, it is claimed that the chief difficulty lies in the very high number of lexical items in comparison to the limited number of units to be dealt with in phonology and in grammar. In his contribution to the 8th International Congress of Linguists, L. Hjelmslev has indicated a theoretical possibility for overcoming the above-mentioned difficulty: 'Une description structurale ne pourra s'effectuer qu'à condition de pouvoir réduire les classes ouvertes à des classes fermées' (1958:653). E. Coseriu emphasizes that the vast proportions of vocabulary (and thus the great number of lexical items to be studied) present no difficulty of principle for research. Rather, he claims, it is a matter of practical difficulty. Like Hjelmslev, he also strives to reduce the complex material to be analyzed: but, with Coseriu, this reduction is achieved on the basis of a series of distinctions to be enumerated briefly below. 4.2.2.1 Necessary preliminary distinctions. Via a succession of seven distinctions, E. Coseriu arrives at the desired homogeneous object of investigation, which can only then be subjected to a structural semantic analysis. 4.2.2.1.1 Distinction between extralinguistic reality (objects) and language (words). The chief difficulty consists in the proximity of lexical function to the reality designated by the lexemes, for vocabulary is the last linguistic stratum before the transition to reality itself; that is, it represents the linguistic stratum having an immediate connection with extralinguistic reality. Therefore, it is sometimes difficult, but always essential, to distinguish between that which belongs to linguistic meaning and that which belongs to a knowledge of the objects. (Cf. Coseriu 1970b.) The matter of technical vocabulary, of terminology, belongs in this framework. Technical vocabulary is simply a nomenclature and as such not structured on the basis of language" but rather on the basis of extralinguistic reality, on the basis of the objects of the discipline in question. Terminology thus presents an objective classification constructed on logical, i.e. exclusive distinctions: A || Not-A; e.g. acid || base in chemistry. Linguistic oppositions, on the other hand, are very often inclusive: NotnA 99

, e.g.

E. Coseriu sees his basic assumption that at least a large part of vocabulary is structured confirmed by the well-functioning linguistic communication even in extensive linguistic communities (1966:178-9).

LINGUISTICS AND SEMANTICS

141

Since, in technical usage the words are really the representatives of the 'objects', signification and designation (cf. 4.2.2.1.7) coincide in this case whereas in the domain of the 'natural' language they must necessarily be separated. Therein lies also the reason for the translatability, or better, the substitutability of terminologies in a 1 : 1 ratio within the language communities having approximately the same state of knowledge in corresponding sciences: En réalité on connaît les 'signifiés' des terminologies dans la mesure où l'on connaît les sciences et les techniques auxquelles elles correspondent, et non pas dans la mesure où l'on connaît la langue: . . . (Coseriu 1966:183).

The examples often cited as particularly clear cases for the structuring of a wordfield — the evaluative scale (J. Trier) or the designations of military rank (R. M. Meyer) — are not instances of linguistic — i.e. semantic — articulations, but rather of conventionally set up, artificial classifications. Mais l'important est qu'on reconnaisse que, dans ce qu'on appelle le 'lexique' d'une langue, il y a de larges sections purement 'désignatives', et où la seule 'structuration' possible est l'énumération, et d'autres qui sont structurées,...: qu'il y a un lexique structuré, linguistique, et un lexique 'nomenclateur' et terminologique {ibid.: 184).

Thus, in a structural view of vocabulary, everything belonging to terminology and nomenclature is eliminated at the outset. By this means one simultaneously achieves, with regard to the analysis, a considerable reduction of the almost unlimited number of lexical items making up the total vocabulary. In this context are also to be placed these associations which are believed to exist between certain lexical contents, but which in reality obtain on the basis of relations between the designated objects; on this point cf. certain relations in Ch. Bally's 'champ associatif'. The question of the relationship between linguistic structurings and the structures of extralinguistic reality turns out to be especially important. The fact that a certain air-temperature is judged as 'cool' by one person and as 'warm' by another, or the fact that there are, in extralinguistic reality, no clearly defined boundaries between what is expressed e.g. by the common linguistic contents 'young' - 'old' is often interpreted as a mark of the subjective and imprecise nature of linguistic content-structurings. Herein lies a fundamental error: les valeurs linguistiques sont des valeurs conceptuelles qui se définissent par leurs oppositions et par leur fonctionnement, et non pas par des critères 'réels' et par les limites, précises ou imprécises, entre les phénomènes de la réalité (ibid.: 186-7).

a) Difficulties which present themselves in the separation of classes of real phenomena are not difficulties affecting the distinction between the corresponding concepts; quite the contrary: such difficulties show that the concepts are clearly separated. Thus e.g. the fact that in extralinguistic reality there are no clear boundaries between day and night does not mean that the concepts 'day' and 'night' are

142

EUGENIO COSERIU AND HORST GECKELER

unclear as concepts. Here, therefore, the precise delimitation of the concepts stands in opposition to an imprecise delimitation of the phenomena conditioned by the nature of the extralinguistic data. b) Lack of agreement in the usage of lexical elements with respect to a specific state of affairs does not imply a lack of agreement of these elements with each other content-wise. Illustrations such as: Ce café est chaud. - Non, il est froid; Vous êtes riche. - Non, je suis pauvre; Vous êtes jeune. - Non, je suis vieux do not prove inconsistency in the content of these adjectives, but rather variance in the estimation or evaluation of the respective state of affairs. The lack of agreement in the use of these adjectives does not concern their meaning - the fact that one can argue about it proves after all that the same contents are meant -, it rather concerns the question of whether one is to designate a specific state of affairs with one adjective or another. Thus, for example, one person finds a temperature of +5°C to be 'cold', but another does not find it so until -15°C. In fact, such cases of lack of agreement do not affect linguistic contents, but rather presuppose them. c) Language does not choose only distinctions which coincide with boundaries in extralinguistic reality. It establishes boundaries in areas which exist as a continuum (e.g. the color adjectives in the domain of the color-spectrum) and distinguishes relationships (e.g. big - little) and combinations of continua and relationships (e.g. young - old) which as such do not exist at all in extralinguistic reality. These distinctions cannot, therefore, be traced back to structures in extralinguistic reality; they must be conceived of as structurings which human interpretation imposes, via language, on reality. Therefore questions such as 'At how many degrees does a temperature begin to be "hot"?' and 'With what number of years does "(old) age" begin?' (Fr. 'A quel âge commence la "vieillesse"?') are totally irrelevant for the content-analysis of 'hot' respectively '(old) age' ('vieillesse'). d) Language may, however, dispense with distinctions in cases where a distinction is clearly present in extralinguistic reality. Thus e.g. the objects 'ladder (Fr. échelle)' and 'stairway (escalier)' are clearly distinguished in reality. In the Italian, Spanish, Portuguese, and Rumanian languages, however, there exists only one sign (scala, escalera, escada, scarâ) corresponding to both objects while in German, English and French the distinction is made linguistically. -The case is exactly reversed in the distinction made by the Rumanian verbs a zice - a spune ('to say'). As a principle, it can be stated that linguistic distinctions may, but need not coincide with objective boundaries in reality. Le langage classe la réalité, mais il le fait selon des intérêts et des attitudes humaines. . . . Dans ce sens la 'subjectivité' est constitutive du langage et elle est un fait linguistiquement objectif. Mais on ne doit pas la confondre avec l'appréciation subjective (individuelle ou traditionnelle) non 'lexématisée' (ou 'grammaticalisée') (Coseriu 1966 : 188).

At some points within the functioning of the vocabulary, however, knowledge of extralinguistic reality or the opinions with respect to it play an important role.

LINGUISTICS AND SEMANTICS

143

Thus e.g. in the interpretation of certain compounds or derivations, whose designatory function may be polyvalent from the point of view of the language-system. We refer in this context to the well-known examples German Strassenhàndler and French bananier, liseuse, etc. Here, too, belongs the systematic locating of probability of the occurrence of words in linguistic contexts. Since linguistic contexts may express real contexts, there exists a certain probability that one will come across, in a linguistic context, the designations for the things which occur together in a real context, as e.g. boeuf, labour, charrue, joug, etc. in Ch. Bally's example. It is clear that this probability is conditioned not linguistically, but by the co-presence of the objects. B. Pottier's category of the virtuème must also be understood in this sense. 4.2.2.1.2 Distinction between language (primary language) and meta-language. Le 'langage primaire' est le langage dont l'objet est la réalité non linguistique; le 'metalangage' est un langage dont l'objet est à son tour un langage: les 'choses' désignées par le métalangage sont des éléments du langage primaire (ou, en général, d'un langage) (Coseriu 1966 :190).

Each element on the level of expression (signifiant) of the primary language can be used metalinguistically and is substantivized for this purpose. Metalinguistic usage constitutes an infinite possibility of discourse (parole). It does not include any semantic structuring, for we are here concerned with an unlimited nomenclature in which every element stands in contrast to every other element. The metalinguistic sphere is therefore to be eliminated from the viewpoint of structural semantics. 4.2.2.1.3 Distinction between synchrony and diachrony. The distinction introduced by F. de Saussure (probably under the influence of G. von der Gabelentz100) between synchrony (better: language description) and diachrony (better: language history) is generally known in linguistics and has been thoroughly discussed; cf. especially E. Coseriu 1958). This distinction is above all methodologically important: the two points of view of language study must not be confused. Since 'la lengua funciona sincrónicamente y se constituye diacrónicamente' (1958:154), we must, when we wish to describe functional language-structures, conduct our investigation in synchrony. Within synchrony we must make still finer differentiations and distinguish the synchrony of structures from the synchrony of the language, for des structures fonctionnelles peuvent se maintenir plus ou moins longtemps dans le temps, ce qui signifie que leur synchronie interne dépasse leur simultanéité avec d'autres structures de la langue (Coseriu 1966:192).

Therefore, while certain structures are maintained in time, others undergo a linguistic change: thus we see that it is not the entire language which changes as a 100 E. Coseriu's study (1967d) treats the question of the dependence of a series of ideas which up to now have been considered as typical for F. de Saussure on the conception of language of G. von der Gabelentz, cf. the latter's chief work (1891).

144

EUGENIO COSERIU AND HORST GECKELER

single system, but that linguistic change is always accomplished within partial systems (or microsystems). — The synchrony of the language must always be related to a specific language-stage (état de langue). In addition, it must be noted that an 'état de langue' need not be absolutely synchronic. Diachronic facts are present up to a certain degree in synchrony and are also known to the speakers. This diachronic interference in synchrony can be noticed especially in languages with a significant cultural and literary tradition, for instance in the intentional use of archaisms. But also in linguistic communities with a minor tradition background, as in the case of dialects, one can observe a certain diachronic awareness on the part of the speakers. The speakers assign certain linguistic phenomena which deviate from their own usage to a specific generation (to a younger or an older). In order to be able to take this state of affairs into account, E. Coseriu recommends for the principles of analysis: Chaque structure doit être établie dans sa 'synchronie' propre, c'est-à-dire, dans son fonctionnement, et non pas dans l'état de langue tout entier, parce que ceci signifierait confondre ou identifier arbitrairement des structures différentes, des modalités fonctionnelles autonomes... En principe, la description de chaque structure sera donc strictement synchronique. Par contre, la description d'un état de langue ('simultanéité des structures fonctionnelles') devra, dans ce cas, constater la pluralité des 'synchronies' qui y sont impliquées,c'est-à-dire, les différences diachroniques connues et utilisées (ou utilisables) par les sujets parlants. Une description agencée et complète implique, à cet égard, qu'on décrive une 'synchronie' choisie comme fondamentale et qu'on consigne parallèlement les autres 'synchronies', c'est-à-dire, les différences diachroniques co-existant dans le même état de langue, pour tous les cas dans lesquels ces différences existent et fonctionnent (1966:194). 4.2.2.1.4 Distinction between technique of discourse ('technique du discours') and repeated discourse ('discours répété'). Under this distinction, valid within synchrony, technique of discourse means the freely available elements and procedures of a language, whereas the term repeated discourse embraces everything that, in a linguistic tradition, appears only in fixed form: fixed expressions and locutions, idioms, proverbs, 'refranes', Wellerisms, quotations (even from other languages), etc. In repeated discourse we are dealing with a kind of collage of past discourse (du 'déjà parlé'); the elements of this 'discours répété' are not commutable. Therefore, no oppositions are possible between them. But as total entities with a global unified meaning these fixed expressions are commutable. One can distinguish three different kinds, according to the level on which each is commutable: 1) Commutation with entire sentences or texts on the level of the sentences or texts: textemes or phrasemes (e.g. the refranes in Spanish). They are not a part of lexematics. 2) Commutation with syntagms, interpreted on the syntagmatic level: stereotyped syntagms (e.g. Fr. avoir maille à partir, etc.). They do not properly belong to the domain of lexematics, either.

LINGUISTICS AND SEMANTICS

145

3) Commutation with simple words, interpreted on the lexical level: lexical periphrases (e.g. Fr. en un clin d'œil, par cœur, etc.). According to E. Coseriu, these periphrases are a part of lexematics, since they can function in a word-field in opposition to simple words. At any rate, a distinction is to be made between those syntagms which may function as simple units and those which always do. In principle, it must be said that it is at times difficult to distinguish lexical periphrases from stereotyped syntagms. The present state of semantics permits, for the time being, no further reliable statements about this problem. The linguistic phenomena of the technique of discourse are synchronically analyzable and structurable. Therefore it is precisely this technique which forms the object of the further distinctions. 4.2.2.1.5 Distinction between architecture of language and structure of language or between historical language and functional language. The synchronic technique of discourse within a historical language (i.e. a language as for example German, French, etc.) is not of a homogeneous nature. It exhibits three types of internal differences which can be more or less far-reaching: Differences in geographical space: diatopic differences (e.g. dialectal differences); Differences conditioned by the socio-cultural classes of the linguistic community: diastratic differences (concerning language levels or 'niveaux'). Differences in the intention of expression: diaphasic differences (concerning language styles). Dans ce sens, une langue historique n'est jamais un seul 'système linguistique', mais un 'diasystème': un ensemble de 'systèmes linguistiques', entre lesquels il y a à chaque pas coexistence et interférence (Coseriu 1966:199). Influenced by L. Flydal's terminology, E. Coseriu calls this diasystem architecture of language and characterizes it as 'l'ensemble de rapports que comporte la multiplicité des "techniques du discours" coexistantes d'une langue historique' (200). The ideal object for investigation, from a structural view of language, is, however, the so-called functional language, which presents a syntopic (i.e. without differences in space), synstratic (i.e. without differences in the socio-cultural layers) and symphasic (i.e. without differences in the intention of expression) technique of discourse. The structure of language can be determined only within a fully homogeneous technique of discourse, that is, in a functional language. Only within the structure of language can oppositions be set up; it is not the principle of opposition which is dominant in the architecture of language, but that of diversity. This by no means implies that different functional languages are not used side by side in speaking. Quite the contrary; the speakers have various functional languages simultaneously at their command and do use them side by side. For reasons of method, however, one must introduce these distinctions, for the decisive fact is that

146

EUGENIO COSERIU AND HORST GECKELER

every opposition is set up and described in the functional language, to which it actually belongs. In the practice of investigation, then, it will be a matter of selecting and analyzing a middle level as the functional language and of always stating deviations therefrom in relation to this base. E. Coseriu designates this kind of investigation as 'description "à étages'" (1966:203). 4.2.2.1.6 Distinction between type, system, norm and discourse. For the technique of discourse in a functional language, E. Coseriu has proposed, and given reasons for, the distinction of four levels of structuring: type, system, norm, and discourse (1962, 1968c, 1969). With regard to a structural lexicology, we can abstract from type as the unity of the various procedures in a language. On the level of discourse ( = concretely realized speaking) we find, in the domain of vocabulary, the discourse-meanings or lexical variants which may be of a contextual or situational nature. Usually these are enumerated in dictionaries as 'acceptions des mots' under the various entries. Discourse-meanings may be determined in a language with respect to another language as well as with respect to the same language (= internal variants). In the framework of his structural semantics, E. Coseriu assigns the other distinction greatest significance: Mais la distinction qui nous paraît essentielle pour la lexicologie structurale est la distinction entre système et norme de la langue. La norme comprend tout ce qui, dans la 'technique du discours', n'est pas nécessairement fonctionnel (distinctif), mais qui est tout de même traditionnellement (socialement) fixé, qui est usage commun et courant de la communauté linguistique. Le système, par contre, comprend tout ce qui est objectivement fonctionnel (distinctif) (1966:205).

From this definition of the norm as the level of what is merely traditionally fixed and not necessarily functional, and the definition of the system as the functional (or distinctive) level of language, it is clear that structural semantics is only concerned with the level of the system (system understood as system of what is already realized in the language and as system of possible realizations), i.e. with the functional lexical oppositions. Beside the lexicology of the system, however, the significance of a lexicology of the norm must not be forgotten. In its domain of competence belong e.g. problems of the fixation of possibilities offered by the linguistic system, questions of frequency, and others. 4.2.2.1.7 Distinction between signification and designation. This distinction, in principle known already since the Stoics, yet over and over again disregarded in linguistics, is considered by E. Coseriu to be of fundamental importance for structural semantics and for a functional study of language altogether (1966:208-10, 1968a: 3, 1970b: 105-6); 'nur die Bedeutung ist eigentlich sprachlich und kann somit sprachlich strukturiert sein und sprachwissenschaftlich strukturiert werden, nicht aber die Bezeichnung, die an sich mit dem Aussersprachlichen zusammenhangt' (1970b : 105). The signification is determined by means of purely linguistic relations on the content-level, by the relationships of signifiés to one another (similar to Saussure's

147

LINGUISTICS AND SEMANTICS

valeur); the designation on the other hand is the relation of whole linguistic signs to 'objects' of extralinguistic reality.101 Schematic representation: signifianT^ signifié signification

designation

object

designation

object

I

X

signifié signifiant

•J

As illustration one can cite E. Husserl's already classic example: der Sieger von Jena - der Besiegte von Waterloo. Here, one and the same 'object' (Napoleon I) is designated by different, indeed, opposite meanings. In summary: En principe, seulement les rapports de signification sont structurables; les rapports de désignation ne le sont pas. La désignation concrète (d'un objet déterminé) est un fait de 'discours', tandis que la signification est un fait de 'langue' (technique du discours). Aussi les rapports de signification sont-ils constants (du point de vue synchronique), tandis que les rapports de désignation concrète sont inconstants (variables). En outre, la désignation peut être métaphorique, tandis que la signification ne l'est pas, du point de vue s'ynchronique et distinctif (...) (1966:209). Elsewhere, E. Coseriu, in the framework of his conception of philosophy of language, once again considers the relationship of signification and designation: Daher ist auch die Bezeichnung durch die Sprache etwas Sekundäres und Bedingtes, eine Möglichkeit, die erst durch die Bedeutung eröffnet wird. Die Bedeutung kann also als Möglichkeit oder Virtualität der Bezeichnung definiert werden.... Bedeutung und Bezeichnung sind also völlig verschiedene sprachliche Funktionen: die Bedeutung ist begrifflich, die Bezeichnung dagegen gegenständlich (1967a : 14-15). In conclusion, and at the same time in retrospect, the hierarchy of the seven distinctions will be outlined once again: the lexematic structures are concerned with the linguistic contents, not with extralinguistic reality; they are based on the primary language, not the meta-language; they have to do with synchrony, not diachrony; they are determined within the technique of discourse, not in repeated discourse; they are concerned in each case with a functional language and not, globally, with a historical language; they refer to the language system (langue), not to the norm of language; this is a matter of signification, not of designation (i.e. only indirectly). Schematic representation: 101

'Les rapports de signification sont des rapports entre des signifiés, tandis que les rapports de désignation sont des rapports entre des signes tout entiers et les réalités extralinguistiques désignées* (1968a : 3).

148

EUGENIO COSERIU AND HORST GECKELER

objects language

/ \

metalanguage diachrony primary language ' synchrony

/

repeated discourse

/

technique s of discourse

historical language type

designation

language V \ norm

signification

/ functional-system

\

discourse

Only after careful application of these seven preliminary distinctions do we arrive finally at the lexematic structures. E. Coseriu conceives of the organization of these structures in the following manner (1968a : 7):

Lexematic Structures

Paradigmatic Structures (oppositional) . I 1 Secondary Primary Structures Structures

1 Syntagmatic Structures (= Solidarities) (combinatorial)

- Lexical field

- Modification

- Affinity

- Lexical class

— Development

— Selection

- Composition

- Implication

4.2.2.2 The paradigmatic structures (oppositional). 4.2.2.2.1 The primary structures. 4.2.2.2.1.1 The lexical field. As the preceding schema illustrates, E. Coseriu defines the lexical field within the lexematic structures as a primary paradigmatic structure. Paradigmatic means that the lexemes which can be chosen at a specific point in the chaîne parlée make up a paradigm, i.e. a system of oppositions.

LINGUISTICS AND SEMANTICS

149

chaîne parlée

e.g. Fr.

'vieux' 'âgé' 'ancien' 'jeune' 'nouveau' 'récent'

Primary means that the lexemes are a part of the 'primary vocabulary', i.e. they do not imply other words, but correspond to immediate experience — as opposed to the secondary structures, which constitute the further development of a primary element (domain of word-formation). The lexical field is defined by E. Coseriu in the following manner: Ein Wortfeld ist in struktureller Hinsicht ein lexikalisches Paradigma, das durch die Aufteilung eines lexikalischen Inhaltskontinuums unter verschiedene in der Sprache als Wörter gegebene Einheiten entsteht, die durch einfache inhaltsunterscheidende Züge in unmittelbarer Opposition zueinander stehen (1967c : 294).102 This conception of the lexical field in no way contradicts the Trier-Weisgerber lexical field theory; rather, it means a further development in a structural direction of this theory which was based to a too large extent on intuition: Nous croyons plutôt que la théorie des champs a besoin d'être approfondie et développée et qu'une des directions dans lesquelles elle pourrait l'être est précisément la direction structurale. Aussi croyons-nous que la théorie des champs conceptuels doit être combinée avec la doctrine fonctionnelle des oppositions linguistiques (qui, du reste, y est implicite) et que l'épreuve de la commutation doit être appliquée également aux rapports lexicaux, non pas pour identifier les unités, qui y sont données, mais pour établir les traits distinctifs qui les caractérisent et, par là, les oppositions de contenu dans lesquelles les unités mêmes fonctionnent. Ce n'est que par l'existence des oppo10! Cf. also 1966:212: 'Un champ lexical est un ensemble de lexèmes unis par une valeur lexicale commune (valeur du champ), qu'ils subdivisent en des valeurs plus déterminées, en s'opposant entre eux par des différences de contenu lexical minimales ('traits distinctifs lexématiques' ou sèmes).* 10 ® As far as we know, E. Coseriu (1964 : 158-9) was the first to have enlarged and extended the concept of opposition for the lexical domain by carrying over into lexematics various oppositional types which had been developed by the Prague school for phonology, as Trubetzkoy's privative, gradual, and equipollent oppositions.

150

EUGENIO COSERIU AND HORST GECKELER

sitions distinctives10s que la 'configuration sémantique' d'un champ devient une véritable 'structure linguistique' (1964:157). This advance is reflected in the introduction of a coherent terminology and particularly in the application of a method based on linguistic procedures. The basic concepts, constitutive elements of the lexical field are these: lexeme, archi-lexeme, seme. 'Jede in der Sprache als einfaches Wort gegebene Einheit ist inhaltlich ein Lexem' (1967c : 294).104 Lexemes are therefore the units functioning within a lexical field, e.g. 'vieux', 'ancien', 'âgé', 'jeune', 'neuf', and others in the field of adjectives concerning age in modern French. 'Eine Einheit, die dem ganzen Inhalt eines Wortfeldes entspricht, ist ein Archilexem' (1967c : 294).105 The archilexeme may be realized as a lexical unit in a determinate language, but need not be. Thus, in French there is no archi-lexematic word covering the field vieux, ancien, âgé, jeune, neuf, etc. The content-differentiating features in the analysis of lexemes may be called semes.1"9 Thus e.g. B. Pottier analyzes the content of fauteuil into the following semes: 'avec dossier', 'sur pieds', 'pour 1 personne', 'pour s'asseoir', 'avec bras', 'avec matériau rigide' (1963:16). (The archi-lexeme for the field chaise, fauteuil, tabouret, canapé, (pouf) would be siège.) In addition to the basic terminological concepts explicitly indicated by E. Coseriu for the analysis of the lexical field, we would like to introduce the concept of dimension, which we have taken over from F. G. Lounsbury. By a dimension we understand a viewpoint of lexical articulation (Gliederung) which is operative in a lexical field and which so to speak furnishes the scale for the oppositions functioning between determinate lexemes of the field (comparable to A.-J. Greimas' 'axe sémantique'); within a dimension, the concept pole can be meaningfully incorporated and applied. Various dimensions can function in a lexical field: thus e.g. the dimensions 'individual age' (Eigenalter) and 'location on the axis of historical time' (zeitliche Einordnung) in the lexical field of the adjectives of age in present-day French.107 In summary are given below a series of negatively determined characteristics of the lexical field according to E. Coseriu: 1) Lexical fields do not represent taxonomies, i.e. they are not scientific classifications of extralinguistic reality. 2) Lexical fields are not 'thing-spheres' (Sachbereiche) of an objective kind. The distinctive features which are constitutive for the content need not coincide with the features necessarily present in the 'thing' itself for the identification of the 'thing' 104

'Une unité de contenu lexical exprimée dans le système linguistique (par exemple le contenu "senex" en latin) est un lexème' (1968a : 8). 105 'Un lexème dont le contenu est identique au contenu d'un champ lexical tout entier est un archilexème' (1968a : 8). 1M 'Les traits distinctifs constituant les lexèmes peuvent être appelés des sèmes (...)' (1968a :8). 107 The most comprehensive analysis to date of a lexical field on the basis of the method discussed above is that of Geckeler (1971a).

LINGUISTICS AND SEMANTICS

151

designated. Content-differentiating features may be of a sort that is not found as such in extralinguistic reality, e.g. in the case of 'beautiful' — 'ugly', 'comfortable' — 'uncomfortable.' On the other hand, the very non-existence of an objective feature may have linguistically a differentiating function, thus e.g. in B. Pottier's field 'siège', in which items are differentiated by means of the fact that the nonpresence of the back on the designated object constitutes, with respect to content, a positive, i.e. pertinent feature (as in the case of 'tabouret' and 'pouf'). 3) Lexical fields are not fields of associations. Associative fields are centrifugal; they spread out uncontrollably, whereas lexical fields have a centripetal character. A lexical field represents a lexematic system whose structuring is established on the basis of the semantic differences of its members. 4) Lexical fields have nothing to do with a word's range of application; more precisely with that of a signifiant (as in the case of the semantic theory of Katz and Fodor). There is no field which embraces only one lexeme. In a lexical field the meanings of the individual units are mutually delimited. 5) Lexical fields are not identical with conceptual fields. 108 Every lexical field is a conceptual field, but not every conceptual field is necessarily a lexical field, for a conceptual field may also be a terminological field. Every lexeme corresponds to one concept, but not every concept is necessarily rendered by one single lexeme. A concept can also be expressed by means of a combination which comprises several words: e.g. la guerre de Trente Ans is the expression for one specific concept. The following is a schematic representation of the relation between lexical field (LF) and conceptual field (CF):

4.2.2.2.1.2

The lexical class. E. Coseriu defines he lexical class as follows:

Eine Klasse ist die Gesamtheit der Lexeme, die unabhängig von der Wortfeldstruktur durch einen gemeinsamen inhaltsunterscheidenden Zug zusammenhängen. Klassen manifestieren sich durch ihre grammatische und lexikalische 'Distribution'; d.h. die Lexeme, die zu derselben Klasse gehören, verhalten sich grammatisch, bzw. lexikalisch analog: sie können grammatisch gleiche Funktionen übernehmen und erscheinen in grammatisch, bzw. lexikalisch analogen Kombinationen (1967c: 294-5). As examples of classes in a determinate language, e.g. in French, one can cite for the substantives 'living being' and 'non-living being', 'person' and 'non-person'; for the verbs, 'transitive' (possibly with additional subcategories) and 'intransitive'. 108 In the early stages of lexical field research, in the work of J. Trier, Begriffsfeld and Wortfeld are not yet precisely differentiated as terms.

152

EUGENIO COSERIU AND HORST GECKELER

E. Coseriu characterizes the classeme108 as follows: 'Der Inhaltszug, durch den eine Klasse definiert wird, ist ein Klassem' (1967c: 295).110 Classemes are a specific kind of seme which are able to function also outside of lexical fields or throughout a series of lexical fields. Whether we are dealing with a seme or a classeme in an individual case can be determined only by a comparison of the analyses of various lexical fields. The classemes are general determinations in the vocabulary, so that one is inclined to regard this classification as a kind of grammar of the vocabulary. Class and classeme must not be confused. Thus, the lexical class together with the lexical field is reckoned among the primary paradigmatic structures of the vocabulary, while classeme and seme belong to the distinctive features of lexematic content. E. Coseriu distinguishes two kinds of classes: determining and determined classes: Determining classes are classes characterized by means of classemes, as e.g. the classes 'persons', 'animals', etc. Determined classes are classes characterized by distinctive features such as 'refers to class X'; thus e.g. Rum. a se insura — a se mSrita, It. ammogliarsi — maritarsi are respectively determined by the distinctive feature 'refers to the class "men" (males)' — 'refers to the class "women"'; cf. also German Mund - Maul, essen - fressen, trinken - saufen ('refers to the class "persons" ' — 'refers to the class "animals" '). As concerns the relationship of classes and lexical fields, one can observe that lexical fields may exhibit three different kinds of relations to the classes: a) an entire lexical field functions within a determinate class: Class A LFi

Classeme

Class B LFa

Thus e.g. the lexical field of kinship relations is located in its entirety within the class 'persons'. b) A determinate classeme functions within a lexical field, i.e. it subdivides the field: Classeme

109

The term classeme has been introduced into semantics by B. Pottier. An intuition of the classeme can be discerned already in Hattori (1956). 110 Cf. also: 'un classème ... étant un trait distinctif fonctionnant dans toute une catégorie verbale (ou, du moins, dans toute une classe déjà déterminée par un autre classème), d'une façon en principe indépendante des champs lexicaux' (1968a : 11) and 'Les valeurs d'ordre très général, fonctionnant dans des séries de champs (par exemple "animé", "inanimé", "personne", "animal"), peuvent être appelées classâmes1 (1966 :212).

LINGUISTICS AND SEMANTICS

153

Thus e.g. the lexical field of adjectives of age in Modern French undergoes a division by the classeme 'for persons'. c) A lexical field is crossed by a determinate classeme, but its relationship to the classeme is that of indifference. In each case the classification results only through the context, i.e. in discourse: Classeme

L

F

Thus e.g. in the case of Fr. débarquer, it is only the context which determines whether it belongs to the class 'transitive' or 'intransitive'. Are we then, in the case of lexical fields and lexical classes dealing with two different kinds of lexical organization? B. Pottier answers this question in the affirmative, for he sees lexical field analysis under the perspective 'de 1'infiniment petit aux classes généralisantes' and classematic analysis from the point of view 'de 1'infiniment grand aux classes particularisantes' (1963:10-26). Another question, namely whether all content-distinguishing features are at the same time classemes or not, can be answered only on the basis of the results of a large number of successfully carried out analyses. If this were the case, then the lexical field would be a structure of intersections of various classes. Despite the fact that this is theoretically possible, E. Coseriu considers it to be dubious; he too considers lexical field and lexical class to be two different kinds of lexematic structures, although his opinion is not based on precisely the same considerations as is B. Pottier's. In conclusion let us quote one more statement by E. Coseriu on the distinction between lexical field and lexical class: Die Klassen dürfen nicht mit den Wortfeldern verwechselt werden. Ein Wortfeld ist ein lexikalisches Inhaltskontinuum, eine Bedingung, die dagegen für eine Klasse nicht notwendig ist Ferner manifestieren sich zwar auch Wortfelder durch ihre lexikalischen Kombinationen; bei den Klassen aber können diese Kombinationen auch grammatischer Natur sein. Schliesslich gehört jedes Appellativum zwar immer zu einem Wortfeld, nicht aber unbedingt zu einer Klasse; so z.B. ist Wecker klassematisch an sich unbestimmt, es kann sowohl ein lebendes Wesen als auch eine Sache bezeichnen (1967c: 295).

4.2.2.2.2 The secondary structures (domain of word-formation).111 According to the respective grammatical determination of the implicit primary lexical units, three types of secondary structures112 can be distinguished: 111

Cf. especially 1968a. "s 'Du point de vue lexématique, elles [les structures secondaires] se distinguent par le fait qu'elles impliquent toujours la transformation irréversible d'un terme primaire existant en tant que lexème de contenu et d'expression dans la langue. C'est-à-dire qu'un terme primaire reçoit

154

EUGENIO COSERIU AND HORST GECKELER

4.2.2.2.2.1 Modification. 'Modification' corresponds to an 'inactual' grammatical determination, i.e. to a determination which does not imply any sentencefunction of the modified primary lexical unit. The word-class (pars orationis) undergoes no alteration. In general, in modification we are dealing with a quantifying of the primary vocabulary element, that is, with diminutive and collective formations, or with prefix formation in the case of verbs, e.g. Fr. maison - maisonnette, It. cavallo - cavallino, Lat. rufus - subrufus, Fr. crier - criailler, pleurer pleurnicher, rouge — rougeâtre, It. quercia - querceto, Fr. venir — revenir, voir — prévoir. 4.2.2.2.2.2 Development. 'Development' corresponds to a grammatical determination which does comprise a sentence-function of the primary lexical unit, whereby the word-class changes in each case. Thus e.g. Fr. beau + predicative function beauté ('le fait d'être beau'); partir + predicative function -» départ ('le fait de partir'); en barque -» embarquer; de la barque -> débarquer. In certain languages there can also be whole series of developments, e.g. Fr. riche -» enrichir -» enrichissement; nation -> national -> nationaliser nationalisation-, and also combinations of modification and development, e.g. It. passeggiare h> passeggiata (dev.) -> passeggiatina (mod.); G. gehen durchgehen (mod.) -» Durchgang (dev.); Fr. voir -» revoir (mod.) -» révision (dev.). In the framework of the procedure of development (cf. Sp. blanco - blancura blanquear - blancamente) internal linguistic gaps in the vocabulary can be detected from the point of view of the language-system as effectively realized. Thus e.g. in French the substantive and the adverb which would correspond to the adjective neuf are lacking, as are the substantive and verb to the adjective récent, etc. Development involves a kind of 'déconcentration' in the meaning of the lexical item developed, with regard to the meaning of the primary lexeme. Cf. It. giornata d'invernó ('winter day') - giornata invernale ('winter day' and 'wintery day'); cf. also the two interpretations of Ch. Bally's examples chaleur tropicale, héroïne cornélienne. 4.2.2.2.2.3 Composition. 'Composition' always implies the presence of two basic elements standing in a grammatical relation to one another. Two types of composition can be distinguished: I o The generic or 'pronominal' composition, type: Fr. pomme pomm - ier 1 2 Sp. Limón limon - ero 1 2 where one of the two combined elements (2) is not identifiable with a lexeme existing in the language in question. une détermination grammaticale et, avec cette détermination grammaticale implicite, il est rendu de nouveau au lexique (dans le sens qu'il peut recevoir les déterminations grammaticales explicites des termes primaires)' (1968a : 13).

LINGUISTICS AND SEMANTICS

155

2° The specific or 'nominal' composition, type: G. kaufen + Mann ('Mann, der kauft') -» Kaufmann, where both combined elements represent lexemes. Type 1° of this new classification is usually referred to as 'derivation' and type 2° as traditional 'composition'. The two types may also appear combined; e.g. G. Kindergärtnerin (spec. comp. [Kindergarten] + gen. comp.), Schullehrer, Fr. coupepapier, etc. 4.2.2.3 Syntagmatic Structures (combinatorial) (1968a, and esp. 1967c) Les structures lexématiques syntagmatiques sont des solidarités entre des lexèmes motivées par leur valeur de langue. Dans une solidarité, il y a toujours un terme déterminant et un terme déterminé, ce dernier impliquant en tant que trait distinctif l'applicabilité à la classe ou au champ du terme déterminant, ou bien à ce même terme déterminant en tant que tel (1968a : 15). us Among these solidarities, which correspond to the 'essential meaning-relations' (wesenhafte Bedeutungsbeziehungen) or to the 'elementary semantic fields' (elementare Bedeutungsfelder) of W. Porzig (cf. 3.3.2.3.), three types can be distinguished: 'affinity', 'selection', 'implication'. 4.2.2.3.1. In affinity, the class of the determining lexemes functions as a distinctive feature in the determined lexemes; that is e.g. the relationship between the class 'women' and Lat. nubor, between the class 'persons' and Lat. senex or between the class 'animals' and Fr. gueule. 4.2.2.3.2 In selection, the archi-lexeme of the determining lexemes functions as a distinctive feature in the determined lexemes; thus in the case of German Schiff, Zug, Auto etc. with respect to fahren. That is, the archi-lexeme of Schiff, Zug, Auto, namely 'vehicle', functions as a distinctive feature in fahren ('to propel oneself by means of a vehicle'). 4.2.2.3.3 In implication, an entire determining lexeme functions as a distinctive feature in the determined lexeme; thus e.g. in the case of Fr. alezan, rouan, It. baio, sauro; Rum. roib, which are used only for horses, or Dutch fietsen 'to ride a bicycle'. 4.2.2.4 After the discussion of this to date most comprehensive and most coherent attempt towards a classification of lexematic structures, the limitations of this semantics must be indicated. These limitations are conditioned by the methodological reduction consisting in the application of the preliminary distinctions outlined in 4.2.2.1. The areas of vocabulary there eliminated have not, until now, been assigned their proper place in a total semantic system. Thus, a considerable number of problems from the lexical domain must remain open for the present. These questions must be tackled after the definitive constituting of structural semantics, for example the problems of the metaphorical sphere, of lexical 119

Cf. also: 'Eine lexikalische Solidarität kann nunmehr als inhaltliche Bestimmung eines Wortes durch eine Klasse, ein Archilexem oder ein Lexem definiert werden, und zwar in der Hinsicht, dass eine bestimmte Klasse, ein bestimmtes Archilexem oder ein bestimmtes Lexem im Inhalt des betreffenden Wortes als unterscheidender Zug funktioniert' (1967c : 296).

156

EUGENIO COSERIU AND HORST GECKELER

periphrases, and others. However, it seems to us that at the moment priority must be given to content-oriented structural semantics.

5.0 The various developments of a structural semantics discussed in Section 4 are exclusively descriptive-synchronic. By comparison, scholars have done far less in the area of diachronic structural semantics. 5.1.0. Modern structural semantics is only now developing a method for the descriptive-synchronic study of vocabulary. Thus, it cannot be expected that there will be much more than isolated attempts towards a structural approach to diachronic semantics. Two such attempts will be mentioned here. 5.1.1 J. Trier, the real founder of lexical field theory, has from the very beginning opened up the diachronic perspective of word-field research to scholars (cf. the entire title of his monograph of 1931: Der deutsche Wortschatz im Sinnbezirk des Verstandes. Die Geschichte eines sprachlichen Feldes, Band 1: Von den Anfängen bis zum Beginn des 13. Jahrhunderts). He conceives of 'Feldgliederungswandel' and of the 'Strukturgeschichte' of a field as a 'komparative Statik, d.h. als eine sprungweise von Querschnitt zu Querschnitt fortgehende, stets und immer von neuem das Gesamtfeld ins Auge fassende zeitlich rückwärts und vorwärts vergleichende Beschreibung', whereby the 'Dichtigkeit der angelegten Querschnitte' (i.e. of the field descriptions, each of them carried out in a determinate synchrony) determines the 'Grad der Annäherung an den tatsächlichen Fluss des Werdens' (1931 :13). Whether or not this 'komparative Statik' can really grasp the essence of historical dynamics is extremely dubious (cf. Coseriu 1958). Trier sees, in the field-history he proposed, a unification of the descriptive and the historical study of language (1932a: 426) and, with his idea of 'Umgliederung', took part (1934a: 184) in the famous dispute on 'das Ineinandergreifen deskriptiver und historischer Sprachwissenschaft' (cf. W. von Wartburg). Trier was already familiar with the distinction, which is of fundamental importance for a structural diachronic semantics, between two kinds of linguistic change: 1° material innovation with unchanged meaning,114 and 2° conceptual (or content-) redistribution (e.g. 1938 : 92). In the light of this relevant differentiation he gives a new interpretation to Wartburg's well-known discussion of the southwest French homophonic clash in gat (1934a: 177ff.). 5.1.2 In a very substantial contribution (1964), E. Coseriu presented the theoretical foundation for a structural diachronic semantics. The most important condition for such a diachronic lexematics is, for him, 'de se placer au point de 114

In this case, Trier speaks of the simple 'Kämmerchenwechseln*, and refers directly to the 'Antike Bedeutungsfelder' of A. Jolles, to whom this distinction was apparently already familiar, too (1934b: 447).

157

LINGUISTICS AND SEMANTICS

vue du contenu en tant que tel - . . . - et de considérer l'expression justement comme "expression", c'est-à-dire uniquement comme manifestation, et garantie de l'existence, des distinctions sémantiques,..(1964 : 148). The object of structural diachronic semantics is defined as follows: c'est le développement historique des 'champs conceptuels' considérés comme structures lexicales de contenu. Et, puisque structure veut dire avant tout opposition distinctive, la sémantique structurale diachronique aura à établir, à étudier et, si possible, à expliquer (motiver) le maintien, l'apparition, la disparition et la modification, au cours de l'histoire d'une langue, des oppositions lexicales distinctives (159-60). E. Coseriu also clarifies the position of structural diachronic semantics within the other lexicological disciplines and in particular its relationship to traditional semantics. The three following problems of fundamental importance are then treated: a) The problem of the delimitation of lexical change: Here, the (in principle) decisive distinction between non-functional and functional lexical change is applied and also terminologically firmly established: The former is termed 'remplacement (changement sémasiologique ou onomasiologique)'; the latter 'modification (changement sémantique proprement dit)' (170).116 As an illustrative example from French language history is cited the replacement of the signifiant 'ive' first of all by the signifiant 'cavale', then by 'jument' (thus, double 'remplacement'). In each case the content 'female horse' remains the same. To illustrate 'modification' the following real lexical change is cited: chef

chef

tête

b) The problem of the types of lexical change: By analogy to phonology and grammar, the following statement can be considered as valid for lexicology, too: 'un changement de structure ou fonctionnel est en principe l'apparition ou la disparition d'un trait distinctif et, par là, l'apparition ou la disparition d'une opposition (en phonologie: "phonologisation" et "déphonologisation")' (1964:173). As illustration (173, 175): U

Ui

Ui

U*

U*

U

115 'C'est une distinction radicale entre deux ordres de faits entièrement différents: un "remplacement" ne concerne que le signifiant (ou le lien signifiant-signifié); une "modification" concerne au contraire le signifié en tant que tel. Dans le cas d'un "remplacement" rien ne se produit, en principe, dans les rapports des contenus lexicaux; dans le cas d'une "modification", ce sont précisément ces rapports qui changent' (1964:170).

158

EUGENIO COSERIU AND HORST GECKELER

Lat.

Sp.

'avis'

'ave'

'pâjaro'

Lat.

Fr.

'albus'

'Candidus'

'blanc'

c) The problem of the 'regularity' proper to lexical change: E. Coseriu distinguishes two types of regularity within functional lexical change (178 ff.): 1° 'régularité': affects only one single unit of the system and its usage; 2° 'systématicité': affects an entire lexical field. In conclusion, the range of application of this semantics will be outlined: Nous pensons qu'une sémantique diachronique structurale, . . . , pourrait-indépendamment de l'intérêt qu'elle présenterait pour elle-même — se révéler fructueuse à plusieurs égards, tout particulièrement pour l'étymologie, pour l'étude des contacts interlinguistiques, pour la typologie linguistique et pour la comparaison des langues (1964: 183). (Cf. Geckeler 1971c.) Unfortunately, linguists have as yet scarcely responded to these stimuli, and the tasks outlined here have not yet been tackled seriously.

6.0 To conclude our survey of linguistic, especially functional semantics, we will present, in the following paragraphs, some suggestions and orientations for further necessary research work in the domain of structural semantics, and will also attempt a prospect of the possible unity of linguistics. 6.1.0 Below we list, in simple outline form, a selection of problems the study and eventual solution of which seems to us very important for the progress of structural semantics. The posing of these questions may also be regarded as a suggestion for the orientation of future research in the field of structural semantics. 6.1.1 The question of the delimitation of lexical fields needs to be studied more thoroughly; further criteria for delimitation must be found, in addition to those listed by J. Lyons and E. Coseriu. The questions of the hierarchy of lexical fields with respect to one another and possible neutralizations (Coseriu 1964:158; Geckeler 1971a: ch.V.3) between them must be further investigated. 6.1.2 Another problem is the distinction of possible types of lexical fields. How, for example, can L. Weisgerber's attempts to differentiate various kinds of lexical fields be structurally interpreted (cf. 3.3.2.1)? 6.1.3 It would also be interesting to pursue a still unpublished idea of E. Coseriu's, according to which the vocabulary of a language is made up of Bedeutungsfelder (i.e. our lexical fields) and Bezeichnungsfelder (designational fields, e.g. the field of the playing of musical instruments in French: jouer, sonner, toucher). 6.1.4 The problematics of lexical classes, which was discovered and first

LINGUISTICS AND SEMANTICS

159

treated in broad outlines by B. Pottier and E. Coseriu, needs further differentiation. The distinction between determining and determined classes introduced by E. Coseriu represents a first effort in this direction. For example, can determining classes be established among adjectives? 6.1.5 Another extensive complex of problems is concerned with the further determination of the grammaticalization of vocabulary and of its types among the 'secondary' lexematic structures (cf. 4.2.2), i.e. this has to do with the domain of word-formation. Which types of sentence-semantic structures underlie the secondary lexical structures? The most recent research has gotten away from assuming simple, actualized sentences, so-called 'kernel sentences' as a basis; thus e.g. H. E. Brekle in a recent publication (1970: 57ff.) no longer proceeds from 'aktuale Satzstrukturen', but from a 'Satzbegriff' (propositional concept), i.e. from the semantic kernel of an actualized declarative sentence, freed from a number of modal relations as e.g. assertion, quantification, negation, modes, tenses,116 but not, as it would seem, e.g. from 'Aktionsart'. Additional studies in this direction would be most welcome. 6.1.6 The structural method developed in phonology with its elaborate technique was transferred as a model to the study of vocabulary (cf. Coseriu 1964 : 150-5), especially to the analysis of lexical fields. At first, in an attempt to justify this transfer, analogies between the phonematic and the lexematic structures were particularly emphasized. However, it remains the task of future research to study and to emphasize precisely what is specific in the structuring of the vocabulary as opposed to the structuring in the phonic domain.117 6.1.7 An additional problem which has likewise been too little studied is the degree of material regularity of the level of expression with respect to the contentlevel in vocabulary (cf. Coseriu 1964: 166-70): 'Il s'agit de la motivation analogique, par laquelle à des contenus semblables peuvent correspondre des expressions semblables' (167). 118 This material regularity is characteristic for grammar, and by comparison very rare in vocabulary outside the domain of word-formation (cf. e.g. the fairly regular formation of designations for fruit trees in French: pomme ->• pommier, poire ->• poirier, etc.). In this connection, L. Weisgerber's content-oriented views of the phonic shape of words fit in well. 6.1.8 The problematics of structural levels (Coseriu 1952, 1968c, 1969) of language in their application to structural semantics also needs further investigation. Structural semantics as we understand it is as such concerned exclusively with the

116

'Ein Satzbegriff ist demnach neutral in bezug auf Wahrheit oder Falschheit des durch ihn bezeichneten Sachverhalts* (Brekle 1970 : 57). 117 E. Coseriu has given a provisory catalogue of analogies and differences between the phonematic and the lexematic structures in his above-mentioned course: Romanische (insbesondere französische) Semantik; a résumé of this catalogue can be found in Geckeier 1971a: Ch. IV.2. 118 ' "Semblables", naturellement, non aux contenus, mais à d'autres expressions correspondant à des contenus analogues' (Coseriu 1 9 6 4 : 1 6 7 , fn. 40).

160

EUGENIO COSERIU AND HORST GECKELER

level of the system (level of functional oppositions). But how does the transition function from the level of the system on the one hand to those of norm and discourse, and on the other hand to the type? Typologically oriented questions119 as e.g. that of analogy and differences in the lexematic structuring of different languages have been neglected by linguists till now.120 6.1.9 At the end of our selection of still unsatisfactorily studied problems we would like to mention the question, which has already been raised by scholars, as to whether or not the content-differentiating features (be they semes or classemes) possess the status of linguistic universals. If so, then is there a definite catalogue of such content-differentiating features (cf. Leibniz) for all languages, as is assumed for the distinctive features in phonology by R. Jakobson and M. Halle (1956)? The whole controversy about linguistic universals, however, suffers from a failure to distinguish with respect to the concept 'universal' (cf. Coseriu 1970b: 119, fn.l) among: a) 'universal' as something essential and conceptually necessary; b) 'universal' as something universally possible; c) what is not actually 'universal', but merely empirically general. 6.2 As a prospect for future research in linguistics we are confronted with the possibility of the union of the analytic-structural and the transformational-generative points of view. Precisely with respect to structural semantics, P.M. Postal (1966:179, fn.10) has emphasized the essential unity of the components from componential analysis, of the semantic markers of Katz and Fodor and the minimum units of content of Hjelmslev. It seems to us rather more exact to draw attention to the possibility of an identity between the semantic markers of TG and our classemes and between the distinguishers of TG and our semes. In the field of word-formation theory, too, approaches such as those of R. B. Lees (1963) and E. Coseriu are potentially unifiable, since after all the same facts are being treated but from different angles. There is no lack of indications of a union of structuralfunctional and transformational-generative points of view, thus e.g. in H. E. Brekle (1970), in M. Gross (Paris) and his team, who are occupied with the classematics of the French verb, and in part also in the work of S. M. Lamb. Thus the goal of striving for a unified linguistics does not seem altogether unattainable to us.

REFERENCES

D. 1965. Pseudo-procedures in linguistics (1963). Studies in phonetics and linguistics, 114-19. London.

ABERCROMBIE,

118

A series of such questions is cited in Coseriu 1966 :205. 12» Multilateral linguistic comparison, as carried out with good results by M. Wandruszka and his students in Tübingen, is not however concerned with the level of the language-type; cf. Wandruszka 1969.

LINGUISTICS AND SEMANTICS

161

E. 1969. Gramática estructural (Según la Escuela de Copenhague y con especial atención a la lengua española). Madrid. AMMER, K . 1958. Einführung in die Sprachwissenschaft 1. Halle. BAHNER, W. 1962. Grundzüge der Feldtheorie von Jost Trier. Eine kritische Analyse. WZUL 11.593-8. BALDINGER, K. 1950. Kollektivsuffixe und Kollektivbegriff. Ein Beitrag zur Bedeutungslehre im Französischen mit Berücksichtigung der Mundarten. Berlin. . 1952. Die Gestaltung des wissenschaftlichen Wörterbuchs. Historische Betrachtungen zum neuen Begriffssystem als Grundlage für die Lexikographie von Hallig und Wartburg. Romanistisches Jahrbuch 5.65-94. . 1957. Die Semasiologie. Versuch einer Überblicks. Berlin. . 1960. Alphabetisches oder begrifflich gegliedertes Wörterbuch? ZRPh 76.521-36. . 1964. Sémasdologie et onomasiologie. RLR 28.249-72. BALLY, CH. 1940. L'arbitraire du signe. Valeur et signification. F M 8 . 1 9 3 - 2 0 6 . . 1965. Linguistique générale et linguistique française. 4th ed. 2nd definitive ed. 1944. Berne. BARTHES, R . 1964. Eléments de sémiologie. Communications 4 . 9 1 - 1 4 4 . BASILIUS, H. 1952. Neo-Humboldtian ethnolinguistics. Word 8 . 9 5 - 1 0 5 . BAUMGÄRTNER, K. 1967. Die Struktur des Bedeutungsfeldes. Satz und Wort im heutigen Deutsch, 165-97. Düsseldorf. BENDIX, E. H. 1966. Componential analysis of general vocabulary: The semantic structure of a set of verbs in English, Hindi, and Japanese. Bloomington and The Hague. BENVENISTE, E. 1954. Problèmes sémantiques de la reconstruction. Word ALARCOS LLORACH,

10.251-64. 1954. Zur Überprüfung des Feldbegriffes. KZ 7 1 . 1 8 9 - 9 8 . BIERWISCH, M. 1967. Some semantic universals of German adjectivals. FL 3.136. BLACK, M . 1949. Language and philosophy, studies in method. Ithaca, N . Y . BLOCH, B., and G . L . TRAGER. 1942. Outline of linguistic analysis. Baltimore. BLOOMFIELD, L . 1943. Meaning. Monatshefte für Deutschen Unterricht 35.101-6. . 1960. Linguistic aspects of science. 7th impression. Chicago. ( = International Encyclopedia of Unified Science 1/4.) . 1965. Language. 8th impression. London. BRÉAL, M. 1883. Les lois intellectuelles du langage; fragment de sémantique. Annuaire de l'association pour l'encouragement des études grecques en France 17.132-42. . 1897. Essai de sémantique. Science des significations. 3rd ed. 1904. Paris. BETZ, W .

162

EUGENIO COSERIU AND HORST GECKELER

E. 1970. Generative Satzsemantik und transformationelle Syntax im System der englischen Nominalkomposition. München. BUSTOS TOVAR, E . DE. 1967. Anotaciones sobre el campo asociativo de la palabra. Problemas y principios del estructuralismo lingüístico, 149-70. Madrid. CARNAP, R. 1942. Introduction to semantics. Introduction to semantics + Formalization of logic. 2 vols (in 1). 2nd ed. 1961. Cambridge, Mass. BREKLE, H .

CARROLL, J . B .

1959.

Lg 35.58-77.

. 1961. The study of language, a survey of linguistics and related disciplines in America. 4th printing. 1st ed. 1953. Cambridge, Mass. CASARES, J. 1 9 4 2 . Diccionario ideológico de la lengua española. Desde la idea a la palabra; desde la palabra a la idea. Barcelona. CONKLIN, H. C . 1962. Lexicographical treatment of folk taxonomies. Problems in lexicography, ed. by F. W. Householder and S. Saporta, 119-41. Bloomington, Ind. COSERIU, E . 1 9 5 2 . Sistema, norma y habla. Montevideo. Reprinted 1 9 6 2 : Teoría del lenguaje y lingüística general, 1 1 - 1 1 3 . Madrid. . 1954. Forma y sustancia en los sonidos del lenguaje. Montevideo. Reprinted 1962, Teoría del lenguaje y lingüística general, 115-234. Madrid. . 1955-56. Determinación y entorno. Dos problemas de una lingüística del hablar. Romanistisches Jahrbuch 7.29-54. Reprinted 1962, Teoría del lenguaje y lingüística general, 282-323. Madrid. . 1958. Sincronía, diacronía e historia. El problema del cambio lingüístico. Montevideo. Reimpresión fotomecánica, Tübingen 1969. . 1962. Logicismo y antilogicismo en la gramática. Teoría del lenguaje y lingüística general, 235-60. Madrid. . 1964. Pour une sémantique diachronique structurale. TraLiLi 2/1.13986.

. 1966. Structure lexicale et enseignement du vocabulaire. Actes du premier colloque international de linguistique appliquée, 175-217. Nancy. . 1967a. Das Phänomen der Sprache und das Daseinsverständnis des heutigen Menschen. Pädagogische Provinz 1-2.11-28. . 1967b. Zur Vorgeschichte der strukturellen Semantik: Heyses Analyse des Wortfeldes 'Schall'. To honor Roman Jakobson: Essays on the occasion of his seventieth birthday, 489-98. The Hague and Paris. . 1967c. Lexikalische Solidaritäten. Poética 1.293-303. . 1967d. Georg von der Gabelentz et la linguistique synchronique. Word 23.74-100. . 1968a. Les structures lexématiques. ZFSL Beiheft (Neue Folge) 1.3-16. . 1968b. Coordinación latina y coordinación románica. Actas del III congreso español de estudios clásicos III. 35-57. Madrid. . 1968c. Sincronía, diacronía y tipología. Actas del XI Congreso Internacional de Lingüística y Filología Románicas, 1.269-83. Madrid.

LINGUISTICS AND SEMANTICS

163

. 1969. Sistema, norma e 'parola'. Studi linguistici in onore di Vittore Pisani 1.235-53. Brescia. . 1970a. Semantik, innere Sprachform und Tiefenstruktur. Sprache — Strukturen und Funktionen, XII Aufsätze zur Allgemeinen und Romanischen Sprachwissenschaft, 213-24. Tübingen. [Also in FLing 4.30—40.] . 1970b. Bedeutung und Bezeichnung im Lichte der strukturellen Semantik. Sprachwissenschaft und Übersetzen, ed. by P. Hartmann and H. Vernay, 104-21. München. DARMESTETER, A. 1887. La vie des mots étudiée dans leurs significations. (English translation, London 1886.) Paris. DORNSEIFF, F. 1938. Das'Problem des Bedeutungswandels'. ZDPh 63.119-38. DUBOIS, J. 1 9 6 0 . Les notions d'unité sémantique complexe et de neutralisation dans le lexique. Cahiers de Lexicologie 2 . 6 2 - 6 . . 1964. Distribution, ensemble et marque dans le lexique. Cahiers de Lexicologie 4.5-16. DUCHÂCEK, O . 1 9 5 9 . Champ conceptuel de la beauté en français moderne. Vox Romanica 1 8 . 2 9 7 - 3 2 3 . . 1960a. Les champs linguistiques. Philologica Pragensia 3.22-35. . 1960b. Le champ conceptuel de la beauté en français moderne. Prague. . 1962. L'homonymie et la polysémie. Vox Romanica 21.49-56. D U D E N . 1959. Grammatik der deutschen Gegenwartssprache, Mannheim. ERDMANN, K. O. 1910. Die Bedeutung des Wortes. Aufsätze aus dem Grenzgebiet der Sprachpsychologie und Logik. 2nd. ed. 1st ed. 1900. Leipzig. FIRTH, J . R. 1930. Speech. London. . 1937. The tongues of men. London. . 1957a. The technique of semantics. (Reprint of 1935 edition.) Papers in Linguistics 1934-1951, 7-33. London. . 1957b. Papers in linguistics 1934-1951. London. . 1962. A synopsis of linguistic theory, 1930-1955. Studies in linguistic analysis. Special volume of the Philological Society [1st printing 1957], 1-32. Oxford. . 1964. The tongues of men and speech. [Reprint of 1930 and 1937.] London. FISCHER-J0RGENSEN, E. 1952. Glossematics. Lecture to the Washington Linguistic Club, 26 March. FOWLER, R. 1965. A note on some uses of the term 'meaning' in descriptive linguistics. Word 21.411-20. FREI, H . 1 9 5 4 . Critères de délimitation. Word 1 0 . 1 3 6 - 4 5 . . 1961. Désaccords. CFS 18.35-51. FRIES, C H . C . 1 9 5 4 . Meaning and linguistic analysis. Lg 3 0 . 5 7 - 6 8 . GABELENTZ, G . VON DER. 1891. Die Sprachwissenschaft, ihre Aufgaben, Metho-

164

EUGENIO COSERIU AND HORST GECKELER

den und bisherigen Ergebnisse. 2nd ed. 1901. [3rd ed. 1969 Tübingen = enlarged reprint of the second ed.] Leipzig. GAMILLSCHEG, E. 1951. Französische Bedeutungslehre. Tübingen. GARDINER, A. 1951. The theory of speech and language. 2nd ed. 1st ed. 1932. Oxford. GAUGER, H . - M . 1 9 6 9 . Die Semantik in der Sprachtheorie der transformationeilen Grammatik. Linguistische Berichte 1.1-18. GECKELER, H . 1970. Review of Lyons 1968. Poetica 3.594-600. . 197,1a. Zur Wortfelddiskussion, Untersuchungen zur Gliederung des Wortfeldes 'alt-jung-neu' im heutigen Französisch. München. . 1971b. Strukturelle Semantik und Wortfeldtheorie. München. . 1971c. Lexikalische Strukturen im Vergleich: Kontrastive Skizze zur Strukturierung des Wortfeldes 'alt-jung-neu' im heutigen Italienisch, Spanisch und Französisch. Festschrift Wandruszka 123-37. Tübingen. GEORGE, F. H. 1964. Semantics. The Teach Yourself Books. London. GIPPER, H. 1959. Sessel oder Stuhl? Ein Beitrag zur Bestimmung von Wortinhalten im Bereich der Sachkultur. Sprache — Schlüssel zur Welt: Festschrift für Leo Weisgerber, 271-92. Düsseldorf. GIPPER, H., and H . S C H W A R Z . 1962ff. Bibliographisches Handbuch zur Sprachinhaltsforschung. Teil 1: Schrifttum zur Sprachinhaltsforschung in alphabetischer Folge nach Verfassern mit Besprechungen und Inhaltshinweisen. Köln-Opladen. GLEASON, H . A., J R . 1 9 6 1 . An introduction to descriptive linguistics. Rev. ed. New York. GODEL, R. 1 9 4 8 . Homonymie et identité. C F S 7 . 5 - 1 5 . GOODENOUGH, W. H. 1956. Componential analysis and the study of meaning. Lg 32.195-216. GREBE, P. 1 9 6 7 . Der semantisch-syntaktische Hof unserer Wörter. Satz und Wort im heutigen Deutsch, 109-14. Düsseldorf. GREENBERG, J. H. 1963. Concerning inferences from linguistic to nonlinguistic data. Language in culture: Conference on the interrelations of language and other aspects of culture, ed. by H. Hoijer, 3-19. 5th impression. 1st ed. 1954. Chicago and London. GREIMAS, A.-J. 1966. Sémantique structurale. Recherche de méthode. Paris. GUIRAUD, P . 1956. Les champs morpho-sémantiques (Critères externes et critères internes en étymologie). BSL 52.265-88. . 1962. La sémantique. Que sais-je? 655. 3rd ed. Paris. HALL, R . A . , JR. 1964. Introductory linguistics. Philadelphia and New York. HALLIG, R . , and W . VON WARTBURG. 1963. Begriffssystem als Grundlage für die Lexikographie. Versuch eines Ordnungsschemas. 2nd ed. 1st ed. 1952. Berlin. HAMP, E. P . 1 9 6 6 . A glossary of American technical linguistic usage 1 9 2 5 1950. 3rd ed. Utrecht and Antwerp.

LINGUISTICS AND SEMANTICS

165

Z. S. 1954. Distributional structure. Word 10.146-62. . 1963. Structural linguistics. 6th impression. Chicago and London. [Originally entitled, 1951, Methods in structural linguistics.] HATTORI, SH. 1 9 5 6 . The analysis of meaning. For Roman Jakobson. Essays on the occasion of his sixtieth birthday, 2 0 7 - 1 2 . The Hague. HAUGEN, E. 1951. Directions in modern linguistics. Lg 27.211-22. HAYAKAWA, S. I. 1949. Language in thought and action. New York. . 1954. Semantics, general semantics, and related disciplines. Language, meaning, and maturity: Selections from ETC: A Review of General Semantics 1943-1953, ed. by S. I. Hayakawa, 19-37. New York, Evanston, and London. HEGER, K. 1963. Homographie, Homonymie und Polysemie. ZRPh 79.471-91. . 1964. Die methodischen Voraussetzungen von Onomasiologie und begrifflicher Gliederung. ZRPh 80.486-516. French tr. 1965, TraLiLi 3/1.7-32. . 1969. Die Semantik und die Dichotomie von Langue und Parole. Neue Beiträge zur theoretischen Standortbestimmung von Semasiologie und Onomasiologie. ZRPh 85.144-215. HEYSE, K. W. L. 1856. System der Sprachwissenschaft, ed. by H . Steinthal. Berlin. HILL, A . A . 1 9 5 8 . Introduction to linguistic structures: From sound to sentence in English. New York. HJELMSLEV, L . 1943. Omkring sprogteoriens grundlaeggelse. Copenhagen. Tr. by F. J. Whitfield, 1963, as Prolegomena to a theory of language, Madison, Wisconsin. . 1958. Dans quelle mesure les significations des mots peuvent-elles être considérées comme formant une structure?, Proceedings of the Eighth International Congress of Linguists, 636-54. Oslo. (= Pour une sémantique structurale. Essais linguistiques, by L. Hjelmslev, 96-112; TCLC 12, Copenhague, 1959). . [1936] 1959. Essai d'une théorie des morphèmes. Essais linguistiques, 152-64. [= TCLC 12], Copenhague. . 1966. Le langage, une introduction. Tr. from the Danish by M. Olsen. Paris. HOCKETT, CH. F. 1 9 5 8 . A course in modern linguistics. New York. HOLT, J. 1 9 4 6 . Rationel Semantik (Pleremik) (= Acta Jutlandica XVIII/3). K0benhavn. . 1961. Order of content entities. Language and society. Essays presented to Arthur M. Jensen on his seventieth birthday, 65-72. Copenhagen. . 1964. Beiträge zur sprachlichen Inhaltsanalyse. Innsbrucker Beiträge zur Kulturwissenschaft 21.1-16. Innsbruck. IKEGAMI, Y. 1962. Structural semantics: Its assumptions and problems. The Tsuda Review 7.1-15. . 1967. Structural semantics: A survey and problems. Linguistics 33.49-67. IPSEN, G. 1924. Der Alte Orient und die Indogermanen. Stand und Aufgaben HARRIS,

166

EUGENIO COSERIU AND HORST GECKELER

der Sprachwissenschaft: Festschrift für Wilhelm Streitberg, 200-37. Heidelberg. . 1932. Der neue Sprachbegriff. Zeitschrift für Deutschkunde 46.1-18. JAKOBSON, R., and MORRIS HALLE. JOLLES, A .

1934.

1956.

Fundamentals of language. The Hague.

Antike Bedeutungsfelder. P B B 58.97-109.

Joos, M. 1958. Semology: A lingustic theory of meaning. SIL 13.53-70. . 1962. Structure in meaning. MSLL 13.41-8. KATZ, J. J. 1966. The philosophy of language. New York and London. . 1967. Recent issues in semantic theory. FL 3.124-94. KATZ, J. J., and J. A . FODOR. 1963. The structure of a semantic theory. Lg 39.170-210. Reprinted 1964 in their The structure of language: Readings in the philosophy of language, 479-518. Englewood Cliffs, N.J. KOCH, W. A . 1963. Zur Homonymie und Synonymie. Eine kritische Zusammenfassung. A L H 13.65-91. KORZYBSKI, A. 1933. Science and sanity: An introduction to non-Aristotelian systems and general semantics. Lancaster, Pa. KRONASSER, H. 1952. Handbuch der Semasiologie. Kurze Einführung in die Geschichte, Problematik und Terminologie der Bedeutungslehre. Heidelberg. KURYEOWICZ, J. 1936. Dérivation lexicale et dérivation syntaxique (contribution à la théorie des parties du discours). BSL 37.79-92. Reprinted in Readings in linguistics II, ed. by E. P. Hamp, F. W. Householder, and R. Austerlitz, 42-50. Chicago and London 1966. . 1949a. Actes du sixième Congrès International des Linguistes, 175-7. Paris. . 1949b. La notion de l'isomorphisme. Recherches structurales 1949 ( = TCLC 5), 48-60. Copenhague. LAMB, S. M. 1963. The sememic approach to structural semantics. (Mimeo.) Berkeley. . 1966. Outline of stratificational grammar. Washington, D.C. LÁZARO CARRETER, F. 1962. Diccionario de términos filológicos. 2nd. ed. Madrid. LEES, R. B. 1963. The grammar of English nominalizations. The Hague. LEISI, E. 1953. Der Wortinhalt, seine Struktur im Deutschen und Englischen. 3rd ed. 1967. Heidelberg. LEROT, J. 1967. Zur formalen Bedeutungslehre I-II. (Dissertation) Löwen. LOUNSBURY, F. G. 1956. A semantic analysis of the Pawnee kinship usage. Lg 32.158-94. . 1964. The structural analysis of kinship semantics. PICL 9.1073-90. LYONS, J. 1963. Structural semantics. An analysis of part of the vocabulary of Plato. Publications of the Philological Society 20. Oxford. . 1966. Firth's theory of 'meaning'. In memory of J. R. Firth, 288-302. London.

LINGUISTICS AND SEMANTICS

167

. 1968. Introduction to theoretical linguistics. Cambridge. MALINOWSKI, B . 1960. The problem of meaning in primitive languages. The meaning of meaning, by Ch. K. Ogden and I. A. Richards, supplement I, 296-336. 10th ed. [1st ed. 1923.] London. MARCHAND, H. 1966. On attributive and predicative derived adjectives and some problems related to the distinction. Anglia 84.131-49. . 1969. The categories and types of present-day English word-formation. A synchronic-diachronic approach. 2nd ed. München. [1st ed. 1960, Wiesbaden.] MAROUZEAU, J. 1961. Lexique de la terminologie linguistique — français, allemand, anglais, italien. 3rd ed. Paris. MATORÉ, G. 1 9 5 1 . Le vocabulaire et la société sous Louis-Philippe. GenèveLille. . 1953. La méthode en lexicologie. Domaine français. Paris. MEYER, R. M . 1910a. Bedeutungssysteme. KZ 43.352-68. . 1910b. Die militärischen Titel. Zeitschrift für Deutsche Wortforschung 12.145-56. MORRIS, C H . W. 1938. Foundations of the theory of signs. International Encyclopedia of Unified Science 1/2. Chicago. . 1955. Signs, language, and behavior. 1st ed. 1946. New York. NAERT, P. 1961. Limites de la méthode distributionnelle. Studia Linguistica 15.52-4. NIDA, E. A. 1951. A system for the description of semantic elements. Word 7.1-14. . 1962. Morphology: The descriptive analysis of words. 2nd ed. 1949. Ann Arbor. . 1964. Toward a science of translating with special reference to principles and procedures involved in Bible translating. Leiden. ÖHMAN, S. 1 9 5 1 . Wortinhalt und Weltbild. Vergleichende und methodologische Studien zu Bedeutungslehre und Wortfeldtheorie. Stockholm. . 1953. Theories of the 'linguistic field'. Word 9.123-34. OKSAAR, E. 1958. Semantische Studien im Sinnbereich der Schnelligkeit. Plötzlich, schnell und ihre Synonymik im Deutsch der Gegenwart und des Früh-, Hoch- und Spätmittelalters. Stockholm and Uppsala. OSGOOD, C. E. 1959. Semantic space revisited. Word 15.192-200. OSGOOD, C. E., G. J . SUCI, and P . H . TANNENBAUM. 1957. The measurement of meaning. Urbana. OTTO, E. 1965. Stand und Aufgabe der allgemeinen Sprachwissenschaft. 2nd ed. 1st ed. 1954. Berlin. PEI, M. 1966. Glossary of linguistic terminology. Garden City, N.Y. PIKE, K . L. 1 9 5 4 - 6 0 . Language in relation to a unified theory of the structure of

168

EUGENIO COSERIU AND HORST GECKELER

human behavior. Parts I-III. Preliminary ed. Glendale, Calif. 2nd ed. 1967. The Hague. PORZIG, W. 1930-31. Die Leistung der Abstrakta in der Sprache. Blätter für Deutsche Philosophie 4.66-77. . 1934. Wesenhafte Bedeutungsbeziehungen. PBB 58.70-97. . 1967. Das Wunder der Sprache. Probleme, Methoden und Ergebnisse der modernen Sprachwissenschaft. 4th ed. [Ist ed. 1950.] Bern-München. . 1959. Die Einheit des Wortes. Ein Beitrag zur Diskussion. Sprache — Schlüssel zur Welt. Festschrift für Leo Weisgerber, 158-67. Düsseldorf. POSTAL, P. M. 1966. Review of Elements of general linguistics, by André Martinet (tr. by Elisabeth Palmer), Chicago 1964. FL 2.151-86. POTTEER, B. 1963. Recherches sur l'analyse sémantique en linguistique et en traduction mécanique. Nancy. . 1964. Vers une sémantique moderne. TraLiLi 2.107-37. . 1965. La définition sémantique dans les dictionnaires. TraLiLi 3.33-9. . 1967a. Présentation de la linguistique. Fondements d'une théorie. Paris. . 1967b. Rehabilitación de la semántica. Problemas y principios del estructuralismo lingüístico, 187-92. Madrid. PRIETO, L. J . 1956. Contributions à l'étude fonctionnelle du contenu. TIL 1.23-41. . 1957. Figuras de la expresión y figuras del contenido. Estructuralismo e historia. Miscelánea Homenaje a André Martinet 1.243-9. La Laguna. . 1964. Principes de noologie. Fondements de la théorie fonctionnelle du signifié. Janua Linguarum, series minor, 35. The Hague. QUADRI, B. 1952. Aufgaben und Methoden der onomasiologischen Forschung. Eine entwicklungsgeschichtliche Darstellung. RH 37. Bern. READ, A. W. 1 9 4 8 . An account of the word 'semantics'. Word 4 . 7 8 - 9 7 . REGNÉLL, H . 1958. Semantik. Stockholm. REISIG, K. 1839. Vorlesungen über lateinische Sprachwissenschaft. Ed. and notes by Friedrich Haase, Leipzig. REY, A. 1965. Les dictionnaires: Forme et contenu. Cahiers de Lexicologie 7.65-102.

. 1969. Remarques sémantiques. Langue Française 4.5-29 (= La sémantique). RICKEN, U. 1961a. Onomasiologie oder Feldmethode? Bemerkungen zur Methode und Terminologie einiger neuerer wortkundlicher Arbeiten. WZUL 10.833-40. . 1961b. Bemerkungen zur Onomasiologie. WZUL 10.409-19. RODRÍGUEZ ADRADOS, F . 1 9 6 7 . Estructura del vocabulario y estructura de la lengua. Problemas y principios del estructuralismo lingüístico, 1 9 3 - 2 2 9 . Madrid. Reprinted 1 9 6 9 , in his Estudios de lingüística general, 2 5 - 6 0 . Barcelona.

LINGUISTICS AND SEMANTICS ROSIELLO,

L.

1962.

La semantica: Note terminologiche ed epistemologiche.

169 AGI

47.32-53.

Cours de linguistique générale. 1st ed. 1 9 1 6 . Paris. Die Wortfeldtheorie. ZDA 7 9 . 2 4 9 - 7 2 . SCHLAUCH, M. 1943. The gift of tongues. London. . 1946. Early behaviorist psychology and contemporary linguistics. Word 2.25-36. SCHWARZ, H. 1959. Leitmerkmale sprachlicher Felder. Ein Beitrag zur Verfahrensweise der Gliederungsforschung. Sprache — Schlüssel zur Welt. Festschrift für Leo Weisgerber, 245-55. Düsseldorf. SCUR, G . S. 1 9 6 9 . On the associative principle and field in linguistics. Studi linguistici in onore di Vittore Pisani II. 9 3 7 - 6 2 . Brescia. SEIFFERT, L . 1968a. Wortfeldtheorie und Strukturalismus. Studien zum Sprachgebrauch Freidanks. Stuttgart. . 1968b. Neo-Humboldtian semantics in perspective: 'Sprache und Gemeinschaft'. JL 4.93-108. SIERTSEMA, B. 1965. A study of glossematics. Critical survey of its fundamental concepts. 2nd ed. The Hague. SLAMA-CAZACU, T . 1 9 6 1 . Langage et contexte. Le problème du langage dans la conception de l'expression et de l'interprétation par des organisations contextuelles. Janua linguarum, series maior, 6. The Hague. SPENCE, N. C. W. 1961. Linguistic fields, conceptual systems and the Weltbild. T P H S 1961.87-106. SPERBER, H . 1965. Einführung in die Bedeutungslehre. 3rd ed. 1st ed. 1923. Bonn. STERN, G. 1931. Meaning and change of meaning with special reference to the English language. (Reprint.) Bloomington. TODOROV, Tz. 1966. Recherches sémantiques. Langages 1.5-43. TOGEBY, K. 1965. Grammaire, lexicologie et sémantique. Cahiers de Lexicologie 6.3-7. TOLLENAERE, F . DE. 1 9 6 0 . Lexicographie alphabétique ou idéologique. Cahiers de Lexicologie 2 . 1 9 - 2 9 . TRAGER, G . L . 1950. The field of linguistics. S . I . L . Occasional Papers 1. 2nd ed. 1st ed. 1949. Norman, Oklahoma. . 1963. Linguistics is linguistics. S.I.L. Occasional Papers 10. Buffalo, N.Y. TRIER, J. 1 9 3 1 . Der deutsche Wortschatz im Sinnbezirk des Verstandes. Die Geschichte eines sprachlichen Feldes I: Von den Anfängen bis zum Beginn des 13. Jahrhunderts. Heidelberg. . 1932a. Sprachliche Felder. Zeitschrift für Deutsche Bildung 8.417-27. . 1932b. Die Idee der Klugheit in ihrer sprachlichen Entfaltung. Zeitschrift für Deutschkunde 46.625-35. SAUSSURE, F . DE.

SCHEIDWEILER, F .

1964.

1942.

170

EUGENIO COSERIU AND HORST GECKELER

. 1934a. Deutsche Bedeutungsforschung. Germanische Philologie, Ergebnisse und Aufgaben, Festschrift für Otto Behaghel, 173-200. Heidelberg. . 1934b. Das sprachliche Feld. Eine Auseinandersetzung. Neue Jahrbücher für Wissenschaft und Jugendbildung 10.428-49. . 1938. Über die Erforschung des menschenkundlichen Wortschatzes. Actes du quatrième Congrès International de Linguistes, 92-8. Copenhague. TRUBETZKOY, N. S. 1967. Grundzüge der Phonologie. 4th ed. [1st ed. 1939.] Göttingen. ULLMANN, S. 1946. Language and meaning. Word 2.113-26. . 1951. Le mot 'sémantique'. FM 19.201-2. . 1957. Historical semantics and the structure of the vocabulary. Estructuralismo e Historia. Miscelánea Homenaje a André Martinet 1.289-303. La Laguna. . 1962. Semantics, an introduction to the science of meaning. Oxford. . 1963. The principles of semantics. 3rd impression. Oxford and Glasgow. WALLACE, A . F . C., and J . ATKINS. 1960. The meaning of kinship terms. A M A 62.58-80. WANDRUSZKA, M. 1968. Der Ertrag des Strukturalismus. Verba et Vocabula: Ernst Gamillscheg zum 80. Geburtstag, ed. by H. Stimm and J. Wilhelm, 619-38. München. . 1969. Sprachen — vergleichbar und unvergleichlich. München. WARTBURG, W. VON. 1937. Betrachtungen über die Gliederung des Wortschatzes und die Gestaltung des Wörterbuchs. ZRPh 57.296-312. . 1962. Einführung in Problematik und Methodik der Sprachwissenschaft. 2nd ed. corrected and expanded in collaboration with S. Ullmann. 1st ed. 1943. Tübingen. WEINREICH, U. 1958. Travels through semantic space. Word 14.346-66. . 1959. Rejoinder to Osgood 1959. Word 15.200-1. . 1962. Lexicographic definition in descriptive semantics. Problems in lexicography, ed. by F. W. Householder and S. Saporta, 25-43. Bloomington, Ind. . 1963a. On the semantic structure of language. Universals of language, ed. by J. H. Greenberg, 114-71. Cambridge, Mass. . 1963b. Lexicology. CTL 1.60-93. . 1966. Explorations in semantic theory. CTL 3.395-477. . 1967. On arguing with Mr. Katz: A brief rejoinder. FL 3.284-7. WEISGERBER, L. 1942. Die volkhaften Kräfte der Muttersprache. 2nd ed. [1st ed. 1939, 3rd ed. 1943.] Frankfurt a.M. . 1954. Die Sprachfelder in der geistigen Erschliessung der Welt. Festschrift für Jost Trier zu seinem 60. Geburtstag am 15. Dezember 1954, 34-49. Meisenheim/Glan.

LINGUISTICS AND SEMANTICS

171

. 1956-57. Die Erforschung der Sprach"zugriffe", 1: Grundlinien einer inhaltbezogenen Grammatik. Wirkendes Wort 7.65-73. . 1957. Sprachwissenschaftliche Methodenlehre. Deutsche Philologie im Aufriss, ed. by W. Stammler, I. columns 1-38. 2nd revised ed. Berlin. . 1962a. Grundzüge der inhaltbezogenen Grammatik. 3rd revised ed. Düsseldorf. . 1962b. Die sprachliche Gestaltung der Welt. 3rd revised ed. Düsseldorf. . 1963. Die vier Stufen in der Erforschung der Sprachen. Düsseldorf. . 1964. Das Menschheitsgesetz der Sprache als Grundlage der Sprachwissenschaft. 2nd revised ed. Heidelberg. WELLS, R. 1 9 5 7 . A mathematical approach to meaning. C F S 1 5 . 1 1 7 - 3 6 . ZAUNER, A. 1902. Die romanischen Namen der Körperteile. Eine onomasiologische Studie. Erlangen. Reprinted 1903, RF 14.339-530. ZVEGINCEV, V. A. 1957. Semasiologija. Moscow.

SPEECH ACTS

L. JONATHAN COHEN

I. THE AUSTINIAN HERITAGE

1. The Main Issues Recent philosophical interest in speech-acts stems mainly from attempts to come to grips with certain problems that have arisen within what may broadly be termed the analytic movement in philosophy. Characteristic of this movement is the policy of treating all philosophical questions, as far as possible, as questions about the meanings of the linguistic expressions involved. But in various ways such questions have led to questions about speech-acts. i. Reluctance to construe moral judgments as being mere tautologies, or mere descriptions of empirical, or supernatural, facts has fostered the supposition that, essentially, they are not assertions at all. This supposition has, in turn, provoked an active interest in those non-assertive forms of locution (exclamations, commands, commendations,1 etc.) which might be the vehicles of moral self-expression. Nor has moral philosophy been the only source of interest in non-assertive locutions. Epistemology has also contributed to this development. When J. L. Austin compared (1946:170) the use of the expression I know with the use of I promise, he was suggesting that recognition of a quasi-performative element in the use of this expression would help to resolve some of the many philosophical perplexities about knowledge. So a stream of questions arise: how many of these non-assertive forms of locution are there? What are they? How are they related to one another? What makes them comparable with one another? What structure is there in this motley domain? ii. Wittgenstein noticed that confusion was often created in philosophy by a tendency to identify meaning with the relation between a name and what it names, or to treat the meanings of expressions as counterpart entities, e.g. as mental imagery or possible states of affairs. To avoid such confusions he recommended philosophers to ask not for the meaning of a word but for its use (e.g. 1953 : § 43). 1 For a comparison of moral utterances with exclamations see Ayer 1946:107; for a comparison with commands see Carnap 1935 : 24; and for a comparison with commendations see Hare 1952 : 127ff.

174

L. JONATHAN COHEN

In this way better justice might be done to the immense variety of ways in which language functions. Unfortunately, however, the 'use' terminology turns out to be capable of creating its own, not unimportant confusions. For example, Wittgenstein remarked (1953 : § 246) that under normal conditions it would be odd to say, when I have a pain, I KNOW / am in pain. But perhaps this oddity tells us nothing of interest about the meaning of know, since it may be due just to the oddity of saying things that are too obvious to be worth saying. The 'use' terminology tends to confuse problems about the conditions under which a concept is applicable with problems about the conditions under which it is appropriate to make utterances involving the concept.2 But if the former problems are problems about meaning, what are the latter problems about? W e seem to need a theory which is not about linguistic abstractions like words or sentences, but about concrete events of speechutterance. iii. A question then arises about the nature of such a theory. Should it be conceived of as a branch of linguistics? Alongside phonology, syntax, and semantics, should we envisage a branch of linguistics that is concerned with speech-acts? Is the force or role of an utterance something that is somehow co-ordinate with, as well as distinguishable from, its meaning? In what follows I shall begin by discussing the views of three philosophers who have made interesting and important contributions to the resolution of these problems. This will involve both a highly selective approach to the literature and also a somewhat summary approach to the views of those philosophers themselves. However, my purpose in beginning with such a critical discussion is not to write an exhaustive survey of what others have already written on the problem, but to suggest the need for making a new start. Up to now discussion of the problem has been largely dominated by Austin's scheme of ideas and has been mainly preoccupied with proposing corrections, modifications, or developments of it. What I aim to show in §§ 2-4 is that this method of tackling the problems will not pay off. In §§ 5-7 I shall propose another method which at least circumvents the difficulties discussed in § § 2-4.

2. J. L. Austin's

Account

Austin's outstandingly original contribution to the subject is unfortunately available only in the form of a posthumously published version of his William James Lectures, carefully put together by J. O. Urmson (Austin 1962). One has therefore to attribute the flaws and inconsistencies in this book to Austin himself, even though he might well have disowned them had he survived. According to Austin every act of speaking, except perhaps (1962:132) a mere exclamation like Damn! or Ouch!, is both a locutionary and an illocutionary act. 1

For some other examples see Searle 1969 :141ff.

SPEECH ACTS

175

Qua locutionary it is, as it were, three acts in one (92 ff.): the (phonetic) act of uttering certain noises; the (phatic) act of uttering certain vocables or words belonging to and as belonging to a certain vocabulary, in a certain grammatical construction, with a certain intonation, etc.; and the (rhetic) act of using those variables with a certain more or less definite sense and reference. Austin takes 'sense and reference' here on the strength of current views, as he puts it (148), and all he says about them directly is that together they are equivalent to'meaning'. But he does remark also that phatic acts may be reported by direct quotation, as in He said 'The cat is on the mat', while rhetic acts are reported in indirect discourse. The product of a phatic act is a unit of language, and its typical fault is to be meaningless, while the product of a rhetic act is a unit of speech, and its typical fault is to be vague or void or obscure, etc. Austin claims that to perform such a three-in-one locutionary act is in general also to perform an illocutionary act. In order to determine what illocutionary act is so performed, says Austin, we must determine in what way we are using the locution. E.g. are we asking or answering a question? Are we giving some information or assurance or warning? Are we announcing a verdict or an intention? Are we pronouncing sentence? Are we making an appointment or an appeal or a criticism? Or are we identifying, classifying, or describing? Austin gives the name of 'illocutionary forces' to those different types of function that language has in the performance of an illocutionary act. It is as essential, he suggests, to distinguish force from meaning, as it is to distinguish sense from reference within meaning. Moreover, in addition to the performance of the locutionary act of saying something, and the performance of an illocutionary act in saying something, we may at the same time perform a perlocutionary act by saying something. Austin calls an act of speaking perlocutionary so far as it produces certain intended or unintended effects upon the feelings, thoughts, or actions of the audience, of the speaker, or of other persons. E.g., while He urged me to shoot her would describe an illocutionary act, He persuaded me to shoot her would describe a perlocutionary one. Austin claims that, just as expressions like 'use of sentence' can blur the distinction between locutionary and illocutionary acts, so too to speak of the 'use' of language can blur the distinction between the illocutionary and perlocutionary act. Speaking of the 'use of language for arguing or warning' looks just like speaking of the 'use of language for persuading, rousing, alarming'. Yet the former may, for rough contrast (103), be said to be conventional, in the sense that at least it could be made explicit by a performative formula, while the latter could not. Thus we can say, performatively, I argue that or I warn you that, but we cannot say I convince you that or I alarm you that. Again, what we do import, he says, by the use of the nomenclature of illocution is a reference, not to the consequences (at least in any ordinary sense) of the locution, but to the conventions of illocutionary force as bearing on the special circumstances of the utterance (114). Any or almost any perlocutionary act is liable to be brought off, in sufficiently

176

L. JONATHAN COHEN

special circumstances, by the issuing, with or without calculation, of any utterance whatsoever. But the range of illocutionary acts that may be brought off by a given utterance is restricted by the conventions of illocutionary force. Austin does insist, however, that the performance of an illocutionary act involves 'the securing of uptake' (115ff.). A man cannot be said to have warned an audience, Austin claims, unless it hears what he says and takes what he says in a certain sense. Also an illocutionary act may take effect in certain ways other than the bringing about of changes in the natural course of events. Thus naming a ship Queen Elizabeth has the effect of putting out of order any later references to it by another name. However, Austin does not think that his locutionary-illocutionary-perlocutionary classification exhausts its domain. He points out (121) that the expression 'the use of language' can cover other matters even more diverse than what he calls illocutionary and perlocutionary acts. We may speak of the use of language in poetry or for joking, and we can also use language to make insinuations, to show off, or to express our feelings (as in swearing). There are many weaknesses in Austin's account of speech-acts. I shall mention only four of them here.3 The first weakness is its proliferation of speech-acts. An apparently single act of linguistic utterance turns out to be at least five acts: phonetic, phatic, rhetic, illocutionary, and perlocutionary. Not that Austin thinks of these acts as spatiotemporally distinct events, in the way that the act of eating a piece of an apple may be thought to consist of the three successive acts of biting, chewing, and swallowing, respectively, or the act of driving a car may at times consist of, say, four simultaneous acts, e.g. the act of applying brakes with one foot, the act of depressing the clutch pedal with the other foot, the act of steering to the sidei with one hand, and the act of sounding the horn with the other hand. Rather, Austin's five acts are just different aspects of a single event, in the way that the man who turns his ignition key may also be, at the same time and place and with precisely the same movement, switching on the starting-motor of his car, beginning a long journey, and breaking off an old friendship. Austin's way of individuating acts is thus not without many parallels in everyday usage. But his terminology is nevertheless unfortunate. The conception of phonetic, phatic, and rhetic acts is complementary to that of illocutionary and perlocutionary ones but adds nothing to our understanding of the phenomena. It is just a reification and telescoping of customary distinctions between phonetic, morpho-phonemic, syntactic, and semantic modes of description. In other words Austin's theory sprouts novel and redundant terminology that cuts off his account of speech from any obvious continuity with current linguistic theory. It might have been much clearer to begin by listing current modes of speech-description and then go on and talk s

For a discussion of other difficulties in Austin's account, cf. Cohen 1964. But my own treatment of the problem there ignored some of the complexities that I try to illuminate in §§ 5-7 below.

SPEECH ACTS

177

methodologically about what other modes of description are requisite, rather than talk ontologically about what other speech-acts take place. Austin has curiously abandoned here the standard preference of analytic philosophy for what Carnap (1937 : 239) called the 'formal' as distinct from the 'material' mode of discourse. Not surprisingly, once this step has been taken, the entities soon begin to multiply. Searle, as we shall see (§ 4), in effect splits Austin's rhetic act into two, an act of referring and predicating, and by parity of reasoning, as will be argued, he should have conjured up several more such acts. Moreover, once one begins to look more closely at the illocutionary scene one can see that on some occasions at least an indefinitely large number of illocutionary acts are encapsulated in a single speech-act. In saying to a politician There is a plot to oust you, a man may be committing himself to the politician's cause, in thus committing himself he may have been condemning his own past actions, in condemning his own past actions he may have been repudiating his previous beliefs, in repudiating his previous beliefs he may have been drawing the obvious conclusions from events, and so on, and so on. When acts begin to multiply like this it looks as though there is work for Occam's razor to do. The second main weakness in Austin's account of speech-acts is the obscurity of the distinction he wishes to draw between meaning and illocutionary force. All he says about the former is that by 'meaning' he intends 'sense and reference' as currently viewed (and even this does not prevent him from — quite inconsistently — listing refer as a verb describing an illocutionary act (1962 : 162)). Perhaps he had in mind here Frege's distinction between sinn and bedeutung (Geach and Black 1952:68) or Strawson's distinction between the referring and descriptive role of linguistic expressions (Strawson 1952:145), or his own earlier distinction between demonstrative and descriptive conventions (Austin 1950:116). But however one construes Austin's account of meaning here, and therewith his account of rhetic and locutionary acts, it is difficult to avoid concluding that his definition of 'meaning' is too restrictive and consequently leaves more aspects of a speechact to be classed under the rubric of its 'illocutionary force' than would otherwise be desirable. For example, Austin lists asking and questioning as illocutionary acts. Yet one would normally speak of an utterance of the sentence Is the door shut? as having a different meaning, not just a different force, from an utterance, with falling intonation, of The door is shut. So it becomes tempting to suspect that Austin has exaggerated the importance of illocutionary force by ignoring certain widely accepted dimensions of meaning. This suspicion is confirmed by the variety of linguistic devices that Austin thinks (1962:13 ff.) are concerned with making clearer the force of an utterance, as distinct from its meaning. Among these he lists not only the explicit performative (as in I ask you whether the door is shut), but also verb mood, cadence, adverb or adverbial phrases (like without fait), and connecting particles (like therefore or although). But on this view there can be no difference

178

L. JONATHAN COHEN

of meaning at all between such utterances as It must have rained, because the streets are wet and It must have rained, therefore the streets are wet. Yet most ordinary speakers of English, let alone linguists or interpreters, would be very surprised indeed to hear that such a pair of utterances have the same meaning. Indeed there is no reason to suppose that particles like therefore must have different functions in the utterance of categorical sentences from those they have in the utterance of conditional ones. But in the antecedent of a conditional sentence, as in the second part of The price of helicopters has come down: if I therefore buy one now, I shall be able to get to work quicker, we cannot substitute a performative occurrence of I conclude that for therefore. We must suppose that therefore has some other use or function here than to clarify the illocutionary force of uttering the conditional sentence. It has a meaning that connects the antecedent clause of the conditional with the preceding sentence, and whatever meaning it has there it can just as well have also in It must have rained, therefore the streets are wet. It is not that we use therefore with the force of I conclude that, as Austin asserts (1962:75), but rather we use I conclude that with the meaning of therefore.* Just the same point can be made against what Austin says about the role of verb mood. You may shut it, he says (1962:74), resembles the performative I give permission, I consent, to your shutting it. But nothing like that can be said of If you may shut the queen's door, you are well trusted. Again it looks not as though may is here being used with the force of / permit that, but rather as though I permit that can sometimes be used with the meaning of may. No doubt the utterance of I permit that can in appropriate contexts give a speech-act permissive force, but it seems to be able to do this in virtue of that phrase's having permissive meaning. The third main weakness in Austin's account of speech-acts is his tendency to blur the distinction between what he calls 'illocutionary' and 'perlocutionary' acts. He described a speech-act as being perlocutionary rather than illocutionary as far as it produced certain intended or unintended effects upon the feelings, thoughts, or actions of people. Yet he also held that 'the performance of an illocutionary act involves the securing of uptake' (1962:116). The illocutionary act will not have been happily, i.e. successfully, performed unless it brings about the under4

A n analogus point may be made against exclusively performative accounts of is true. Cf. Cohen 1950 : 136ff., and more generally Geach 1960 : 221ff. and Searle 1962 : 423ff. W. D . Ross made a similar point (1939 :33f.) against Carnap's 'command' theory of moral judgment. Austin's concept of meaning also assigns the same meaning to certain mutually contradictory sentences (Cohen 1973). For he (1962:161) lists affirmation and denial as illocutionary acts. So for him I affirm that the door is shut may render explicit the utterance of The door is shut, while I deny that the door is shut may render explicit The door is not shut. Presumably, therefore on Austin's view we should regard the word not in The door is not shut as being indicative of illocutionary force, not meaning, and as having neither sense nor reference. The two mutually contradictory sentences The door is shut and The door is not shut then emerge as having the same meaning.

SPEECH ACTS

179

standing of the meaning and force of the locution. So in Austin's view, if I shout across to a farmer Your haystack is on fire and he does not hear me because he is deaf, then I have failed to warn him. But there are at least two serious difficulties in this doctrine. Certainly we do sometimes use warn in a strong sense in which it would be true to say, in regard to the above-mentioned circumstances, that I tried to warn the farmer and failed. But there is also another weaker sense in which it would be equally true to say, in regard to just the same circumstances, that I warned the fanner though he did not hear me. In the latter, weaker sense what is required is not that my utterance should actually have achieved uptake but that, whether or not it actually achieved this, it could reasonably have been expected to do so. So one difficulty with Austin's view is that it leaves us with no rubric for the large variety of speech-acts that are often described in such a weaker sense, whether these be warnings, greetings, prayers, apologies, challenges, entreaties, recommendations, or anything else. These acts slip right through the mesh of Austin's terminology, since, if it is left quite open whether or not uptake is achieved, such acts are neither locutionary, illocutionary, nor perlocutionary. The other serious difficulty in this feature of Austin's doctrine is that it is not easy to see why uptake should be thought necessary for the performance of some of the illocutionary acts which Austin himself lists (152 ff.). Reckoning, assessing, diagnosing, calculating, analyzing, postulating, distinguishing, defining, and formulating, for instance, all seem to be wholly unconcerned with uptake in any of their senses. Where uptake may be relevant we can say, in the strong sense, such things as I tried to warn the farmer, but failed because he was too deaf to hear me. But, if a man tries to calculate the sum of his debts and fails, it must be for other reasons than merely that he has not been heard. It is tempting to suggest that Austin could have got out of these difficulties by treating the strong senses of verbs like warn, entreat, greet, etc., as being descriptive of a perlocutionary act. But he wanted to distinguish between those acts which consist solely in the achievement of a certain effect, like persuading or humiliating, and those for which the achievement of a certain effect is merely a necessary condition, like warning or greeting (in the strong sense). Only the former was he willing to regard as perlocutionary. No doubt that was at least partly because he thought of illocutionary acts as being primarily regulated by conventions and perlocutionary ones as being primarily non-conventional in character (118). But this brings us up against a fourth major weakness in Austin's theory. The fourth weakness is the often reiterated thesis that 'there cannot be an illocutionary act unless the means employed are conventional' (118; cf. also 105, 108, 115, 120, 121, 127). The trouble here is that some of the things Austin calls illocutionary acts are far more conventionbound than others. The act of naming for example, in the sense of assigning a name, is very often regulated by ceremonial conventions. When babies are named, for example, they may be sprinkled with

180

L. JONATHAN COHEN

consecrated water, and ships with champagne. But the act of naming, in the sense of mentioning a name, is not normally made into a ceremony. In writing this paper I have so far named several philosophers. But it has been done quite without fuss or ritual. A builder's estimate for work to be done on a house may have to be set out in the proper form, otherwise it becomes invalid. But what conventions have ever invalidated an observer's estimate of the speed of an aeroplane? Again acquittals or convictions in a court of law may have to be pronounced in a conventional formula by the foreman of the jury, and a judge who sentences to death may have to wear a black cap. But descriptions, answers, and conclusions are not normally thus regulated by extra-linguistic conventions. True, conclusions may be regulated by laws of logic, but these are hardly as conventional as naming rituals. Nor, for that matter, is a form of acquittal conventional if it is precisely prescribed by statute. At one point (103) Austin implies that by calling the use of language in an illocutionary act 'conventional' he means that 'at least it could be made explicit by the performative formula'. But this does not get rid of the trouble at all, because some performative formulas are closely regulated by extra-linguistic convention, like I name this ship . . . , while others, like I warn you that... are not. Sailors are notoriously prone to be superstitious and generally do not want their ships named in unconventional ways. But anyone who really needs a warning is glad to have it uttered however unconventional the circumstances. No doubt a good case can be made out for saying that language itself is a convention-governed form of communication, in the sense that each speaker of a language conforms to certain regularities because of his expectation that the others will do likewise (cf. Lewis 1969). But what this explains is how sentences come to have meanings. To use Austin's terminology, it elucidates how locutionary acts are possible. It tells us nothing about other conventions that have to exist for illocutionary acts to be possible. And, of course, so far as meanings are governed by linguistic conventions, it will certainly be conventional that certain types of speech-act are described as 'warnings'. But this is irrelevant to the issue. For if it justified us in calling illocutionary acts conventional it would equally justify us in regarding any describable event whatever, under its proper description, as conventional. In sum, Austin's account of speech-acts is unsatisfactory on at least four counts. It multiplies acts unnecessarily, its distinction between locutionary and illocutionary acts is obscure, its distinction between illocutionary and perlocutionary acts is equally troublesome, and its thesis about the essential conventionality of illocutionary acts is quite untenable. I shall now consider two unsuccessful attempts to construct a better account of speech-acts upon more or less Austinian foundations. 3. P. F. Strawson's Modification of Austin's Strawson's starting-point (1964) 5

5

Account

in his discussion of Austin's account of illo-

My summary of Strawson's argument in this important paper inevitably does less than justice to the scrupulous clarity and precision with which it is stated.

SPEECH ACTS

181

cutionary acts is that 'there are many cases in which the illocutionary force of an utterance, though not exhausted by its meaning, is not owed to any conventions other than those which help to give it its meaning', and he thinks it possible to discuss the problem this raises without concerning himself directly with Austin's account of meaning. To suppose that there is always and necessarily a convention to which entreaties, for example, conform 'would be like supposing that there could be no love affairs which did not proceed on lines laid down in the Roman de la Rose or that every dispute between men must follow the pattern specified in Touchstone's speech about the countercheck quarrelsome and the lie direct'. Rejecting the view that Austin has just mistakenly generalized from some cases to all, he prefers to suppose that Austin's use of the term 'conventional' here is idiosyncratic and seeks to elucidate this use. He takes as a clue Austin's rather dark saying that the performance of an illocutionary act 'may be said to be conventional in the sense that at least it could be made explicit by the performative formula'. Whatever it is that leads Austin to call illocutionary acts in general 'conventional', must, according to Strawson, 'be closely connected with whatever it is about such acts as warning, entreating, apologising, advising, that accounts for the fact that they at least could be made explicit by the use of the corresponding first-person performative form'. So we have to explain what this is. Strawson takes as a basis for his explanation H. P. Grice's concept of someone's non-naturally meaning something by an utterance. According to Grice (1951) a person S non-naturally means something by an utterance x (which can be linguistic or non-linguistic) if S intends (¿1) to produce by uttering x a certain response (r) in an audience A, and also intends (4) that A shall recognise S's i'i intention, and also intends (13) that this recognition on the part of A of S's h intention shall function as /4's reason, or a part of his reason, for his response r. However, Strawson argues that it is possible to imagine a situation in which Grice's conditions would be satisfied by a person S and yet, in a sense of communicate fundamental to any theory of meaning, S could not be said to be trying to communicate by means of his production of x with the person A in whom he was trying to produce the response r. This would be a situation in which A would take S to be trying to bring it about that A is aware of the fact that p but would not take S as trying to tell him, in the colloquial sense of tell, that p, e.g. where S deliberately arranges evidence that p which A will both see and know to have been so arranged. Accordingly, Strawson concludes, we must add to Grice's conditions the further intention (¿4), on the part of S, that A should recognise S's ¿2 intention. Strawson admits the possibility that further argument could be produced to show that even adding this condition is not sufficient to constitute the case as one of attempted communication. But he does insist that at least this addition is necessary. Strawson goes on to suggest how, so far as his modified version of Grice's analysis is correct, an account may also be given of what it is for A to understand

182

L. JONATHAN COHEN

something by an utterance x. This is that S's U intention and hence his 4 intention must be fulfilled, though the i'i and 13 intentions need not be. Strawson then proposes to equate Austin's notion of uptake with this quasi-Gricean concept of understanding. But he rejects Austin's thesis that the securing of uptake is essential to any illocutionary act and holds instead that the aim, if not the achievement, of securing uptake is essentially a standard, if not an invariable, element in the performance of such an act. So to say something with a certain illocutionary force, Strawson holds, is at least (in the standard case) to have a certain complex intention of the ¿4 form. Strawson next applies this account of illocutionary force to the task of elucidating two other features of Austin's doctrine. One of these features is Austin's assertion that the production of an utterance with a certain illocutionary force is a conventional act in the sense of being generally suitable for being made explicit with the help of an explicitly performative formula. Strawson argues that in general a man can speak of his intention in performing an action with a kind of authority which he cannot command in predicting its outcome, and that the author of a speechact has a particular motive for making that intention clear. A speaker, Strawson says, 'will not have secured understanding of the illocutionary force of his utterance, he will not have performed the act of communication he sets out to perform, unless his complex intention is grasped'. Hence if there exists any conventional linguistic means of doing so, the speaker has both a right to use, and a motive for using, those means. And what better means can there be than an explicitly performative formula? Austin also held (1962:104) that there are several things which we might be doing in saying things but which are not illocutionary acts, like showing off or insinuating. Strawson points out how his account of illocutionary acts excludes such activities, since in their case it is no part of the speaker's intention to secure his intention by means of the hearer's recognition of the intention to secure it. Similarly the further effects a speaker may intend his utterance to have — e.g. the influence he intends to exert on his hearers by telling them something — is also excluded by Strawson's requirement that the speaker should have intention 14. Thus in Strawson's view those illocutionary acts that are not essentially conventional (as are judges' sentences, juries' acquittals, umpires' verdicts, etc.) are conventional in the sense that they may be linguistically conventionalized (by the use of a performative formula) right up to the point at which illocutionary force is exhausted by meaning. The two types of illocutionary act can now be compared more revealingly. In the case of an illocutionary act of a kind not essentially conventional, the act of communication is performed if uptake is secured, just so far as the utterance is taken to be issued with the complex overt intention with which it is issued. 'But even though the act of communication is performed, the wholly overt intention which lies at the core of the intention complex may, without any breach of rules or conventions, be frustrated. The audience response

SPEECH ACTS

183

(belief, action, or attitude) may simply not be forthcoming.' But it is different with the utterance which forms part of a wholly convention-governed procedure. 'Granted that uptake is secured, then any frustration of the wholly overt intention of the utterance (the intention to further the procedure in a certain way) must be attributable to a breach of rule or convention. The speaker who abides by the conventions can avowably have the intention to further the procedure in the way to which his current linguistic act is conventionally appropriate only if he takes it that the conventional conditions for so furthering it are satisfied and hence takes it that his utterance will not only reveal his intentions but give them effect. There is nothing parallel to this in the case of the illocutionary act of a kind not essentially conventional. In both cases, we may say, speakers assume the responsibility for making their intentions overt. In one case (the case of the conventionconstituted procedure) the speaker who uses the explicitly performative form also explicitly assumes the responsibility for making his overt intention effective. But in the other case the speaker cannot, in the speech-act itself, explicitly assume any such responsibility . . . Whether it is effective or not is something that rests with his audience.' Strawson views these two types of illocutionary acts as opposite poles of a spectrum rather than as being mutually exclusive and jointly exhaustive classifications. Nevertheless, he thinks we can perhaps extract something which is common to both and to all the other types that lie between them. 'For the illocutionary force of an utterance is essentially something that is intended to be understood. And the understanding of the force of an utterance in all cases involves recognising what may be called broadly an audience-directed intention and recognising it as wholly overt, as intended to be recognised' (though Strawson fully admits, of course, that a speaker who offers information, instructions, or advice may yet be overtly indifferent as to whether or not his information is accepted as such, his instructions followed, or his advice taken, just so long as his communication is understood). Thus Strawson's account of illocutionary acts is a proposal for remedying the third and fourth of the four weaknesses in Austin's account that I mentioned in § 2. Strawson obviously does not regard the first — the proliferation of 'acts' — as any weakness at all, and he expressly sets aside the problem of Austin's account of the rhetic act. We can sum up his theory by saying that it draws a sharper line between illocutionary and perlocutionary acts by altering Austin's requirement that the former should actually secure uptake to the condition that in standard cases uptake should at least be intended, and that it then claims to be able to interpret Austin's thesis that all illocutionary acts are conventional in such a way as to make that thesis fit the facts fairly closely. It is not clear why Strawson should prefer to think of himself as reinterpreting Austin's doctrine rather than correcting it in regard to the problem of conventionality, and as correcting rather than reinterpreting it in regard to the requirement of uptake. But this is not an important issue. What

184

L. JONATHAN COHEN

is important is that even in the form in which Strawson presents it Austin's doctrine of illocutionary acts is still radically incoherent. The trouble is that Austin lists many types of illocutionary acts which turn out neither to be essentially regulated by conventions or other non-linguistic rules nor to be done with the complex, fourfold intention described by Strawson. These acts are not the intermediate cases mentioned by Strawson (like ordering), which may owe something to conventions at the same time as they also owe something to their complex intentional structure. Instead they are cases that owe nothing to either of these two factors, and so fall quite outside the spectrum of illocutionary act-types that Strawson adumbrates. Let us look more closely at those speech-acts which, though listed by Austin as being illocutionary, do not normally owe anything to extra-linguistic conventions or rules of any kind. These types of speech-act may be divided into three groups. There are those that are nearly always conceived as having an intentional structure rather like the one that Strawson describes, those that are often conceived as having such a structure and are often not so conceived, and those that are rarely if ever so conceived. Into the first group fall such speech-acts as entreating, commiserating, advising or reporting. It seems to be part of what is meant by describing an act as one of these that the author of the act has at least the intentions ii, it, and is of the Grice-Strawson analysis. Certainly it would be very odd or unusual to speak of someone as unintentionally entreating, unintentionally commiserating, unintentionally advising, or unintentionally reporting. Of course, a man may unintentionally entreat the wrong official, commiserate with the wrong victim, give the wrong advice, or report the wrong game. But what is unintended here is the object of his activity, not its nature. So it looks as though intentions z'i, 4, and h at least must be present. However, I am not sure that intention k also has to be present in every type of case. Perhaps the earnestness that is essential to an entreaty's being more than merely half-hearted, or the sincerity that is essential to the genuineness of a commiseration requires an h intention in the speaker's mind. But I am not sure that h intentions have to be present in acts of advising or reporting. What is distinctive of both these, for example, is rather the linguistic means the speaker employs to produce a certain response in his audience. Into the second group fall such speech-acts as warning, answering, instructing, or revealing. It is not at all odd or unusual to speak of a conspirator's unintentionally warning his opponents by what he says, or of a criminal suspect's unintentionally answering the questions the police would like to have answered, or of a factory foreman's unintentionally instructing an industrial spy, or of a political speaker's unintentionally revealing the plans of his government. Certainly when these activities are intentional, as they often are, they can be described as being achieved either in saying that so-and-so or by saying this, whereas when they are unintentional the latter description seems more appropriate. So in terms

SPEECH ACTS

185

of one of Austin's criteria for distinguishing between illocutionary and perlocutionary acts it might seem arguable that when these acts are unintentional they are really perlocutionary, and not illocutionary at all. But this objection will not succeed. For first, Austin himself admits that the in/by criteria is somewhat shaky (121 ff.). And, secondly, he regards an act as being perlocutionary only so far as it actually produces certain effects at the time, whereas this may not happen in cases of the types of speech-act under consideration. The conspirator's unintentional warning may remain unnoticed by those against whom he is plotting; the criminal suspect's unintentional answer may escape the attention of the thickwitted policeman; the foreman's unintentional instructions may not be heard by the industrial spy because of the din in the factory; and the politician's unintentional revelations may not be spotted until the history of the period comes to be written much later. Moreover, it should be emphasised that what is unintended here need not be just the object of the activity. The very nature of the speechact actually performed can be unintended, where this is an act like warning, answering, instructing, or revealing. Strawson recognises that speech-acts which are 'essentially conventional' (regulated by extra-linguistic conventions or rules) may in exceptional cases be unintentional. But his intentionalist account of illocutionary acts that are not 'essentially conventional' does not allow any room for the not uncommon cases in which these too are unintentional. There is also a third group of Austin's illocutionary acts in regard to which the speaker's intentions vis-à-vis an audience seem largely irrelevant. Into this group fall such speech-acts calculating, assuming, deducing, or analyzing. Some of these can be done altogether unintentionally, as when by addressing a young woman as 'Miss . . . ' a man may unintentionally assume that that she is unmarried. Others, like calculation, can hardly ever be done unintentionally. But what characterizes all the members of this group, and distinguishes them from members of the two groups previously mentioned, is that their nature does not consist at all in the bearing they are intended to have, or might be taken as having, on the responses of an audience. A solitary speaker would be either just silly or at least rather whimsical to engage in any entreating, commiserating, advising, reporting, warning, answering, instructing, or revealing. But he could quite sensibly be engaged in voiced or silent soliloquy in the course of which he made calculations, assumptions, deductions, or analyses. Mathematicians and philosophers are constantly doing this, with no thought whatever about an audience. It follows that there are at least two large groups of Austin's illocutionary acts which cannot be said to be 'conventional' even in the very idiosyncratic use of that term which Strawson's intentionalist interpretation imputes to Austin. Nor could a defender of Strawson's interpretation object that this only shows another respect in which Austin's thesis about conventionality was oversimplified. It is not sufficient to admit that, besides illocutionary acts that are essentially conventional and those that are conventional in the intentionalist sense, there are also several

186

L. JONATHAN COHEN

types of illocutionary act that are neither. For, it must be remembered, Strawson's interpretation was primarily based on Austin's assertion that the performance of an illocutionary act 'may be said to be conventional in the sense that at least it could be made explicit by the performative formula'. Strawson claimed that his interpretation was supported by its ability to elucidate this assertion. But the trouble is that even calculating, postulating, deducing, analysing, etc., are capable of being made explicit by the appropriate performative formulas, and, if these speech-acts do not characteristically involve intentions vis-à-vis an audience, the having of such intentions cannot be invoked in these cases to explain why a speaker's illocutionary act can be made explicit by a performative formula. There must be some other reason why a performative formula can render the illocutionary act explicit in these cases, and if for them then why not also for all the other types of cases too? So instead of constituting a general elucidation of the connection between illocutionary acts and performative idiom the intentionalist account looks rather as though it can do no more than provide part of the analysis of entreaties, commiserations, and a few similar speech-acts. 4.

/. R. Searle's Modification and Development of Austin's

Account

Like Strawson, Searle has sought to combine elements of Grice's theory of meaning with elements of Austin's theory of speech-acts. In a normal utterance, he holds (Searle 1969:23 ff.), a speaker performs, at one and the same time, acts of four different kinds. He performs an utterance act by uttering words, morphemes, or sentences; he performs a propositional act by referring and predicating; he performs an illocutionary act by stating, questioning, commanding, promising, etc.; and he performs a perlocutionary act by achieving some effect on the actions, thoughts, etc., of his hearer. These various speech-acts are in general made possible by, and are performed in accordance with, certain rules for the use of linguistic elements (1969:16). Speaking a language is thus engaging in a form of behavior that is governed by constitutive rules (37). More specifically, uttering a sentence and meaning it is a matter of (a) intending to get the hearer to know (recognize, be aware) that certain states of affairs specified by certain of the rules obtain, (6) intending to get the hearer to know (recognize, be aware of) these things by getting him to recognize intention a, and (c) intending to get him to recognize intention a in virtue of his knowledge of the conventional rules for the sentence uttered (48). For example, in Searle's view (57 ff.), given that a speaker S utters a sentence T in the presence of a hearer H, then in the literal utterance of T, S sincerely and non-defectively promises that p to H if and only if 1. Normal input and output conditions obtain (no deafness, joking, playacting, etc.). 2. S expresses the proposition that p in the utterance of T. 3. In expressing that p S predicates a future act A of S.

SPEECH ACTS

187

4. H would prefer S's doing A to his not doing A and S believes H would prefer his doing A to his not doing A. 5. It is not obvious to both S and H that S will do A in the normal course of events. 6. S intends to do A. 7. S intends that the utterance of T will place him under an obligation to do A. 8. S intends to produce in H the knowledge that the utterance of T is to count as placing S under an obligation to do A, and so on, as with intentions b and c above. 9. The semantic rules of the dialect spoken by S and H are such that T is correctly and sincerely uttered if and only if conditions 1-8 obtain. Searle has here stated rules for the performance of the typically illocutionary act of promising. He goes on to explore the nature of the rules for the nonillocutionary speech-acts of referring and predicating. Given that S utters an expression R in the presence of H in a context C then in the literal utterance of R S successfully and non-defectively performs the speech-act of singular identifying reference if and only if (94 f.) 1. Normal input and output conditions prevail. 2. The utterance of R occurs as part of the utterance of some sentence (or similar stretch of discourse) T. 3. The utterance of T is the purported performance of an illocutionary act. 4. There exists some object X such that either R contains an identifying description of X or S is able to supplement R with an identifying description of X. 5. S intends that the utterance of R will pick out or identify X to H. 6. S intends that the utterance of R will identify X to H by means of H's recognition of S's intention to identify X and he intends this recognition to be achieved by means of H's knowledge of the rules governing R and his awareness of C. 7. The semantic rules governing R are such that it is correctly uttered in T and C if and only if conditions 1-6 obtain. For predicating P of an object X Searle's (126 f.) first three conditions are the same as those for referring. In addition he requires that 4. The utterance of T involves a successful reference to X. 5. X is of a type or category such that it is logically possible for P to be true or false of X. 6. S intends by the utterance of T to raise the question of the truth or falsity of P of X. 7. S intends to produce in H the knowledge that the utterance of P raises the question of the truth or falsity of P of X, by means of H's recognition of this intention; and he intends this recognition to be achieved by means of H's knowledge of the meaning of P. 8. The semantic rules governing P are such that it is correctly uttered in T if and only if conditions 1-7 obtain.

188

L. JONATHAN COHEN

Searle claims that his theory of speech-acts illuminates several problems in contemporary philosophy, and in particular that it shows up three common fallacies. First, there is what he calls 'the naturalistic fallacy fallacy', viz. the thesis that it is impossible to derive evaluative from descriptive statements (132). Searle argues (148) that this thesis arose from a failure to distinguish between the use of certain sentences, i.e. the illocutionary force of their utterance, and their meanings. E.g. the truth conditions of the proposition The premisses of the argument X entail the conclusion of X are also necessary and sufficient for the truth of the proposition The argument X is valid. So an utterance of the former proposition has just the same meaning as an utterance of the latter. But whereas the illocutionary act performed in expressing and uttering the former proposition would normally be a descriptive one, the illocutionary act performed in uttering the latter proposition would normally be an evaluative one. Secondly, there is what Searle calls 'the speech-act fallacy' (136 ff.), viz. the thesis that the meanings of certain words, like good, true, etc., can be given by an account of certain speechacts of recommending, endorsing, etc., that they are supposed to be primarily used to perform. Searle argues, in substance, that these meanings cannot be so given because the words often occur with demonstrably the same meaning in contexts where no such speech-acts are at issue. And thirdly, there is what Searle calls 'the assertion fallacy' (141 ff.), whereby philosophers, he thinks, have confused conditions for the illocutionary act of asserting a sentence in which such words as know, voluntary, etc., occur, with truth-conditions for the predication of these words. It is clear that Searle also, like Strawson, has abandoned Austin's thesis that illocutionary acts need to secure uptake. His theory of speech-acts is therefore free from the third of the four above-mentioned weaknesses in Austin's account, viz. the blurring of the distinction between illocutionary and perlocutionary acts. Also his theory, if it is correct, has a quick way of settling the trouble about conventionality — the fourth weakness in Austin's account. Searle does not seek to distinguish like Strawson between illocutionary acts that are essentially conventional, because bound by extra-linguistic rules, and acts that are not. For him all illocutionary acts that are accomplished by the use of language are conventional, because language is constituted by conventional rules that regulate illocutionary acts as well as utterance acts and propositional acts. For example, 'in the use of speech acts performed within a language', he says, 'it is a matter of convention — as opposed to strategy, technique, procedure, or natural fact — that the utterance of such and such expressions under certain conditions counts as the making of a promise' (37). I shall, however, reserve for § 5 my reasons for rejecting the thesis that what Austin calls illocutionary acts are normally regulated by the constitutive rules of natural languages like English or French. In the present section I shall confine myself to a criticism of Searle's attempt to remedy the second of the four weaknesses that I earlier pointed out in Austin's

SPEECH ACTS

189

account of speech-acts, viz. the obscurity of of Austin's distinction between meaning and illocutionary force. According to Austin meaning is conveyed by the rhetic act and consists in sense and reference, on which he does not expand. Searle apparently seeks to develop this Austinian account of meaning, and elucidate the concepts of sense and reference, by supposing two further types of speech-act, an act of predicating and an act of referring, which together constitute a propositional act. But it is very doubtful whether predication and reference admit of a speech-act analysis that will distinguish them adequately from illocutionary acts and so serve the underlying purpose of distinguishing meaning from illocutionary force. Consider predication first. In every act of predicating an expression P, Searle claims, the speaker intends to 'raise the question' whether P is true or false of an object, and utterance of P counts as raising this question (127). But this is just not true. By surrounding the utterance of P with suitable emphasis one can intend to dismiss the question, not raise it. Or, if someone else has raised the question and one does not want to dismiss it, one can take up the question oneself. No doubt it would be possible for Searle to replace 'raise the question' here by 'raise, take up, or dismiss the question'. But as soon as there are seen to be at least those three, mutually exclusive possibilities, the act of raising, taking up, or dismissing the question looks as thought it ought to be regarded as an illocutionary act. To define predication by reference to the disjunction of these acts looks very much like defining an essential component of the propositional act in terms of a certain dimension of illocutionary force — a dimension that includes pursuing, exploring, twisting, settling, disposing of, resurrecting, and doing many other things to questions. So this kind of definition can hardly serve as a basis for distinguishing between rhetic or propositional acts, on the one side, and illocutionary acts, on the other. Searle's propositional act of referring is just as suspect. Even Austin lists the act of referring as an illocutionary one (1942:162). After all, to describe what a man referred to involves describing something of what he achieved with his utterance, which seems to be a generic feature of all descriptions of illocutionary acts. And it is also worth considering how acts of referring are individuated. In Searle's view there has to be at least one distinct object for each act of reference, and the instrument for each act of reference is a distinct part of the sentence uttered. So presumably in uttering George loves Mary a man has performed two acts of referring — one act of referring to George and another act of referring to Mary. But though his utterance undoubtedly contained references to two people, and might even be said to have contained two references to people, and though he would certainly be said to have referred to two people, it is extremely doubtful whether he would ever be said to have referred twice to people in the course of such an utterance. If I tell you in the course of a long conversation, first that my tailor lives in Oxford, and later that he is married, and later still

190

L. JONATHAN COHEN

that he is a keen gardener, I shall have referred three times to my tailor. But if I say to you just My tailor took the coat that he had made for me and gave it to the apprentice whom he liked best, then though there are three references to the tailor in my utterance it would seem odd to say that I had referred three times to him. Rather, I could be said to have referred to him in three different respects. A whole sentence is the normal unit of speech, and we so often need to know the whole sentence in order to know what has been said that it seems rather queer to envisage an act of speech that is essentially smaller than an act of uttering a whole sentence. To break down an act of speech into locutionary, illocutionary, and perlocutionary acts, as Austin does, is to treat an act of sentence-utterance as a unitary whole which has three main describable aspects. But to adopt Searle's conception of an act of referring and an act of predicating is to abandon altogether the conception of an act of sentence-utterance as a unitary whole. Moreover, Searle's decomposition of a propositional act into an act of referring and an act of predicating is far from adequate to the complexity of meaning. If this kind of analysis is to be made at all, then at least three other types of act need to be recognized that are co-ordinate with these alleged acts of referring and predicating. First, an act of quantifying would need to be supposed, in order to distinguish between the propositional act performed in uttering, say, All the people here are philosophers, and that performed in uttering Some of the people here are philosophers. Secondly, an act of connecting would need to be supposed in order to distinguish between, say, It's either rainy or windy on the island and If it's rainy on the island, it's windy, or between It snowed after it rained and It rained after it snowed. Thirdly, an act of modalizing would need to be supposed in order to distinguish between, say, If you do shut the queen's door, you are well trusted, If you may shut the queen's door, you are well trusted, and If you ought to shut the queen's door, you are well trusted. None of these differences lend themselves easily to representation as differences of predication or reference, as many logical studies have shown. Still less can they be represented as differences of illocutionary force, since they are differences in what is said, not in what is achieved with what is said. But when one sees that at least these three other types of speech-act need to be postulated as soon as Searle's acts of referring and predicating are admitted one can also see that, unless they are mixed up with illocutionary elements, such acts are like shadows cast by the terms of logical analysis: predicate, denotation, quantifier, connective, modality, etc. To suppose a special type of speech-act for each of these analytical abstractions adds nothing to our understanding of human speech. Indeed,6 if an act of quantification is one of the constituent acts that goes to compose certain propositional acts, then presumably another component act on such occasions must be an act of schema-utterance rather than an act of pred•

I owe this point to a conversation on the subject with Donald Davidson.

SPEECH ACTS

191

icating in Searle's sense. Every 3x-type act must be accompanied by a corresponding Fx-type one. So in general Searle's concept of a 'prepositional act' is scarcely plausible. Terms like proposition and predicate normally orginate, as tools of logical analysis, when we decide to consider the content of an utterance in abstraction from its author, date, and physical manifestation. To talk of a prepositional act, or an act of predication, seems rather like trying both to carry out such an abstraction and to refrain from carrying it out at one and the same time. Further difficulties therefore arise in Searle's theory when we press his thesis (1969:30) that we can distinguish two (not necessarily separate) elements in the syntactic structure of a sentence, which he calls the prepositional indicator and the illocutionary force indicator. 'The illocutionary force indicator', he says, 'shows how the proposition is to be taken, or to put it another way, what illocutionary force the utterance is to have; that is, what illocutionary act the speaker is performing in the utterance of the sentence'. Illocutionary force indicating devices in English according to Searle, include at least: word order, stress, intonation contour, punctuation, the mood of the verb, and explicit performatives. But once Searle's theory of the prepositional act begins to crumble we do not know where to draw the line, in his view, between prepositional indicators and illocutionary force indicators. For example, since in Searle's view, as in Austin's, the same utterance may constitute the performance of several different illocutionary acts, how can we tell whether to regard an utterance of George will marry Mary as being just an assertion, say, of the proposition George will marry Mary or as being both an assertion and a prediction of the proposition George (timelessly) marries Mary? I.e. how can we tell whether or not prediction, retrodiction, etc., are to count as illocutionary acts? Similarly, utterance of the sentence Smith is coming to London might be construed to involve not only the illocutionary act of assertion but also the illocutionary act of talking about movement towards the speaker, while the proposition expressed would be just that Smith is moving to London. Or again Smith's utterance to Jones of the sentence I am listening to you might be construed to involve illocutionary acts of autobiographizing and addressing — with the personal pronouns functioning as illocutionary force indicators as well as referring devices. At times (69 f.) Searle seems to suggest that the possibility of achieving an explicit rendering by means of a performative formula is to be taken as a criterion of whether or not an illocutionary force is expressible in a language. For example, if an illocutionary verb rubrify existed in English, meaning to call something red, some English utterances would have the illocutionary force of rubrification and I hereby rubrify it would just mean the same as It's red. 'Analogously', says Searle, 'we happen to have an obsolete verb macarise, meaning to call someone happy'. But by this criterion there would be many anomalies. The past and future tenses of English would function as illocutionary force indicators since the

192

L. JONATHAN COHEN

verbs predict and retrodict exist, whereas the present tense would not so function since no corresponding illocutionary verb exists. Or again, the second person pronoun would function as an illocutionary force indicator, since the verb address exists, whereas the first and third person pronoun would not so function, since no corresponding illocutionary verbs exist. It is difficult to accept a theory of illocutionary force that, on the one hand, supposes every normal utterance to have an illocutionary force and, on the other hand, imputes anomalies of this kind to the range of forces that English utterances can have. The situation is not much better, however, if we suppose that Searle does not insist on taking the possibility of an explicit performative rendering as a criterion of whether or not an illocutionary force is expressible in a language. For then there will be an illocutionary act to correspond with every predicatable expression — an act of rubrification to correspond with is red, an act of altification to correspond with is high, an act of volification to correspond with flies, an act of imputification to correspond with imputes, and so on plus indefinitely more acts of rubrificatifiation to correspond with calls red, etc. Under these circumstances every propositional act of predicating P in uttering a categorical sentence would also ipso facto be an illocutionary act of calling something P. Austin's quite long, but definitely bounded, list of illocutionary act-types would be extended to infinity, while at the same time the typology of illocutionary forces would become largely indistinguishable from the typology of predicate-meanings. I conclude that Searle has not shown either predication or reference to admit of speech-act analysis that will distinguish it adequately from an illocutionary act and so serve the underlying purpose of distinguishing meaning from illocutionary force. II. A NEW START

5. Should the Study of Speach-Acts Form a Branch of Linguistics? It is clear that neither Strawson nor Searle succeeded in remedying the weaknesses in Austin's account of speech-acts, and rather than essay some further patching-up of that doctrine it may well be more profitable to make a new start. What follows is an attempt to do just this. I take over from Austin, as my datum or startingpoint, none of his terminology or distinctions but only his lists of illocutionary verbs, and I conceive my task as that of trying to elucidate the role of such verbs along with such other verbs as those Austin discusses in connection with them. What is common to all these words is that each can be used to describe something that we sometimes achieve with linguistic utterances in virtue of their contents. But these achievements are very various and correspondingly the relation of the utterance to the achievement varies quite widely. Acquittals, warnings, promises, apologies, denials, deductions — all of these have somehow to fall into the net. Moreover, if the concept of the description of an utterance's achievement is as wide

SPEECH ACTS

193

as this, we obviously cannot exclude it from covering such verbs or verb-phrases as joke, hint, insinuate, show o f f , play-act, etc., which Austin regarded as describing neither illocutionary nor perlocutionary acts. Nor can we exclude it from covering numerous other verbs and verb-phrases, like lie, misrepresent, misname, perorate, parody, speak ironically, etc., that Austin did not mention. Now the first question about these verbs that I propose to tackle is this: do any of them belong, as such, in the terminology of linguistics? I.e., should the full description of a natural language, like English or French, be conceived to include descriptions of any of the various types of things achievable with utterances in that language? There is an undercurrent of an affirmative answer to this question in Austin's account of speech-acts. His claim that illocutionary force is distinguishable from meaning, but somehow co-ordinate with it, seems to suggest that he may at times be thinking of his account of speech-acts as a pioneering contribution to linguistic theory. His thesis that illocutionary acts are to be compared with phonetic, phatic, and rhetic ones, seem to suggest that alongside the phonology, morphology, syntax, and semantics of a language we need to study the illocutionary forces that its utterances can have. Certainly Searle makes it quite plain that he conceives the constitutive rules of a natural language to include rules that combine to constitute the illocutionary forces of utterances in that language. At one point (1969:61 ff.) for example, he lists five rules for the use of any illocutionary force indicator for promising. If a given language has these rules in regard to a particular idiom i, then the presence of i in an utterance normally serves to indicate that the illocutionary force of that utterance is that of a promise. My own answer, however, to this question — the question whether speech-act descriptions have any place in linguistics — would be in the negative, and for two reasons. First, very many of the activities that Austin calls illocutionary acts seem to involve the tenure of rather rare offices or the possession of rather specialist abilities. For example, it is difficult to regard the linguistic competence that every native speaker acquires in childhood as including the abilty to excommunicate, reprieve, veto, or nominate. A child may learn what to describe as an excommunication, reprieve, veto, or nomination, insofar as it learns the meanings of the corresponding words, but that is not the same as learning to do these things. A child learns what to describe as riding a bicycle long before it can actually do this, and similarly the ability to describe an act as one of excommunication is quite distinct from the ability to perform an act that actually has the force of excommunicating someone. Indeed a man's ability to excommunicate, reprieve, veto, or nominate may lapse, be restricted, be subordinated, or be altogether taken away, while his linguistic competence remains unimpaired. Again explaining, inferring, advising, commiserating, thinking, or apologizing are things done much better by some people than by others. Children have to be taught when to say

194

L. JONATHAN COHEN

Thank you and when to apologize, and some never learn at all. But to teach them such things is to teach manners, not language. Some people get better at advising or inferring as they grow older, and some get worse. But neither change need affect their status as fluent speakers of their native language. Since inferring and advising are skills that people exercise with language, one no more needs to describe them when describing the language, than one needs to describe a grandmaster's strategies for winning when one describes the rules of chess. Moreover, most of these rather specialist abilities, like the skills of inferring, advising, analyzing, defining, sympathizing, assessing, calculating, distinguishing, etc., are readily transferred from one language to another. If you're good at them in one language, and then learn to speak another perfectly, you'll be just as good in the other. But if you're bad at them in your native language, you'll be just as bad in another language, however perfectly you learn to speak it. So the full description of a natural language should not embrace what its speakers can achieve with their utterances at least where this achievement depends on the tenure of rather rare offices or the possession of rather specialist abilities. There is also another reason for rejecting the view that what Austin called illocutionary forces are co-ordinate with meanings and form a subject-matter for linguistics. The view tends to obstruct recognition of, and distinction between, two important dimensions of meaning. I have previously ( § 4 ) referred to one of these as the modality of a sentence. This feature of a sentence's meaning is commonly conveyed in English by such auxiliary verbs as is, may, ought, or must, by the use of indicative or subjunctive moods or by adverbs like probably, necessarily, or possibly. What compels acceptance of this as a feature of meaning, not of illocutionary force, is that it often occurs in subordinate clauses — within, say, the antecedent of a conditional sentence or in oratio obliqua. A n utterance of the sentence If you may shut the queen's door, you are well trusted means that you are well trusted if you do have permission to shut the door: it does not itself have the force of giving permission for anything. Certainly, in determining the correct description of an utterance's achievement the modality of its main clause is nearly always one of the cardinal factors. A n occurrence of ought suggests advice, recommendation, assessment, warning, etc.; an occurrence of may suggests permission, agreement, conjecture, etc., and so on. But the modality of a sentence never suffices to determine uniquely the description of what the sentence's utterance achieves. You ought to go may be uttered to advise, warn, prescribe, recommend, instruct, urge, or admonish. It is important to distinguish the modality of a sentence or clause's meaning from another feature of it that also has a very close bearing on the correct description of what an utterance achieves. This other feature I shall call a clause's or sentence's semantic force. It is commonly conveyed in English indirect discourse by the choice between declarative, imperative, interrogative, and optative forms, and in direct discourse by the choice between any one of these and exclamatory

SPEECH ACTS

195

or performative forms. 7 (By a 'performative form' I m e a n a main clause occurrence of a verb in that f o r m — generally first person present indicative — in which its utterance can perform that which in other forms the verb can only describe. A paradigmatic use is I promise in I promise to repay you next week. Less typical cases are I shall now reveal in / shall now reveal that I am a police constable, or You are hereby warned in You are hereby warned that anything you say may be used in evidence against you. Of course, some morphophonemic strings may be ambiguous between performative and declarative.) Performative forms may be conjoined but not disjoined, collocate with hereby but not with often, etc. It may well b e that an adequate grammar of English could shorten this list by reducing some of the semantic forces mentioned to others. But that point is quite unimportant for m y present purpose. Some such distinctions seem to be present 7

A number of papers have also been written recently by linguists in which arguments are advanced for the thesis that the underlying syntactic structure of every sentence contains a performative verb, real or abstract, which determines its illocutionary force. If these arguments were valid for that thesis they would count against my own view here. But what these arguments in fact seem to show, as far as they are valid, is just that the underlying syntactic structure of every sentence contains some elements that determine its semantic modality and semantic force. To suppose that a performative verb is present in this structure is to suppose a far more specific determination of that modality and force than is justified by the evidence. For example, there is an inadequacy in Robin Lakoff's argument for claiming that in this way both an abstract verb of ordering and an abstract verb of wishing can function as performatives in the underlying structure of sentences. Her argument does not show that these alleged abstract verbs are verbs of ordering and wishing, respectively, any more than that they are verbs of requesting, advising, recommending, commanding, prescribing, directing, counselling, instructing, entreating, or enjoining, on the one hand, or verbs of desiring, fancying, preferring, hoping, hankering, longing, craving, wanting, aspiring, or coveting, on the other. What the argument shows is just that in its underlying structure a sentence may, among other possibilities, be (undifferentiatedly) imperative or (undifferentiatedly) optative. Cf. R. Lakoff (1968) and G. Lakoff (1971). Similarly, J. R. Ross's arguments for the thesis that an abstract performative of the form I say to you that occurs in the underlying structure of all declarative English sentences fail to show why this performative has to be a verb of saying rather than one of telling, declaring, asserting, informing, asseverating, stating, etc. Cf. J. R. Ross (1970). The are also other invalidities in Ross's arguments. For example, he argues that since As for myself, I shan't be invited is non-deviant and As for herself, Glinda won't be invited is deviant, we have to suppose a latent performative I say that... prefixed to the first of these two sentences, to correspond with non-deviant contexts of as for herself, like Glinda knows that, as for herself, she won't be invited. But I cannot see anything syntactically deviant in As for herself, Glinda won't be invited; for example, the utterance of such a sentence might be followed by But her boy-friend will be invited. Compare Glinda herself won't be invited, but her boy-friend will be. For more detailed criticisms of Ross's thesis cf. B. Fraser (1970) and S. Anderson (1970). At the same time it is clear that performativeness has to be regarded as co-ordinate with declarativeness, imperativeness, optativeness, etc., and not as a general feature of all sentences, since many sentences occur with no other verb than the performative one - e.g. I hereby dismiss you, I name this ship Queen Elizabeth, etc. - and these sentences therefore cannot be classified as being declarative, imperative, optative, exclamatory, or interrogative. I say to you that it is raining has a performative main clause and a declarative subordinate clause, and so does I deny that it is raining. C. L. Baker (1970) argues that English questions, both direct and indirect, contain initial question-morphemes in their deep structure, and this argument wholly accords with my own - more general and less rigorously argued - thesis about semantic force.

196

L. JONATHAN COHEN

in every language, even if they are only marked by international contours. And what distinguishes semantic force quite clearly from modality is that, unlike modality, it is normally either cancelled or excluded by conditionalization. The sentences You are shutting the queen's door and You are well trusted both have a declarative semantic force. But neither the subordinate nor the main clause in If you are shutting the queen's door, you are well trusted has one, though the sentence as a whole does, derived from the semantic force of the sentence that forms its main clause. Similarly Give me the key has an imperative semantic force, and so does If you are shutting the queen's door, give me the key. But the clause give me the key in the latter sentence does not have imperative semantic force on its own because the meaning of the whole sentence is such as to make the giving conditional on the shutting. Furthermore, there is also another reason for distinguishing semantic force, as I have here been calling it, from modality. Namely declarative, interrogative, exclamatory, and optative semantic forces may be combined with one or another of several different modalities. We can say (interrogatively) both May I go? and Must I go?; (exclamatorily) both What a few options I may exercise! and What a lot of things I must not do! and (optatively) both Would that he must go! and Would that he goes! In determining the correct description of what a sentence's utterance does on a particular occasion the sentence's semantic force is obviously at least as cardinal a factor as the sentence's modality. It often restricts quite sharply the range of descriptions that might fit. But semantic force is not to be confused with what Austin or Searle call illocutionary force. It is a feature of langue, not parole. It is a property of sentences, not of their utterances. A sentence with an imperative semantic force like Tell me the time can be uttered in many different contexts. In some the utterance will be describable as a request, in others as an order, a command, an entreaty, a plea, a prayer, a demand, or a mere question. Similarly on optative like Would that my land were well ploughed! may be uttered as the expression of a wish, desire, hope, prayer, entreaty, anxiety, etc. A sentence with declarative semantic force can be uttered with an even greater variety of what Austin called illocutionary force. The declarative sentence, with indicative modality, You are the kindest person I have met here may be uttered in such a way as to thank, compliment, congratulate, praise, favor, approve, welcome, toast, champion, class, applaud, assess, grade, characterize, describe, evaluate, or entreat a man; to criticize, blame, or resent the behavior of his fellows; to affirm, state, answer, rejoin, report, correct, or concede something; and so on. So even the combination of a particular semantic modality with a particular semantic force is far from sufficient to determine uniquely the description of what a sentence's utterance achieves. A particular sentence with performative semantic force tends perhaps to be more restricted than a particular sentence of any other semantic force in the variety of descriptions that may be appropriate to its utterance. But

SPEECH ACTS

197

even I promise that you will be repaid next week may be uttered to voice a guarantee, assurance (even about someone else's actions), pledge, oath, or undertaking at the same time as a promise, or to recognize a debt, accept an obligation, or agree to a request; and the performative sentence I class you as the kindest person I have met here has just as wide a range of uses as has the assertoric sentence You are the kindest person I have met here. In short, every sentence of a natural language has to be supposed to have both a semantic modality and a semantic force, which combine to play a considerable part in restricting the range of speech-acts the sentence's utterance may be able to achieve This fact creates a standing temptation for philosophers to treat these speech-acts themselves, or a large sub-set of them, as being part of the subjectmatter of linguistics. Graced with the title of 'illocutionary forces' they then come to be regarded as being somehow co-ordinate with the meanings of utterances. But the meaning of a sentence's utterance on a particular occasion is determined by the meaning of the sentence in the 'langue' as disambiguated by the context, and since the meaning of the sentence in the 'langue' must be supposed to include its semantic modality and semantic force these elements are present also in the meaning, stricto sensu, of the utterance. Whatever else affects the so-called illocutionary force of the utterance stems from its context, not from the composition of the sentence uttered. Hence there is no feature whatever of the sentence in the langue that can be related to the illocutionary force of an utterance analogously to the way in which the meaning of the utterance is related to the meaning of the sentence uttered. Illocutionary force, whatever it may be, is not a topic for linguistics. Perhaps it will be objected that I have loaded the dice against the partisans of illocutionary force by including in the meaning of a sentence's utterance the contribution made by its semantic force. Perhaps it will be said that the illocutionary force of a sentence's utterance is to be regarded as a contextual disambiguation of the sentence's semantic force, just as the meaning of the utterance is determined by a contextual disambiguation of other features of the sentence. But if the proposed analogy is carefully scrutinized it will be seen to be very remote. The contextual disambiguation of the other features of a sentence normally consists either in the resolution of homonymy or syntactic ambiguity or in the fixing of demonstratival and pronominal references. In the former case a choice has to be made between alternative sentences (i.e. between one morphophonemic string under a structural description and another): in the latter case no such choice has to be made. But a sentence can be uttered again and again with the same type of illocutionary force, as with the same noun-meanings or compositional structure, whereas after a time many refential words, like now or I, cannot avoid being uttered with different references. So presumably the intended analogy would be with the resolution of homonymy or syntactic ambiguity, not with the fixing of reference. Yet, if all the different illocutionary forces with which the same sen-

198

L. JONATHAN COHEN

tence could be uttered were to be treated as the various possible disambiguations of an ambiguous string of phonemes, at least two very paradoxical conclusions would follow. In the first place we should have to treat a sentence like You are the kindest person I have met here — for the utterance of which I listed above a bare minimum of twenty-seven possible illocutionary forces — as being immensely more ambiguous than such old favorites as, say, The box was in the pen or They heard the shooting of the hunters-, and this seems gratuitously paradoxical. Also, the normal achievement in contextual disambiguation of meaning on a particular occasion is the choice of one, and just one, sentence out of a set of possible alternatives as being the sentence actually uttered on that occasion. But it must be very rare indeed for a particular utterance to have only one so-called illocutionary force. Much more typically one and the same utterance of even very short and simple sentences, like It is raining, would be both a description, say, and a report and a remark and perhaps also a grumble, even though another utterance might be the first three of these and not the fourth. So if illocutionary force were a disambiguation of semantic force it would be a very peculiar kind of disambiguation, since it would normally leave several of the alleged ambiguities unresolved. In the face of these paradoxes it seems reasonable to suppose that the so-called illocutionary force of a sentence's utterance is not a disambiguation of its semantic force. It is particularly easy to misclassify so-called 'illocutionary forces' as a topic for linguistics if the term 'performative' is allowed to slide, as it seems to have done in Austin's usage, from being an adjective applicable to components of sentences or of sentence-meanings to being an adjective applicable to utterances. The word itself does not occur in that portion of the paper (1946) in which Austin first published a discussion of performative usage. But he did discuss phrases and expressions there rather than utterances and it is in this application that Strawson attributed to him the use of the term 'performatory' (1949). Austin's later application of the term 'performative' to whole utterances, in 19558, was an unfortunate deviation that may have played a substantial part in making his theory of illocutionary forces seem plausible. For as soon as one sees that the whole of the communication achieved by saying I promise to go may also, in appropriate contexts, be achieved by saying I shall go, it is reasonable to suppose that if the former utterance is to be called performative in virtue of the promise it makes, so too must the latter be. The only difference will be that the former is explicitly performative and the latter inexplicitly so. By parity of reasoning, if a witness's utterance I state that he hit her is performative, then his statement He hit her is also performative, though again inexplicitly so. But if even statements like He hit her are performative then all normal utterances are. The old contrast between performative 8

This is the date of Austin's William James Lectures at Harvard, which were posthumously published (1962) under the title How to do things with words. Cf. also his 1956 broadcast 'Performative utterances' which has also been published posthumously (1961).

SPEECH ACTS

199

and descriptive verb-use has quite disappeared, and the classificatory value of the term 'performative' has been eroded by a typically philosophical inflation. Compare what happens with terms like mental or material. If all events turn out to be mental (or all material), a new and difficult issue arises as to what can then be meant by the term mental (or material), since within its domain of predicability no contrast is available between the mental and the non-mental (or between the material and non-material). Similarly 'performative' can now no longer serve as a classificatory term, such that some but not all members of its domain may be correctly labelled with it. No wonder then that the need is felt to introduce a new term — a term with a different domain. Inflation often leads to the introduction of a new currency. If all utterances turn out to be performative, then instead of distinguishing some utterances from others it looks as though we have to distinguish one aspect of every utterance from another. Every utterance is said to be really at least two speech-acts, not just one, and new terms are introduced, 'locutionary' and 'illocutionary', with these symbiotic acts as their domain. Finally, to round out the dualism, meaning is associated with locutionary acts and force with illocutionary ones. But all this can be avoided if we keep 'performative' as a term applicable to certain verb-uses within that dimension of meaning-description in which the contribution of a performative verb-use to a sentence's meaning may be compared or contrasted with that of a declarative, imperative, interrogative, exclamatory, or optative verb-use. There is a small, finite variety of semantic forces, and performative force is one of them. But, if a language can have infinitely many sentences, there r an infinite variety of speech-acts and perhaps even of illocutionary forces (cf. § 4, ad fin.), if any clear sense can be attached to the phrase 'illocutionary force'. It is obvious that within a linguistic theory of meaning the distinction between semantic modality and semantic force needs a very great deal more elaboration than my brief sketch of it affords.9 But my aim here has not been to construct a framework for the semantics of natural languages. I have employed the concepts of semantic modality and semantic force solely in order to show that a study of the so-called illocutionary forces of English (or French) utterances is not to be thought of as a branch of linguistics co-ordinate with English (or French) phonology, morphology, syntax, and semantics. 6. How is the Study of Speech-Acts to be Structured? For the reasons given in § 5, we do not need a new branch of linguistics to deal with the phenomena discussed by Austin. The complete linguistic description of a natural language should not be taken to include a description of the variety of • R. M. Hare's distinction between what he calls 'tropics' and 'neustics', respectively, in "Meaning and speech acts" (1970), corresponds fairly closely with my distinction here between semantic modality and semantic force.

200

L. JONATHAN COHEN

acts that we achieve with our utterances. Yet there is a way in which one part of a familiar branch of linguistics may shed a great deal of light on the problem of speech-acts. For much of this problem would be resolved by an adequate theory of the markers that are needed in a certain area of semantics, viz. in the semantics of words like promise, protest, assume, name, appoint, persuade, insinuate, etc. If we move from the material to the formal mode of speech, as it were, it becomes quite easy to integrate one important strand in the controversy into linguistics. Instead of discussing speech-acts and their empirically detectable varieties, we may well do better to discuss the terms used in describing speech-acts and survey the empirically detectable varieties of ways in which these terms function and the empirically detectable varieties of truth-conditions for applying them. And such a discussion may be conceived as a contribution to semantic theory, rather than just to the semantics of English, since, though the list of distinctive features that would result might not be complete, it would presumably be applicable, so far as it went, to the speech-act-describing vocabularies of other languages besides English. After all, the words used in describing what people do with their utterances are as much in need of semantic investigation as is any other portion of our total vocabulary, and in order to describe their meanings in the way in which an ideal dictionary would describe them we need to know the variables, or dimensions of feature, that are relevant for testing hypotheses about them. By engaging in that kind of enquiry, when we shift from the material to the formal mode of speech here, we are not in any way abandoning science in favor of speculation. Our enquiry is fundamentally inductive in its logical structure, as it is elsewhere in linguistics or experimental science (cf. Cohen and Margalit 1970, and also Cohen 1970a and b). Note too that the various different categories of speech-acts have to be distinguished in any case. This way we achieve such distinctions within an old and familiar branch of linguistics instead of having to create a new and unfamiliar one. Moreover, if we talk about the various ways of describing speech-acts rather than about the various acts co-existent in any single speech-act, we shall resist the tendency to defy Occam's razor and multiply acts more than is necessary, which was one of the four main weaknesses in the Austinian heritage. At the same time the other three main difficulties in Austin's account are also avoided. The problem of how to distinguish between meaning and illocutionary force disappears, because this force is not to be thought coordinate with meaning as a topic for linguistic study; and a proper regard can now be paid to the complexities of the facts which falsify excessively schematic theories of speech-acts. I shall therefore list here some ten of the main categories of semantically distinctive features for words and phrases used in describing what people achieve with their utterances. But I do not regard my list as being either exhaustive or irreducible. It is just a sketch of the kind of distinctive feature matrix that, if

SPEECH ACTS

201

it could ever be made adequate to its task, would introduce some degree of order and system into this field of human vocabulary. What will emerge, as is common with distinctive feature matrices in linguistics, is that hardly any two of the distinctions that need to be drawn co-incide in extension with one another or are such that one member of the pair is subordinate to the other. It is simply a mistake to try and treat some particular dimension of classification (e.g. Austin's illocutionary/perlocutionary distinction) as the fundamental one, with each of its two or three fundamental genera sub-dividing into species, and each of these in turn into sub-species, and so on. Anyone who adheres to such a tree-structure model in this field, as Austinians seem to do, is forced either to ignore the numerous counter-examples, or to impose some Procrustean misdescription on them, or to conclude that any systematic classification is impossible. First, then, there is the question whether or not the meaning of the word or phrase is such that the speech-achievement it describes can be rendered explicit by a performative use of the word. For example, warn is to be marked as performative, but alarm, insult, hint, perorate, and drink to are not. Note here that, while alarm describes a perlocutionary act in Austin's view, warn, drink to, and, presumably, perorate describe illocutionary acts and insult and hint describe neither. {Drink to is not performative10 because it describes what you do when your drinking takes place after, not while, you say I drink to you. Utterance of I drink to you declares how a certain other act is to be understood, and that act is an act of drinking, not speaking. You cannot quench your thirst with speech.) Secondly, there is the question whether or not the meaning of the word or phrase is such as to imply that the speech-achievement it describes is intended. For example, entreat and blackmail are to be marked as intentional, but assume and embarrass are not: assumptions may be made, and embarrassment created, either intentionally or unintentionally. Note here that while entreat and assume are performative, blackmail and embarrass are not. So the performative/non-performative distinction cuts right across the intentional/non-intentional one. Thirdly, there is the question whether or not the meaning of the word or phrase is such that it implies the utterance it describes should not be taken literally. For example, joke, pretend, show o f f , speak ironically, tell an allegory, and talk in parables, should be marked as not literal in contrast with bore, insult, warn, or name. But note that, while to describe a man as joking or pretending is to imply that the joke or pretence, respectively, was intentional, both irony and allegory can be either intentional or unintentional. Fourthly, there is the question whether or not the meaning of the word or phrase is such that it can introduce indirect discourse, or more precisely (since there is no reason why our matrix should be confined to binary distinctions) the question whether it typically introduces declarative, interrogative, imperative, or optative indirect discourse, or no indirect discourse at all, like report, enquire, 18

Nevertheless Austin lists it as an illocutionary verb (1962 : 159).

202

L. JONATHAN COHEN

order, hope, and resign, respectively. Note that this distinction cuts across each of the three preceding ones. Report and resign are both performative, hope and bore are not. Entreat and blackmail are both intentional, assume and oppose are non-intentional. Warn and resign are both literal, pretend and speak ironically are not. Fifthly, there is the question whether or not the meaning of the word or phrase is such as to imply that the speech-achievement it describes is regulated by extralinguistic rules of some kind. For example, deduce implies the relevance of logical principles, excommunicate implies the relevance of religious canon, command implies the relevance of a hierarchy of authority, bequeath implies the relevance of laws of inheritance, and so on. But perorate, remark, beg, and take it do not imply anything of the kind. Note here that while deduce or remark introduces declarative indirect speech and command or beg introduces imperative indirect speech, no indirect speech at all is introduced by excommunicate or perorate. So this fifth distinction at least cuts across the fourth.11 Sixthly, there is the question whether or not the meaning of the word or phrase is such as to imply that the utterance has some causal effect on the course of events or more precisely the question whether the only effect implied is the securing of uptake, or whether some alternative or additional effect is implied or whether no effect at all is implied. For example, warn (in one sense), persuade, and pray differ from one another in this respect. But, whereas each member of this trio can introduce indirect discourse, it is also easy to find another trio of which no member can do this, e.g. greet (in one sense), embarrass, and resign. Again sometimes the effect achieved must be intended, as with blackmail, and sometimes it need not be, as with embarrass. Seventhly, it is possible to distinguish between those words that imply the speechact feature they describe to be covert and those that do not, like hint, insinuate, imply, convey, deal in innuendo, or talk guardedly, on the one hand, and praise, express feelings, command, or persuade, on the other; and this distinction obviously cuts across the distinction between those words that can introduce indirect discourse and those that cannot. Also, while convey and persuade imply an effect, imply and praise do not. Eigthly, one can distinguish between describing what is achieved in virtue of the content of an utterance, as with praise, imply, or lie, and what is achieved in virtue of its non-rule-governed choice of vocabulary, style, accent, etc., as with pun, mimic, blaspheme, or parody. Either type of achievement can be covert (imply, encode) or overt (praise, blaspheme). Ninthly, there is the distinction, in Austin's terminology, between verdictives, 11

Thus M. Furberg is wrong to think that his distinction between what he calls illocutionary acts and archetypal performatives, respectively, coincides with the distinction between utterances that do involve indirect discourse and utterances that do not; instead, the two distinctions cut across one another (Furberg 1969 : 459).

203

SPEECH ACTS

exercitives, commissives, behabitives, and expositives, like acquit, appoint, promise, greet, and postulate, respectively. But this again cuts across other distinctions. For example, pace Austin, it cuts across the performative/non-performative distinction, because though greet and postulate are both performative, both the behabitive drink with and the expositive perorate are not. Similarly, while neither the exercitive excommunicate nor the commissive champion can introduce indirect discourse, both the exercitive pray and the commissive guarantee can do so. Tenthly, there is the question whether or not the meaning of the word or phrase is such as primarily to classify the utterance's achievement or to evaluate it. For example, while assert, question, warn, name, persuade, etc., are classificatory, prove, demonstrate, invalidate, refute, misname, lie, quibble, perjure, equivocate, blaspheme, misdescribe, misrepresent, mislead, misjudge, beg the question, talk nonsense, argue invalidly, declare inappositely, comment irrelevantly, speak insincerely, etc., are evaluative. Note that this distinction too cuts across a number of others, contrary to what is suggested by Austin's confining his discussion of what he calls 'felicity' and 'infelicity' to illocutionary acts, i.e. to convention-bound acts that are capable of explicitly performative execution (1962 : 135ff.). Some members of both groups imply intention (e.g. entreat, lie), while some do not (e.g. assume, misdescribe). Some members of both groups introduce indirect discourse (e.g. argue, prove), while some do not (speak, lie). Some members of both groups imply a causal effect (e.g. persuade, mislead), while some do not (e.g. argue, prove). And members of both groups are verdictives (e.g. grade, misjudge), exercitives (e.g. name, misname), commissives (e.g. swear, perjure oneself), behabitives (e.g. curse, blaspheme), and expositives (e.g. argue, prove). No doubt there are many other distinctions that also need to be drawn in this field of vocabulary. But the various ways in which the ten above-mentioned ones intersect with one another will at least suffice to show the inapplicability of schematic dichotomies or trichotomies in the field, like Austin's illocutionary/perlocutionary contrast. There is no Aristotelian tree-structure here. 7. Some Remaining

Problems

There are obviously some questions about what we achieve with speech-acts that are not answerable from the information provided by a distinctive feature matrix of this kind. The semantics of terms used for describing such achievements cannot answer every general question that these achievements raise. In particular it cannot answer factual questions about the occasions on which speech-acts of a specified type are performed in this or that community or about the nonlinguistic rules that may regulate them. It cannot tell us, for instance, what precise speech-act of acceptance a bride must perform in the course of an Anglican wedding ceremony nor what precise speech-act of repudiation a husband must perform in the course of a Moslem divorce. These are matters for civil or religious jurisprudence or for some

204

L. JONATHAN COHEN

form of anthropology or sociology. Similarly if we are interested in the rhetorical or psychological, rather than the legal or religious, effects of certain types of speechacts, we shall have to turn to the appropriate factual study of speech-behavior. Also, some terms for describing speech-acts are rather rich in deductive liaisons and constitute a territory for logical, or quasi-logical, exploration. For example, the verb imply may be used in describing utterances, both in virtue of relations of logical implication or entailment that hold between the contents of actual or possible utterances and also in virtue of what H. P. Grice has called the 'implicatures' of speech-acts, as when if I say It's raining I imply that I have reason to believe that it is raining. We obviously need a theory of such implicatures and their connection with the standing presumptions of human speech, whether this theory be built on the foundation proposed by H. P. Grice (1961),12 or on foundations like those suggested by J. Hintikka's work on the concepts of knowledge and belief (1962) or by N. Rescher's (1968 : 250 ff.) and L. Apostel's (1972) work on the concept of assertion. By coupling together, as Apostel suggests, assertion logics, action logics, tense logics, erotetic logics, and deontic logics of various kinds it may become possible to define several other speech-act terms and to explore their deductive liaisons. Or at least it may become possible to explore the consequences of classifying a speech-act term as a verdictive, exercitive, commissive, behabitive, or expositive, and to give each of these semantic markers a rather more precise sense than Austin gave them. But, though this is an interesting and potentially fruitful field for research, such an approach to the subject should be viewed as complementary or ancillary to the distinctive-feature-matrix methodology rather than as an alternative to it. For certainly there are some important dimensions of speech-act description that seem not to lend themselves so easily or so profitably to formalization, e.g. the performative/non-performative distinction,13 or the literal/non-literal one. No doubt some kind of axiomatized formulation of these concepts is possible. But I am inclined to suspect that such a formulation could not be achieved without introducing a good deal of ad hoc machinery, so that it would be relatively uninteresting. Finally, there are various philosophical questions the answers to which seem to invoke an account of speech-acts at one stage or another. For example, there are the three issues described above (§ 4), which Searle calls 'the naturalistic fallacy', 'the speech-act fallacy', and 'the assertion fallacy', respectively. Of these alleged fallacies, however, as I have argued elsewhere (Cohen 1970c), the first 12

Grice has claimed that his theory of implicatures confirms the truth-functionality of the logical particles of natural language. But this is a mistake (Cohen 1972). In the present article I do not concern myself at all with any of the general problems that arise about the classification of semantic features: compare, for example, Bierwisch (1971). 13 Lemmon (1962) argued that a performative sentence was merely a declarative one that was verifiable, in certain circumstances, by its own utterance. But if you are verifiably promising, thanking, naming, etc., there is no point in your also declaring at the same time that you are. Hence Lemmon's interpretation of performative sentences was rather an implausible one, and to that extent a formalization of performatives that is based on Lemmon's theory, like Aqvist (1971), is not relevant here.

SPEECH ACTS

205

is not a fallacy at all and has nothing in particular to do with speech-acts, if what it claims is just that no ethical proposition can be deduced from non-selfcontradictory premises that do not include at least one ethical proposition. And what Searle calls 'the assertion-fallacy' is easily elucidated in terms of my proposed distinctive-feature-matrix methodology for dealing with speech-acts. In Searle's view this fallacy occurs when conditions for the illocutionary act of asserting a sentence in which such words as known, voluntary, etc., occur have been confused with truth-conditions for the predication of these words. In my view the fallacy occurs when features that would be mentioned in the semantics for an actual or possible word describing a certain type of speech-act (e.g. the act of claiming knowledge), have been mistakenly read into the semantics for a quite different word (e.g. know) which might be uttered, in a grammatically predicative role, as part of a performance of that speech-act. In regard to what Searle calls 'the speech-act fallacy', however, the situation is somewhat more complicated. The name has been given by Searle, it will be remembered, to the thesis that the meanings of certain words like good, true, etc., can be given by an account of certain speech-acts of commending, endorsing, etc., that they are supposed to be primarily used to perform. Searle's argument is that these meanings cannot be so given because the words often occur with demonstrably the same meanings in contexts where no such speech-acts are at issue (e.g. in the antecedents of conditional sentences like / / this is good, I'll buy it). Hare's rejoinder (1970:23-4) to Searle is to distinguish between what he calls 'tropics' and 'neustics', which are his names for expressions that signify what I have called (§5) the 'modality' and 'semantic force' respectively, of a sentence or clause. The neustic of a sentence disappears when the sentence is embedded as the antecedent of a conditional sentence, but its tropic remains. Hence in Hare's view it is possible for the meaning of good to be analyzed in terms that include a mention of other tropics besides the indicative, even when good occurs in the antecedent of a conditional sentence. But Hare's rejoinder to Searle is not fully cogent unless two further points are established. It has to be demonstrated first that at least one non-indicative tropic is implicit in good, and secondly, that the modality signified by this non-indicative tropic, or complex of non-indicative tropics, is best elucidated by reference to the speech-act of commending. It would probably be wise to keep an open mind about the possibility or impossibility of demonstrating the first of these two points. The semantics of natural languages is as yet far too undeveloped for any certainty about such an issue to be reasonable. But there is a reason for doubting whether any particular modality or complex of modalities is likely to be best elucidated by reference to a single speech-act. For normally, as we have already seen, neither semantic modality nor semantic force, nor yet a combination of the two, suffices to determine uniquely the correct contextual description of a sentence's utterance. More specifically, the sentence This is good can be uttered to accept, endorse, authenticate, recommend, praise, welcome, applaud, and so on, as well as to commend.

206

L. JONATHAN

COHEN

Perhaps it will be claimed instead that, though the meaning of a word may never be elucidatable by reference to some uniquely correct speech-act description for any utterance of a standard sentence in which the word occurs, nevertheless it may be elucidatable by reference to a disjunction of such descriptions.

After

all, if the semantic modality and semantic force, for example, of a sentence do in fact restrict the range of so-called illocutionary acts performable by uttering the sentence, it may well seem possible that a reference to such a restriction would suffice to characterize some aspect of the sentence's meaning. M o r e precisely, it has been argued by W . P. Alston (1964:36) that 'sameness of illocutionary-act potential is what constitutes sameness of meaning for sentences'. But it is obvious that this thesis encounters serious difficulties if such activities as analyzing, or deducing a non-trivial conclusion from, are allowed to count as illocutionary acts. The sentence He is a cousin of mine would normally be attributed the same meaning as One of his two pairs of grandparents is or was the same as one of my two pairs of grandparents, though neither of his parents is or was a parent of

mine.

But it would be true of the latter sentence to say that its illocutionary-act potential includes the act of analyzing the former sentence, though not itself, and true of the former sentence to say that its illocutionary-act potential includes the act of analyzing the latter sentence, though not itself.

So here are two

apparently synonymous sentences that definitely do not share the same illocutionary-act potential.

No

doubt the concept of

illocutionary-act

potential

could be restricted here in some way that would exclude such acts as analyzing or deducing a non-trivial conclusion.

But it is extremely difficult to see how

the requisite restriction could be stated in sufficiently general, catch-all terms without employing the concept of meaning in one form or another, since what one would want to exclude would be any acts involving exposition of meaning. So it looks as though this attempt to escape from problems about meaning into some presumedly unproblematic domain of illocutionary acts only leads us back again to the same problems about meaning. It may be rejoined that, though the concept of sentence-meaning cannot itself be elucidated by reference to the range of possible speech-act descriptions for utterances of a sentence, nevertheless some information is certainly provided about the meaning of such a sentence as This is good by listing a substantial number of the speech-act descriptions that might be appropriate and a substantial number that might not. This is true. But it is a somewhat indirect way 14 of getting at the meaning of a sentence, and seems to rely very much on a person's ability to jump to the right conclusion about what meaning a sentence could have that was equally A more direct way, where possible, would be to offer a synonymous sentence. (There are obviously some words and idioms about which the most revealing semantic hypothesis is the one that describes their normally achieved utterance, e.g. an expression of farewell like Good-bye or a call f o r attention like Hullo. But, as such, these words or idioms are functioning as whole sentences rather than as components of sentences - which is the normal role f o r good, ought, etc.) 14

207

SPEECH ACTS

well capable of performing such-or-such different acts. It gives clues to the solution of a puzzle about some word or sentence's meaning rather than an actual solution to the puzzle.

REFERENCES

ALSTON, W. P. 1964. Philosophy of language. Englewood Cliffs, N J . ANDERSON, S. 1970. On the linguistic status of the performative/constative distinction. Mathematical Linguistics and Automatic Translation Report No. NSF-26 to the National Science Foundation, 1-37. Cambridge, Mass. APOSTEL, L. 1972. Elocutionary forces and the logic of change. Mind 81.208-24. AQUIST, LENNART. 1971. Performatives and verifiability by the use of language. Brown University, Rhode Island, and University of Uppsala, Sweden. Mimeo. AUSTIN, J. L. 1946. Other minds. Proceedings of the Aristotelian Society, supp. series 20.169-75. . 1950. Truth. Proceedings of the Aristotelian Society, supp. series 24.111-28. . 1961. Philosophical papers. Oxford. . 1962. How to do things with words, ed. by J. O. Urmson. Oxford. AYER, A. J. 1946. Language, truth and logic. London. BAKER, C. L. 1970. Notes on the description of English questions: The role of an abstract question morpheme. FL 6.197-219. BIERWISCH, MANFRED. 1971. On classifying semantic features. Semantics, ed. by D. D. Steinberg and L. A. Jakobovits, 410-35. Cambridge, England. CARNAP, R. 1935. Philosophy and logical syntax. London. . 1937. The logical syntax of language. London. COHEN, L. JONATHAN. 1950. Mr. Strawson's analysis of truth. Analyisis 10.136-40. . 1964. Do illocutionary forces exist? Philosophical Quarterly 14.118-37. Reprinted 1969 in Symposium on J. L. Austin, ed. by K. T. Fann, 420-44. London. . 1970a. The implications of induction. London. . 1970b. Some applications of inductive logic to the theory of language. American Philosophical Quarterly 7. . 1970c. Searle's theory of speech acts. Philosophical Review 79.545-57. . 1972a. Some remarks on Grice's views about the logical particles of natural language. Pragmatics of natural languages, ed. by Y. Bar-Hillel. Dordrecht, Holland. . 1973. The non-existence of illocutionary forces: A reply to Mr. Burch. Ratio 15.125-31. COHEN, L. JONATHAN, and AVISHAI MARGALIT.

1970.

T h e role of inductive rea-

soning in the interpretation of metaphor. Synthese 11.220-38. FRASER, B. 1970. A reply to 'On declarative sentences'. Mathematical Linguistics

208

L. JONATHAN COHEN

and Automatic Translation Report No. NSF-24 to the National Science Foundation, 305-22. Cambridge, Mass. j FURBERG, M. 1969. Meaning and illocutionary force. Symposium on J. L. Aus tin, ed. by K. T. Fann, 445-67. London. GEACH, P. T. 1960 Ascriptivism. Philosophical Review 69.221-5. GEACH, P. T . , and M . BLACK, eds. 1 9 5 2 . Translations from the philosophical writings of Gottlob Frege. Oxford. GRICE, H. P. 1957. Meaning. Philosophical Review 67.377-88. . 1961. The causal theory of perception. Proceedings of the Aristotelian Society, supp. series 35.123-41. HARE, R. M. 1952. The language of morals. Oxford. . 1970. Meaning and speech acts. Philosophical Review 79.3-24. HINTIKKA, J. 1 9 6 2 . Knowledge and belief. Ithaca, N . Y . LAKOFF, G. 1971. On generative semantics. Semantics — an interdisciplinary reader in philosophy, linguistics, anthropology and psychology, ed. by D. D. Steinberg and L. A. Jakobovits, 232-96. Cambridge, Eng. LAKOFF, R . 1 9 6 8 . Abstract syntax and Latin complementation. Cambridge, Mass. LEMMON, E . J. 1 9 6 2 . On sentences verifiable by their use. Analysis 2 2 . 8 6 - 9 . LEWIS, D. K. 1969. Convention: A philosophical study. Cambridge, Mass. RESCHER, N. 1968. Topics in philosophical logic. Dordrecht. Ross, J. R. 1970. On declarative sentences. Readings in English transformational grammar, ed. by R. A. Jacobs and P. S. Rosenbaum, 222-72. Waltham, Mass. Ross, W. D. 1939. Foundations of ethics. Oxford. SEARLE, . R. 1 9 6 2 . Meaning and speech acts. Philosophical Review 7 1 . 4 2 3 . 1969. Speech acts. Cambridge. STRAWSON, P. F . 1949. Truth. Analysis 9.83-97. . 1952. Introduction to logical theory. London. . 1964. Intention and convention in speech acts. Philosophical Review 73.439-60. Reprinted 1969 in Symposium on J. L. Austin, ed. by K. T. Fann, 380-400. London. WITTGENSTEIN, L . 1 9 5 3 . Philosophical investigations. Oxford.

PART TWO

SEMIOTICS

SEMIOTICS: A S U R V E Y OF T H E STATE OF T H E A R T THOMAS A. SEBEOK

INTRODUCTORY NOTE

The contents of this article represent a drastically condensed version of my forthcoming book on Semiotics (Sebeok, forthcoming). Specifically, three of the chapters have been omitted here altogether: Ch. 3, on Anthroposemiotics, Ch. 4, on Zoosemiotics, and Ch. 6, on Pedagogic implications and applications. Further, the third and concluding section of Ch. 5, Toward a classification of signs and sign systems — dealing with their evolution (particularly in light of the ethological concept of ritualization), as well as their development in ontogeny — has also been left out. Some of these topics are treated in various detail, throughout these pages, by specialists far more qualified than I am. My own views of what I have called 'anthroposemiotics' — briefly, the congeries of man's species-specific signaling systems — are outlined elsewhere (Sebeok 1970); a partial listing of my publications on 'Zoosemiotics' has appeared in Sebeok and Ramsay (1969: 227-228), and this is a subject to which I am committed to return in the near future (cf. Sebeok 1972). 1. THE SCOPE OF SEMIOTIC

The subject matter of semiotics1 — ultimately a mode of extending our perception of the world — is the exchange of any messages whatever and of the systems of signs which underlie them; hence, alongside such fields as social anthropology 1 Throughout this paper, semiotics will be used as my preferred synonym among the family of terms to which, among still others, semeiotic, semeiotics, semiotic, semiology, (rarely, especially in Lady Welby's treatment) signifies, and also, sometimes confusingly, semantics (properly a cover term for both the theory of meaning and the theory of reference), belong in the English usage of various authors at various times. (See further, fn. 33, below.) There are strong scholarly predilections in this matter, variously rationalized; for example, Earl W. Count (1969:76, fn. 2) defends his choice of semeiotics by arguing that 'The spelling is better etymology than semiotics, and it avoids the ambiguity of semi-. Semi-otics would be nonsense . . C e r t a i n authors (e.g., Louis Hjelmslev, and currently Julia Kristeva) use more than one of these terms, with different meanings. Cf. the general discussion of the terminological quandary (in Italian, but by and large applicable to French and, mutatis mutandis, to English) in Eco 1968 (383-87). I have recently devoted a paper to a detailed exploration of this subject, under the title, " 'Semiotics' and its congeners", to appear in a Festschrift for Archibald A. Hill.

212

THOMAS A. SEBEOK

(which deals with the exchange of mates), and economics (which deals with the exchange of utilities, viz., goods and services), semiotics is most commonly regarded as a branch of the communication disciplines.2 Some investigators prefer to emphasize the study of the systematic — rather than the transactional — aspects of the repertory of signs (noumena vs. phenomena).8 Whichever approach one may favor, however, the key concept of semiotics remains always the 'sign'. Messages may be emitted and/or received either by inorganic objects, such as machines;4 or by organic substances, for instance, animals, including man, or by some of their component parts (e.g., ribonucleic acid, mRNA, that serves as an information-bearing tape 'read' by particles, called ribosomes, that travel along it, carrying amino acid sequence information (leas 1969: 8)); one may also speak of information, for instance, in cardiovascular functioning, where messages are conveyed from peripheral vessels to the brain, relayed thence to the heart and back to the brain (Adey 1967: 21). The interaction of organic beings with inorganic things (such as communication between a man and a computer) can also be treated as a semiotic problem. However, this survey will, in the main, be restricted to the semiotic manifestations of whole living systems. A further distinction could provisionally be drawn between terrestrial and extraterrestrial communication. Exploration of the latter is shared by fields such as mathematics, exobiology, and radioastronomy with science-fiction.6 Since, at present, the study of cosmic message exchanges remains a discipline without a subject matter, this survey must be further confined to semiotic behavior on earth. Message interchanges are most conveniently — if anthropocentrically — treated in two vast domains: in man; and in other living systems; (also, of course, in their interactions, for instance, as regards semiotic processes that enter into the training, 2

Cf. Lévi-Strauss (1958:326): 'Dans toute société, la communication s'opère au moins à trois niveaux: communication des femmes; communications des biens et des services; communication des messages. Par conséquent, l'étude du système de parenté, celle du système économique et celle du système linguistique offrent certaines analogies. Toutes trois relèvent de la même méthode; elles diffèrent seulement par le niveau stratégique où chacune choisit de se situer au sein d'un univers commun.' 3 This differential emphasis between empirical and analytic approaches is discussed by Valesio (1970). Prieto also notes that there are two tendencies in contemporary semiotic researches: one emphasizing communication, the other signification (his own work following the former); see his 1968 article, and his four contributions cited ibid., p. 144. Marxist philosophers naturally stress the communication process as the starting point in the analysis of the sign; cf. i.a., Schaff 1962:155-211; and Resnikow 1968:38. 4 Cf. Wiener 1950:85-88: ' . . . it is quite possible for a person to talk to a machine, a machine to a person, and a machine to a machine . . . the new machines will not stop working merely because we have discontinued to give them human support'. For technical details, see, e.g. Gorn 1968. s On extraterrestrial semiotic processes as a scientific problem, see the summary of the literature in David Kahn 1967, Ch. 26 and p. 1130. On semiotic processes in science-fiction in general, see Krueger 1968; a particular example, by a well-known linguist (whom Krueger failed to mention), is Hockett's "How to learn Martian" (1955). A n Institute for Extraterrestrial Linguistics exists, since 1969, in Heidelberg, as reported in Linguistische Berichte, N o . 4, p. 88 (1969).

SEMIOTICS: A SURVEY

213

taming, and domestication of animals«). By 'other living systems', usually are meant animals other than man, but plants and fungi can also be said to be the sources and/or destinations of messages, notably in their interrelations with certain animals7 further, the biological status of such eminently communicative Protozoa as cellular slime molds (Acrasiales) is still in doubt (Bonner 1963; see also Bleibtreu 1968: Ch. 7). Anthroposemiotics, that is, the totality of man's species-specific signaling systems, was the first domain concretely envisaged and delineated, under the designation semiotic. For most investigators, from 1960 to this day, both notions still remain synonymous. The second domain, zoosemiotics, which encompasses the study of animal communications in the broadest sense, was named and comprehensively outlined only in 1963. It would now seem more accurate to consider anthroposemiotics and zoosemiotics, separately and conjointly, as two principal divisions of semiotics, having in common certain essential features but differing especially as to the fundamental and pervasive role that language plays in the former in contradistinction to the latter. A third domain, endosemiotics — which studies cybernetic systems within the body — will not be examined in detail on this occasion; it is enough to say now that, in this field, the genetic code plays a role comparable to that of the verbal code in anthroposemiotic affairs (for reasons specified in Masters 1970), 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 (G. Bateson 1970); the field of transducer physiology, which studies the conversion of outside signals to their initial output inside, is, as yet, a totally undeveloped science. 2. HISTORICAL OUTLINE OF THE FIELD

The discipline of semiotics — referring in earliest usage to medical concerns with the sensible indications of changes in the condition of the human body — constituted one of the three branches of Greek medicine. For the Stoics, the term acquired a broader meaning as a basic division of philosophy, including logic and 9

Cf., e.g., Hediger 1968, esp. Chs. 8 and 9, and Zeuner 1963. Hediger, p. 120, for example, writes: Training a circus animal means teaching it, by specialized handling and continuous use of effective cues, until it performs certain actions at a special personal signal. These actions, in their elements, are well known to the animal, but in freedom they would never be caused by the same stimuli, and never performed under the same conditions.' Again, p. 128, he writes: 'During the course of training the animal learns a lot of human signals and expressions and, conversely, the trainer develops an understanding of the animal's expression and signals.' In semiotic terms, then, to tame and train an animal is to come to share, at least in part, its repertory of signs, and vice versa; in domestication, the diachronic dimension is superimposed. 7 One particularly interesting community of interacting plants and animals involves milkweeds belonging to the family of Asclepiadaceae, a group of insects, the Danaidae, and blue jays, in a complex and subtle semiotic ecosystem; cf. Brower, Brower, and Corvino 1967. 8 For a partial sketch of the history of semiotics, cf. the Appendix (pp. 285-310) to Morris 1946. A full history of this field is yet to be written.

214

THOMAS A. SEBEOK

the theory of knowledge. Hellenistic philosophy — not only among the Stoics, but also in its Epicurean and the Skeptic varieties — centered around a theory of signs, and was anticipated by Aristotle, who, in turn, was influenced by Plato, the Sophists, and several Greek physicians (such as Asclepiades, Erasistratus, Herophilus, all mentioned by Sextus9). In the Middle Ages,10 a comprehensive and very keen theory of signs (known as scientia sermocinalis), embracing grammar, logic, and rhetoric, was elaborated by a number of scholars. 'Within this development two streams can be distinguished: the dominant tendency was to interpret signprocesses within the framework of the Platonic and Aristotelian metaphysics; this was countered by a progressively growing attempt to assimilate semiotic to empirical science and philosophy. The first direction was carried on by Leibniz, the second by the British empiricists' (Morris 1946: 286). Leibniz gave his main attention in semiotics to the syntactical study of signstructures. He regarded even the calculus as a part of an all-embracing theory of signs and as an illustration of the fruitfulness of semiotic. His ideas of a universal system of signs (the characteristica universalis) have been variously carried forward by symbolic logicians and others such as Boole, Carnap, Frege, Gomperz, Husserl, Lambert, Peano, Russell, Tarski, Whitehead, and notably Peirce.11 The British empiricists (Francis Bacon, Bentham, Berkeley, Hobbes, Hume, and notably Locke), on the other hand, were variously concerned with the semantic dimension primarily. The Stoic term semiotic was reintroduced, in 1690, into English philosophical discourse by John Locke, in his Essay concerning humane understanding. Locke declared the 'doctrine of signs' as that branch of his tripartite division of all sciences 'the business whereof is to consider the nature of signs, the mind makes use of for the understanding of things, or conveying its knowledge to others'. For communication and for recording of our thoughts, 'signs of our ideas are . . . necessary: those which men have found most convenient, and therefore generally make use of, are articulate sounds. The consideration, then, of ideas and words as the great instruments of knowledge, makes no despicable part of their contemplation who would take a view of human knowledge in the whole extent of it. And perhaps if they were distinctly weighed, and duly considered, they would afford us another sort of logic and critic, than we have been hitherto acquainted with' (Locke 1690: Bk. IV, Ch. XXI, §4). The preceding passage contains lingering echoes of the introductory paragraphs of Antoine Arnauld's and Claude Lancelot's Grammaire générale et raisonnée, first 8

Cf. the account, summarized after Sextus Empiricus, of "Aenesidemus' theory of signs", constituting Appendix C (pp. 266-268) of Ogden and Richards 1938. On sign theories of the Stoics and Epicureans, see Stough 1969:102-104, 125-128. 10 On medieval linguistic study, see Bursill-Hall 1971. 11 On Peirce, see below; on Husserl, see Ogden and Richards 1938:269-272, and Schaff 1962: 162-173; on Russell, Schaff 1962:273; on Frege, Schaff 1962:273-274; on Gomperz, Schaff 1962: 274-277. Karl Söder's 1964 dissertation, Beiträge J. H. Lamberts zur formalen Logik und Semiotik, is known to me only through Albrecht 1967:68-69, 253 (fn. 25).

SEMIOTICS: A SURVEY

215

published in 1660. In a third, revised edition, that appeared in 1676, the Port-Royal grammarians had written: 'Parler, est expliquer ses pensées par des signes, que les hommes ont inventez à ce dessein. On a trouvé que les plus commodes des ces signes, estoient les sons et les voix . . . . on a inventé d'autres signes pour les rendre durables et visibles, qui sont les caractères de l'ecriture . . . Ainsi l'on peut considérer deux choses dans ces signes: La premiere; ce qu'ils sont par leur nature, c'est à dire, en tant que sons et caractères. La seconde, leur signification: c'est à dire, la maniéré dons les hommes s'en servent pour signifier leurs pensées' (Brekle 1966: 5). The pertinence and import of this capital workshop for the history of semiotics — not only of the Port-Royal Grammar, but also and especially of Arnauld's and Pierre Nicole's Logic (1662) — are only now coming to be appreciated (cf. Brekle 1964; and see fn. 86, below). The real founder and first systematic investigator of modern semiotic, however, was the subtle American philosopher, Charles Sanders Peirce (1839-1914).12 Convinced that many passages of Locke's Essay 'make the first steps in profound analyses which are not further developed' (II, 649), Peirce took over from him the term semiotic with its definition as the 'doctrine of signs', and devoted a 'lifelong study' to the nature of signs, that is, 'the doctrine of the essential nature and fundamental varieties of possible semiosis'. He viewed himself as 'a pioneer, or rather a backwoodsman, in the work of clearing and opening up . . . semiotic', and added: 'I find the field too vast, the labour too great, for a first-comer. I am, accordingly, obliged to confine myself to the most important questions' (Peirce V, 488). Peirce explicitly connected sign-processes with processes involving mediation or 'thirdness', as in the following passage: 'It is important to understand what I mean by semiosis. All dynamical action, or action of brute force, physical or psychical, either takes place between two subjects . . . or at any rate is a resultant of such actions between pairs. But by "semiosis" I mean, on the contrary, an action, or influence, which is, or involves, a cooperation of three subjects, such as a sign, its object, and its interprétant, this tri-relative influence not being in any way resolvable into actions between pairs . . . . my definition confers on anything that so acts the title of a "sign" ' (Peirce V, 484). Because of his idealism, the complexities of his argument, and especially the fact that his terminology is not constant — 'raising doubts as to whether or not he has changed his thesis with his terms' (Weiss and Burks 1945) — Peirce's writings are somewhat hard to understand. This difficulty has given impetus to an evergrowing secondary exegetical literature.13 12 On Peirce's semiotics in historical perspective, see also Morris 1946:287-291; and Ogden and Richards 1938:279-290, where pertinent extracts from his correspondence with Lady Welby are also included. On the latter, see also Lieb 1953. For Peirce's review of Lady Welby's What is meaning? (1903), see Peirce VIII, 171-175. (In citing Peirce's Collected Papers, Roman numerals refer to volumes, Arabic to paragraphs.) 15 Cf., i.a., Fitzgerald's selected bibliography in his 1966 book.

216

THOMAS A. SEBEOK

In the semiotic domain, it was basically his contention that 'the entire universe . . . is perfused with signs, if it is not composed exclusively of signs' (Peirce V, 448n), and that every 'thought . . . is in itself essentially of the nature of a sign' (ibid., 594) or, at least, that thinking never occurs without the presence of some feeling, image, conception, or other representation which serves as a sign14). In his view, furthermore, 'a law is in itself nothing but a general formula or symbol' {ibid.., 107), and man himself 'is the thought', or, in other words, a human being 'is a sign himself' (ibid., 314; VI, 344). A sign is to be understood only 'in terms of some other more comprehensive and rational sign which welded the totality of things and beliefs together into a fixed and perfect whole' (Weiss 1940), and, 'Finally, as what anything really is, is what it may finally come to be known to be in the ideal state of complete information, so that reality depends on the ultimate decision of the community; so thought is what it is, only by virtue of its addressing a future thought which is in its value as thought identical with it, though more developed' (Peirce V, 316). Peirce's ultimate classification (c.1906) of signs into sixty-six different kinds developed slowly but persistently over some forty years. It was in 1867, in his paper "On a new list of categories", that he first stated his now famous fundamental triad that — with intermediate and hybrid forms determined subsequently (cf. Wells 1967, esp. #2a) — has proved to be of such high utility in several recent studies of both human and animal communication (for anthroposemiotic applications, see Jakobson 1964, 1965, 1967; for zoosemiotic applications, Sebeok 1967c). Peirce initially asserted that there were three kinds of signs (or, as he then called them, 'representations'): (1) likenesses (a term soon abandoned in favor of icons), or 'those whose relation to their objects is a mere community in some quality', the Platonic notion that the signifier 'imitates' the signified; (2) indices, or 'those whose relation to their objects consists in a correspondence in fact', that is, pointing, ostension, or deixis as a mode of signification, as the footprint that Crusoe found in the sand 'was an Index to him of some creature'; and (3) symbols (which are the same as general signs), or 'those the ground of whose relation to their objects is an imputed character', which he later called 'laws', meaning conventions, habits, or natural dispositions of its interprétant or of the field of its interprétant.15 Some three and a half decades afterwards, he incorporated this trichotomy (B) into a broader scheme,16 yielding the following sets of sign relations (Table 1): 14

Peirce V, 283: 'When we think . . . we ourselves, as we are at that moment, appear as a sign.' By 'interprétant' Peirce appeared to mean the interpreter's reaction to the sign, that is, the sign's transmutation into an inferred neural code. On information processing in the nervous system in general, see the Proceedings of the International Union of Physiological Sciences, Vol. 3 (see Bibliography). More particularly in regard to the mechanisms of speech, see Brain 1961; Luria 1967 and in this volume. 16 On the three trichotomies yielding ten classes of signs, plus what Peirce called 'degenerate' signs, see Peirce II, 233-273. This is also discussed by Bense, 1965, and further elaborated in his 1967 book. 15

SEMIOTICS: A SURVEY

217

TABLE 1

(2) Sinsign 1

(3) Legisign

(1) Qualisign '

(A) Signs in themselves (3) Symbol ' (2) Index - • (1) Icon

-(B) Signs in relation to objects

(C) Signs interpreted to represent

(3) Argument (2) Dicent (1) Rheme

It is fair to remark that the more Peirce elaborated and refined his scheme," the more it became a challenge to his successors to devise applications to particular semiotic systems, most of all to the pattern of language. One notably useful distinction in his scheme, however, was that the very same word could stand for different varieties of (A) signs in themselves: As it is in itself, a sign is either of the nature of an appearance, when I call it a qualisign-, or secondly, it is an individual object or event, when I call it a sinsign . . . ; or thirdly, it is of the nature of a general type, when I call it a legisign. As we use the term 'word' in most cases, saying that 'the' is one 'word' and 'an' is a second 'word', a 'word' is a legisign. But when we say of a page in a book, that it has 250 'words' upon it, of which twenty are 'the's, the 'word' is a sinsign. A sinsign so embodying a legisign, I term a 'replica' of the legisign. The difference between a legisign and a qualisign, neither of which is an individual thing, is that a legisign has a definite identity, though usually admitting a great variety of appearances. Thus, &, and, and the sound are all one word. The qualisign, on the other hand, has no identity. It is the mere quality of an appearance, and is not exactly the same throughout a second. Instead of identity it has great similarity, and cannot differ much without being called quite a different qualisign.18 17 Where Rheme is defined as 'a Sign which, for its Interprétant, is a Sign of qualitative Possibility, that is, is understood as representing such and such a kind of possible Object', that is, roughly, a term or prepositional function; a Dicent is 'a Sign, which, for its Interprétant, is a Sign of actual existence', or roughly, a proposition-, and an Argument 'is a Sign which, for its Interprétant, is a Sign of law'. See Peirce II, 250-252. The English philologist J. P. Postgate, in his Inaugural Address on "The science of meaning", unsuccessfully attempted to introduce rheme as a technical term in semantics, which he then proposed to relabel Rhematology; see Bréal (1900:329). 18 Peirce, in a 1904 letter to Lady Welby, cited in Ogden and Richards 1938:282; also Peirce VIII, 334. Some of the distinctions made there were then developed in a 1906 article under the title, "Prolegomena to an Apology for Pragmaticism" ( = Peirce IV, 530-572). A 'sinsign' (II, 245) is perhaps better known as a token (IV, 537) (which Peirce sometimes called an Actisign), in contrast to type, or 'legisign' (earlier Famisign; cf. VIII, 363) (see further Section 3, below). In 1906, Peirce (TV, 537) also named a legisign he called tone, which he defined as 'an indefinite significant character such as a tone of voice', and which he regarded as neither a token nor a type; this may constitute the earliest explicit recognition of paraphonetic features as a part of semiotics.

218

THOMAS A. SEBEOK

The above triple trichotomies yield ten classes of signs, labelled and illustrated by Peirce (Table 2): TABLE 2

A-B-C Name of Sign 1-1-1 Qualisign 2-1-1 Iconic Sinsign 2-2-1 Rhematic Indexical Sinsign 2-2-2 Dicent Sinsign 3-1-1 Iconic Legisign 3-2-1 3-2-2 3-3-1 3-3-2 3-3-3

Rhematic Indexical Legisign Dicent Indexical Legisign Rhematic Symbol (or Symbolic Rheme) Dicent Symbol Argument

Example A feeling of 'red' An individual diagram A spontaneous cry A weathercock A diagram, apart from its factual individuality A demonstrative pronoun A street cry A common noun Proposition Syllogism

Ten classes of signs, summarized in tabular form by Weiss and Burks 1945, after Peirce II, 254-263.

It is indeed difficult to appreciate the utility of assigning a demonstrative pronoun in any natural language, for instance, to the category of a Rhematic Indexical Legisign, which 'is any general type or law, however established, which requires each instance of it to be really affected by its Object in such a manner as merely to draw attention to that Object'. This formulation ignores the evident fact that every demonstrative pronoun constitutes a different nodal point in a special network of verbal signs (Benveniste 1969). Wells thinks, moreover, that Peirce's notion of symbol, though original, is fruitless, because of the way in which he generalizes his concept of the mind, which 'permeates and infects what he says of the symbol'. It is only his notion of index that Wells (1968: #2b) deems both novel and fruitful.19 On the other hand, such passing strictures are not to be mistaken for an adequate critique of Peirce's work in semiotic. As another important contributor to semiotic remarked later, Peirce's 'classification of signs, his refusal to separate completely animal and human sign-processes, his often penetrating remarks on linguistic categories, his application of semiotic to the problems of logic and philosophy, and the general acumen of his observations and distinctions, make his work in semiotic a source of stimulation that has few equals in the history of this field' (Morris 1946: 290). 18 The notion of the icon is related to the Platonic process of mimesis, which Aristotle broadened (in the Poetics, IV) from a chiefly visual representation to embrace all of cognitive and epistemological experience. For semiotic implications of this axiological category, cf. Morawski 1970.

SEMIOTICS: A SURVEY

219

Since the bulk of Peirce's pertinent work, completed soon after the turn of the century, was published posthumously, it could hardly have come to the attention of his contemporary, Ferdinand de Saussure (1857-1913), the Swiss linguist who, at about the same time, conceived the need for a general science of signs which he tentatively designated sémiologie.20 This he thought indispensable for the interpretation of language, as well as of all other systems of signs in their interrelation with language: 'Puisqu'elle n'existe pas encore, on ne peut dire ce qu'elle sera: mais elle a droit à l'existence, sa place est déterminée d'avance.... Par là, non seulement on éclairera le problème linguistique, mais nous pensons qu'en considérant les rites, les coutumes, etc., comme des signes ces faits apparaîtront sous un autre jour, et on sentira le besoin de les grouper dans la sémiologie et de les expliquer par les lois de cette science' (Saussure 1960: 33, 35). Saussure's originality consisted of the recognition of the vital importance for linguistics of a comparative analysis and classification of the different sign systems, since such a confrontation is alone capable of revealing 'what properties are shared by verbal signs with some or all other semiotic systems and what the specific features of language are' — in brief, that which can alone enable us to discriminate what is necessary from what is contingent in various systems of communication, notably in language (Jakobson 1969: 78; cf. Sebeok 1968, esp. p. 11). The task Saussure set for himself was not to decide whether linguistics was more closely allied to psychology or to sociology, or even to find its place among existing disciplines, but to raise the issue to an entirely new level and to pose the problem in fresh terms, as shown in the following key citation: La langue est un système de signes exprimant des idées et par là, comparable à l'écriture, à l'alphabet des sourds-muets, aux rites symboliques, aux formes de politesse, aux signaux militaires, etc., etc. Elle est seulement le plus important de ces systèmes. On peut donc concevoir une science qui étudie la vie des signes au sein de la vie

sociale-, elle formerait une partie de la psychologie générale; nous la nommerons sémiologie du grec sëmeîon 'signe'). Elle nous apprendrait en quoi consistent les signes, quelles lois les régissent.... La linguistique n'est qu'une partie de cette science générale, les lois que découvrira la sémiologie seront applicables à la linguistique, et celle-ci se trouvera ainsi rattachée à un domaine bien défini dans l'ensemble des faits humains. C'est au psychologue à déterminer la place exact de la sémiologie;21 la tache du 80

The notion of semiotics, and its designation as la sémiologie, were first recorded in a note of Saussure's dated November, 1894; (cf. Godel 1957:275). Prieto's remark that Saussure seemed 'Le premier à concevoir cette science' (1968:93), is, of course, absurdly inaccurate. Saussure was even anticipated, in spirit if not in terminology, by Rudolf Kleinpaul, who published a wellreasoned and abundantly illustrated 456 page work on non-verbal communication a few years earlier (1888). 81 Here Saussure refers to Adrien Naville, Nouvelle classification des sciences (1901:104), who recorded this early version of his Geneva colleague's views on the subject: Saussure insists on the importance 'd'une science très générale, qu'il appelle sémiologie, et dont l'objet serait les lois de la création et de la transformation des signes et de leur sens. La sémiologie est une partie essentielle de la sociologie. Comme le plus important des systèmes de signes c'est le langage conventionnel des hommes, la science sémiologique la plus avancée c'est la linguistique ou science des lois de la vie du langage.' Cf. Godel 1957:181.

220

THOMAS A. SEBEOK

linguiste est de définir ce que fait de la langue un système spécial dans l'ensemble de» faits sémiologiques . . . si pour la première fois nous avons pu assigner à la linguistique une place parmi les sciences, c'est parce que nous l'avons rattachée à la sémiologie. (Saussure 1960: 33-34) In contrast to Peirce — who was heir to the entire tradition of philosophical analysis of signs — Saussure's point of departure and constant center of attention was language; for him, the notion of the sign was primarily a linguistic fact, that somehow expanded to encompass the other processes of human, in particular social, signaling behavior. As for Locke before him, and for many other scholars after him — for instance, Bloomfield, who asserted that 'Linguistics is the chief contributor to semiotic' (1939: 55), or Weinreich, who called natural human language 'the semiotic phenomenon par excellence' (1968: 164) — for Saussure language occupied pride of place among all semiotic systems. A contemporary critic, Roland Barthes, going a step further, deems it necessary to 'face the possibility of inverting Saussure's declaration: linguistics is not a part of the general science of signs, even a privileged part, it is semiology which is a part of linguistics: to be precise, it is that part covering the great signifying unities of discourse. By this inversion', adds Barthes (1968: 11), 'we may except to bring to light the unity of the research being done in anthropology, sociology, psycho-analysis and stylistics round the concept of signification.'22 The validity of this enthusiastic reversal of the semiotic hierarchies can be contemplated, if at all, only if zoosemiotics — the signaling behavior of all the speechless creatures — is left out of acoount; we shall, therefore, reject it without further discussion, as not fitting to the scope of this article. What is the precise nature of the relationship of language to the other human semiotic systems? It appears that, for Saussure, the answer to this question lay in his well-known conception of the 'arbitrary' character of the linguistic sign. In general, he asserts that the prime target of semiotics will be 'l'ensemble des systèmes fondés sur l'arbitraire du signe', and it is just in consequence of this that language acquires its position of preeminence: 'on peut . . . dire que les signes entièrement arbitraires réalisent mieux que les autres l'idéal du procédé sémiologique; c'est pourquoi la langue, le plus complexe et le plus répandu des systèmes d'expression, est aussi la plus caractéristique de tous; en ce sens la linguistique peut devenir le patron général de toute sémiologie, bien que la langue ne soit qu'un système particulier' (Saussure 1960:110). Thus, while Saussure clearly insisted on the notion that linguistics had a necessary relationship with semiotics — a relationship, moreover, in which semiotics was superordinate — he failed to define its exact " Genette explicates this reversal of Saussure's formula in his 1969 article. On semiotics as unification of science, see further Morris 1946:223-227, where this idea appears in two senses: semiotics is said to provide a comprehensive language for talking about sign phenomena, instead of looking at them piecemeal, and to provide an instrument for analysis of the relations between all special scientific languages; cf. also Hjelmslev 1953. On the Unity of Science movement, the associated International Encyclopedia and Institute, see Feigl 1969; for glimpses of the role of Morris, cf. Feigl 1969:647-648, 656-657, 659-660.

SEMIOTICS: A SURVEY

221

character beyond asserting the principle of the 'arbitrariness of signs', which he regarded as a pervasive feature of all systems of expression, especially of language. In sum, for Saussure, semiotics remained a programmatic science, and its salient traits were those that he identified with his linguistic model (le patron général). For several decades after the epoch of Peirce and Saussure, the further development of the science of signs was stimulated by the joint and separate work of C.K. Ogden and I.A. Richards, who were under the direct influence of Jeremy Bentham. Their volume, The meaning of meaning (193 8),23 proposed to distinguish the signification of signs represented in scientific discourse (referential or 'symbolic' terms) from those characteristic of non-scientific discourse (emotive or expressive terms), an emphasis which has proved fruitful in the consideration of semioti'c issues in connection with aesthetics in general and literary and other art theory in particular. They also introduced the so-called context theory of meaning (for a critical review see Hotopf 1965:253-264), which brought mental process into the definition of the sign. This was done as follows (according to a later formulation by Richards): 'A sign . . . works by being a member of a certain kind of interpretative context in the mind — let us write it abcq. When abc ... recurs without q — such is the fortune and overwhelming important characteristic of these interpretative contexts — the effect is in certain respects as though q also had recurred. And q, then, is what a (the sign) is of, what is signifies, or represents' (Richards 1933); in other words, whenever an object experienced together with another causes us to think of that other, then it acts as a sign of the other, and these objects are said then to form a context. Richards' still later statement, that 'meaning is delegated efficacy' (1936:32), moves more in a behavioristic direction, but a detailed account of how the sign exercises this delegated efficacy was left for the next major contributor to semiotics, Charles Morris, to supply. Richards himself concluded, in 1936 (p. 34), that 'no one knows' how signs work. In passing, it should also be pointed out that the collaborative work of Ogden and Richards, and, after the death (in 1957) of his senior partner, the later work of Richards especially, fell 'mainly in the sphere of the educator's interest in semiotic' (Morris 1964:296). Morris, in a succession of articles and books, has consistently aimed to lay the foundation for a comprehensive and fruitful science of signs.24 Most concise and lucid in form, his exposition of 1938, Foundations of the theory of signs, remains " This book - the substance of which first appeared in periodical form during 1920-22 passed through numerous editions (1925 1 , 1938 5 ), with constant revisions; references are made, throughout, to the 1938 s edition. Supplements included Malinowski's influential piece on "The problem of meaning in primitive languages", and the interesting but hitherto neglected essay by Crookshank, on "The importance of a theory of signs and a critique of language in the study of medicine". There is a brief account of Ogden and Richards' views on the modes of signifying in Morris 1946:69-72. 24 For a complete listing of Morris' relevant publications, see Morris 1964:92-93. On Morris and his works, see Rossi-Landi 1953. [Morris' writings on the general theory of signs appeared in 1972, as Approaches to Semiotics 16, The Hague, Mouton.]

222

THOMAS A. SEBEOK

one of the classics of the field. In this monograph, he introduced his 'very useable and useful trichotomy', which 'has a Peircean basis' (Wells 1967:#3), namely, the division of semiotics into syntactics, semantics, and pragmatics.25 When there is signification, there is (a) a sign, which has (b) a meaning or an object, and which has (c) this meaning or this object for an interpreter of the sign. Accordingly, he defines syntactics as 'that branch of semiotic that studies the way in which signs of various classes are combined to form compound signs', that is, syntax treats signs in abstraction from their objects and their users. He next defines semantics as 'that branch of semiotic which studies the signification of signs', that is, semantics takes signs and objects into account, but not interpreters. At last, he defines pragmatics as 'that branch of semiotic which studies the origin, the uses, and the effects of signs', that is, pragmatics takes all three factors into account.26 Logicians usually claim that the pragmatic relationship presupposes the semantic and the syntactic, and that the semantic presupposes the syntactic, but that, on the other hand, 'the syntactic relationship does not presuppose the semantic and the pragmatic', and that 'the semantic relationship may be studied without reference to the pragmatic' (e.g. Bochenski 1968: 33). In the language development of the child, for example, the semantic aspect always implies the syntactic, but this hierarchy seems to be just the opposite in the case of Washoe, the young female chimpanzee that has been taught to communicate by gestures based on the standard American Sign Language of the deaf (Gardner and Gardner 1969). In other words, the eidetic meaning of a sign invariably implies an operational meaning along with it for the human, but not necessarily so for the animal; David McNeil has summarized the situation aptly: 'A child will retain grammar even when speaking utter nonsense, whereas a chimpanzee apparently will exclude grammar even when conveying perfect meaning' (in Pfeiffer 1969: 399; cf. Bronowski and Bellugi 1970, and Brown 1970). Another trichotomy introduced by Morris, cross-cutting the former (cf. Sellars 1950) distinguishes pure, descriptive, and applied semiotics. Pure semiotics is said to elaborate 'a language to talk about signs', and Morris' own books may be regarded as outstanding exemplars of this genre. Descriptive semiotics studies actual signs; almost all zoosemiotics is descriptive in this sense. Applied semiotics 'utilizes knowledge about signs for the accomplishment of various purposes'. These distinctions were found to be eminently applicable not only to sign-processes at the level of human behavior, but also to those that occur in the speechless animals. The possibility of such an extension of semiotics was explicitly envisaged, but not elaborated, by Morris himself (1946:52-55, V). This was tried much later by the zoologists Peter Marler (1961) and W. John Smith (1968) (also anticipated by Yerkes and Nissen, 1939), and was also attempted, in a substantially wider frameis This aspect of semiotics was illuminatingly applied by Greenberg (1964) to the many-faceted relations between ethnology and linguistics. Weinreich (1968:169), on the other hand, thought that this tripartition 'turned out to be of little use in connection with natural sign phenomena'. *• For definitions of these and other terms employed by Morris, see the Glossary (pp. 345-356) appended to his Signs, language and behavior (1946).

SEMIOTICS: A SURVEY

223

work, in a series of papers, by myself (e.g. Sebeok 1972). Marler held that the application of syntactics to animal communication is particularly clear, that semantics is of doubtful value in this sphere; but he was most interested in animal pragmatics, which he helped develop (although he was anticipated in this effort, in some ways, by J. B. S. Haldane) with the aid of still another set of distinctions drawn by Morris, namely, that among identifiors (that signify a location in space or time), designators (that signify characteristics or stimulus-properties of stimulus-objects), appraisors (that signify something as having a preferential status for behavior), and prescriptors (that signify the requiredness of certain response-sequences). These (non-exhaustive) categories of signs, that influence the behavior of their interpreters in different ways, were then applied by Marler (1956) in an analysis of vocal communication in the chaffinch: 'in essence, given a knowledge of the response of other animals to the signal, and of the other circumstances in which that same response is given', Marler could 'infer the nature of the "message" transmitted by the signal'. Morris was close to two of his linguist colleagues at the University of Chicago: Manuel J. Andrade and Leonard Bloomfield. His basic view of the relation between semiotics and linguistics was that the former provides the metalanguage for the latter, 'and that the terminology of linguistics is to be defined by linguists on the basis of the terms of semiotic' (Morris 1946:221). In 1946, he also introduced the very general notion of a lansign-system (ibid.: 35ff.), roughly equivalent to language, but applicable as well to such speech surrogates as script, more or less formalized transforms such as symbolic logic and mathematics, and 'perhaps to the arts' (1964:60). In 1964, he observed that while recent work in linguistics had been in the syntactical study of certain lansign-systems (viz., the natural languages), he now sensed — correctly, as it turned out — a growing movement to extend the study of language to take into account its semantic and pragmatic aspects as well; in this, he saw a clear parallel in the development of logic in this century. He also reemphasized that linguistics is a part of semiotics, citing in support of this hierarchy, besides Bloomfield's, the authority of Saussure, Hjelmslev, and Greenberg. In addition to his monograph cited above, Morris carried forward Peirce's dictum that, to determine the meaning of any sign, 'we have . . . simply to determine what habits it produces'. Accordingly, his book, Signs, language and behavior, was produced in the attempt to describe and differentiate signs in terms of the dispositions to behavior which they cause in their interpreters. In other words, semiotics is here tied to a particular theory of behavior — although it also draws upon the results of the insights of logicians, especially of Rudolf Carnap — which stresses the response to the stimulus, while the stimulus situation is disregarded: Morris' starting point here seems to have been 'an analysis of indexical signs, as opposed to conventional symbols. The modification consists in this that a definite response to an utterance ceases to be considered the meaning of that utterance; it is replaced in that role by the type of response, i.e., a regular and potential response. The theory is, as it were, a behaviourist, external aspect of associationism: the association of thoughts

224

THOMAS A. SEBEOK

and/or ideas is replaced by a relationship between acts or types of behaviour' (Pelc 1969). Morris' latest statement, Signification and significance, while retaining his commitment to a variety of behavioristic psychology, aims at a synthesis of semiotics and axiology, the theory of value. It explores the relation of signs and values, and assesses their place in human action. At this point, I must record my over-all agreement with the late Uriel Weinreich's assessment (1968:166) that 'the reduction of sign phenomena to some more general kind of behavioral phenomenon have produced no marked success in either theory or research practice. A sign type is not always — perhaps only rarely — correlated with a class of specific stimuli or of overt responses.' If this has a certain application to Morris' efforts to couple semiotics to behaviorism, the criticism holds a fortiori regarding the attempts of experimental psychologists to specify the meaning of individual signs in terms of attitude taken towards them, emotions induced by them, productive of subjective connotations of strength of weakness, warmth or coldness, friendliness or unfriendliness, or further signs automatically evoked by them — by such techniques as a semantic differential or a free-association test, the former reviewed at length by Weinreich.27 In the decades that followed the publication of Saussure's Cours, linguists were too busy with problems internal to their discipline — particularly those attendant on the development of structuralism — to pay more than perfunctory attention to semiotics, although some linguistically oriented scholars, working in adjacent disciplines and concerned with questions of foundations, made fertile exploratory contributions into sign systems other than languages (e.g., a system of garments, certain of the verbal arts, etc.; I shall return to some of these in Section 3), more or less inspired by general principles of semiotic analysis. One of the rare exceptions was the Belgian linguist Eric Buyssens, who undertook a serious, original, and reasonably comprehensive essay (1943) to realize 'le voeu de Saussure', to confront languages with other (human) sign systems.28 Buyssens introduces the term seme, defining this as 'tout procédé idéal dont la réalisation concrète permet la communication', the adjective semic, and the phrase semic act, meaning the concrete embodiment of a seme. Semes - not signs,29 which according to Buyssens, cannot be isolated, and can be defined solely as semic function — are then classified by a double criterion: according to the sensory modality 27

See Osgood, Suci, and Tannenbaum 1957, and Weinreich's review article, 1958; also Snider and Osgood 1969. On behaviorism and semiotics in general, see also Wells 1967, esp. II. 28 Whether he was successful was challenged by representatives of the Geneva school, e.g. Frei 1963. See also Buyssens 1967, the first part of which constitutes a somewhat modified and augmented edition of Buyssens 1943; and, in the same tradition, Pohl 1968, especially the first volume, Le symbole, clef d'humain. 29 But cf. Prieto 1968:125: 'le terme "signe" . . . est sans doute synonyme de "sème"'. Nevertheless, he finds 'sign' an ambiguous notion and advocates, with Buyssens, the adoption of 'seme' instead. On Saussure's conception of sème, see Engler 1968:44.

SEMIOTICS: A SURVEY

225

involved, such as auditory, visual, and the like, or modalities (since semes like etiquette may combine words, gestures, perfumes, décors); and according to their etiology, which depends on the kind of relation between the form and the meaning of a given seme. This can be intrinsic, 'existant antérieurement entre deux faits', which I take to be iconic a n d / o r indexic; or extrinsic, where the relation exists 'uniquement en vue de la communication', or is 'arbitrary', in Saussure's sense. Some semes may have a mixed character. Among intrinsic semes, Buyssens cites the following: l'art, la publicité imagée, certains symboles, certaine gestes, les histoires sans paroles, les dessins qui à l'envers de certains postes récepteurs de radio indiquent l'endroit où il faut relier l'antenne, la prise en terre, le pick-up ou le réseau; les dessins qui, sur les machines-outils destinées à l'exportation, indiquent la fonction des manettes, pédales, leviers, boutons, interrupteurs; les enseignes parlantes; les dessins qui, par example, sur le ferry-boat Londres-Istamboul, indiquent à l'entrée de chaque salle sa destination: un serveur pour le restaurant, une dame sur un tabouret pour le bar, etc. Among extrinsic semes, he cites: le discours et ses transcriptions...; les signalisation routière, ferroviaire et maritime; les sonneries de clairon, de trompette ou de cor; les batteries de tambour; les signaux auditifs par sifflet, sirène, cloche, gong, canon; les signaux optiques par fusées, lampes, phares, bouées, fumées; les symboles des mathématiciens, physiciens, chimistes, logiciens; les notations commerciale, musicale et prosodique; les modèles de quantité de documents officiels (billets de banque, pièces d'identité, affiches, etc.); les taches de couleurs par les quelles on marque dans les campagnes les itinéraires touristique; la croix rouge sur fond blanc des services sanitaires; le coup de sonnette du visiteur; la sonnerie de téléphone et son signal 'occupé', etc., etc. [Buyssens 1943 : 46] A mixture of intrinsic and extrinsic semes is found, among other structures, in etiquette, and in maps, blueprints, and the like. In concluding, Buyssens claims to have demonstrated that there is n o difference at all, in principle (de nature), between (natural) languages and other sémies, that is, sets of semes opposed to each other by normal and significant differences; the superiority of speech is, for him, merely a quantitative matter. The well-argued, adequately illustrated, if somewhat quaint booklet of Buyssens has been virtually ignored by other linguists, possibly because of its appearance during World War II; nevertheless, it still repays careful reading. Curiously enough, however, when, some thirteen years later, Buyssens himself described (1956) the system of 460 gestures used by members of a Cistercian religious order, he abandoned all of his theoretical apparatus and treated it conventionally as a system of signs. 80 T o date, the most explicit and formalized exposition of the Saussurean heritage, 50 Cf. Clelia Hutt's recent study on the Trappists (1968). Partly in the tradition of Buyssens, to whom he owes the observation 'qu'il existe des procédés de signalisation a-systématiques', are several articles by Mounin; see, e.g. 1959.

226

THOMAS A. SEBEOK

however, is the theory of Glossematics, as embodied in the writings of two Danish linguists, Louis Hjemslev (1953) and H. J. Uldall (1957), in which semiotic notions played a leading role, and with all sorts of ramifications (which will not be discussed here). Glossematic theory is based on the over-all assumption that the properties of all semiotic systems — accounted for in terms of a moderately complex algebraic apparatus — are based on a small set of primes, such as class and component, function, necessary and not necessary functive, both/and and either/or functions, and the empirical assumption of two non-conforming planes — content and expression — that consist, each, of two nonconforming hierarchies, form and substance. In its final stages, the algebra of functions proposed by both Hjelmslev and Uldall was conceived as a highly general and uniform framework for the description of human behavior. In her Introduction to the second edition of Uldall's 1957 work, Eli Fischer-J0rgensen (1967) emphasized the differences between Hjelmslev's and Uldall's conception of the scope of glossematics. She points out that for Uldall glossematics was 'a formal theory, which is not defined by any specific material, but designed explicitly to be used for all human activity', whereas for Hjelmslev glossematics was a linguistic theory, 'which, implicitly, may serve also as a model for other humanistic disciplines' (1967: x-xi). At first blush, all this brings to mind Noam Chomsky's skepticism toward such universality; how worthwhile is it to rise to a level of abstraction where there are 'plenty of other things incorporated under the same generalizations which no one would have regarded as being continuous with language or particularly relevant to the mechanisms of language'? (Chomsky 1967:73). It is, indeed, one of the crucial questions of semiotics whether such generalizations, represented by the glossematicians (and again in such works as Barthes', e.g. 1968), merely express a methodological attitude, at best providing a uniform frame of reference for some or all of the behavioral sciences, or incorporate genuine empirical insights. The glossematicians would presumably argue that a systematic confrontation of language with other — both less and more complex — semiotic systems would reveal peculiarities of linguistic structure, that a study of this kind presupposes a uniform frame of reference, and that the issue boils down to the care one exercises in the selection of a common point of view so that 'these sciences are concentrated around a linguistically defined setting of problems' (Hjelmslev 1953:69). In the glossematic conception, natural languages are distinguishable and distinguished from non-language by specific properties, as follows: they belong to the class of semiotic systems exhibiting two planes with a diverging form; further, the content of languages can be manifested by all possible content substances, or, in other words, a language is that semiotic system which is able to express everything that can be expressed at all. Accordingly, says Hjelmslev: In practice, a language is a semiotic into which all other semiotics may be translated — both all other languages, and all other conceivable semiotic structures. This translatability

SEMIOTICS: A SURVEY

227

rests on the fact that languages, and they alone, are in a position to form any purport whatsoever; in a language, and only in a language, we can 'work over the inexpressible until it is expressed' (Kierkegaard). It is this quality that makes a languable usable as a language, capable of giving satisfaction in any situation There is no doubt that it rests on a structural peculiarity, on which we might be able to cast better light if we knew more about the specific structure of non-linguistic semiotics. It is an all but obvious conclusion that the basis lies in the unlimited possibility of forming signs and the very free rules for forming units of great extension (sentences and the like) which are true of any language... in general, a language is independent of any specific purpose. [Hjelmslev 1953 : 69] As Manfred Bierwisch (1972) has rightly pointed out, it is at this juncture that Hjelmslev and Chomsky converge in the tradition of Cartesian linguistics, and come to grips with what the latter has sometimes called the 'creative aspect of language use', which is thus claimed to be the specific property that distinguishes language from all other semiotic systems; ('creativity', in Chomsky's sense, seems to refer to the appropriateness of language to any new situation). Hjelmslev, in his concluding statement, makes the claim that 'semiotic structure is revealed as a stand from which all scientific objects may be viewed' (1953:81), and indeed, by the middle of the past decade, we witnessed, as Jakobson has recently observed, 'a spontaneous and rapid international development of the new discipline which encompasses a general theory of signs, a description of the different sign systems, their comparative analysis and classification' (1969:78). Scientific workshops in several countries31 — particularly in France, Italy, Poland, the U.S.A., and the U.S.S.R. — are now conducting researches in a wide variety of semiotic problems, involving multifarious academic disciplines. (It must be admitted, however, that certain of these — somewhat in the manner of Monsieur Jourdain — have pursued semiotic goals, 'il y a plus de quarante ans', without knowing it!) In the United States, the first conference on semiotics32 was held on May 17-19, 1962 (Sebeok, Hayes, and Bateson 1964). Initially convened to discuss merely the twin topics of paralinguistics and kinesics, the inquiry and debate were soon widened to take in other areas of non-verbal communication, although there was throughout a centripetal tendency to view problems with constant reference to language. It was only in the final discussion session of this three-day meeting — in the last minutes of the conference, in fact — that Margaret Mead, in her usual incisive fashion, 31

Hjelmslev's line of thinking about semiotic and allied notions is currently carried forward, in Europe, by Greimas (and his students, especially in France and in Italy) - see fifteen of his collected essays (1970), six of which have also appeared in Italian as Modelli semiologici ([1968]); and, in the U.S., by L a m b and his students - see e.g., his Outline of stratificational grammar ([1966]) - and by Chafe, w h o reviewed (1968) Lamb's Outline, and w h o gives further references (p. 603) to Lamb's works and his own; see also Fleming, "Stratificational theory: A n annotated bibliography" (1969). 32 T h e conference, sponsored by the U.S. O f f i c e of Education, was held on the Bloomington campus of Indiana University.

228

THOMAS A. SEBEOK

succeeded, in crystallizing the thoughts of the some sixty scholars, of most diverse backgrounds, around the 'semiotics'33 label. 'It would be very nice', she said, if we could go away from here with at least a preliminary agreement on the use of some phrase that we could apply to this whole field. Kinesics and paralinguistics, after all, are two We are, I think, conceivably working in a field which in time will include the study of all patterned communication in all modalities, of which linguistics is the most technically advanced. If we had a word for patterned communications in all modalities, it would be useful... many people here, who have looked as if they were on opposite sides of the fence, have used the word 'semiotics'. It seems to me the one word, in some form or other, that has been used by people who are arguing from quite different positions.... A lot of people are not going to spend their time on this field unless it is set up for exploration at all levels and in all modalities, but with a great deal of help from linguists. [Mead 1964 : 275] This pioneer interdisciplinary conference was deliberately organized as a pentad. Its five foci were: medicine, that is, communication without words between patient and doctor (Ostwald 196434); psychology, in particular, research on extralinguistic phenomena (Mahl and Schulze 1964); pedagogical perspectives in paralinguistics and kinesics (Hayes 1964); cross-cultural perspectives (La Barre 1964); and emotive language (Stankiewicz 1964 s5). The published transactions also included a general statement by Miss Mead, summarizing the highlights of the symposium and pointing the way for 'continuing work on the multi-modal analysis of communication'. She affirmed that, 'in inaugurating the new science' of semiotics, we must act on behalf of man, by which she meant to disavow any manipulative applications (Mead 1964; see also Mead 1969 for a later version of her views on semiotics). In the very year when this initial systematic American effort appeared in book form, Soviet scholars were holding the first of a continuing series of colloquia on semiotics in Kàâriku (near Tartu, in Estonia), under the leadership of Juri M. Lotman.36 While detailed information about this meeting is lacking,37 a report of the next one, which was held on August 16-26, 1966 (and attended also by two U.S. participants), has been published (Venclova 1967). According to this account, the meeting was devoted, in the main, to the area of semiotics that has been called in French 'systèmes secondaires modelants', that are (in the narrow sense) complex 53

Editorial comment on the word 'semiotics' appears in the Preface, Sebeok, Hayes and Bateson 1964. 34 A revised version of this paper has appeared (1968). 35 Stankiewicz is currently expanding this paper into a monograph on Expressive language. 39 The principal writings of Soviet semioticians centered around Lotman (who is Chairman of the Russian Literature Department at the State University of Tartu) have hitherto appeared under his editorship in Trudy po znakovym sistemam, Vols 2, 3 (cf. also Venclova 1967), 4, and 5 (see bibliography). See also Julia Kristeva's 1967 (reprinted 1969b, with 10 other essays) essay, which takes the writings of the Soviet semioticians as her point of departure. For a panoramic view of semiotics in the USSR, see Faccani and Eco 1969; Papp 1969; and especially the survey by Segal, in Juri M. Lotman and Boris A. Uspenskij, eds., Ricerche semiotiche: Nuove tenderize delle scienze umane nell'URSS (Torino, 1973), 452-70. 37 Three of the papers presented at the Tartu colloquium of May 10-20, 1968 have appeared in Semiotica, Vol. 1: Ivanov 1969; Kristeva 1969a, and Lotman and Pjatigorskij 1969.

SEMIOTICS: A SURVEY

229

macrostructures composed of a natural language as their infrastructure and having at least a text coded with prefabricated units and constructions as their secondary superstructure.38 The products of verbal art evidently belong here and elevate the study of the poetic function of language to the status of semiotic inquiry; likewise, the semiotic interpretation of myths and other kinds of oral tradition fall within this sphere. The approximately thirty communications presented tended to divide into eight groups: typology of texts; semiotics of folklore; poetics; typology of cultures; problems of mythology; semiotics of space and time in modeling systems; artistic models in a communicative frame; and general problems of semiotics. In 1965, a gathering was convened in Warsaw, but the first full-scale international conference on semiotics — with representatives from Czechoslovakia, Denmark, France, the German Democratic Republic, Italy, Poland, the U.S.A., and the U.S.S.R. in attendance — took place on September 12-18, 1966, at Kazimierz nad Wisla (in Poland). By that time, the Collège de France's Laboratoire d'Anthropologie sociale had established its Section de Sémio-linguistique, led by A. J. Greimas, with the blessing and collaboration of Lévi-Strauss, Benveniste, and Barthes. Polish leadership was provided by Stefan Zolkiewski and his associates, especially Maria Renata Mayenowa.39 A remark of Mayenowa's reflected the mood of this conference; she later wrote: it is only recently that semiotics has begun to constitute itself as an autonomous field of research superimposed on the several disciplines dealing with specific sign systems. However, since these disciplines, with the sole exception of linguistics, are themselves of recent origin, more or less contemporary with semiotics, we cannot as yet be said to have developed adequate and universally accepted theories for sign systems, other than those developed in linguistics for the natural languages [1967: 59-64; cf. Sebeok 1967a]. 88

'Systèmes secondaires modelants' is a translation from the Russian of an expression proposed by A. A. Zaliznjak, V. V. Ivanov, and V. N . Toporov, in reference to the framework used in their collaborative work on Belorussian myths and rites, Slavjanskie jazykovyie modelurujuiZie semioticeskie sistemy (1965:6-8). The notion of a secondary modeling system, in the broad sense, refers to an ideological model of the world where the environment stands in reciprocal relationship with some other system, such as an individual organism, a collectivity, a computer, or the like, and where its reflection functions as a control of this system's total mode of communication. A model of the world thus constitutes a program for the behavior of the individual, the collectivity, the machine, etc., since it defines its choice of operations, as well as the rules and motivations underlying them. A model of the world can be actualized in the various forms of human behavior and in its products, including linguistic text - hence the emphasis on the verbal arts - social institutions, movements of civilization, and so forth. If I understand the notion fully, it is akin to the thematic principle that some American and other anthropologists have - with more artistic insight than scientific rigor - referred to as ethos, the integrating or summative pattern of a culture; cf. Kroeber and Kluckhohn 1952, esp. pp. 165-171. For broad applications, especially in the sociology of literature, see further ¿olkiewski 1969. The relevant literature is reviewed, with particular reference to folklore, in Voigt 1969. 39 U.S. participants were: Henry Hiz, Roman Jakobson, Krystyna Pomorska, Meyer Schapiro, and myself; N o a m Chomsky submitted a written communication. The proceedings of the conference, including the full texts of the papers delivered or sent, have appeared as Vol. 1 in the Series Maior of Janua Linguarum (The Hague: Mouton & Co.).

230

THOMAS A. SEBEOK

Nevertheless, the conferees were confident that the time had come to advance semiotics by two practical and interrelated actions: the creation of an International Association for Semiotics; and provision for a periodical publication. It was decided to hold a new meeting in Warsaw in 1967, to finalize arrangements for the proposed organization, and to immediately authorize the setting up of a suitable publication outlet. Unfortunately, the 1967 meeting had to be postponed to 1968, at which time an international political crisis prevented the holding of more than a rump session in Warsaw. Eventually, however, an organizational meeting was called in Paris, on January 21, 1969, and there the Association Internationale de Semiotique (International Association for Semiotic Studies) was born.40 Its officers are, at present (1972), the following: President: Emile Benveniste (France) Acting President: Cesare Segre (Italy) Vice-Presidents: Roman Jakobson (U.S.A.), Juri M. Lotman (U.S.S.R.), A. Ludskanov (Bulgaria), D. Pignatari (Brazil) Treasurer: Jacques Geninasca (Switzerland) Secretary-General: Umberto Eco (Italy) Executive Secretary: Julia Kristeva (France) Editor-in-Chief: Thomas A. Sebeok (U.S.A.) The Association has announced three aims: to promote semiotic researches in a scientific spirit; to advance international cooperation in this field; and to collaborate with local associations (which had already been formed in France, Italy, Poland, and the U.S.S.R.). It proposes to achieve these goals by organizing national and international meetings, and through its sponsorship of an international quarterly, Semiotica (now appearing in three volumes, each of four issues, annually). Semiotica — which is doubly co-sponsored by UNESCO'S Conseil International des Sciences Sociales and its Conseil International de Philosophie et des Sciences Humaines — represents a direct outgrowth of Studies in Semiotics — Recherches Semiotiques, authorized by the Kazimierz assembly and constituting a regular section throughout eleven issues (Vol. 6, Nos, 2-6, 1967; Vol. 7, Nos. 1-6, 1968) of the bilingual periodical, Social Science Information — Information sur les Sciences Sociales. It was produced under the joint auspices of a Standing Committee on Publications, established at Kazimierz, entrusted to the editorial care of A. J. Greimas (Editor-in-Chief, 1967), Juri M. Lotman, Thomas A. Sebeok (Editor-in-Chief, 1968), and M. W. Skalmowski (1967; replaced by Jerzy Pelc in 1968). Effective 40

The address is: 23 via Melzi d'Eril, Milano, Italy. Its directors listed include, in addition to the officers named in the text, E. Ver6n (Argentina), R. Lindekens and N . Ruwet (Belgium), M. Yanakiev (Bulgaria), P. Bouissac (Canada), Ivo OsolsobS (Czechoslovakia), Holger S. S0rensen (Denmark), Erhard Albrecht and Manfred Bierwisch (German Democratic Republic), HansHeinrich Lieb and Hansjakob Seiler (German Federal Republic), J. Cohen and H. W. Thorpe (Great Britain), Ivan Fonagy and Gyorgy Szepe (Hungary), B. Hrushovski (Israel), Umberto Eco and Aldo Rossi (Italy), Sh. Kawamoto (lapan), T. A. van Dijk (The Netherlands), Jerzy Pelc and Stefan ¿olkiewski (Poland), Mihai Pop (Romania), S. Bayrav (Turkey), Henri Hiz (U.S.A.), and S. K. Saumjan (U.S.S.R).

SEMIOTICS: A SURVEY

231

1969, this relatively modest initial effort has been transformed into a full-scale independent journal, renamed Semiotica, edited by Thomas A. Sebeok (U.S.A.), with an Editorial Board including Roland Barthes (France), Umberto Eco (Italy), Henry Hiz (U.S.A.), Julia Kristeva (France), Juri M. Lotman (U.S.S.R.), Jezy Pelc (Poland), Nicolas Ruwet (Belgium), Meyer Schapiro (U.S.A.), and Hansjakob Seiler (G.F.R.). The publisher is Mouton, of The Hague, The Netherlands. Furthermore, the journal is being supplemented by Approaches to Semiotics, to accommodate book-length contributions, such as original monographs, collections, relevant conference proceedings, important reprintings and translations. The fourth volume in this series is a republication of the most consequential contents of the eleven issues of Studies in Semiotics — Recherches Semiotiques, published in order to make them conveniently available under a single cover.40a The founding of the Association, with the concurrent establishment of both publication outlets, will undoubtedly serve as a rallying point and perhaps as a catalyst for all interested individuals and groups who have a serious concern for the notion of the sign; the year 1969 may thus well turn out to have been a milestone in the history of semiotics. 3. TOWARDS A CLASSIFICATION OF SIGNS AND SIGN SYSTEMS

It must have become apparent from the foregoing that while numerous attempts were made by philosophers and philosophically inclined linguists, throughout the history of semiotics, to classify signs or systems of signs,41 none of them has as yet enjoyed universal and permanent acceptance. What follows is a non-exhaustive survey of what seem to me important criteria that must be taken into account in working toward a reasonably holistic categorization — surely a task to be completed in the distant future. The interrelationships among the classes that are suggested here are clearly of the kind that are convertible into Euler-Venn diagrams: that is, they overlap; this approach to classification seems to me unavoidable at this stage in the development of semiotic (cf. Lenneberg 1969). The ensuing attempt differs from previous ones chiefly insofar as it lays heavy emphasis on semiotically relevant data from current ethological literature (Sebeok 1972). In biology, as Ralph Gerard has suggested, it is useful to conceive of a material system — an org — in one of three ways: its 'being', or structure, that is, its en4

°a Essays in Semiotics/Essais de semiotique, eds. Julia Kristeva, Josette Rey-Debove, Donna Jean Umiker (The Hague: Mouton, 1971). 41 One of the more thoughtful attempts was Bally's "Qu'est-ce qu'un signe?" (1939). SpangHanssen's "Recent theories on the nature of the language sign" (1954; but completed in 1948), is the most comprehensive survey of 'recent' psychological sign theories (Ogden and Richards, Karl Britton, Bertrand Russell, Charles W. Morris), as well as of 'recent' linguistic theories (Ferdinand de Saussure, Leo Weisgerber, Alan H. Gardiner, Karl Buhler, Eric Buyssens, Leonard Bloomfield, Louis Hjelmslev). See also Mahmoudian's entry on "Signe" in Martinet 1969. Jakobson has recently devoted a special study to the classification of sign systems (1970) - as announced in Jakobson 1969:80, fn.

232

THOMAS A. SEBEOK

during status in a synchronic sense; its 'behaving', or function, 'a repetitive perturbation' along a secular trend; and its 'becoming' or history, representing cumulative changes in the longitudinal time section (Gerard 1957, 1960). A semiotic system — a sign — can likewise be fruitfully examined from each of these three points of view: we can ask, what is a sign, how does the environment and its turbulences impinge upon it, and how did it come about? Functional classifications of the sign are empirical, extrinsic; they are based upon variations at different nodal points in an expanded communication model. A structural identification of the sign is analytic, intrinsic; it utilizes types of associations potentially inherent in the architecture of the sign itself. The problem of becoming introduces diachronic considerations of two sorts: we can focus on the evolution of sign systems in phylogeny, or consider their development in ontogeny. What follows is organized according to this scheme, but diachronic considerations are omitted from this article. 1. Function

In delineating the scope of the field at the outset, there was an underlying pragmatic implication: when proposing to restrict the discussion to 'the semiotic manifestations of whole living systems' (p. 212 above), attention was deliberately focused upon the origin and/or effects of signs or, more generally, the sorts of relationships that can prevail between the source of a message and its destination. The first basic classificatory criterion can, therefore, be constructed in terms of the nature of all possible sources of signs, for it is reasonable to begin to classify where the coding itself begins — at the input end. According to the distinctions introduced in Section 1, these could be provisionally diagrammed as in Table 3: TABLE 3 SOURCES OF SIGNS

Inorganic objects Natural

Manufactured

Organic substances Extraterrestrial I Homo sapiens

Components

Organisms

Terrestrial Speechless creatures

Components

Organisms

This excessively simplistic scheme becomes at once more complicated as the nature of the sign-receivers — the other end of the feedback loop in the transactional chain — is taken into additional consideration. Man's conspecific messages can, for

SEMIOTICS: A SURVEY

233

example, be differentiated as intrapersonal, requiring only one participant engaged with a prospective or imaginary interlocutor (cf. the phenomenon of 'inner speech',42 or 'internal dialogue';43 interpersonal (dialogue, requiring an alter besides the ego)," or pluripersonal45 — in Eskimo oral tradition, far example, 'the myth-teller speaks as many-to-many, not as person-to-person (E. Carpenter 1960). A human message may be directed at a machine,48 or at a personified supernatural, as in an incantation or a prayer addressed to a deity (Sebeok 1962b); communication with ancestral spirits is prevalent and commonplace in many cultures, for example, in New Guinea (Eilers 1967:34—36). Vice versa, animates may receive signs from the environment — cf. the phenomenon of echolocation (Griffin 1968) — or fancy receiving them, 'as in some of the epigrams of Callimachus and of his imitators, the stone is thought of as carrying on a brief dialogue with the passerby' (Hadas 1954:50-51), or again from the location of stars and planets, the length and intersection of lines in the hand, the entrails of sheep, the position of dregs in a teacup — in brief, by those pseudo-semiotic divinatory techniques that are known variously as augury, astrology, palmistry, haruspication, and the like (Kleinpaul 1883: Ch. Ill; D. Kahn 1967:92). Animal senders and receivers of signs are either conspeoifics, or they belong to two or more different species, one of which may be man (cf. Hediger 1965, 1967). Accordingly, zoologists tend to classify sign systems first of all into intraspecific vs. interspecific varieties;47 an animal can direct its messages, moreover, at a particular member of its own species, or at an individual of another species.48 42

On the genesis and function of inner speech, cf. Egger 1904, and especially Vygotsky 1962. Cf. Peirce, IV, 6: ' . . . thinking always proceeds in the form of a dialogue - a dialogue between different phases of the ego - so that, being dialogical, it is essentially composed of signs . . . ' Also, id., VI, 338: 'All thinking is dialogic in form.' 44 Matson and Montague 1967, contains some useful, if recherché, selections viewing human communication as a dialogue. 45 A pluripersonal group may be centrifugal (one-to-many) or centripetal (many-to-one); see Table D, on p. 277, in Ruesch and Bateson 1951, and Ch. 11 generally, for a review of different types of coexistent networks of communication. For a discussion of 'pluripersonal* communication in alloprimates, see Altmann 1967. 4 * Cf. fn. 4. Dewan 1969 describes the author's experiments to voluntarily control his own alpha rhythm and to send a Morse code in this way by processing the resultant EEG by means of filters and computer program so that these messages are converted into typewritten outputs. For a characteristically pessimistic view of "The future of man-machine languages", see BarHillel 1968. 47 Thus Wynne-Edwards writes (1962:24): . . the primary subdivision in the functions of communication should clearly fall between signals directed at other members of the same species and those . . . directed at other kinds of animals; these divisions are respectively intra- and interspecific'. See further the tabular outline of animal behavior, based on the origin and destination of signals, in Wenner 1969. 48 For a remarkable example of the former, cf. the mutual auditory display of shrikes of the genus Laniarius: each pair of the species aethiopicus 'tends to develop a particular repertoire of duet patterns by which they can be distinguished from other pairs in the neighbourhood', as shown in Thorpe 1963; more generally, see Thorpe 1968. The extremely complex relation between picarian birds and their mammalian symbionts (ratels, baboons, or humans) furnishes an even more striking illustration of the latter; see Friedmann 1955. 45

234

THOMAS A. SEBEOK

Contact among emitters and receivers is established and maintained by miscellaneous flow-processes that link them across space and time, and our classification of sign systems becomes still further refined as this operationally crucial third factor — the medium of transmission — is taken into account. In principle, any form of energy propagation or transfer of matter can serve as a sign carrier, depending on an animal's total perceptual equipment. Sign systems are thus also distinguishable in terms of the channel or channels connecting the input side with the output side. If an animal's sensory capacity allows for the parallel processing of information through multiple input channels, calculable redundancies will be found to prevail (an effect sometimes referred to as 'the law of heterogeneous summation') (Sebeok 1968a:36), and the application of the rules for switching from one subassembly to another, under particular circumstances, will yield different hierachical arrangements that render the classification even subtler in consequence of contextual effects. From the semiotic point of view — to say nothing of an evolutionary standpoint — it is very important to appreciate fully both the advantages and the disadvantages of every channel (Table 4) utilized for successful communication in and across all species. TABLE 4 CHANNELS

Matter Gases

Liquids

I Optical Reflected daylight

Energy Solids

Proximal 1 Tactile

Bioluminescence

Chemical

Physical

Distal 1 Acoustic Air

1 Electric

Water

1 Thermal

1 Etc.

Solids

Yet the foundations for a comparative analysis of this kind have barely begun to be laid.49 For every semiotic system, it would be useful indeed to have a lucid and concise account, comparable with the following characterization of speech: 'Speech is produced with the human body alone, without any tool; it is independent from light and can be used day and night; it fills the entire space around the source and does not necessitate a straight line of connection with the receiver; it can also be 49

A typology is adumbrated, in summary fashion, in Sebeok 1967b. See further Geldard 1960, as well as Jakobson 1964, 1967; Sebeok 1967c.

SEMIOTICS: A SURVEY

235

greatly varied from intimate whisper to long distance shouting; and it involves a very small amount of energy (Lotz 1950). Among the criterial design features of speech just cited, Lotz singled out for first mention that signs encoded in this mode can be produced by the human body alone, without recourse to any tools. The use of both ready-made objects and shaped artifacts as tools is known to occur variously throughout the animal world, ranging from moths and spiders through birds, otters, and primates.50 Chimpanzees of the Gombe Stream Reserve, for instance, build nests, fold selected leaves to facilitate drinking or to wipe their body, use sticks, twigs, and grasses to get termites, ants, or honey, and use them, as well, as olfactory probes; they also use stones to break nuts, and employ both sticks and stones in agonistic displays (van Lawick-Goodall 1968). The instrumental behavior of these chimpanzees exemplifies a twofold function of tool using in animals and man: a (presumably) primary amplifying function, and a (presumably) secondary semiotic function. When the chimpanzee uses a stone to break nuts, it draws material supplies from its environment to adaptively extend and improve, with relatively moderate expenditure of energy, the manipulatory systems necessary for its existence in its living space. When, on the other hand, it uses a stone in agonistic displays, the primary amplifying function is endowed with a secondary sign function — the behavior has, in the parlance of the ethologists, become ritualized. The instrumental — as opposed to the merely somatic — production of animal signs sometimes takes exceedingly bizarre forms, as in those 'wonderful arena birds called bower birds, with their houses and ornamented gardens and their courtship displays that replace plumage with glittering natural jewelry' (Gilliard 1963).51 Both human and animal sign systems can thus be classified into organismal, according to whether they are produced by the body alone, or artifactual, produced by the body amplified. Diverse surrogates for spoken languages (cf. Stern 1957) can be contrasted in this way (among others, of course): for example, the transposition of speech into whistles is organismal, into drumbeats artifactual, although all these are manifested in the same acoustic substance; or into relatively autonomous graphic systems, or Braille (all in different substances, viz., for the addressee,52 50

On the use of tools by animals in general, see Hall 1963 and the comments that followed, in Current Anthropology 7:215-216 (1966); in moths and spiders, Wickler 1968:58, 131-132; in birds (viz., Egyptian vulture, Neophon peromopterus) reported in Time, January 6, 1967, under the heading, "Birds that throw stones"; in the California sea otter, Enhydra lutris, Hall and Schaller 1964; in primates, Hall 1963, and Kortland and Kooij 1963. 51 These birds and their display have evolved, along quite different lines, essentially into builders of avenue-type bowers, and builders of maypole-type bowers, but both types 'use displayobjects of various circumscribed colours which are collected with great discrimination . . . ' , according to Marshall 1954:170. 52 For the addresser, of course, the situation may be quite different: a written sign may be encoded chemically (e.g., in ink), but will be decoded visually; cf. Sebeok 1967c:1778. Signs encoded visually may be decoded by the blind through the skin of the back; cf. Bach-Y-Rita, et al. 1969. 53 On speech surrogates in general, see Donna Jean Umiker's contribution to this volume, and Sebeok and Umiker, 1974.

236

THOMAS A. SEBEOK

auditory, visual, tactile).58 A transitional grouping of implements is constituted by detachable parts of an animal's body itself; thus a particle of dung, or the trace of secretion on the territorial marking place, can function as a sort of synecdoche, with the animal's separated pars — that can continue to be efficacious even after the temporary departure or permanent extinction of the source — standing pro toto.M The identical semiotic process is the basis for contagious — in contrast to homeopathic — magic, where the intended victim's nail parings, or the like, act as the indexic signs.65 Objects used for semiotic display can, moreover, be distinguished further according to whether they are 'Found Objects', like a plucked flower, or matter, as precious metal, wrought into floral shape, to be worn as a brooch; prehistoric artists are said to have discovered, by a semiotic technique that is sometimes oalled 'épouser les contours' (Giedion 1960), in natural rock formations the images of the animals they sought.56 Another important distinction, also highly relevant to the problem of classification, results from contrasting 'pure' with other sorts of semiotics. Let us recall that Morris specified the task of pure semiotics to be the elaboration of 'a language to talk about signs' (cf. p. 222 above). Since it is obviously necessary to use signs for referring to signs in discourse about signs, and since, for this purpose, a signifier (or signans, or sign vehicle; see below) is commonly employed as an index of its own sign, such 'a language to talk about signs', or a semiotic system specialized for communicating about another semiotic system, can be said to constitute a metasemiotic (by analogy with 'metalanguage'). The object semiotic (by analogy with 'object language') then becomes any 'other' semiotic system communicated about in the metasemiotic.57 Morris assigned the study of such actual sign systems to descriptive semiotics. Confusingly, however, other scholars refer to the study of phenomena which arise when signs are actually used in the process of communication as applied semiotics (one might call this 'performance'), as against pure semiotics, which then becomes the study of the underlying rules which regulate the 54

Bilz 1940:195 develops the pars pro toto principle in a double sense: 'einmal als Möglichkeit sprachlicher Mitteilungen, indem der Erzähler eine der Teilfunktionen herausgreift und für das Ganze setzt, d. h. mit der Komponente das volle Funktionsbild meint oder das ganze Objekt bezeichnet, zum anderen aber verfährt auch die Natur in der gleichen Weise, indem sie das ganze Funktionsbild tatsächlich in seine Teilfunktionen aufsplittern kann und nur die eine oder die andere Komponente statt des Ganzen mobilisiert'. 55 Frazer, The golden bough (1951:43): The most familiar example of Contagious Magic is the magical sympathy which is supposed to exist between a man and any severed portion of his person, as his hair or nails . . 56 On semiotic organization by chance, see further O'Brien 1969. Communication by readymade objects as signs 'where as the vehicle of information serves the presented reality itself, has been called ostension by Osolsobë (1967:101), who explored the significance of 'showing things and persons actually existing and present in time and space . . . limited to what is at our disposal at the time of communication just now and just there' in the arts, particularly in the context of the theater. 57 Variants of this pair of terms were operationally used by Hjelmslev; cf. 1943, Def. 102: metasemiotic, 'a scientific semiotic one or more (two) of whose planes is (are) (a) semiotic(s)', and Def. 103: object semiotic, 'semiotic that enters as a plane into a semiotic'.

SEMIOTICS: A SURVEY

237

objective structure of signs ('competence'). Again, according to a distinction introduced by Jakobson, the object of pure semiotics is the linguistic sign, whereas applied semiotics deals with such systems of signs as may be embodied, for instance, in cuisine, garments, the products of gardening, or architectural constructs; (as he remarked, 'we don't inhabit signs').58 However, still others prefer to confine the latter expression to merely pedagogical applications. Morris himself intended by applied semiotics to circumscribe the field that utilizes 'knowledge about signs for the accomplishment of various purposes', a definition which I then extended to applied zoosemiotics. Since I have found Morris' trichotomy useful, especially in heuristic comparisons of human with animal communication, I prefer iit to a bifurcated model and will, at least provisionally, continue to use all three terms — pure, descriptive, and applied — much in the way he has defined them for semiotics and, by implication, mutatis mutandis, for zoosemiotics. Higher organisms are so constructed, both centrally and peripherally, that they are able to — although they need not — draw simultaneously upon two or more repertoires of sign systems. This capability, notably characteristic of man, allows for a high number and many kinds of admixtures of expressions. It also provides another criterion for classifying sign systems, as ranging from 'simple', homogeneous structures to syncretic formations of varying degrees of complexity, the components of which, furthermore, may or may not be patterned symmetrically. This last point can be illustrated as follows: if I assert 'yes', and, at the same time, nod my head affirmatively, my verbal sign is symmetrical — as well as mutually redundant — with my kinesic sign. In a Franz Lehar operetta, a Chinese mandarin sings an aria vigorously disclaiming his alleged susceptibility to bribery, while, simultaneously, he keeps, thrusting his palm out behind his back: the audience is given to clearly understand that, of the two asymmetrical, in fact, wholly contradictory message components, the kinesic one is to be interpreted as true and the acoustic one as false.69 What Lotz (1950:717) has called a 'ribbon concept' of communication — a notion which, of course, overlaps, yet is not identical, with what Birdwhistell (e.g. 1968) and others have called 'multi-modal' communication60 — seems to charac58

Lecture on "Language in relation to other communication systems" delivered on October 14, 1968, reported in de Mauro and Grassi 1969, and witnessed by the author. 59 A striking illustration of asymmetry between an acoustic and a visual message component was cited by Sebeok in 1962a: fn. 8: 'A patient was told under hypnosis she must raise her right index finger to indicate an affirmative answer, left for negative. Brought out of trance, she was questioned by her doctor. Unable to face her emotional problem, she shook her head vigorously in manifest negation, but her right finger shot up, showing that the correct answer was "yes".' The problem area was recently reviewed and discussed in Ekman and Friesen 1969, but the classic treatise on the subject remains Quintilian's Institutes of Oratory (see esp. Book XI, Ch. Ill), from the first century A.D. 40

Mary Catherine Bateson writes (1968:10-11) that 'we must assume that coded linguistic communication is concurrent with little known types of communication in other modalities, and that the overall system includes rules about the way in which the codes interact and what kinds of code switching are possible'. See also the remarks by several participants in the discussions

238

THOMAS A. SEBEOK

terize most of normal human signaling behavior. Were this imaginary ribbon to be sectioned at any point in time, it would, much more often than not, reveal verbal, paraphonetic, kinesic, proxemic, olfactory, and perhaps other strands of signs mixed together but stacked in a hierarchy suited to the requirements of the global message involved. Signs are combined in appropriate — linguists would say 'grammatical' — ways to yield a compound designatum (or signatum, that is, content that is signified; see below). The emotional attitude of a speaker towards his verbal message can be conveyed either by features — emotive, expressive — coded by the language itself (Stankiewicz 1964), or by accompanying non-verbal features — paraphonetic, kinesic, etc. — that serve to support or belie it (contributing to effects like irony or sarcasm). The revealing character of the non-verbal features when the verbal performance is latent was underscored by Freud (1959:94) in an often-cited remark: 'He that has eyes to see and ears to hear may convince himself that no mortal can keep a secret. If his hps are silent, he chatters with his finger-tips; betrayal oozes out of him at every pore.' The semiotic systems called 'secondares modelants', or macrosystems — that is, those that imply, by definition, a natural linguistic infrastructure (as discussed in fn. 38) — all constitute elevated of cases syncretism, including those with a predominantly folkloristic and artistic function. Here belong all genres of discourse composed of prefabricated texts, such as myths and the other products of oral tradition; the verbal arts in general and poetry in particular; such hybrid formations — labeled with transitional awareness — as vocal music, circus acts, dramatic performances, sound film; and blended structures of the highest intricacy, as opera. To illustrate: a sound film can partake of four sign systems, one visual and three auditory — language, music, and other sound-effects ('bruit');61 an acrobatic act partakes of at least five, namely, the performer's dynamic behavior, his social behavior, his costume and other accessories, the linguistic accompaniment, and the musical accompaniment (Bouissac 1968, 1969, 1970). As to the art of the opera,62 which, in Stender-Petersen's formulation, irreversibly presupposes the reported in Sebeok, Hayes and Bateson 1964, passim. The logician, Bar-Hillel (1970:3), likewise insists 'on the fact that, in general, human communication proceeds in more than one channel at a time, that each channel has its own specific properties, and that their interconnection creates possibilities whose number is greater than the sum of the possibilities of each channel taken separately, a fact which, of course, also raises questions as to the precise implications of this unusual situation'. " Cf. various essays by Metz, in particular 1968:113-114. On the semiotic of film, see, i.a., the following works: Metz 1968; Bettetini 1968; Garroni 1968; Wollen 1969 (esp. Ch. 3, on "The semiology of the cinema"); and Worth 1969; and the special issue of Sprache im Technischen Zeitalter, No. 27 (1968), entirely devoted to "Zeichensystem Film - Versuche zu einer Semiotik". 62 Cf. also Sebeok 1970:114-115, including the following illustration: 'Imagine, if you will, a production of Don Giovanni, which is, to begin with, coded in a natural language, say, Italian. The secondary code would be its libretto, the drama Lorenzo da Ponte created out of the old Spanish legend. In the third place, there is the score, the way Mozart enhanced the drama musically. Then, in performance, there is a whole variety of visual effects, partially controlled by the lighting, involving the scenery, and a host of costuming devices, helping keep apart in

SEMIOTICS: A SURVEY

239

literary, the musical, and the scenic arts, it is a semiotic product that can, indeed, achieve 'un degré très élevé d'indépendance raffinée' (Stender-Pedersen 1949).

2. Structure Rulon Wells (1967) tells us that 'Semiotic has two groups of affinities. It is connected, on the one hand, with communication, and, on the other, with meaning.' If, in this article, I have so far focused on phenomena of usage, particularly those that pertain to miscellaneous transactional relationships, this was not for lack of interest in such facts as intention;68 therapeutic explorations of man's paysage intérieur by Freud and by Jung (Jung et al. 1968), Lacan (1966, 1968, Bär 1971), or Shands (1969, 1970, 1971), and other psychiatrists; in those symbolic forms in the dimensions of our milieu extérieur studied, among many others, by Ernst Cassirer (1923, 1924, 1929, 1944) and his epigones (e.g. Langer 1948), or the dialectic 'dramatism' of Kenneth Burke (1966; reviewed by Hymes 1968) and those who would apply his model of symbolic action in the various social sciences;64 nor in the fascinating problems of Weltanschauung,K as advanced by Wilhelm von Humboldt (cf. Miller 1968: Ch. 2, the influence of Humboldt on Cassirer), reexamined by appearance the distinct personalities, for example, of the bourgeois Donna Anna from the selfsacrificing Donna Elvira, and both from the peasant girl Zerlina. There is the cunning culinary effect of the banquet scene, and, if you are sitting close enough to the stage, there are olfactory effects achieved by the smell of the greasepaint. The genius of Mozart finally orchestrated this whole masterpiece into an eloquently tight and unified macrostructure. Don Giovanni is a very complex construct, with many components, but its infrastructure is still a natural language.' 63 The place of 'intention' - or, more broadly, goal-orientation - in a communication model constitutes an entangled and controversial problem. In the sense of self-awareness - so-called 'subjective teleology' - the notion may be criterial in the definition of all anthroposemiotic systems, and notably characterizes language (as D. M. MacKay is wont to insist, e.g., in his 1961 article, where he points out that semantic questions can find their natural place in information theory if and only if language is pictured in the shape of a 'goal-directed self-adaptive' diagram). Jakobson (1970:10), for example, speaks of symptoms as 'mere unintended indexes as a subspecies of a vaster semiotic class', and of 'unwitting indexes' as a variety of signs. But this would hardly be pertinent to zoosemiotic systems, as implied in Verplanck 1962. Evans and Bastian 1969 convincingly argue against the 'stultifying effects' of the groundless conception of intentionality in the social behavior of the bottle-nosed dolphin and other marine mammals. On the other hand, in the sense of implying explanations - so-called 'objective teleology' - to account for directively organized structures, mechanisms, and patterns of behavior, the concept may not only be appropriate but is perhaps indispensable for the study of communication in all organisms; cf. Ayala 1968; for the use of teleological conceptions in the analysis and description of behavior generally, see also Miller, Pribram, and Galanter 1960. On the two kinds of teleology, see the twin articles by Hofstadter 1942a and b. 64 Sociological reflexes, exemplified by Duncan 1968, reviewed by Hymes 1969; anthropological reflexes, exemplified by the new University of Chicago Press series in Symbolic Anthropology, edited by David W. Crabb - see Peacock 1968, with a foreword by Hymes, setting the series firmly in the Burke tradition; kindred in spirit are the works of Turner, e.g. 1967 and 1969. 65 For a recent thorough review of the Weltanschauung problem as a whole, see Rossi-Landi 1968.

240

THOMAS A. SEBEOK

Edward Sapir (1929) and popularized by Benjamin Lee Whorf (Whorf 1956), resisted by Max Black (1868:71-75), and inverted by Adam Schaff, with the assertion that 'Reality shapes language...' (1962:349). It was, rather, because I share Weinreich's judgment that semantic analysis quickly embroils all venturers 'in some of the more inconclusive epistemological controversies of social science' (Weinreich 1968:165s6), and because the theories of both reference and meaning remain as opaque and intractable as ever.67 And if semantic theory is incoherent, hardly reliable when applied to natural languages — save, to a limited extent, in the domain of highly systematic grammatical elements, less so for the more loosely structured vocabulary, but practically not at all for compound signs (idioms, sentences in general) — zoosemantic theory, properly speaking cannot even be asserted to exist.68 Nevertheless, the nature of the coupling between the moieties that are traditionally and universally recognized as the minimal components of any sign has perennially been a crucial issue for anyone seriously interested in semiotics, and therefore at least a rudimentary acquaintance with the composition of signs is a prerequisite for even their mere classification. A sign, then, by all accounts — from Stoic philosophy to contemporary thinking — is conceived of as constituted of two indispensable halves, one sensible, the other intelligible: the signifier, a perceptible impact on at least one of the sense organs of the interpreter; and the content signified. Dante incisively formulated this, in 1305, in his unfinished. De vulgari eloquentia (1-3), as follows: 'hoc equidem signum sensuale quid est, in quantum sonus est; rationale vero, in quantum aliquid significare videtur ad placitum' ('this sign . . . being, as to sound, sensible by nature, but rational [i.e. intelligible] in so far as it carries some meaning by convention [i.e., arbitrarily]') (1957:18; 1890:6). (In medieval Latin, the corresponding pair of terms for signifier and signified was signans and signatum, rendered by Saussure as signifiant and signifié, by Morris as sign vehicle and designatum, etc.). The distinction between the two can be crudely illustrated by the following anecdote: a Danish photographer's apprentice is sent to America to learn to improve his technique in portraiture; there he is told to always have his subjects say cheese; 66

Would that S0rensen's remark (1967) were true in a serious sense: 'Since the notion of meaning is a basic notion, one would think that everyone must know the meaning of "meaning". This, in point of fact, is what everyone does.' •7 In spite - or just because - of this, I have spared no editorial efforts to obtain adequate coverage of these topics, wherever pertinent, in the Current Trends in Linguistics series; the following major chapters have appeared: in Vol. 1, Soviet and East European linguistics, Weinreich, on Soviet "Lexicology" (1963); in Vol. 3, Theoretical foundations, id., "Explorations in semantic theory" (1966); in Vol. 9, Linguistics in Western Europe, Ullmann, "Semantics" (1972); in the present volume, 12, Linguistics and adjacent arts and sciences, three separate chapters, to wit, by Coseriu and Geckler, "Semantics and comprehension"; by Clark, "Semantics"; and by Werner et al., "Ethnographic semantics". •8 Two recent papers by W. John Smith (1969a, 1969b) are especially important for emergent zoosemantic theory. Jakob von Uexküll's 1940 Bedeutungslehre (1970) is the classic treatise on zoosemantics.

SEMIOTICS: A SURVEY

241

upon his return to Denmark, he instructs them to say ost — the joke turns on his confusion of the bilingual signifiers with the common signified. In various systems of signs, notably in language, a sign vehicle can sometimes — when the contextual conditions are appropriate — signify by its very absence, occur, that is, in zero form.9» A particular occurrence of a sign — what Peirce labelled a 'sinsign' — is now more commonly called a token, whereas the class of all occurrences of the sign — Peirce's 'legisign' — is called a type. Let us reapply his own illustration: if a page in a book has 250 words upon it, this is the number of word tokens, whereas the number of different words on the page is the number of word types.70 Now the two principal sets of questions that have preoccupied most students of the sign are these: (1) how do particular sign tokens refer, how do sign types have the constant capacity to mean-, and (2) what, precisely, is the distinction between the relation of reference, or denotation, and the relation of meaning, or designation?71 The theory of verbal reference and the theory of verbal meaning are usually treated together under semantics-,72 the word zoosemantics was recently coined to accomodate presumably corresponding processes among animals (Sebeok 1965). Although an understanding of denotata and designata, and of the differences between reference and meaning, are absolutely essential to explain a vast number of facts of semantics and zoosemantics, there is no consensus among philosophers on any theory according to which the data accumulated by linguists, over many centuries, from the languages of the world, and ethograms, gathered by ethologists, " Linguists who employ the expression 'zero sign' (viz. zero phoneme or allophone, zero morpheme or allomorph, and the like), mean, of course, 'zero sign vehicle', or, rarely, 'zero designatum', but never both; if taken literally, the notion of a 'zero sign' would be oxymoronic. On the use of zero in linguistics, see, e.g., Jakobson 1966 (article originally published in 1939); or Jakobson 1940; Frei 1950; Godel 1953; Haas 1957. (Pohl 1968:34-36, remarks that civilian clothing functions as a zero when worn in a context of uniforms; this, however, confounds the unmarked/marked opposition with the realized/zero opposition, a better illustration of which might be clothing when worn in a nudist camp!) The role of zero sign vehicles in animal communication systems has never been analyzed; one line of possible inquiry is suggested by the fact that, in several species of fireflies, pulse interval is a significant element in stimulating females, and that the pulse intervals are different in Photinus consanguineus and macdermotti-, in the related lineellus, furthermore, the pulse number is variable, 'which further indicates the significance of pulse interval' - see Lloyd 1966:78. Ardrey (1970:75) claims that the African 'elephant's alarm call is silence'. 70 Cf. fn. 18. For a recent exploration of this distinction, see Richards 1969. 71 Although most theorists have distinguished the capacity of signs to refer from their actual referring, this particular distinction between Sinn and Bedeutung derives from Gottlob Frege; see his 1892 article; cf. Husserl's Bedeutung vi. Bezeichnung, Mill's connotation vs. denotation, Paul's Bedenkung vs. Benutzung, Saussure's valeur vs. substance, etc., etc. For a critical discussion of Frege's sense-reference distinction, see Ch. Ill in Grossman 1969. (Another important question, not considered below, is indicated by Rulon Wells in the title of his article exploring the subject, "Meaning and use" [1954].) 78 But see the strictures of Quine 1963:130: 'When the cleavage between meaning and reference is properly heeded, the problems of what is loosely called semantics become separated into two provinces so fundamentally distinct as not to deserve a joint appellation at all.'

242

THOMAS A. SEBEOK

over several decades, in a growing number of species,78 can be coherently analyzed, ordered, compared, and generally comprehended. A few of the many linguistic, ethnolinguistic, and psycholinguistic attacks on the problems of reference and/or meaning — which could, of course, also be supplemented by the works of countless philosophers and logicians74 — fall into the following (partially overlapping) groups: A) a whole spectrum of 'traditional' approaches, as practiced mainly in Continental Europe and Great Britain, comprehensively reported in several works of Stephen Ullmann;76 B) the so-called 'structural' semantics of discourse, as exemplified by the recent book of the semiotician A. J. Greimas78 (Greimas 1966; reviewed by Ullmann 1967, Baumann 1969); 73

Niko Tinbergen fixed the 'ethogram' as the starting point of ethological investigation when he proclaimed that 'special emphasis should be placed on the importance of a complete inventory of the behavioral patterns of a species' (1951:78). 74 Analytic philosophers typically assign the theory of truth and the theory of logical deduction to semantics, on the ground that truth and logical consequence are concepts based on the relation of designation and hence semantic concepts (Carnap 1942:10); this view is developed, for instance, in a series of papers by Donald Davidson (most recently, 1970). According to Ryle 1957, 'Answers to this highly abstract question, What are meanings? have, in recent decades, bulked large in philosophical and logical discussions. Preoccupation with the theory of meaning could be described as the occupational disease of twentieth-century Anglo-Saxon and Austrian philosophy.' S. J. Schmidt 1969 is a recent survey of philosophical trends, aware, however, of linguistic contributions as well. 75 Most recently, in Ullmann 1972; q.v. for references to his other, book-length treatments, including especially his Semantics: An introduction to the science of meaning (1962). See also Todorov 1966; Alain Rey 1969 (with additional bibliographical indications on pp. 124ff.), and Coseriu and Geckeler, this volume. 76 The word 'structural', used by Greimas in his title, has only approximate kinship with the same word as used by John Lyons in the title of his book, Structural semantics: An analysis of part of the vocabulary of Plato (1963). Lyons's 'structural' semantics is an attempt to update the concept of Wortfelder, of lexical subsystems, developed by J. Trier (cf. Trier 1931), and others, and is loosely coupled to Chomsky's transformational model (see also Lyons 1968, esp. Chs. 9 and 10.) The relationship of Weisgerber to structuralism is reviewed in Schaeder 1969. On the other hand, as already suggested in fn. 31, Greimas appears to have been principally influenced by Hjelmslev, particularly perhaps by the latter's "Pour une sémantique structurale", written in 1957 and published in 1959. Cf. also the Plenary Session, devoted to the question "To what extent can meaning be said to be structured?", at the Eighth International Congress of Linguists, with important contributions by Hjelmslev and by Wells, followed by an interesting discussion, as published in the Proceedings, pp. 636-704 (1958). A second important source of Greimas's semantics was the first English edition of Propp's Morfologija skazki (Morphology of the fairytale) (1958, 1968 2 ; Russian originals, 19281, 19692), as pointed out by Eleasar M. Meletinskij, in his afterword to the second Russian edition (Moscow: Nauka), and again in his 1969 article. The word structure, and its derivatives, structural, structuralism, are among the most overburdened clichés of contemporary social thought (especially of France). A few among the seemingly endless references - selected here because of their partial pertinence in the context of semiotic studies - are: Bastide 1962; Viet 1965; Ehrmann 1966; Bierwisch 1966; Piaget 1968; Ducrot, Todorov, Sperber, Safouan, and Wahl 1968; Macksey and Donato 1970. Three elementary introductions may also be consulted: Auzias 1967; Fages 1968, and Lane 1970. In all this welter of publications on 'structuralism', and despite the fact the 'structuralism' is intimately linked with modern semiotics, no exhaustive treatment exploring the interrelationship of the two

SEMIOTICS: A SURVEY

243

C) the semantic differential (developed by C. E. Osgood), allied free-association techniques, and psycholinguistic analyses more generally;77 D) the method of 'back translation', and related techniques derived from anthropological fieldwork;78 E) methods of cognitive anthropology:79 F) methods aiming at compatibility with generative-transformational approaches to grammar, particularly to syntax.80 A noticeable discrepancy between what a sign type designates and the denotatum of one of its tokens may be responsible, on various levels, for the linguistic processes known as 'figures of speech' (e.g., metaphor), as well as perhaps kindred phenomena found in animals.81 This is also the mechanism involved in lying, which — certain opinions notwithstanding — corresponds to various forms of deception found throughout the animal kingdom.82 exists; see, however, the essay by Segre 1969. A book on Structuralism around the world is currently in preparation under Sebeok's editorship, for the Approaches to Semiotics series. 77 The principal references to the work of Osgood and his associates were given in fn. 27. In general, see the Journal of Verbal Learning and Verbal Behavior, Vols. 1 (1962) onwards, passim; Brown 1958, especially Ch. 3, on "Reference and meaning"; Carroll 1964, section on "Meaning as a problem for psychology", pp. 33-42; and Clark, this volume. 78 Phillips 1959-60; Nida 1964: Chs. 3-5; Nida and Taber (in preparation); and Nida's article on the "Science of translation" (1969), with further references given in his fn. 4. [Nida's Componential analysis of meaning will also be published, in 1974, in the Approaches to Semiotics series.] Some of the problems involved in this approach were anticipated by Susan M. Ervin, in 1954 [1965]. 79 This expression covers roughly the same area in anthropology as 'formal analysis', 'componential analysis', 'folk taxonomy', 'ethnoscience', and 'ethnosemantics'. For the latest collection of readings, containing several of the classic contributions to this approach to semantic analysis, see Tyler 1969; also, Bendix 1966; and Werner et al., this volume. 80 Examples of various approaches to the semantic interpretation of syntactic structures are: Katz 1966, 1967; Weinreich 1966; Bierwisch 1969. McCawley 1968 represents a position intermediate between Weinreich's and Katz's, roughly, that deep grammatical structure is the same thing as semantic structure and that, therefore, a separate semantic component of the type envisaged by Katz (and some of his collaborators) is superfluous. Chomsky's "Deep structure, surface structure, and semantic interpretation" (1971) convincingly reemphasizes that properties of surface structure do play a role in determining semantic interpretation. Further references will be found in Katz's bibliography and several of the aforementioned papers. Krenn and Miiller 1970 assembles a bibliography of 146 items of 'generative' semantics to accompany their note on the subject. 81 The problem was posed in Bronowski 1967, thus: 'Does any animal language have figures of speech?*, and he observes that 'This is an interesting question because the answer is not straightforward, and throws light on how we ourselves couple the ambivalent meanings in human language'. If this question means, rephrased in more formal terms, 'whether an animal ever uses the same gesture in two different meanings', I am convinced that the answer is yes, and commonly so, for the reasons suggested by Smith, namely, that, although the number of displays in any species of birds and mammals is highly restricted (from about 15 to 45 per species), contextual information 'greatly extends the set of events concerning which there can be communication by means of displays' (Smith 1969b: 149). 82 Cf. Caws 1969: 'truth . . . is a comparative latecomer on the linguistic scene, and it is certainly a mistake to suppose that language was invented for the purpose of telling it.' Sturtevant 1947:48 speculated that 'language must have been invented for the purpose of lying'; Hockett asserted that 'Lying seems extremely rare among animals', in connection with his design feature

244

THOMAS A. SEBEOK

Recognition of the manifold possible relations between signifier and signified has led to corresponding classifications of signs, among which Peirce's maximal scheme, with sixty-six varieties (Weiss and Burks 1945), was surely the most exuberant. Today, only some half a dozen types of signs are regularly identified and commonly employed (with but roughly comparable definitions). It should be clearly understood, furthermore, that it is not signs that are being classified here at all, but rather aspects of signs. In other words, a sign may exhibit more than one aspect, so that we may recognize differences in gradation. For instance, a symbol, such as an imperative, may also be endowed with signal value; an emblem, such as the U.S. flag, is also partially iconic; or a primarily indexic sign, as a directional arrow, may possess a discernible symbolic component in addition.88 Aspects co-occur in a definite hierarchy; thus the sign is legitimately, if loosely, labeled after the aspect that ranks predominant. The following terms seem to occur most frequently in contemporary semiotics: 1) SIGNAL. When a sign token mechanically or conventionally84 triggers some action on the part of the receiver, it is said to function as a signal. Writing in connection with alloprimates, C. R. Carpenter defines signaling behavior generally, in many qualities, forms, and patterns, as 'a condensed stimulus event, a part of a longer whole, which may arouse extended actions. Signaling activity, in its simplest form, is produced by an individual organism; it represents information; it is mediated by a physical carrier, and it is perceived and responded to by one or more individuals. Like the stimulus event, of which signaling behavior is a special case, this kind of behavior RELEASES more energy than is used in signaling' (Carpenter 1969:44; see also Resnikow 1968). Examples: the exclamation, 'Go!' or, alternatively, the discharge of a gun starting a footrace; sonic bursts emitted by an echolocating marine mammal or a bat.85 #14, as enumerated in (1966); there is abundant evidence to the contrary (Altmann 1967:353-355, under "Prevarication"). On lying as a linguistic problem, see Weinrich 1966. For the concept of lying in animals in general, see i.a., Ch. 8, on "Lügenerscheinungen im Tierreich", in Kainz 1961; Hediger 1968:150f.; and Lorenz and Leyhausen 1968:370 (by Leyhausen). A recent report, furnishing a clear instance of this phenomenon in foxes, will be found in Riippell 1969, so that Pfeiffer's claim (1969:265), that pretense or deceit 'is peculiarly primate behavior; it is seen in no other species', is certainly wrong. 83 Cf. Count 1969:102: 'Symbolization . . . is supposable as a matter of a continuous (qualitative) degree.' 84 'Signals m a y . . . be provided by "nature", but they also be produced artificially', as pointed out by Kecskemeti 1952:36. 85 Let us recall here, in passing, that, according to the well-known and widely influential 'organon model' of Karl Bühler (1934:28 and passim), the signal appeals to the destination, whose behavior it governs (i.e., it is an elicitor or inhibitor of action); the symptom has to do with the source, whose inner behavior it expresses; and the symbol relates to the designation ('Es ist Symbol Kraft seiner Zuordnung zu Gegenständen und Sachverhalten, Symptom (Anzeichen, Indicium) Kraft seiner Abhängigkeit vom Sender, dessen Innerlichkeit es ausdrückt, und Signal Kraft seines Appels an den Hörer, dessen äusseres oder inneres Verhalten es steuert wie andere Verkehrszeichen'). For further animal examples of a signal, consult, e.g., Burkhard, Schleidt, Altner, et ah, 1967, Sebeok 1968, and Sebeok and Ramsay 1969.

SEMIOTICS: A SURVEY

245

( 2 ) SYMPTOM. A compulsive, automatic, nonarbitrary sign, such that the signifier is coupled with the signified in the manner of a natural link. A SYNDROME is a configuration of symptoms. (Both terms have strong, but not exclusively, medical connotations.86) Examples: fever is a symptom of disease; Hediger (1968:144-45) remarks of the excrement of giraffes that, 'normally, the falling of faeces should give a typical rustling sound [but] if the excrement is voided in shapeless, pattering portions, this is an important guide to the [zoo] keeper' as to the animal's state of health.87 ( 3 ) ICON. 88 A sign is said to be iconic when there is a topological similarity between a signifier and its denotata. Examples: a photograph, an algebraic formula; 86

Cf. Ostwald 1964. The denotata of symptoms and syndromes are generally not the same for the addresser (viz., 'the patient') and the addressee (viz., 'the physician'); for some Freudian implications of this observation, cf. Brown 1958:313; cf. also Kecskemeti 1952:61. Two fundamental books in this classic area of semiotics are Ruesch 1957 and 1961. It is interesting to note that the subtle Port-Royal logicians drew a distinction between 'ordinary' symptoms and what physicians call 'vital signs', on the basis of an essentially quantitative criterion; thus they observe that 'il y a des signes certains . . . comme la respiration l'est de la vie des animaux; et it y en a qui ne sont que probables . . . comme la pâleur n'est qu'un signe probable de grossesse dans les femmes ' I n other words, the terms 'compulsive, automatic', in my definition, are subject to a probabilistic refinement, for, although the denotation of a symptom is always equivalent to its cause, some symptoms are effectively connected with an antecedent condition 'for sure', whereas the link of other symptoms with the foregoing state of affairs is merely assumed with varying degrees of likelihood. Arnauld and Nicole, furthermore, recognize the emblem ('Il y a des signes joints aux choses, . . . comme l'arche, signe de l'église, étoit jointe à Noé et à ses enfants . . .*), and, at least in my reading, both the icon and perhaps the index as well, as in the following remarkable passage: 'La troisième division des signes est, qu'il y en a de naturels qui ne dépendent pas de la fantaisie des hommes, comme une image qui paroit dans un miroir est un signe naturel de celui qu'elle représente et qu'il y en a d'autres qui ne sont que d'institution et d'établissement, soit qu'ils aient quelque rapport éloigné avec la chose figurée, soit qu'ils n'en aient point de tout ' Citations are from a late edition of La Logique, ou l'art de penser (1816; 16621), after Brekle 1966:110-111, with whose interpretation, however, I cannot fully agree (especially his remark that 'Für die Zeichentheorie . . . ist diese Gattung nicht interessant'); now that the first variorum edition of the Port-Royal Logic is about to be published (see Linguistics, No. 49, p. 78, fn. 7), hopefully the contribution of its authors to semiotics can be definitively assessed. 87 Symptomatic displays - sometimes called 'autonomic effects' - were acutely observed and described by Charles Darwin 1872. In a famous passage, pp. 101-102, he remarked 'that the erection of the dermal appendages', in a variety of vertebrates, 'is a reflex action, independent of the will; and this action must be looked at, when occurring under the influence of anger or fear, not as a power acquired for the sake of some advantage, but as an incidental result, at least to a large extent, of the sensorium being affected. The result, in as far as it is incidental, may be compared with the profuse sweating from an agony of pain or terror' in man. 88 On the notion of the icon, cf. fn. 19; on iconic functions in language, cf. Jakobson 1965, and Valesio 1969. Peirce (II, 277) distinguished three subclasses of icons: images, diagrams, and metaphors. The theory of diagrams loomed large in Peirce's semiotic researches, and became the subject of several recent, still unpublished, U.S. dissertations: Don D. Roberts, The existential graphs of Charles S. Peirce (University of Illinois, 1963), and J. Jay Zeman, The graphical logic of C. S. Peirce (University of Chicago, 1964). 'The investigation of diagrams', Jakobson remarks, 'has found further development in modern graph theory', the implications of which, for semiotics, although unmistakable, have never been scrutinized; cf., however, the provocative piece by Gardner 1968, with further references suggested on p. 144. [Note: a revised version of Robert's book appeared in 1974 in Mouton's Approaches to Semiotics series.]

246

THOMAS A. SEBEOK

the hind end of an aphid's abdomen, which, to an ant worker, signifies the head of another ant, and the kicking of the aphid's hind leg represents an imitation of the antennal movement of an ant (Kloft 1959). (4) INDEX. A sign is said to be indexic insofar as its signifier is contiguous with its signified, or is a sample of it. Examples: a clock; the linguistic categories known as 'deictics', notably the 'shifters' (such as the personal pronouns of English);89 the tail-wagging 'dance' performed by a bee on a horizontal surface.90 5) SYMBOL. The operative words in the foregoing definitions of icon and index were, respectively, 'similarity' and 'contiguity'. A sign without either similarity or contiguity — but only with a conventional link — between its signifier and its denotata, and with an intensional class91 for its designatum is called a symbol. (Admittedly, this is the most abused term in the set. In consequence, it has either tended to be grotesquely over-burdened,92 or reduced to such general kinds of behavioral phenomena as 'stimulus',93 or even to nullity.94 The term EMBLEM is sometimes used as a partial — more restricted — synonym for symbol, usually formalized in the visual modality; thus one might say that the 'Stars and Stripes' is either the symbol or the emblem of America, but one cannot say that H2O is a '•chemical emblem'.95 Symbols are often asserted to be the exclusive property of 89 On the notion of deixis, which is, of course, the Greek word for 'indicating', see Frei 1944; and Lyons 1968:275-281. On the notion of shifters, see Jespersen 1964:123-124; Sturtevant 1947: 135-136; and Jakobson 1963: Ch. 9. Cf. also Peirce II, 289. 00 When the bee (Apis mellifera) dances on a horizontal surface, 'the direction of a waggling run points directly to the goal', that is to say, the sign is indexic. If, however, the dance takes place on the vertical comb surface - as is the case, as a rule, in the dark hive - then 'the dancer transposes the solar angle into the gravitational angle', that is to say, the sign becomes symbolic. Data reported in von Frisch 1967:230-231. 91 An intensionally defined class is one defined by the use of a prepositional function; the denotata of the designation are defined in terms of properties shared by all, and only by, the members of that class. Cf. Reichenbach 1948:193. In the terminology of Lewis 1946:39, intension refers to 'the conjunction of all terms each of which must be applicable to anything to which the given term would be correctly applicable'. For a standard technical account of the method of extension and intension, see especially, Carnap 1956; and cf. Stanosz 1970. See further fn. 98, below. 92 Cf., e.g., the Cassirer, Langer, and Burke references in the bibliography and related citations given in fn. 64, above; also White 1940, who wrote: 'Human behavior is symbolic behavior; symbolic behavior is human behavior. The symbol is the universe of humanity . . . the key to this world and the means of participation in it is - the symbol'. This seems also to be the point of view espoused by the founder of the International Society for the Study of Symbols, Theodore C. Kahn; see his 1969 article. Kantor 1936:63: ' . . . the term symbol is made to do duty for everything the psychologist calls a stimulus.' 94 Cherry 1966:309, for example, writes: 'We avoid the term symbol as far as possible in this book . . . ' In many standard textbooks of linguistics, the term is never mentioned; Landar's Language and culture (1966), and Yuen Ren Chao's Language and symbolic systems (1968), constitute two conspicuous exceptions. 95 Following a proposal of D. Efron's (1941), Ekman and Friesen, in their ' T h e repertoire of nonverbal behavior: Categories, origins, usage, and coding", 1969b, reintroduce and sharpen the notion of emblem as a category of semiotic behavior: 'Emblems differ from most other nonverbal behaviors primarily in their usage, and in particular in their relationship to verbal behav-

SEMIOTICS: A SURVEY

247

humans, but the ability of living things to form intensional class concepts obtains far down in phylogenesis, and this capacity for organizing universals from particulars has been provided with a mathematical-neurological rationalization nearly a quarter of a century ago (Pitts and McCulloch 1947; cf. Arbib 1971). Both according to the above definition of a symbol, and the more common Aristotelian ones resting on the doctrine of arbitrariness, that were promoted in linguistics especially by William Dwight Whitney and, after him, Saussure,96 animals undoubtedly do have symbols. I have previously cited this example of 'arbitrariness' in animals: tail movements in a dog designate friendship, in a cat hostility, but in a horse they merely denote the presence of flies; many other illustrations could be readily rehearsed.97 (6) NAME. A sign which has an extensional

class for its designatum

is called

a

name. Thus individuals denoted by the proper name 'Veronica' have no common property attributed to them save the fact that they all 'answer' to 'Veronica'.99 Names appear to belong to that mode of signifying which Morris labelled identifiers. He distinguishes three kinds of these, of which namors and indicators seem pertinent here. Namors are linguistic identifiors, hence they include singular signs, like proper names. Indicators, on the other hand, are non-linguistic signs; such 'names' are universally incorporated — ex hypothesi — into the calls of birds and mammals.100 ior, awareness and intentionality. Emblems are those nonverbal acts which have a direct verbal translation, or dictionary definition, usually consisting of a word or two, or perhaps a phrase. This verbal definition or translation of the emblem is well known by all members of a group, class or culture . . . . People are almost always aware of their use of emblems; that is, they know when they are using an emblem, can repeat it if asked to do so, and will take communicational responsibility for it.' For a more recondite use of the term, in a different semiotic domain, see Hollander 1959. 86 Even Leonard Bloomfield, in Language (1933:145), reemphasized that 'the connection of linguistic forms with their meanings is wholly arbitrary'. For the controversy engendered by Saussure's principle of the arbitrary nature of the sign, see the references in the headnote (p. 191) to Sechehaye, Bally, and Frei, "Pour l'arbitraire du signe", first published in 1940-1941, as reprinted in Godel 1969. For a recent review of the tenet of arbitrariness in language, see Bolinger 1968:14-17, 240-243, who, long before, wrote an article entitled "The sign is not arbitrary" (1949). For a more complete survey, see Engler 1962. •7 Haldane 1955; Altmann 1967:339: 'With few exceptions, the semantic social signals that have been studied in primates so far are arbitrary representations'; Bronowski 1967:376: 'It might be thought that because only human beings think with arbitrary symbols, they are also alone in speaking with them. But once again, this is not so.' The editor of the international reference work, Bibliographie zur Symbolik, Ikonographie und Mythologie (1968), Manfred Lurker, admits in his Foreword, p. 4, that 'Sicher gibt es Symbolerscheinungen im Reich der Tiere', but elects to restrict its coverage to humans. • 8 An extensionally defined class is one defined by listing the names of the members, or by pointing to every member successively, according to Reichenbach 1948. See also fn. 91, above. •• As Kecskemeti put it (1952:130), 'Considered in terms of its intension . . . a name is simply a blank, unless and until a description referring to the same object is supplied'. On semantic aspects of proper names, including an examination of various theories, see S0rensen 1963. IOO On various kinds of identifiors, see Morris 1946:76-77. On the universality of the identification message in all vertebrate displays, see Smith 1969a-b. For signs indicating individual iden-

248

THOMAS A. SEBEOK

In summary, we can say that the following formal attributes are critical for the aforementioned categories of sign-aspects: the denotation of a SYMPTOM is tantamount to its cause within the emitter, whereas that of a SIGNAL causes an alteration in the behavior of the receiver; an ICON entails similarity between signifier and signified, whereas the components of an INDEX are in a status of contiguity; a SYMBOL requires the ooncept of an intensional class, whereas a NAME requires that of an extensional class; finally, an EMBLEM is opposed to a symbol as a category marked by the channel in which it is manifested.

REFERENCES

ADEY, W. ROSS. 1967. Historical review. Information and control processes in living systems, ed. by Diane M. Ramsey, pp. 20-22. New York, New York Academy of Sciences, Interdisciplinary Communications Program. ALBRECHT, ERHARD. 1967. Sprache und Erkenntnis: Logisch-linguische Analysen. Berlin, VEB Deutseiher Verlag der Wissenschaften. ALTMANN, STUART A . , ed. 1967. Social communication among primates. Chicago, University of Chicago Press. ARBIB, MICHAEL A . 1971. How we know universals: Retrospect and prospect. Mathematical Biosciences 11.95-107. ARDREY, ROBERT. 1 9 7 0 . The social contract: A personal inquiry into the evolutionary sources of order and disorder. New York, Dell Publishing Co. ARMSTRONG, EDWARD A . 1 9 6 3 . A study of bird song. London, Oxford University Press. ARNAULD, ANTOINE and PIERRE NICOLE. 1 6 6 2 . La Logique, ou l'art de penser. Paris, A. Delalain, 1830. AUZIAS, J.-M. 1967. Clefs pour le structuralisme. Paris, Seghers. AYALA, F. J. 1968. Biology as an autonomous science. American Scientist 56.20721. BACH-Y-RITA, P . , et al. 1 9 6 9 . Vision substitution by tactile image projection. Nature 2 2 1 . 9 6 3 - 6 4 .

R. H. and W. E. SCHEVILL. 1966. Physeter clicks. Whales, dolphins, and porpoises, ed. by Kenneth S. Norris, pp. 510-27. Berkeley and Los Angeles, University of California Press.

BACKUS,

tity, see the references to Thorpe in the bibliography and note the following passages - cited from Thorpe 1967: . . when the partners were absent, the remaining bird would use the sounds normally reserved for his partner, with the result that the said partner would return as quickly as possible as if called by name . . . . Field observations suggest at times that one bird is calling its partner back "by name"'; for further particularly good examples in a large variety of birds, see Armstrong 1963:9-12; for scattered illustrations in primates, see Marler 1965, and von LawickGoodall 1968:270. On the apparent distinctiveness of individual whale click trains ('signature'), see Backus and Schevill 1966.

SEMIOTICS: A SURVEY

249

BALLY, CHARLES. 1939. Qu-est-ce qu'un signe? JPsych 1939.161-74. BÄR, EUGEN. 1971. The language of the unconscious according to Jacques Lacan. Semiotica 3.241-68. BAR-HILLEL, Y . 1968. The future of man-machine languages. Purposive systems: Proceedings of the First Annual Symposium of the American Society for Cybernetics, ed. by Heinz von Foerster, et al., pp. 141-52. New York, Spartan Books. BARTHES, R. 1968. Elements of semiology. New York, Hill and Wang. BASTIDE, R., ed. 1962. Sens et usages du terme structure dans les sciences humaines et sociales. The Hague, Mouton. BATESON, G. 1970. Form, substance and difference. Contribution No. 65. Hawaii, Oceanic Institute. BATESON, M. C. 1968. Linguistics in the semiotic frame. Linguistics 39.5-17. BAUMANN, H.-H. 1969. Über französischen Strukturalismus: Zur Rezeption moderner Linguistik in Frankreich und in Deutschland. STZ 30.157-83. BEER, C. G. 1963-1964. Ethology — the zoologist's approach to behavior. Tuatara 2.170-77 and 3.16-39. BENDIX, E. H. 1966. Componential analysis of general vocabulary: The semantic structure of a set of verbs in English, Hindi, and Japanese. Bloomington, Indiana University. BENSE, M. 1965. Semiotik und Linguistik. Grundlagenstudien aus Kybernetik und Geisteswissenschaft 6.97-108. . 1967. Semiotik: Allgemeine Theorie der Zeichen. Baden-Baden, Agis-Verlag. BENVENISTE, E. 1969. Sémiologie de la langue. Semiotica 1.1-12, 127-35. BETTETINI, G. 1968. Cinema: Lingua e scrittura. Milan, Bompiani. BIERWISCH, MANFRED. 1966. Strukturalismus. Geschichte, Probleme und Methoden. Kursbuch 5.77-152. . 1969. On certain problems of semantic representation. F L 5.153-84. . 1972. Generative grammar and European linguistics. CTL 9.313-42. BILZ, RUDOLF. 1940. Pars pro toto: Ein Beitrag zur Pathologie menschlicher Affekte und Organfunktionen. Leipzig, Georg Thieme Verlag. BIRDWHISTELL, R. L. 1968. Kinesics: Inter- and intra-channel communication research. SocScil 7/6.9-26. BLACK, M. 1968. The labyrinth of language. New York, Frederick A. Praeger. BLEIBTREU, J. N. 1968. The parable of the beast. London, Victor Gollancz. BLEST, A. D. 1961. The concept of 'ritualization'. Current problems in animal behaviour, ed. by W. H. Thorpe and O. L. Zangwill, pp. 102-24. Cambridge, The University Press. BLOOMFIELD, L. 1933. Language. New York, Henry Holt. . 1939. Linguistic aspects of science. International Encyclopedia of Unified Science, Vol. 1, no. 4, p. 55. Chicago, The University of Chicago Press.

250

THOMAS A. SEBEOK

J. M. 1968. The methods of contemporary thought. New York, Harper & Row. BOCK, KENNETH E . 1 9 6 6 . The comparative method in anthropology. Comparative Studies in Society and History 8 . 2 6 9 - 8 0 . BOLINGER, DWIGHT. 1949. The sign is not arbitrary. B I C C 5.52-62. . 1968. Aspects of language. New York, Harcourt, Brace and World. BONNER, JOHN T. 1963. How slime molds communicate. SA 209.84-94. BOUISSAC, PAUL A . R . 1 9 6 8 . Volumes sonores et volumes gestuels dans un numéro d'acrobatie. Langages 1 0 . 1 2 8 - 3 1 . . 1969. Le numéro d'acrobatie en tant que structure narrative. Oral communication delivered at the Troisième Symposium International sur l'étude des Structures Narratives, August 1969. . 1970. Pour une sémiotique du cirque. Semiotica 3.93-120. BRAIN, Lord. 1961. The neurology of language. Brain 84.145-46. BRÉAL, MICHEL [ 1 9 0 0 ] , Semantics. Studies in the science of meaning. New York, Henry Holt & Co. BREKLE, HERBERT E. 1964. Semiotik und linguistische Semantik in Port-Royal. IF 69.103-21. , ed. 1966. Grammaire générale et raisonnée, ou La grammaire de PortRoyal, par Claude Lancelot et Antoine Arnauld. Critical edition, facsimile printing of the 3rd ed. of 1676. Stuttgart-Bad Cannstatt, Friedrich Frommann Verlag. BRONOWSKI, J. 1 9 6 7 . Human and animal language. To honor Roman Jakobson, 1 . 3 7 4 - 9 4 . The Hague, Mouton. BRONOWSKI, J. and URSULA BELLUGI. 1 9 7 0 . Language, name, and concept. Science BOCHENSKI,

168.669-73. BROWER, LINCOLN P . , JANE VAN ZANDT BROWER, a n d JOSEPH M . CORVINO.

1967.

Plant poisons in a terrestrial foodchain. Proceedings of the National Academy of Sciences, 5 7 . 8 9 3 - 9 8 . Washington, D.C. BROWN, ROGER. 1 9 5 8 . Words and things. Glencoe, Illinois, The Free Press. . 1970. The first sentences of child and chimpanzee. Psycholinguistics, pp. 208-31. New York, The Free Press. BÜHLER, KARL. 1 9 3 4 . Sprachtheorie. Jena, Fischer. BURKE, KENNETH. 1966. Language as symbolic action: Essays on life, literature, and method. Berkeley and Los Angeles, University of California Press. BURKHARD, D . , W . SCHLEIDT, H. ALTNER, et al. 1 9 6 7 . Signals in the animal world, New York, McGraw-Hill. BURSILL-HALL, GEOFFREY. 1 9 7 1 . Speculative grammars of the Middle Ages: The doctrine of partes orationes of the Modistae. Approaches to semiotics, Vol. 11. The Hague, Mouton. BUYSSENS, ERIC. 1943. Les langages et le discours: Essai de linguistique fonctionnelle dans le cadre de la sémiologie. Brussels, J. Lebègue.

SEMIOTICS: A SURVEY

251

. 1956. Le langage par gestes chez les moines. Revue de l'Institut de Sociologie de Bruxelles 29.537-45. . 1967. La communication et l'articulation linguistique. Brussels, Presses Universitaires de Bruxelles. CARNAP, RUDOLF. 1 9 5 6 . Meaning and necessity: A study in semantics and modal logic. Chicago, The University of Chicago Press. CARPENTER, CLARENCE R. 1969. Approaches to studies of the naturalistic communicative behavior in nonhuman primates. In Sebeok and Ramsay, eds., 1969 (q.v.), pp. 40-70. CARPENTER, EDMUND S. 1960. Eskimo. Toronto, University of Toronto Press. CARROLL, JOHN B., ed. 1 9 5 6 . Language, thought and reality: Selected writings of Benjamin Lee Whorf. Cambridge, Mass., The M.I.T. Press. . 1964. Language and thought. Englewood Cliffs, N.J., Prentice-Hall. CASSIRER, ERNST. 1 9 2 3 . Die Philosophie der Symbolischen Formen. Berlin, Bruno Cassirer, 1923, 1924, 1929. . 1944. An essay on man: An introduction to a philosophy of human culture. New Haven, Yale University Press. CAWS, P E T E R . 1969. The structure of a discovery. Science 166.1375-80. CHAFE, WALLACE. 1968. Review of Sydney M. Lamb, Outline of stratificational grammar. Lg 44.593-603. CHAO, YUEN R E N . 1 9 6 8 . Language and symbolic systems. Cambridge, The University Press. CHERRY, COLIN. 1966. On human communication: A review, a survey, and a criticism. Cambridge, Mass., The M.I.T. Press. CHOMSKY, NOAM. 1 9 6 7 . The general properties of language. Brain mechanisms underlying speech and language, ed. by F. L. Darley, p. 73. New York, Grune and Stratton. . 1971. Deep structure, surface structure, and semantic interpretation. Studies in general and Oriental linguistics presented to Shirô Hattori on the occasion of his sixtieth birthday, ed. by Roman Jakobson and Shigeo Kawamoto, pp. 5 9 91. Tokyo, TEC Company. CLARK, HERBERT. 1 9 7 2 . Semantics and comprehension. This volume. COSERIU, EUGENIO, and HORST GECKLER. 1 9 7 2 . Semantics. This volume. COUNT, EARL W. 1969. Animal communication in man-science: An essay in perspective. In Sebeok and Ramsay, eds. 1969 (q.v.), pp. 71-130. CROOKSHANK, F. G. 1925 [1938], The importance of a theory of signs and a critique of language in the study of medicine. Supplement to Ogden and Richards 1938 (q.v.). DAANJE, A . 1 9 5 0 . On locomotory movements in birds and their intention movements. Behaviour 3 . 4 8 - 9 9 . DANTE ALIGHIERI. 1957. D Ë vulgarî eloquentiâ, ed. by Aristide Marigo. Firenze,

252

THOMAS A. SEBEOK

Felice Le Monnier. English translation by A. G. Ferrers Howell, London, Kegan Paul, Trench, Trübner & Co., Ltd., 1890. DARWIN, CHARLES. 1 8 7 2 . The expression of the emotions in man and animals. London, John Murray. DAVIDSON, DONALD. 1970. Semantics for natural languages. Linguaggi nella società e nela tecnica 177-88. Milano, Edizioni di Communità. DEWAN, EDMOND M. 1969. Rhythms. Science and technology (Jan.) 1969. 20-28. DUCROT, O., T . TODOROV, D . SPERBER, M. SAFOUAN, and F. WAHL. 1968. Qu'est ce que le structuralisme? Paris, Editions du Seuil. DUNCAN, HUGH D . 1968. Symbols in society. New York, Oxford University Press. Eco, UMBERTO. 1968. La struttura assente: Introduzione alla ricerca semiologica. Milan, Bompiani. EFRON, DAVID. 1941. Gesture and environment. New York, King's Crown Press. EGGER, V. 1904. La parole intérieure. Paris, Alcan. EHRMANN, JACQUES. 1 9 6 6 . Structuralism. Yale French Studies, nos. 3 6 - 3 7 . New Haven, Conn. EILERS, FRANZ-JOSEF. 1967. Zur Publizistik schriftloser Kulturen in NordostGuinea. St. Augustin, Steiger Verlag. EKMAN, P. and W. V. FRIESEN. 1969a. Nonverbal leakage and clues to deception. Psychiatry 32.88-106. . 1969b. The repertoire of nonverbal behavior: categories, origins, usage, and coding. Semiotica 1.49-98. ENGLER, RUDOLF. 1 9 6 2 . Théorie et critique d'un principe Saussurien: l'arbitraire du signe. CFS 19.5-66.

. 1968. Lexique de la terminologie Saussurienne. Utrecht and Anvers, Spectrum. ERVIN, SUSAN M . 1 9 6 5 2 . Information transmission with code translation. Psycholinguistics: A survey of theory and research problems, ed. by Charles E. Osgood and Thomas A. Sebeok, pp. 185-92. Bloomington, Indiana University Press. EVANS, WILLIAM E . and JARVIS BASTIAN. 1 9 6 9 . Marine mammal communication: Social and ecological factors. The biology of marine mammals, ed. by Harald T. Andersen, pp. 425-75. New York, Academic Press. FACCANI, REMO, and U . Eco, eds., 1 9 6 9 . I sistemi di segni e lo strutturalismo Sovietico. Milan, Bompiani. FAGES, J . - B . 1 9 5 8 . Comprendre le structuralisme. Paris, Privat. FEIGL, HERBERT. 1969. The Wiener Kreis in America. Intellectual migration: Europe and America, 1930-1960, ed. by Donald Fleming and Bernard Bailyn, pp. 630-673. Cambridge, Mass., Harvard University Press, The Belknap Press. FISCHER-J0RGENSEN, ELI. 1 9 6 7 . Introduction to Uldall 1 9 5 7 (q.v.), T C L C 1 0 . FITZGERALD, JOHN J. 1966. Peirce's theory of signs as foundation for pragmatism. The Hague, Mouton.

SEMIOTICS: A SURVEY FLEMING, ILAH.

1969.

Stratificational theory: An annotated bibliography.

253 JL

3.37-65.

G. 1951. The golden bough: A study in magic and religion. New York, Macmillan. FREGE, GOTTLOB. 1 8 9 2 . Über Sinn und Bedeutung. Zeitschrift für Philosophie und philosophische Kritik 100.25-50. FREI, HENRI. 1944. Systèmes de déictiques. AL 4.111-29. . 1950. Zéro, vide et intermittent. ZPhon 4.161-191. . 1963. Le signe de Saussure et le signe de Buyssens. Lingua 12.423-28. FREUD, SIGMUND. 1 9 5 9 [ 1 9 0 5 ] . Fragment of an analysis of a case of hysteria. Collected Papers, Vol. 3. New York, Basic Books. FRIEDMANN, HERBERT. 1955. The honey-guides. Washington, D.C., Smithsonian Institution. FRISCH, KARL VON. 1967. The dance language and orientation of bees. Cambridge, Mass., Harvard University Press, The Belknap Press. GARDNER, MARTIN. 1 9 6 8 . Combinatorial problems involving 'tree' graphs and forests of trees. SA 218.118-20. GARDNER, R . ALLEN and BEATRICE T. GARDNER. 1969. Teaching sign language to a chimpanzee. Science 165.664-72. GARRONI, EMILIO. 1 9 6 8 . Semiotica ed estetica. Bari, Editori Laterza. GELDARD, FRANK. 1 9 6 0 . Some neglected possibilities of communication. Science

FRAZER, JAMES

131.1583-88.

1969. The reverse side of the sign. SocScil 8/4.169-82. GERARD, RALPH. 1 9 5 7 . Units and concepts of biology. Science 1 2 5 . 4 2 9 - 3 3 . . 1960. Becoming: the residue of change. Evolution after Darwin, ed. by Sol Tax, pp. 255-67. Chicago, The University of Chicago Press. GIEDION, SIGFRIED. 1 9 6 0 . Space conception in prehistoric art. Explorations in communication: An anthology, ed. by E. Carpenter and M. McLuhan, pp. 71-89. Boston, Beacon Press, 1966. GILLIARD, E. THOMAS. 1963. The evolution of bowerbirds. SA 209.38-46. GODEL, ROBERT. 1 9 5 3 . La question des signes zéro. C F S 1 1 . 3 1 - 4 1 . . 1957. Les sources manuscrites du Cours de linguistique générale de F. de Saussure. Geneva, Droz. . ed. 1969. A Geneva School reader in linguistics. Bloomington, Indiana University Press. GORN, SAUL. 1968. The identification of the computer and information sciences: Their fundamental semiotic concepts and relationships. FL 4.339-72. GREENBERG, JOSEPH H . 1 9 6 4 . Linguistics and ethnology. Language in culture and society, ed. by Dell Hymes, pp. 27-31. New York, Harper & Row. GREIMAS, A . J. 1 9 6 6 . Sémantique structurale. Paris, Librairie Larousse. . [1968]. Modelli seomiologici. Urbino, Argalia Editore. . 1970. Du sens: Essais sémiotiques. Paris, Éditions du Seuil.

GENETTE, GÉRARD.

254

THOMAS A. SEBEOK

, et al., eds. 1970. Sign . Language . Culture. JanL, series maior 1. The Hague, Mouton. GRIFFIN, DONALD R . 1 9 6 8 . Echolocation and its relevance to communication behavior. In Sebeok, ed., 1968 (q.v.), Ch. 9. GROSSMAN, REINHARDT. 1969. Reflections on Frege's Philosophy. Evanston, Northwestern University Press. HAAS, W. 1957. Zero in linguistic description. Studies in linguistic analysis, special volume of the Philological Society of London, pp. 33-53. Oxford, Basil Blackwell. HADAS, MOSES. 1 9 5 4 . Ancilla to classical reading. New York, Columbia University Press. HALDANE, J. B. S. 1955. Animal communication and the origin of human language. Science Progress 43.387. H A L L , K . R . L . 1 9 6 3 , 1 9 6 6 . Tool-using performances as indicators of behaviorial adaptability. CAnthr 4.479-94, and 7.215-16.

R. L. and GEORGE B. SCHALLER. 1964. Tool-using behavior of the California sea otter. Journal of Mammalology 45.287-98.

HALL, K .

HARTSHORNE, CHARLES, PAUL W E I S S , a n d ARTHUR W . B U R K S , e d s . 1 9 6 5 - 1 9 6 6 .

Col-

lected papers of Charles Sanders Peirce. 6 Vols. Cambridge, Mass., Harvard University Press, The Belknap Press. HAYES, A L F R E D S. 1 9 6 4 . Paralinguistics and kinesics: Pedagogical perspectives. In Sebeok, Hayes, and Bateson, eds., 1964 (q.v.), pp. 145-75, with discussion, pp. 175-90.

1965. Man as a social partner of animals and vice versa. Symposia of the Zoological Society of London, no. 14, pp. 291-300. . 1967. Verstehens- and Verständigungsmöglichkeiten zwischen Mensch und Tier. Schweizerische Zeitschrift für Psychologie und ihre Anwendungen 26.234-55. . 1968. The psychology and behaviour of animals in zoos and circuses. New York, Dover Publications. HESS, E. H. 1962. Ethology: An approach toward the complete analysis of behavior. New directions in psychology, ed. by R. Brown et al., pp. 157-266. New York, Holt, Rinehart and Winston. HJELMSLEV, LOUIS. 1 9 4 3 . Prolegomena to a theory of language. Baltimore. IUPAL, 1953. [Danish original 1943.] . 1959. Pour une sémantique structurale. TLCP 12.96-112. HOCKETT, CHARLES F. 1 9 5 5 . How to learn Martian. Astounding Science Fiction, May, 1 9 5 5 . . 1966. The problem of universals in language. Universals of language, ed. by J. H. Greenberg, pp. 1-29. Cambridge, Mass., The M.I.T. Press. HOFSTADTER, ALBERT. 1942a. Objective teleology. The Journal of Philosophy 38.29-39. HEDIGER, H .

SEMIOTICS: A SURVEY

. 1942b. Subjective teleology.

255

Philosophy and Phenomenological Research

2.88-97. HOLLANDER, JOHN. 1959. The metrical emblem. Kenyon Review 21.279-296. HOTOPF, W.H.N. 1965. Language, thought and comprehension: A case study of the writings of I. A . Richards. Bloomington, Indiana University Press. HÜTT, CLELIA. 1968. Etude d'un corpus: Dictionnaire du langage gestuel chez les Trappistes. Langages 10.107-18. HYMES, DELL, ed. 1964. Language in culture and society. New York, Harper & Row. . 1968a. Foreword to James L . Peacock, Rites of modernization: Symbolic and social aspects of Indonesian proletarian drama. Chicago, University of Chicago Press, Series in Symbolic Anthropology. . 1968b. Review of Kenneth Burke, Language as symbolic action: Essay on life, literature, and method. L g 44.664-69. . 1969. Review of H . D. Duncan, Symbols in society. Science 164.695-96. ICAS, M . 1969. The biological code. Amsterdam, North-Holland. IVANOV, V . V . 1969. L'évolution des signes-symboles. Semiotica 1.218-21. IVANOV, V . V . and V . TOPOROV. 1965. Slavjanskie jazykovyie modeluruju,scie semioticeskie sistemy. Moscow, Nauk. JAKOBSON, ROMAN. 1939. Signe zéro. Reprinted in Readings in linguistics II, ed. by E. Hamp, F. Householder, and R. Austerlitz, pp. 109-15. Chicago, University of Chicago Press. 1966. . 1940. Das Nullzeichen. B C L C 5.12-14. . [1963]. Essais de linguistique générale. Paris, Les Editions de Minuit. . 1964. On visual and auditory signs. Phonetica 11.216-20. . 1965. Quest for the essence of language. Diogenes 51.21-37. . 1967. About the relation between visual and auditory signs. Models for the perception of speech and visual form, ed. by W. Wathen-Dunn. Cambridge, Mass, M.I.T. Press. . 1968. Language in relation to other communication systems. Address before conference on Language in society and in the technical world, Milan, Olivetti. Reported in de Mauro and Grassi, 1969. (q.v.) .

1969. Linguistics in relation to other sciences. P I C L 10.1.78.

. 1970. Language in relation to other communication systems. Linguaggi nella società e nella tecnica: Convegno promosso dalla Ing. C. Olivetti & C., S. p. A . per il centanario della nascita di Camillo Olivetti, pp. 3-16. Milano, Edizioni di Communità. JESPERSEN, OTTO. 1922. Language, its nature, development, and origin. New York, W . W . Norton, 1964. JUNG, CARL G. et ed., eds. 1968. Man and his symbols. New York, Dell. KAHN, DAVID. 1967. The codebreakers. New York, Macmillan.

256

THOMAS A. SEBEOK

C. 1969. Symbols and man's nature. The International Journal of Symbology 1.5-6. KAINZ, FRIEDRICH. 1 9 6 1 . Die 'Sprache' der Tiere: Tatsachen-Problemschau-Theorie. Stuttgart, Ferdinand Enke Verlag. KANTOR, J. R . 1 9 3 6 . An objective psychology of grammar. Bloomington, Indiana University Publications, Science Series No. 1. KATZ, JERROLD J. 1 9 6 6 . The philosophy of language. New York, Harper & Row. . 1967. Recent issues in semantic theory. FL 3.124-94. KECSKEMETI, PAUL. 1 9 5 2 . Meaning, communication, and value. Chicago, University of Chicago Press. KLEINPAUL, RUDOLF. 1 8 8 8 . Sprache ohne Worte: Idee einer allgemeinen Wissenschaft der Sprache. Leipzig, W. Friedrich. [Reprinted as Approaches to Semiotics 1 9 , The Hague, Mouton, 1 9 7 2 . ] KLOFT, W. 1959. Versuch einer Analyse der trophobiotischen Beziehungen von Ameisen zu Aphiden. Biologisches Zentralblatt 78.863-70. KORTLAND, ADRIAAN and M . KOOIJ. 1 9 6 3 . Protohominid behaviour in primates. Symposia of the Zoological Society of London, No. 1 0 , pp. 6 1 - 8 8 . KRENN, HERWIG and KLAUS MÜLLER. 1 9 7 0 . Generative Semantik. Linguistische Berichte 5 . 8 5 - 1 0 6 . KRISTEVA, JULIA. 1967. L'expansion de la semiotique. SocScil 6/5.169-81. [Reprinted in Kristeva 1969b.] . 1969a. La sémiologie comme science des idéologies. Semiotica 1.196-204. . 1969b. Semiotikè: Recherches pour une sémanalyse, pp. 43-59. Paris, Editions du Seuil. KROEBER, A . L . and CLYDE KLUCKHOHN. 1 9 5 2 . Culture: A critical review of concepts and definitions. Cambridge, Mass., Peabody Museum. KRUEGER, JOHN R . 1 9 6 8 . Language and techniques of communication as theme or tool in science-fiction. Linguistics 3 9 . 6 8 - 8 6 . LA BARRE, WESTON. 1 9 6 4 . Paralinguistics, kinesics, and cultural anthropology. In Sebeok, Hayes, and Bateson, eds. 1 9 6 4 (q.v.), pp. 1 9 1 - 2 2 0 , with discussion, pp. 2 2 1 - 3 7 . LACAN, JACQUES. 1 9 6 6 . Ecrits. Paris, Editions du Seuil. . 1968. The language of the self: The function of language in psychoanalysis, trans, with notes and commentary by A. Wilden. Baltimore, John Hopkins University Press. LAMB, SYDNEY M. n.d. [1966]. Outline of stratificational grammar. Washington, D.C., Georgetown University Press. LANDAR, HERBERT. 1 9 6 6 . Language and culture. New York, Oxford University Press. LANE, MICHAEL, ed. 1970. Introduction to structuralism. New York, Basic Books. LANGER, SUSANNE K. 1942. Philosophy in a new key: A study in the symbolism of reason, rite, and art. New York, Penguin Books, 1948. KAHN, THEODORE

SEMIOTICS: A SURVEY

257

1968. The behaviour of free-living chimpanzees in the Gombe Stream Reserve. Animal Behaviour Monographs, Vol. 1, Part. 3. LENNEBERG, E R I C H. 1969. Problems in the systematization of communicative behavior. In Sebeok and Ramsay, eds. 1969 (q.v.) Ch. 5. LÉVI-STRAUSS, CLAUDE. 1 9 5 8 . Anthropologie structurale. Paris, Plön. LEWIS, C. I. 1946. An analysis of knowledge and valuation. La Salle, 111., Open Court Publishing Co. LIEB, IRWIN C., ed. 1 9 5 3 . Charles S. Peirce's letters to Lady Welby. New Haven, Whitlock's. LLOYD, JAMES E. 1966. Studies on the flash communication system in Photinus fireflies. Ann Arbor, Museum of Zoology, University of Michigan, no. 130. LOCKE, JOHN. 1 6 9 0 . Essay concerning humane understanding. LORENZ, KONRAD. 1 9 6 5 . Über tierisches und menschliches Verhalten: aus dem Werdegang der Verhaltenslehre, Vols. 1-2. Munich, R. Piper and Co. LORENZ, KONRAD and PAUL LEYHAUSEN. 1 9 6 8 . Antriebe tierischen und menschlichen Verhaltens, Munich, R. Piper and Co. LOTMAN, JURI M., ed. 1965, 1967, 1969, 1971. Trudy po znakovym sistemam, II, III, IV, V. Tartu, Transactions of the Tartu State University, Vols, 189, 198, 236, 284. LOTMAN, JURI M. and A. M. PJATIGORSKIJ. 1 9 6 9 . Le texte et la fonction. Semiotica 1 . 2 0 5 - 1 7 . LURIA, A. R. 1967. Problems and facts in neural linguistics. International Social Science Journal 19.36-51. . 1972. Basic problems of neurolinguistics. This volume. LURKER, MANFRED, ed. 1 9 6 8 . Bibliographie zur Symbolik, Ikonographie und Mythologie. Baden-Baden, Verlag Librairie Heitz BMBH. LYONS, JOHN. 1 9 6 3 . Structural semantics: An analysis of part of the vocabulary of Plato. Oxford, Basil Blackwell. . 1968. Introduction to theoretical linguistics. Cambridge, University Press. MCCAWLEY, JAMES D . 1968. The role of semantics in grammar. Universals in linguistic theory, ed. by E. Bach and R. T. Harms, pp. 124-69. New York, Holt, Rinehart and Winston. MACKAY, DONALD M . 1961. The informational analysis of questions and commands. Information theory, ed. by Colin Cherry. Washington, D.C., Butterworths. MACKSEY, RICHARD and EUGENIO DONATO, eds. 1 9 7 0 . The languages of criticism and the sciences of man: The structuralist controversy. Baltimore, The Johns Hopkins Press. MAHL, GEORGE F . , and GENE SCHULZE. 1 9 6 4 . Psychological research in the extralinguistic area. In Sebeok, Hayes, and Bateson, eds. 1 9 6 4 (q.v.) pp. 5 1 - 1 2 4 , with discussion, pp. 1 2 5 - 4 3 .

LAWICK-GOODALL, JANE VAN.

258

THOMAS A. SEBEOK

Signe. La linguistique: Guide alphabétique, ed. Martinet, pp. 3 4 5 - 5 3 , 4 2 8 - 2 9 . Paris, Editions Denoël. MALINOWSKI, B. 1925. The problem of meaning in primitive languages. Supplement to Ogden and Richards 1938 (q.v.). MARLER, PETER. 1 9 5 6 . The voice of the chaffinch and its function as a language. Ibis 9 8 . 2 3 1 - 6 1 . . 1961. The logical analysis of animal communication. Journal of Theoretical Biology 1.295-317. . 1965. Communication in monkeys and apes. Primate behavior: Field studies of monkeys and apes, ed. by Irven DeVore, 566-67. New York, Holt, Rinehart and Winston. MARSHALL, ALEXANDER J. 1954. Bower-birds: Their displays and breeding cycles. Oxford, Clarendon Press. MASTERS, ROGER D . 1 9 7 0 . Genes, language, and evolution. Semiotica 2 . 2 9 5 - 3 2 0 . MATSON, FLOYD W. and ASHLEY MONTAGUE, eds. 1967. The human dialogue: Perspectives on communication. New York, The Free Press. MAURO, TULLIO DE, and LETIZIA GRASSI. 1 9 6 9 . I linguaggi nella società e nella tecnica. LeSt 4 . 2 4 9 - 6 3 . MAYENOWA, MARIA R. 1967. Semiotics today: Reflections on the Second International Conference on Semiotics. SocScil 6/2-3.59-64. MEAD, MARGARET. 1964. Vicissitudes of the study of the total communication process. In Sebeok, Hayes, and Bateson, eds. 1964 (q.v.) pp. 277-87. . 1969. From intuition to analysis in communication research. Semiotica 1.13-25. MELETINSKIJ, ELEASAR M. 1969a. Afterword to V . Ja. Propp, Morfologia skazki, Moscow, Nauk. . 1969b. Zur strukturell-typologischen Erforschung des Volksmärchens. DJbVk 15.1-30. METZ, CHRISTIAN. 1968a. Propositions méthodologiques pour l'analyse du film. SoScil 7/14.107-19. . 1968b. Essais sur la signification au cinéma. Paris, Klincksieck. MILLER, GEORGE A . , KARL PRIBRAM, and EUGENE GALANTER. 1 9 6 0 . Plans and the structure of behavior. New York, Holt, Rinehart and Winston. MILLER, ROBERT L. 1968. The linguistic relativity principle and Humboldtian ethnolinguistics: A history and appraisal. The Hague, Mouton. MORAWSKI, STEFAN. 1970. Mimesis. Semiotica 2.35-58. MORRIS, CHARLES. 1 9 3 8 . Foundations of the theory of signs. International Encyclopedia of Unified Science, Vol. 1, no. 2. Chicago, The University of Chicago Press. . 1946. Signs, language and behavior. New York, Prentice-Hall. . 1964. Signification and significance: A study of the relation of signs and values. Cambridge, Mass., The M.I.T. Press. MAHMOUDIAN, MORTÉZA. 1 9 6 9 .

by

A.

SEMIOTICS: A SURVEY

259

1959. Les systèmes de communication non-linguistique et leur place dans la vie du XX e siècle. BSL 54.176-200. 2 NAVILLE, ADRIEN. 1901 . Nouvelle classification des sciences. Paris, Felix Alcan. NIDA, EUGENE A. 1964. Toward a science of translating. Leiden, E. J. Brill. . 1969. Science of translation. Lg 45.483-98. NIDA, EUGENE A . and CHARLES R. TABER. In preparation. Introduction to semantic structures. NORRIS, KENNETH S., ed. 1966. Whales, dolphins, and porpoises. Berkeley and Los Angeles, University of California Press. O ' B R I E N , JAMES F . 1 9 6 9 . Design by accident. New York, Dover Publications. OGDEN, C. K., and I. A. RICHARDS. 1938. The meaning of meaning. London, Kegan Paul, Trench, Trubner. 5th ed. rev. New York, Harcourt, Brave [Original edition 1925.] OSGOOD, CHARLES E., G. J. Suci, and P. H. TANNENBAUM. 1957. The measurement of meaning. Urbana, University of Illinois Press. OSGOOD, CHARLES E . and THOMAS A . SEBEOK, eds. 1 9 6 5 2 . Psycholinguistics: A survey of theory and research problems. Bloomington, Indiana University Press. OSOLSOBÉ, Ivo. 1967. Ostenze jako mezni pripad lidského sdëlovâni a jeji vyznam pro umëni. Estetika 1.2-23,101. OSTWALD, PETER F . 1 9 6 4 . How the patient communicates about disease with the doctor. In Sebeok, Hayes, and Bateson, eds. 1 9 6 4 (q.v.), pp. 1 1 - 3 4 , with disMOUNIN, GEORGES.

cussion, pp. 3 5 - 4 9 .

. 1968. Symptoms, diagnosis and concepts of disease: Some comments on the semiotics of patient-physician communication. SocScil 7/4.95-106. [Rev. version of Ostwald, 1964.] PAPP, FERENC. 1969. Jelentés, kôzlés, szemiôtika a Szovjet nyelvtudomânyban. Nyelv és kommunikaciö, ed. by Tamâs Szecskö and György Szépe, Vol. 2, pp. 95-125. Budapest, A MRT Tömegkommunikacios Kutatoközpontjanak Kiadâsa. PEACOCK, JAMES L . 1 9 6 8 . Rites of modernization: Symbolic and social aspects of Indonesian proletarian drama. Chicago, University of Chicago Press. PEIRCE, CHARLES S. 1906. Prolegomena to an apology for pragmaticism. The Monist 16.492-546. . 1965-1966. Collected papers of Charles Sanders Peirce, ed. by Hartshorne, Weiss, and Burks (q.v.). PELC, JERZY. 1 9 6 9 . Meaning as an instrument. Semiotica 1 . 2 6 - 4 8 . PFEIFFER, JOHN E . 1 9 6 9 . The emergence of man. New York, Harper & Row. PHILLIPS, HERBERT P. 1960. Problems of translation and meaning in fieldwork. Human Organization 18.184-92. PIAGET, JEAN. 1 9 6 8 . Le structuralisme. Paris, P U F . PITTS, WALTER, and WARREN S . MCCULLOCH. 1 9 4 7 . How we know universals:

260

THOMAS A. SEBEOK

The perception of auditory and visual forms. Bulletin of Mathematical Biophysics 9.127-49. POHL, JACQUES. 1 9 6 8 . Symboles et langages. 2 vols. (Vol. I , Le symbole, clef d'humain.) Paris, Editions Sodi. POSTGATE, J . P . 1900. The science of meaning, inaugural address. In: Semantics: Studies in the science of meaning, ed. by Michel Bréal, trans. Mrs. Henry Cust, pp. 311-36. New York, Henry Holt and Co. PRIETO, LUIS J . [ 1 9 6 8 ] , La sémiologie. Le langage (Encyclopédie de la Pléiade, vol. 25), ed. by A. Martinet. Paris, Gallimard. Proceedings of the International Union of Physiological Sciences, Vol 3. n.d. [1964]. Amsterdam, Excerpta Medica Foundation, International Congress Series, no. 49. PROPP, V. J. 1928. Morfologia skazki [Morphology of the fairytale]. First Eng. ed., Bloomington, Indiana University Research Center in Anthropology, Folklore, and Linguistics, 1958; 19682. QUINE, W . V . O . 1 9 6 3 . From a logical point of view. New York, Harper & Row. REICHENBACH, HANS. 1948. Elements of symbolic logic. New York, Macmillan. RESNIKOW, L . O . 1 9 6 8 . Erkenntistheoretische Fragen der Semiotik. Berlin, VEB Deutscher Verlag der Wissenschaften. [Original Russian edition, 1 9 6 4 . ] REY, ALAIN. 1 9 6 9 . Remarques sémantiques. Langue Française 4 . 5 - 2 9 . RICHARDS, I. A. 1933. Preface to a dictionary. Psyche 13.10-24. . 1936. The philosophy of rhetoric. London, Oxford University Press. . 1969. Tipi e campioni. Strumenti Critici 3.187-93. ROBERTS, D O N D . 1 9 6 3 . The existential graphs of Charles S. Peirce. University of Illinois dissertation. [See fn. 88, above.] ROSSI-LANDI, F. 1953. Charles Morris. Rome, Fratelli Bocca Editori. . 1968. Ideologie della relatività linguistica. Ideologie 4.3-69. [English edition in Mouton's Approaches to Semiotics series, 1973.] RUESCH, JÜRGEN. 1957. Disturbed communication: The clinical assessment of normal and pathological communicative behavior. New York, W. W. Norton. . 1961. Therapeutic communication. New York, W. W. Norton. RUESCH, JÜRGEN and G. BATESON. 1951. Communication: The social matrix of psychiatry. New York, W. W. Norton. RÜPPEL, G. 1969. Eine 'Lüge' als gerichtete Mitteilung beim Eisfuchs (Alopex Lagopus L.). Zeitschrift für Tierpsychologie 26.371^. RYLE, GILBERT. 1 9 5 7 . The theory of meaning. British philosophy in mid-century, ed. by C . A. Mace, pp. 2 3 9 - 6 4 . London, Allen and Unwin. SAPIR, EDWARD. 1929. The status of linguistics as a science. Lg 5.207-14. SAUSSURE, FERDINAND DE. I 9 6 0 5 . Cours de linguistique générale. Paris, Payot. [Original ed. 1 9 1 6 . ] SCHAEDER, BURKHARD. 1 9 6 9 . Die Sprachinhaltforschung im deutschen Sprachraum: Wortschatz- und Grammatiktheorie. SG 2 2 . 2 9 4 - 3 0 9 . SCHAFF, ADAM. 1 9 6 2 . Introduction to semantics. Oxford, Pergamon.

SEMIOTICS: A SURVEY

261

J. 1969. Bedeutung und Begriff : Zur Fundierung einer sprachphilosophischen Semantik. Braunschweig: F. Vieweg. SEBEOK, THOMAS A. 1962a. Coding in the evolution of signalling behavior. BSci 7.430-42. . 1962b. The texture of a Cheremis incantation. MSFOu 125.523-27. . 1965. Animal communication. Science 147.1006-14. . 1967a. Colloque international de sémiologie. Linguistic Reporter 9/1-3. Washington, D.C., Center for Applied Linguistics. . 1967b. Discussion of communication processes. In Altmann, ed., 1967 (q.v.), pp. 363-69. . 1967c. On chemical signs. To honor Roman Jakobson, 3.1775-82. The Hague, Mouton. . 1968a. Communication models and signaling behavior. International Encyclopedia of the Social Sciences, 3.33-40. New York, Macmillan and Free Press. . 1968b. Goals and limitations in the study of animal communication. In Sebeok 1968c (q.v.), pp. 3-14. , ed. 1968c. Animal communication: Techniques of study and results of research. Bloomington, Indiana University Press. . 1969. Semiotics and ethology. In Sebeok and Ramsay 1969 (q.v.), pp. 200-231. . 1970. Zoosemiotic structures and social organization. Linguaggi nella société e nella tecnica. Convegno promosso dalla Ing. C. Olivetti & C., S. p. A. per il centenario della nascita di Camillo Olivetti, pp. 113. Milano, Edizione di Comunità. . 1972. Perspectives in zoosemiotics. The Hague, Mouton. . Forthcoming. Semiotics: a survey of the state of the art. The Hague, Mouton. SEBEOK, THOMAS A . , A . S . HAYES, and M. C . BATESON, eds. 1 9 6 4 . Approaches to semiotics. The Hague, Mouton. SEBEOK, THOMAS A . , and ALEXANDRA RAMSAY, eds. 1 9 6 9 . Approaches to animal communication. The Hague, Mouton. SEBEOK, THOMAS A . , and DONNA JEAN UMIKER, eds., 1 9 7 4 . Speech surrogates: Vol. 1, Drum and whistle systems. The Hague, Mouton. SECHEHAYE, ALBERT, CHARLES BALLY, and HENRI F R E I . 1940-1941. Pour l'arbitraire du signe. A Geneva school reader in linguistics, ed. by Robert Godel, pp. 191-95. Bloomington, Indiana University Press (1969). SEGRE, CESARE. 1969. Fra strutturalismo e semiologia. I segni e la critica, ed. by C. Segre, pp. 61-92. Turin, Giulio Einaudi. [English edition in Mouton's Approaches to Semiotics series, 1973.] SELLARS, WILFRID. 1 9 5 0 . Language, rules, and behavior. John Dewey: Philosophy of science and freedom, ed. by Sidney Hook, pp. 2 8 9 - 3 1 5 . New York, Dial press.

SCHMIDT, SIEGFRIED

262

THOMAS A. SEBEOK

Semiotic approaches to psychiatry. The Hague, Mouton. . 1970. Momentary deity and personal myth: A semiotic inquiry using recorded psychotherapeutic material. Semiotica 2.1-34. . 1971. The war with words. The Hague, Mouton. SMITH, W . JOHN. 1 9 6 8 . Message-meaning analysis. In Sebeok, ed., 1 9 6 8 , (q.v.), Ch. 4. . 1969a. Displays and messages in intraspecific communication. Semiotica 1.357-69. . 1969b. Messages of vertebrate communication. Science 165.145-50. SNIDER, JAMES G. and C. E . OSGOOD, eds. 1 9 6 9 . Semantic differential technique: A sourcebook. Chicago, Aldine Publishing Co. SÖDER, KARL. 1964. Beiträge J. H. Lamberts zur formalen Logik und Semiotik. Dissertation Greifswald. S0RENSEN, HOLGER S . 1 9 6 3 . The meaning of proper names: With a definiens formula for proper names in Modern English. Copenhagen, G. E. C. Gad. . 1967. Meaning. To honor Roman Jakobson, 3.1876-99. The Hague, Mouton. SPANG-HANNSEN, HENNING. 1 9 5 4 [ 1 9 4 8 ] . Recent theories on the nature of the language sign. T C L C 9 . STANKIEWICZ, EDWARD. 1 9 6 4 . Problems of emotive language. In Sebeok, Hayes, and Bateson, eds. 1 9 6 4 . (q.v.), pp. 2 3 9 - 6 4 , with discussion, pp. 2 6 5 - 7 6 . . Forthcoming. Expressive language. The Hague, Mouton. STANOSZ, BARBARA. 1 9 7 0 . Formal theories of extension and intension of expressions. Semiotica 2 . 1 0 2 - 1 4 . STENDER-PETERSEN, A. 1949. Esquisse d'une théorie structurale de la littérature. TCLC 5. 277-87. STERN, THEODORE. 1 9 5 7 . Drum and whistle languages: An analysis of speech surrogates. AmA 5 9 . 4 8 7 - 5 0 6 . STOUGH, CHARLOTTE L. 1 9 6 9 . Greek skepticism: A study in epistemology. Berkeley and Los Angeles, University of California Press. STURTEVANT, EDGAR H. 1 9 4 7 . An introduction to linguistic science. New Haven, Yale University Press. THORPE, W. H. 1963. Antiphonal singing in birds as evidence for avian reaction time. Nature 197.774-76. . 1965. Ethology as a new branch of biology. Readings in animal behavior, ed. by Thomas E. McGill, pp. 34-49. New York, Holt, Rinehart and Winston. . 1967. Vocal imitation and antiphonal song and its implications. Proceedings of the XIV International Ornithological Congress, ed. by D. W. Snow, pp. 245-63. Oxford and Edinburgh, Blackwell Scientific Publications. . 1968. Perceptual basis for group organization in social vertebrates, especially birds. Nature 220.124-28. SHANDS, HARLEY C . 1 9 6 9 .

SEMIOTICS: A SURVEY

263

1951. The study of instinct. New York, Oxford University Press. . 1963. On aims and methods of ethology. Zeitschrift für Tierpsychologie 20.410-33. TODOROV, TZVETAN. 1966. Recherches sémantiques. Langages 1.5—43. TRIER, J. 1931. Der deutsche Wortschatz im Sinnbezirk des Verstandes: Die Geschichte eines sprachlichen Feldes, Vol. 1, Von den Anfängen bis zum Beginn des 13. Jahrhunderts. Heidelberg, Carl Winter. TURNER, VICTOR W . 1 9 6 7 . The forest of symbols. Ithaca, Cornell University Press. . 1969. The ritual process: Structure and anti-structure. Chicago, Aldine Publishing Co. TYLER, STEPHEN A., ed. 1969. Cognitive anthropology. New York, Holt, Rinehart and Winston. ULDALL, H. J. 1957. Outline of glossematics, Part I: General theory. Copenhagen, Nordisk Sprog- og Kulturforlag. ULLMANN, STEPHEN. 1 9 6 2 . Semantics: An introduction to the science of meaning. Oxford, Basil Blackwell. . 1967. Review of A. J. Greimas, Sémantique structurale. Lingua 18.296-303. . 1972. Semantics. CTL 9.343-95. UMIKER, DONNA JEAN. 1 9 7 2 . Other speech surrogates. This volume. VALESIO, PAOLO. 1 9 6 9 . Icons and patterns in the structure of language. P I C L 10/1.383-87. . 1971. Toward a study of the nature of signs. Semiotica 3.155-85. VENCLOVA, TOMAS. 1 9 6 7 . Le colloque sémiotique de Tartu. SoScil 6 / 4 . 1 2 3 - 2 9 . VERPLANCK, W I L L I A M S. 1962. Unaware of where's awareness: Some verbal operants — notâtes, monents, and notants. Behavior and awareness: A symposium of research and interpretation, ed. by Charles W. Ericksen, pp. 130-58. Durham, N. Carolina, Duke University Press. VIET, JEAN. 1 9 6 5 . Les méthodes structuralistes dans les sciences sociales. The Hague, Mouton. VOIGT, VILMOS. 1969. Modellâlâsi kisérletek a folklorisztikâban. Ethnographia 80.355-92. VYGOTSKY, L . S . 1 9 6 2 . Thought and language, ed. and trans, by E . Hanfmann and G. Vakar. Cambridge, Mass., M.I.T. Press. WEINREICH, U R I E L . 1 9 5 8 . Review of Osgood, Suci, and Tannenbaum, The measurement of meaning. Word 1 4 . 3 4 6 - 6 6 . . 1963. Lexicology. CTL 1.60-93. . 1966. Explorations in semantics theory. CTL 3.395-477. . 1968. Semantic and semiotics. International Encyclopedia of the Social Sciences, Vol. 14, pp. 164-69. New York, Macmillan and The Free Press. WEINRICH, HAROLD. 1 9 6 6 . Linguistik der Lüge. Heidelberg, Verlag Lambert Schneider.

TINBERGEN, NIKO.

264

THOMAS A. SEBEOK

WEISS, PAUL. 1 9 4 0 .

The essence of Peirce's system. The Journal of Philosophy

37.253-64.

and ARTHUR BURKS. 1 9 4 5 . Peirce's sixty-six signs. The Journal of Philosophy 4 2 . 3 8 3 - 8 8 . WELLS, RULON. 1 9 5 4 . Meaning and use. Word 1 0 . 2 3 5 - 5 0 . . 1968. Distinctively human semiotic. SocScil 6/6.103-24. WENNER, ADRIAN M . 1 9 6 9 . The study of animal communication: An overview. In Sebeok and Ramsay, eds. 1969 (q.v.), pp. 232-43. W H I T E , LESLIE A. 1940. The symbol: The origin and basis of human behavior. Philosophy of Science 7.451-63. WIENER, NORBERT. 1950. The human use of human beings. Boston, Houghton Mifflin. WOLLEN, P E T E R . 1 9 6 9 . Signs and meaning in the cinema. Bloomington, Indiana University Press. WORTH, SOL. 1 9 6 9 . The development of a semiotic of film. Semiotica 1 . 2 8 2 - 3 2 1 . WYNNE-EDWARDS, V. C. 1962. Animal dispersion in relation to social behaviour. New York, Hafner Publishing Co. YERKES, ROBERT M., and HENRY W. NISSEN. 1939. Pre-linguistic sign behavior in chimpanzee. Science 89.585-87. Zeichensystem Film — Versuche zu einer Semiotik. 1968. Special issue of STZ, No. 27. ZEMAN, J . JAY. 1 9 6 4 . The graphical logic of C. S. Peirce. Unpublished dissertation, University of Chicago. ZEUNER, FREDERICK E. 1963. A history of domesticated animals. New York, Harper and Row. 2OLKIEWSKI, STEFAN. 1968. Semiotika a kultura. Bratislava, Nakladatel'stvo Epocha. WEISS, PAUL,

NOTE A D D E D IN PAGE PROOF This chapter was begun in 1969 and completed in 1971. A few essential references were added, in 1972, to the first proofs. Countless articles and some important books have appeared in the meantime, many of the former in Semiotica, now a monthly, and the latter as volumes of the expanding Approaches to Semiotics series. The contents of Section 3 of my article were thoroughly revised, expanded, and published as two separate articles: the materials on pp. 232-39 appeared in a Festschrift for Einar Haugen (1972), and those on pp. 239-48 in a Festschrift for George Ts'ereteli (1974); some of the omitted portion, alluded to on p. 232, was dealt with, in journalistic fashion, in my piece in the Times Literary Supplement, last October (1973). All of these themes, and others, will be discussed anew, in an integrated fashion, in my book on Semiotics, under the imprint of Penguin (1975), and then in a still larger treatise now in preparation for Mouton's Approaches to Semiotics series.

PARALINGUISTICS

DAVID CRYSTAL

Unlike many of the other topics covered by this volume, defining the starting-point for a discussion of paralanguage's currency and trendiness is not a difficult task. There has already been one well-organized attempt to summarise the state of the art, in this subject, with the 1962 Indiana University Conference on Paralinguistics and Kinesics, which was later published under the heading Approaches to semiotics (Sebeok, Hayes, and Bateson 1964). In retrospect, it is clear that this conference performed two major roles. First, it brought together the rather scattered material which had been published or mimeographed in the few years since the stimulating specification of the field by Trager, Smith, and their associates, and attempted to establish explicitly the general principles which seemed to underlie the various approaches. Secondly, in trying to define the subject, and to relate it to the other branches of semiotics, a number of promising lines of enquiry were suggested, and sometimes very specific research topics outlined. In these ways, the conference participants provided considerable bibliographical information, and presented a fairly complete picture of the pre-1960 period in paralinguistic studies. It therefore should not be too much of a simplification to consider current trends in paralanguage as running from the date of that conference to the present. Indeed, much of the research which has taken place over the past ten years has in fact been influenced, directly or indirectly, by the work of this conference. In my view, there are three main trends in paralinguistic study in the sixties: (a) further development of the approach outlined mainly in Trager (1958), including its increasing application to other (non-linguistic) areas; (b) the development of other approaches not within the Trager-Smith framework; and (c), partly as a result of these first two trends, the development of considerable theoretical confusion. It is profitable to begin by suggesting some reasons for the latter. Part of the reason for the confusion which surrounds paralinguistic study at the present time is certainly due to the way in which linguistics developed during the sixties. The early definitions of paralanguage, and the descriptions of paralinguistic effects, as we shall see, were made exclusively within the theoretical and descriptive framework provided by Trager and others (see Trager 1949, Trager and Smith 1951, Smith 1952, Trager 1958); but to follow their account of paralanguage, one had really to understand their whole approach to linguistic analysis. For example, it

266

DAVID CRYSTAL

was not possible to appreciate the basis of their distinction between some types of pitch movement and others (e.g. between overhigh/overlow pitch heights, which were considered paralinguistic, and the four pitch levels, which were phonemic, and hence linguistic) without understanding their 'emic' approach to prosodic features, and this in turn would have to be seen in the context of what in their view constituted 'linguistics proper' (see below). But in the sixties, the theoretical approach begun by Trager and Smith became largely of historical interest within linguistics (though certain aspects of their description continued to be used); and with this, the status of paralinguistic study became unclear. A number of factors contributed to this situation. The absence of any reassessment of the theory in the light of other developments in the course of the past ten years is particularly important. When a theory (in this case, generative theory) is so much in the ascendant, any earlier linguistic approach which does not take cognizance of the claims of that theory, and attempt some comparison with its own claims, is necessarily going to distance itself from the eye of the majority of linguists and become, in effect, of historical interest. In the present case, it would have been beneficial for the development of paralinguistic study if Trager, or someone, had discussed it in the light (or darkness) of the competence/ performance distinction, for its accountability in terms of this dichotomy is by no means clear. Or again, paralinguistic effect could usefully have been brought into the discussion of meaning-relations, in the sense of Fillmore (1968a), or of syntactic presuppositions (cf. Fillmore 1968b, Chomsky 1971), in view of the close relationship between paralanguage and intonation, and the relevance of these effects in determining certain types of structural contrast.1 But as far as I am aware, such matters have not been raised, and the general differences between the views of Trager as presented in The field of linguistics or The outline of English structure, and generative grammar, have simply become more marked with the passage of time. Trager and Smith themselves seem to have recognized the passing of an era, as the rather wistful foreword to the seventh, and final printing of the Outline (in 1966) indicates.2 And neither they nor their colleagues have answered in print the 1

The application of case grammar in the University of Edinburgh (Dept. of Child Life and Health) project on language acquisition allows for paralinguistic features in the interpretation of utterances: 'an utterance needs to be interpreted before it may be categorised, and interpretation may depend on the syntax and semantics of the utterance, on its prosody, on its co-text (discourse relations), or on extralinguistic information'. Paralinguistic features are in fact specifically mentioned in the presyntactic section of their analysis. (Information from a preliminary draft of an Appendix outlining their descriptive apparatus, May, 1968.) 1 T h e hoped-for writing of a completely revised English structure has not come to pass, but a considerable amount of progress has been made and the results of this research is beginning to appear now in various places. Consequently, the sixth printing being exhausted, it has been decided to make this final printing of the Outline of English structure, to meet the continued demand for copies. But users are urged to remember the original date of publication, to search out the comparatively few publications that have appeared since then which stem from the

PARALINGUISTICS

267

fairly steady flow of criticism which has been levelled at their approach over the past twenty years (e.g. by Bolinger [1949, 1951], Stankiewicz [see Sebeok, Hayes, and Bateson 1964:265-7], Crystal and Quirk [1964] or Lieberman [1965]).3 Perhaps part of the reason for the failure to defend the approach lay in its pioneering status, which seemed to generate a certain complacency over theoretical matters. Market's comment at the Indiana Conference is indicative of this, and other examples of the same attitude can easily be collected: 'I think we will find, if we go over the list, that everything we want to call paralanguage is covered by Trager's system. We would also find those things that are not covered are not paralanguage; they are psycholinguistics or metalinguistics.' Trager himself takes a more reasonable view of the limitations of his approach (as the foreword quoted in fn. 2 suggests), referring to it only as a 'guide to observation' (1964:22); but one would never guess this from the subsequent applications of his approximation by others in such fields as psychotherapy or anthropology, where the completeness of the description and the adequacy of the underlying theory is assumed. Trager's 1958 paper is however a first approximation towards a description, not a theory. It is not a 'system' in any theoretical sense (cf. Duncan 1969:119), but rather an inventory and notation of effects made on fairly ad hoc principles. There were other reasons for a restricted interest in paralinguistic studies amongst linguists in the sixties, to the extent that it is given no mention at all in such standard textbooks as Gleason (1961), Robins (1964), Hall (1964), Lyons (1968) or Malmberg (1968). The link with semiotics, for instance, has in a way been a hindrance as well as a help, in that its description along with the relatively indiscrete and unstructured non-vocal modes of communication has tended to minimize attention to any features of a more clearly 'linguistic' character that it might display. And the almost exclusive emphasis on paralanguage's affective function was doubtless a further factor which made many linguists feel that this was not an area which it was of importance to investigate. Most linguists were — and are — of the opinion that paralanguage is at best of marginal significance to linguistics, and equally well or more appropriately studied by other disciplines. The 'marginal' view is of course reinforced by the (rather unfortunate) terms used. Labels like 'para'language (cf. 'supra'segmental) reflect a theoretical viewpoint which sees effects of this kind as 'additional' — as indeed does the classification of paralanguage as a related art/ science in this volume.4 It is by no means easy to wriggle out of this terminological straightjacket in order to argue a more central linguistic view of these phenomena,

same analytical foundation, and to use the book's conclusions not as the final word on anything but as points of departure for further analyses.' 3 Trager's (1964) reaction to Crystal (1963) simply reasserts his original position, and does little more than clarify the extent of the difference between the two positions. 4 Cf. such definitions of 'prosodic features' as Carrell and Tiffany's (1960:260): 'the variables which add the "plus" values to communication . . . ' .

268

DAVID CRYSTAL

but it is important to try; and one of the major trends in paralinguistic study in the sixties has been to present this alternative.5 Paralanguage was of course a marginal field for Trager too, but at least there its relationship with aspects of language which were considered more central was fairly explicit, and the techniques of classification and description were modelled on familiar linguistic lines. The term 'paralanguage' filled a slot in Trager's over-all descriptive framework, entering into the definition of the whole, and receiving its own definition and status from its relationships with other categories of the theory. And once the parent-theory came to be considered inadequate as a view of language, then necessarily paralanguage's status became unclear. The mid-sixties saw the retention and development of many of the descriptive insights of the approach, but with less and less reference being made to the theoretical principles which had earlier been established to account for them. The development of paralinguistic ideas within semiotics, and their application to such fields as psychotherapy, anthropology, or language teaching, was almost entirely a descriptive development, which was of course welcomed by scholars from these fields, who had lacked any means of formalizing these features of behavior hitherto. Similarly, the Trager/Smith description of phonology, and the accompanying notation, continued to be used as a basis for teaching about English and about linguistics in many influential textbooks (e.g. Gleason 1961), and has been given a new lease on life by Chomsky and Halle (1968). But exposition and examination of the theoretical basis of the description has not taken place during this time. One thus finds the development of a situation in which a widely used descriptive framework rests on a largely implicit theoretical foundation; and in the absence of analytic criteria being made explicit, one naturally finds arbitrary descriptive decisions, ambiguous cases being forced into one or another of the set of choices provided by the framework, and, following on this, inconsistency in the use of terms by various scholars. 'Paralanguage' itself, as a term, was particularly affected by the lack of any explicit theoretical foundation. In the mid-sixties, one finds an increasing number of references to the 'language and paralanguage' of a community, without further comment — as if the latter were in some way a reality which existed independently of one's specific approach to language.8 Scholars of radically different theoretical 5

The fallacy of assuming that those areas of language that one's available linguistic techniques cannot handle are therefore less important was stressed by Birdwhistell in a paper in 1959: 'It is all too easy to assume that there is in any social interchange a central, a primary or a real meaning which is only modified by a redundant surround . . . Our temptation so to classify certain aspects of a transaction as the central message and other aspects serving only as modifiers rests upon untested assumptions about communication.' • Cf. Abercrombie (1968 : 55): 'the term paralanguage . . . seems to me potentially misleading: it can give the impression that, because there exists a (more or less) homogeneous entity called language, there must be, existing beside it, a comparably homogeneous entity called paralanguage. I believe this is not s o . . . These non-verbal, though conversational, activities to which the word paralanguage refers are far too diverse, too little codified, too uninvestigated, and too insufficiently understood, to be given the air of unity which a noun confers on them.'

PARALINGUISTICS

269

persuasions, who had cause to refer to paralanguage, would refer to Trager (1958) and assume that the conceptual relationship between that paper and their own approaches was obvious. Significantly, the number of references to Hill's brief remarks (1958:408-9) increases during this period — which is significant, since paralanguage for Hill subsumed kinesics, an extension of the term which most of the early workers in the field did not intend, and which is still very much a minority view. Thus, while the existence and relevance of paralinguistic effects has been recognized by many scholars in various disciplines, in the absence of any theoretical nucleus to provide a baseline for comparing and assessing modifications and extensions in the use of terms, a multiplicity of definitions and characterizations of paralanguage have developed, and produced a general state of confusion. In trying to classify some of these differences, I shall concentrate on those papers which explicitly refer to their subject-matter as falling under the heading of paralanguage. There are of course other studies which deal with similar effects, though not labelling them 'paralinguistic', and I shall refer to these in passing later.7 A preliminary point which has to be made is to indicate the quite remarkable range of subject-matter which is allowed under the heading of paralanguage. In the process of accumulating information for this paper, I wrote to a number of scholars asking them what work, if any, they were doing within this field. The term 'paralanguage' was deliberately not defined in my letter. Excluding those who expressed uncertainty as to whether their work did fall under this heading or not, there was still considerable variability in interpretation. All of the following were suggested: animal vocalization (or some aspect of it), memory restrictions on language, recall ability for language, utterance length, literary analysis, environmental restrictions on language use (accounting for such matters as word or phoneme frequency differences in social groups), glossolalia, and emotional expression in general language disturbance — in effect, a fair proportion of sociolinguistics and psycholinguistics. These were, however, marginal interpretations of paralanguage, compared with the senses classified below, which I would consider to be the most widespread and influential uses of the term. The fact that there are seven distinct 'basic' senses, of course, simply indicates the distance this subject has to travel before its claims to scientificness can be taken seriously. I shall illustrate these senses in terms of the increasing restrictedness of the phenomena allowed in under the heading of paralanguage. 7

For general coverage of this field and related areas in the sixties, see Birdwhistell (1961), Barbara (1963), Ostwald (1963), Kramer (1963, 1964), the papers by La Barre, Hayes, and Mahl and Schulze in Sebeok, Hayes and Bateson (1964), Crystal and Quirk (1964: chapter 2), Austin (1965, in press), Diebold (1965), Ekman (1965), Crystal (1966, 1969a), Egorov (1967), Abercrombie (1967:chapter 6), Weakland (1967), Duncan (1969), Vetter (1969), Markel (1969b), and the papers by Bateson and Hockett in McQuown (in press). M. R. Key, of the University of California, is engaged in the compilation of a collection of papers within the field of paralinguistics; J. Laver and S. Hutcheson of the University of Edinburgh have also edited a collection of readings in which paralanguage figures prominently (Communication in face-to-face interaction, Penguin, 1972).

270

DAVID CRYSTAL

1. Including non-human as well as human vocalization. Austin's (in press) 'soundsignal systems' are illustrated from various animal species as well as man, and referred to as paralanguage. Abercrombie (1968:56) restricts the term to human behavior, but implies a parallel with animal communication. Hockett (1960) refers to paralinguistic phenomena in his discussion of animal communication and the origin of speech, but it is unclear whether the term should be construed in a purely 'human' sense. Increased study of human paralanguage over the past few years has, however, shown more rather than less difference between animal species and man in this area. There is little suggestion in animal communication of the 'systems' of paralinguistic effects postulated for man (see below); nor is there any comparison with the number and complexity of the phonetic variables which enter into the definition of human paralinguistic effects, and the nature of the semantic contrasts which they expound. The few broad similarities which do exist seem trivial by comparison. 2. Including non-vocal as well as vocal features of human communication. This sense is the one normally implied by the use of the term 'non-verbal' in, e.g., social psychology. Specifically, kinesics is brought under the heading of paralanguage. This is an influential sense, as it is the one used by Hill (1958:408-9); and Abercrombie has given some reasons for his use of the term in this way in his 1968 paper (p. 56). But, as Hayes suggested, in discussion at the Indiana Conference (see Sebeok, Hayes, and Bateson 1964:153), this is a potentially confusing sense — and it is one which the title of the Conference carefully avoided. Despite the attempt to foist onto kinesics essentially linguistic categories, it is highly unlikely that kinesic behavior has sufficient structural complexity, discreteness, or semantic organization to warrant its analysis in the same terms as linguistic behavior; and spurious terminological identity is best avoided. 3. Including all non-segmental ('suprasegmental') features and some segmental ones* In this sense, Smith's vocal identifiers (1952), e.g. 'uh-uh', or Trager's vocal segregates (1958), including hesitation features, amongst other things, are allowed under the heading of paralanguage. Any segmental utterance not having the normal 8

The generally held distinction between segmental and non-segmental phonology is retained as a framework for discussion in this paper. Segmental phonology is defined purely in terms of vowels, consonants, and their combinatorial properties. Non-segmental phonology comprises all contrastive sound-effects which have an essentially variable relationship to the segmental or lexical component of utterance, and which are not describable with reference to single segments (except insofar as single segments can be exponents of syllables); specifically, these are effects which operate continuously over a stretch of utterance, minimally one syllable (as in a pitch contour), or which are specified in terms of a number of segments, some adjacent, some not, all of which are affected by a single 'setting' or configuration of the vocal organs, to produce a single perceptual impression and a single semantic interpretation (e.g. labialization, velarization). 'Suprasegmental' features, in the sense of Trager, Smith, and others, thus comprise one sub-set of the totality of non-segmental features of a language. 'Intonation', and 'paralanguage', would be defined with reference to other sub-sets. It is not the purpose of this paper to make any terminological decisions as to how non-segmental phonology may best be compartmentalized, however.

PARALINGUISTICS

271

phonemic structure of a language (e.g. in English) is usually included. This is an unhelpful sense, it seems to me, as it is using the term 'paralanguage' in a catchall, negative way, i.e. for every vocal effect which cannot be accounted for in other, previously defined, more central (in some sense) linguistic categories. In fact, the boundary-line between the more generally accepted paralinguistic features (in the sense of 4 - 6 below) and these segmental vocalizations is unclear, because of their functional overlap: durational variations which are non-segmental require reference to silence as a structural marker; but silence is linear (used sequentially and not concomitantly with segmental (verbal) utterance), and can also be used as an exponent of hesitation; hesitation, however, can be vocalized segmentally (cf. Blankenship and Kay [1964] and footnote 14 below); and once this is allowed in, then a wide range of formally similar utterances can equally well be called paralinguistic. But any such extension produces some terminologically bizarre situations, reflecting a conceptual confusion: for example, a segregate uttered on overhigh pitch would be a paralinguistic feature varied paralinguistically. It would seem to be clearer, as well as more consistent, to restrict paralanguage to the non-segmental component of utterance; though even here there are differences of opinion as to how much should be included under this heading, as the next three senses show. 4. Including voice quality as well as (all or most) non-segmental features. The distinction between voice quality — in the sense of a permanently present background person-identifying vocal characteristic — and language is generally accepted in some form or other within linguistics, though the terminology for the former varies (e.g. 'timbre', 'voice set', 'personal articulatory setting').9 Voice quality is considered idiosyncratic, biologically controlled, irrelevant to the semantic interpretation of the message; language is considered to be shared, culturally controlled, the basis of semantic interpretation.10 Paralanguage subsumes voice quality in e.g. Cammack and Van Buren (1967:7), where it is defined as 'variations in voice quality, tempo, register, and volume';11 or Bronstein and Jacoby (1967 : 83), where it is characterized as 'special qualities of voice or special changes in the tempo or 9

See, for example, Abercrombie (1967 : 91), Laver (1968), and other references there. It is not the purpose of this paper to survey research into voice quality; but insofar as paralinguistic analysis presupposes an ability to distinguish voice quality from other effects, then some awareness of current research in this area will clearly be helpful. Some recent papers on speaker identity, norms of articulation, etc. are the following: Fonagy and Magdics (1960), Kersta (1962), Siertsema (1962), Hargreaves and Starkweather (1963), Berger (1964), Floyd (1964), Voiers (1964), Akiyama and Yumoto (1965), Han (1966), Janota (1967) and Laver (1968). Special emphasis on the physical analysis of voice quality (or some aspect of it) may be found in: Black (1961), Bowler (1964), Garvin and Ladefoged (1963), Laver (1964, 1967), and Wendahl (1966). A review of the literature, and a discussion of its relevance to nonsegmental phonology is to be found in Crystal (1969a). 10 For phenomena which might be considered as falling in between these alternatives, see the discussion on sociolinguistic variation and paralanguage at the end of this paper. 11 They further define it as 'all phenomena of speech that are not language', defining language as 'the phonological, morphological, and grammatical subsystems, together with the peripheral semantic and phonetic effects'.

272

DAVID CRYSTAL

in the loudness or softness that may accompany our speaking'.12 Halliday, Mcintosh, and Strevens (1964:96) seem to identify paralanguage with voice quality ('features, such as voice quality and handwriting, which do not carry formal contrasts'). Duncan (1969:118) also includes speech non-fluencies (defined as uncontrolled vocal effects) as paralanguage, and studies paralinguistic behavior in psychotherapy interaction, where the distinction between controlled, normal behavior and its opposite ceases to be relevant for any definition of the term (cf. Rubenstein and Aborn [1960] for an interpretation in terms of language disturbance). However, the theoretical importance of distinguishing between controlled, contrastive communicative behavior and uncontrolled, non-contrastive behavior within a group is so central that there seems nothing to be gained by putting them both under the same label. 5. Including only non-segmental features, but excluding prosodic phonemes and voice quality. This is the dominant sense of the term, and the one which will be used in the rest of this paper, unless stated otherwise. In this view, the pitch, stress, and juncture phonemes are considered linguistic; other functions of pitch, loudness, and duration, and any other suprasegmental effects, are considered paralinguistic. Markel (1965) defines paralanguage as the 'non-phonemic' aspects of speech; Duncan, Rosenberg, and Finkelstein (1969) say, for example, 'paralinguistic intensity and pitch would be transcribed when these variables in speech exceed the range of variation necessary for conveying phonemic stress and pitch'; Austin (in press) talks of 'the signal system of the nonarticulated vocal tract'. Ostwald (1964: 17), however, seems to allow intonation into his definition of 'paralinguistic acoustic cues', when he illustrates these by reference to 'variously intoned forms of "oh" and the nuances which support or belie overt meaning of words'. Scholars not working within a framework of prosodic phonemes either tend to call everything non-segmental paralinguistic, or (more often) talk about 'intonation and paralanguage'.13 The distinction between phonemic and non-phonemic aspects of non-segmental phonology is of crucial importance in assessing paralinguistic studies; consequently it will be discussed further below. 12

'Qualities' here is not in Trager's sense, but is more like Abercrombie's (1967 :91), as Bronstein and Jacoby's discussion in chapter 2 makes clear. The confusion this particular term can cause can also be seen in, for example, the title of Markel, Meisels and Houck (1964), "Judging personality from voice quality". Here, 'voice quality' is being used in Trager's sense; but the research they refer to in the thirties, at the beginning of their paper, did not use the term in this way, but in the more general sense referred to at the beginning of paragraph 4 above. It would certainly help matters if Trager's voice qualities were always referred to in the plural. 15 The distinction between intonation and paralanguage is however blurred in Abercrombie (1967 : 130), where paralinguistic phenomena are defined as 'features of voice dynamics such as continuity [namely, incidence of pauses], variation in loudness, in tempo, and in tessitura' (namely, a range of notes within which the pitch fluctuation of a speaker's voice falls during normal circumstances). Intonation is part of paralanguage for Herriott (1969:14), as are gesture and most other non-segmental features (cf. pp. 46-7, 151).

PARALINGUISTICS

273

6. Including only a sub-set of non-segmental features other than prosodic phonemes and voice quality. Paralinguistic features here are defined with reference to a description of non-segmental phonology for a particular language to produce a very restricted sense for the term, e.g. Crystal and Quirk (1964) and the work influenced by them. However, a requirement of any theory of paralinguistic phenomena, in whatever sense, is that they be defined with reference to phonetic criteria of a general nature (types of articulatory movement, etc.), and not solely from (in this case) English. The paralinguistic systems of Crystal and Quirk ultimately have to be seen within a more general framework, to allow for cross-cultural comparisons; and it would seem more useful to phrase one's definition of paralanguage in these terms to begin with.14 7. Functional definitions. Paralanguage does not seem to have ever been defined purely in functional terms, but there is usually a functional component in the definitions which are given. This normally amounts to no more than saying that these features are non-linguistic or extra-linguistic (cf. Abercrombie's definition (1968:55) as 'non-linguistic elements in conversation', or Cammack and Van Buren's (1967:7) as 'all phenomena of speech [that] are not language') — approaches which would be vacuous without any explicit definition of language. More specifically and positively, paralinguistic effects are said to be either expressive of emotions or personality (which is the standard account), or a means of identification of social groups (e.g. Lerman and Damste 1969) or language varieties (e.g. Weeks 1970). Lotz's useful article on the structure of speech (1963) uses the term 'paraphonetic' (p. 23) to refer to a channel of speech which gives information about 'emotion, attitude, etc.', though it is not entirely clear to what extent there is an identity with another of his categories mentioned in the same paper, the 'pragmatic' features of speech. Reference to a language independent theory of emotion, personality, or social groups is usually absent, however, and the useful work in this area is limited to experimental analyses of a fairly restricted order (see below). Suggestions as to paralanguage having any kind of structural or denotative function are on the whole absent — though if paralinguistic effects can enter into the exponence of certain types of intonation contrast (as stated in Charleston 1960: chap. 1 and Crystal 1969:137), then a potential structural function must be recognized. 14

Also under section 6 one might list the research into pausal phenomena, which has increased in quantity over the past ten years, and which is regularly considered to be paralinguistic. A critical account of this work is given in Crystal (1969a: 169-70) and Duncan (1969), and a general review of the research at University College London under Goldman-Eisler may be found in her 1968 book. Other recent papers are the following: Barik (1968), Bernstein (1962), Black, et al. (1966), Blankenship and Kay (1964), Boomer (1963, 1965), Boomer and Dittman (1962, 1964), Dittman and Llewellyn (1967, 1968), Cook (1969), Goldman-Eisler (1961a-e, 1967, 1968), Grisina (1969), Henderson, Goldman-Eisler and Skarbek (1965a and b, 1966), Lay and Burron (1968), Levin and Silverman (1965), Levin, Silverman and Ford (1967), Livant (1963), Martin (1967, 1970), Martin and Strange (1968a and b), Matarazzo, et al. (1968), Penge (1970), Simkins (1963), Suci (1967), Tannenbaum, Williams and Hillier (1965), and Tannenbaum, Williams and Wood (1967).

274

DAVID CRYSTAL

The existence of this degree of terminological disunity among scholars, and the absence of any generally recognized set of criteria for paralinguistic analysis, thus means that the most useful research is that which either explicitly relates its hypotheses to a specific model of paralinguistic behavior (such as Trager's), or which concentrates on obtaining experimental evidence for some quite specific hypothesis, the terms of which are given a clear but ad hoc definition. The work of Duncan (1965, 1969) and others in psychotherapy interaction and related matters provides an illustration of this, as his investigations take well-recognized paralinguistic effects and examine their function in relation to clearly defined, specific social tasks. Duncan and Rosenthal (1968) provide evidence to suggest that the way in which an experimenter reads his instructions to subjects can be a determinant of their responses to an experimental procedure; and a subsequent and more rigorous experiment shows the relevance of paralanguage to the concept of experimenter bias more clearly (Duncan, Rosenberg, and Finkelstein 1969). Duncan, Rice, and Butler (1968) show that certain paralinguistic criteria are able to differentiate therapist behavior in peak and poor interaction hours. In a different, but comparably specific connection, Davy and Quirk (1969) show that paralinguistic behavior is of relevance in assessing informants' judgments about the acceptability or otherwise of syntactic structure. Experimental work on the whole is lacking, however, and few of the projects which are currently examining aspects of paralinguistic behavior on a larger scale, as part of a broader communicational framework, have as yet reported anything other than their first speculations: cf. Blasdell and Aram (1969) in relation to first language learning; Austin (in press), who relates his study to animal communication and other non-verbal activities; Lomax and others (1968), in relation to the cross-cultural study of song and dance style (cantometrics, choreometrics); the Tracor project (Pendergraft and Ziehe 1967) on developing a programming system for semiotics; and Ostwald (1963) in connection with psychotherapy interaction. There is a marked lack of descriptive studies. English tends to be the language of illustration still, with very often the same examples cited over and over again. Novelty is provided by Austin's contribution (1965, in press), by Vanderslice and Pierson (1967) on Hawaiian English (their term 'prosodic' subsumes some of what other scholars would call paralinguistic), by Lawrence (1967) on some aspects of Texan English, by Crystal and Davy (1969), who make frequent reference to paralinguistic effect in their study of occupational and other varieties of English, by Pellowe (1970) on Tyneside English, and by the work into English varieties by Osmers (1967), Davy (1968), Benyon (1969), and Kempson (1969). There is little on other languages, either explicitly or implicitly, under the heading of paralanguage, apart from Cammack and Van Buren's illuminating cross-cultural study of aspects of English and Japanese (1967), and Black, et al. (1966). Lawendowski (1970), using a Pittenger and Smith (1957) framework, makes some reference to Polish; and Fonagy and Magdics (1963) refer to Hungarian. Crystal (1971) has surveyed

PARALINGUISTICS

275

the sociolinguistic literature from this point of view, and suggests that paralinguistic effect is of primary importance in the identification of several social categories (specifically age, sex, status, occupation, and speech function). He also points out that relatively little attention has been paid to this in the descriptive linguistic or anthropological literature. There are, fortunately, a few exceptions to this statement. An important publication is the collection of papers edited by Gumperz and Hymes (1964), and, within this, Hymes's emphasis on the need to develop an 'ethnography of speaking', in which he refers to the need for semiotic and related studies. Many of the contributors to this volume underline his point. Albert, for example (1964), refers to the training in tone of voice and its modulation, inter alia, for men in Burundi, and shows its relevance to age, sex, kinship, and other relationships, referring to certain highly conventionalized speech patterns such as those used in visiting formulae, petitioning situations, rules of precedence, and respect patterns. Distinctions are made, many of them paralinguistic, according to the social role of those present, the degree of formality (especially relating to whether the situation is public or private), and the objectives of the speech situation. The term 'paralanguage' is not however used. John Boman Adams mentions the importance of stereotyped pitch patterns and tones of voice in order to establish status between participants in one dialect of Egyptian (1957:226): the villager is ordinarily conditioned to give and receive communications whose content is so stereotyped that he pays little attention to it other than to note that it conforms to the norms of traditional utterance and that the speaker is socially acceptable . . . These statuses are often established in the exchange of stereotyped expressions of esteem and concern that are obligatory whenever two or more persons meet. Since the same expressions are always uttered, interpretations of 'friendliness' or 'enmity' depend upon meanings conveyed by subtle qualities of tone, pitch, and melody. These qualities, in their different modes, are interpretable to one who is acquainted with their culturally defined meanings. In Cayuvava, a language of Bolivia, there is a set of nasal phonemes, but nasalization also occurs with 'honorific' stylistic function (cf. Key 1967:19): an individual of lower social or economic status addresses one of higher rank with a prominence of nasalization for all vowels of the utterance; and similarly with a woman being polite to her husband, or a man asking a favor. According to Ferguson (in Sebeok, Hayes, and Bateson 1964:274), velarization in Arabic indicates, amongst other things, masculinity, whereas avoidance of velarization indicates the opposite. Effeminacy in English seems to be partly a segmental, partly a paralinguistic matter (cf. Crystal 1971 : 189, Lerman and Damste 1969). According to Garbell (1965), many female speakers over 70 of Urmi, a dialect spoken by Jews in N. Persian Azerbaijan, replace practically all 'plain' words by 'flat' words, i.e. words consisting of 'flat' phones, which in Garbell's metalanguage means such features as the strong velarization of all oral consonants, the articulation of all labials with marked lip

276

DAVID CRYSTAL

protrusion and rounding, and pharyngealization. Kelkar (1964), under the heading of 'paraphonology', refers to the extended pitch and loudness characteristics, and the relatively slow and regular speed of baby-talk, in Marathi, and mentions certain general vocal effects, such as pouting and palatalization (cf. also Ferguson 1964). In connection with distinguishing genres of speaking ('speech functions'), Jacobs (1956:127) emphasises the relevance of 'stylized devices such as connectives, pauses, and vocal mannerisms'; and paralinguistic effects are indeed quite often referred to in this area. For example, Devereux (1949:269) refers to the staccato, rapid manner of delivery of a traditional, memorized text in Mohave, which the speakers find very difficult to slow down (sometimes impossible, when the utterance is in front of other people from the same tribe). Samarin (1968) refers in passing to paralanguage in Gbeya insults. Paralinguistic variations in modes of recitation (e.g. of prayers, formulae) are referred to by Fischer (1966:180-1), Conwell and Juilland (1963 : 20), West (1962 : 90), and Crystal (1971 : 195). Fuller discussion of conversational genres referring to paralinguistic effects is found in Conklin (1959) and MigUazza and Grimes (1961). Conklin studies ways of modifying normal speech-patterns for purposes of entertainment or concealment (most frequently as part of voice disguise in customary courting behavior) in Hanunoo. Migliazza and Grimes differentiate between 'one-sided' and 'balanced' types of conversation in Shiriana (1961:36-7). They illustrate the former by reference to 'myths' and 'narratives', and their distinction is worth quoting because of its detail: Phonologically a myth is characterized by an initial period in which only lento pause groups occur, a body in which combinations of lento and andante pause groups occur, many of which contain ideophonic feet, 15 and a termination in which one or two lento pause groups occur, with extra length on the vowel of die final stressed syllable in the contour and at times a voiced breath intake after the end of the final pause group . . . four pitch levels [adequate for normal speech] do not handle the pitch patterns, which range over a wide area and move largely in long glissandos.

As an example of 'balanced conversation', the authors cite the BARGAINING DIALOGUE, which is 'delivered at night by a trading partner from one village to his partner from another in the presence of all hosts and members of a trading party, in which each partner's speech has the general characteristics of a monologue, except that the intonation is replaced by a chant form' (p. 38). They also mention the relevance of crescendo, decrescendo, and laryngealization for the definition of certain speech styles. It should be emphasized that most fieldworkers, even in linguistics, do not make a systematic survey of paralinguistic effect a routine part of their investigations. People still seem unaware of the KIND of phenomenon which they are liable to come 15

An ideophonic foot is a highly conventionalized effect, referring to a rhythm unit 'accompanied either by an anomalous pitch pattern... or by a voice quality that stands out in contrast with that of the rest of the utterance (usually laryngealized or breathy in relation to the overall voice quality)' (p. 35). 'Lento' and 'andante' are differentiated partly in terms of speed, and partly in terms of the number of contours involved.

PARALINGUISTICS

277

into contact with in this area of communicative activity, how they should describe what they hear, or how they should integrate this with other aspects of any linguistic description they may be making. On the whole, the above references are exceptional, and few of them actually use the term 'paralanguage'. Moreover, they are not readily comparable, as, in the absence of any generally-used theory, there is no guarantee that different scholars are using such terms as 'tone of voice' in the same way (it is often obvious that they are not16); there have been few attempts to transcribe utterances in order to indicate the frequency of occurrence and distribution of specific effects; and there have been certain methodological weaknesses (e.g. a failure to distinguish clearly between voice quality and linguistic effect — a point I shall discuss further below — or to provide information about sampling of data or selection of informants). The sporadic, impressionistic nature of many of the above accounts is to be regretted, but the state of affairs is probably unavoidable at the present time, in view of the amorphous theoretical situation which characterizes fundamental research. There is after all little point in accumulating quantities of comparative detail which cannot, in the absence of an explicit theoretical framework, be compared. Within paralinguistics, under the heading of the linguistic communication of emotion, there has been a considerable amount of discussion and experimentation, but once again, in the absence of consistent theoretical distinctions between such terms as 'controlled' and 'uncontrolled' (behavior), 'emotion' and 'personality', 'trait' and 'stereotype', it is often difficult to compare the results of different groups. General reviews of this area are to be found in: Charleston (1960), Starkweather (1961), Dittman and Wynne (1961), Soskin and Kauffman (1961), Sychra (1962), Fónagy and Magdics (1963), Knapp (1963, especially Mahl's contribution), Davitz (1964), and Shapiro (1968). Kramer (1963) is a widely read overview of work on 'nonverbal' properties of speech on the one hand, and personality and emotion characteristics on the other. Crystal (1969a: 62-77) provides another critique on similar lines. On the whole, experimental work has been geared towards determining the extent to which vocal effects are indices of different physical or personality traits or types as viewed 'from the outside', by obtaining reactions of sets of judges to stimuli. The work of Markel and others is of particular importance here. Papers by Markel, Meisels, and Houck (1964) and others have shown very clearly that specific impressions of a speaker's physical characteristics and demeanor are determined by qualities of voice (in Trager's [1958] sense). Markel, Eisler, and Reese (1967) have indicated that regional dialect features in voice quality are significant in judging personality; and in a more comprehensive approach, Markel (1969a) finds further evidence to suggest that a formal profile of voice qualities is an index of personality types. The link between emotional and personality states is suggested in Costanzo, Markel, and Costanzo (1969), where voice quality pro19 Few people would make a systematic distinction between 'tone', 'pitch', and 'melody', as in the quotation from Adams above, for example.

278

DAVID CRYSTAL

files are viewed as representing modes of interpersonal orientation, which they interpret as constituting a bridge between personality dispositions and emotional states. Recent additions to the literature reviewed by Kramer are Ptacek and Sander (1966) on age recognition from the voice; Schwartz and Rine (1968, but cf. Schwartz 1968 also) on sex recognition from aspects of whispered speech; and Lay and Burron (1968) on hesitancy evaluation. Beier (1969) and others are working on the evaluation of cues (one class of which are paralinguistic in character) in judging emotional meaning, and showing that there is consistency in rating. Markel (1965) provides clear evidence of the reliability of rating some types of paralinguistic features. Effects which would be classified as paralinguistic have rarely been referred to in studies of first language acquisition, which is hardly surprising, in view of the methodological difficulties of obtaining information in this area. The role of intonation in language acquisition has been discussed, however (see Crystal, 1969b, for a review of this topic), and in view of the integral relationship postulated between intonational and paralinguistic effects for the child, it is likely that the discussion will be broadened to take account of the latter. Crystal in fact argues that paralinguistic effects are among the first language-specific features discriminated and produced by the child (cf. also Weir 1966 and other references there, where it seems that other vocal effects as well as intonation were involved). Current research into cross-cultural comparison of children's vocalizations during the first year, to determine the point and nature of language-specific divergence, is being carried on by Crystal and others. Blasdell and Aram (1969) are investigating the kinesic, phonetic, and prosodic systems of the first language learner, though they are working with older children (15-30 months). Another survey into the development of language out of infant vocalizations is being carried on by Fargo (1967) and others. Discrimination of parameters of relevance for the identification of paralinguistic effects is studied (but only for children with reduced phonemic systems) by Bergendal and Soderpalm Talo (1969). Carlson and Anisfeld (1969) note various paralinguistic effects in their two-year-old subject. Brooks, Brandt, and Wiener (1969) study the responses of children from middle and low socioeconomic groups to verbal reinforcers communicated with and without tonal inflection (but some paralinguistic effects enter into the stimuli). Levin and Silverman (1965), and Levin, Silverman, and Ford (1967) have studied the varying incidence of hesitations in children's speech, relating it to such tasks as explanation and description. A detailed review of the literature is to be found in Crystal (1969c). It is clear from this that many research groups are investigating language acquisition situations where paralinguistic effects are relevant, but are not controlling for them or paying them sufficient attention. A major departure from the American approaches to paralinguistic study (cf. the second trend cited on p. 265 above) came in the work of various British scholars in the early sixties, and provides the main source of theoretical comment on many of

PARALINGUISTICS

279

the questions raised at the beginning of this paper. A comprehensive bibliographical picture may thus be helpful. Crystal (1963) is an attempt to introduce paralinguistic studies to an audience which on the whole (apart from Abercrombie and Catford) had ignored it. Various criticisms were made and expanded in Crystal and Quirk (1964), which is largely concerned with the formulation of analytic principles, the definition of phonetic parameters for the description of paralinguistic features, and the development of a fresh notation for them. Crystal (1969a) is an attempt to develop an analysis of intonation for English within the framework of a more general theory of non-segmental phonology. This book incorporates discussion of the linguistic status of paralanguage (originally to be found in Crystal 1966), and various experimental work (see Quirk and Crystal 1966, and Crystal's paper to the Xth International Congress of Linguists, 1967). Further illustration of the approach is to be found in Crystal and Davy (1969), where different varieties of English are analyzed in these terms, and a number of theses researched at University College London (Osmers 1967, Davy 1968, Benyon 1969, Kempson 1969, Penge 1969). The approach is also being used in the Tyneside Dialect Survey at the University of Newcastle-upon-Tyne (see Pellowe 1970). There is a brief rejoinder to Crystal's original article in Trager (1964), and further comments on the points of difference are to be found in Rensky (1966) and André (1965, 1966). Further replies to the main criticisms which have been made of Trager's approach are so far lacking, but are expected. The central question, which should have been raised years ago, and which has still no final answer, concerns the linguistic status of the vocal effects being labelled 'paralinguistic'. Trager's position is well-known, and is quite explicit, so I will not present it in detail here. As Trager says (1964:23), 'the very essence of the analysis of paralanguage is in the recognition that paralanguage is NOT a part of language, and that therefore the statements about it are NOT linguistic statements'. This position follows naturally from the assumptions Trager holds: that linguistics proper concerns the analysis of phonological and morphological systems only (microlinguistics), pitch, stress, and juncture phonemes being recognized as part of this description (Trager 1949, Trager and Smith 1951). The arguments against this position hinge on the question of criteria: on what grounds is the distinction between emic and non-emic phonological features based, and why should the 'proper' subject-matter of linguistics be restricted to the former? These arguments are general ones, based on considerations of the nature of language as a whole. They assert that the microlinguistic view of language proper is an arbitrary preconception restricting one's view of other phenomena, which forces an analyst to make a binary 'yes/no' decision as to the status of certain features, where perhaps there is none, or where the analysis is of a 'more/less' kind. This point has been made in various places. Bazell (1954:133) asked why phonemes should be the only criteria of relevance for intonation analysis, and his arguments apply a fortiori to paralanguage. Haas too (1957:159) criticized the 'segmental principle' as being a major prejudice

280

DAVID CRYSTAL

in linguistics. (Cf. also Halliday 1961:252, 275, Crystal and Quirk 1964:36, Crystal 1969a: 186, Bolnger 1949 [and later discussion by him of 'gradience'], and Sebeok 1962:437-9 for a discussion of discreteness and gradience in relation to expressive language.) Stankiewicz also raised the question, at the Indiana Conference (Sebeok, Hayes, and Bateson 1964:266-7), but his point received no answer: 'It seems futile to approach language as a monolith and to exclude from it those phenomena which are not cut to the same pattern . . . The linguistic status of certain features which signal emotion cannot easily be decided, since they do not lend themselves to the kind of systematization that linguists are used to, nor are they easily correlated with other linguistic elements.' 17 He suggests there may be a 'fuzzy periphery' to language, which scholars are now describing more consistently and in 'somewhat different terms from the cognitive or discrete elements of language'; but he leaves as open questions 'the systematicity of the expressive devices and their relationship to the cognitive elements of the code' and the 'grading' of expressive components, as opposed to the 'discreteness' of phonemes and morphemes (p. 267). Crystal (1969a: 190) also argues that there is no reason for judging non-segmental phonological effect by phonemic/morphemic criteria. 'The valuable discreteness of phonemic and morphemic definition, with their amenability to clear-cut substitutability tests of an either/or character, usually free from overlap, does not exclude the fact that there are parts of language which are not amenable to such treatment, but are more accurately and realistically covered by setting up scales of contrastivity, which are of a "more/less" character.. . There seems to be no real reason why these [i.e. prosodie and paralinguistic] features should be excluded from the field of linguistics proper, and why they should not be taken as wholly intralinguistic items, but of a different kind from phonemic and morphemic units, as these are normally understood.' The assumption underlying this argument is that any vocal effect which can be shown to have a systematic, shared, contrastive communicational function is by definition part of the over-all soundsystem of a language, and thus linguistic. In an experimental situation, if the substitution of one vocal effect for another in an utterance would lead judges to consistently rate the two utterances thus produced as 'different' in meaning — and moreover different in the same respect — then these effects are linguistically contrastive. (For reliability in rating paralinguistic features, see Markel 1965.) This procedure is of course essentially the same as that used in traditional segmental phonology. But to restrict its application to segmental units only, or to a sub-set of the suprasegmental ones, is circular, in that any such restric17

Cf. Newman (1946 : 172), whose 'expressive prosody... is not necessarily capable of the same type of systématisation as that which is applicable to usual kinds of morphemes'. Much of the trouble, it has been suggested (e.g. by Kramer 1963, Crystal 1967, 1969a), is due to the uncritical and unsystematic use of vague descriptive labels (e.g. 'harsh', 'precise") in attempting to define the meaning of paralinguistic effects; it is not the effects themselves which are unsystematic and ambiguous, but rather the labels used to refer to them. For an analysis using a large number of such labels, see Voiers (1964).

PARALINGUISTICS

281

tion presupposes the very distinction between linguistic and non-linguistic units which it was the purpose of the substitution test to establish. The procedure has to be applied to ALL vocal effects in the first instance, without introducing naive assumptions taken from early theories of meaning (e.g. suggesting that there is a non-arbitrary distinction between 'intellectual' and 'emotional' information in communication). One then finds that there is certainly systematicity and discreteness in the paralinguistic area, though this is not as marked as in the case of intonation. It is therefore difficult to see what grounds there are for classifying some features of non-segmental phonology as phonemic (or linguistic), and some as non-phonemic (or extra-linguistic).18 To avoid the arbitrariness implicit in this approach, then, it is suggested that a more appropriate model of the non-segmental sound system is to introduce the concept of a SCALE of linguisticness, ranging from 'most' to 'least' linguistic. At the 'most linguistic' polarity would be classified those features of utterance most readily describable in terms of closed systems of contrasts, which have a relatively clear phonetic definition, which display evidence of a hierarchical structure, and which are relatively easily integrated with other aspects of linguistic structure (particularly syntax), e.g. tone-unit boundaries, nuclear tone type and placement, and other 'prosodic' features (in the sense of Crystal 1969a). At the other, 'least linguistic' end would be placed those features of utterance which seem to have little potential for entering into systemic relationships, which are relatively indiscrete, and which have a relatively isolated function and little integrability with other aspects of language structure, e.g. breathy vocal effects in English — effects which are often confused with voice quality characteristics on first hearing. Vocal effects lacking any semantic force would then be considered non-linguistic, and would thus fall under the heading of either voice quality or physiological reflexes (such as coughing). It is, then, largely a terminological matter as to how parts of this scale are labelled: for a mixture of acoustic, linguistic, and historical reasons, it is usual to separate off the effects based on pitch, loudness, and duration, referring to these as prosodic features; but it is doubtful whether it is worth placing any great theoretical weight on such divisions.19 18

In addition to the arguments claiming considerable identity between suprasegmental phonemes and paralinguistic effects in general, one should also remember the arguments against extending the concept of the phoneme to any non-segmental effects - which so far remain unanswered, though Bolinger's early strictures date back to 1949. The reasoning underlying these arguments is essentially the following: that if one examines the physical, structural, and semantic characteristics of vowel and consonant (i.e. segmental) phonemes, on the one hand, and such features as intonation (or, for that matter, paralanguage) on the other, one finds so little in common that to use phonemic terminology to describe the latter is to considerably distort their linguistic identity. For further details, see Bolinger (1949, 1951) and Crystal (1969a : 196ff.). " For these reasons, I doubt whether analyses of paralinguistic behavior in terms of a given pitch-phoneme framework can be valid (cf. Duncan 1969 : 134 and Duncan, Rice, and Butler 1968 : 569, who make reference to their location in phonemic clauses). In view of the arbitrariness of the distinction betweeen suprasegmental phonemes and paralanguage, it is not really surprising that paralanguage should be 'closely coordinated' with intonation (cf. Duncan, Rosenberg, and Finkelstein 1969).

282

DAVID CRYSTAL

In a sense, the value of the above approach is not in the alternative analysis of paralanguage presented, as this might be totally misconceived, but rather in the attention it focusses on the need for analytic criteria to be made explicit. Far too much time has been spent on transcriptional matters and discovery procedures. Duncan (1969:119), for example, which is largely concerned with procedural matters, distinguishes three phases in research into non-verbal behavior: the development of a notation, the analysis of structures between the transcribed behaviors, and the relationships between these and external variables. But the notational contribution of linguistics to paralinguistic study is very much a superficial one, and it should not be stressed at the expense of matters of analysis (a situation curiously reminiscent of the emphases in early discussions of phonemics in the thirties). It is clear from the critical discussion in Crystal (1969a: 75-7, 87-8) and elsewhere that there are a number of theoretical and methodological issues relevant to paralinguistic analysis which have hardly begun to be studied. Of particular importance is the need to develop a more sophisticated phonetic framework for the description of paralinguistic effects, in order to provide a more objective set of correlates for the notional categories of description that are regularly used. Lieberman and Michaels (1962) is one attempt to do this using acoustic criteria. Catford (1964) is another, using articulatory criteria: his study of phonation types is a preliminary survey of basic laryngeal activities in speech in terms of stricture-type and location, vocal-fold length, thickness and tension, upper larynx constriction, and vertical displacement of the larynx. It will also be important to correlate the phonetic 'basis' of paralinguistic effects with that of vowel and consonant phonemes. 'Secondary articulations' such as velarization, palatalization, and labialization are regularly cited in both phonemic and paralinguistic studies, but there does not seem to be any generally recognized means of integrating this information, and there is a basic lack of knowledge about the 'facts'. For example, does a language which uses velarization as a normal phonemic distinctive feature also allow paralinguistic velarization? To what extent are paralinguistic effects reactions against regular phonemic tendencies in a language (cf. Stankiewicz's remarks [1964:246] about the aversion of speakers of languages neighboring on Russian to the Russian feature of palatalization when it was introduced into speech, and similar examples)? Further empirical studies are clearly necessary, as the amount of data on which generalizations about paralanguage are based is extremely small (as mentioned above, largely restricted to English, and moreover to fairly abnormal varieties of English, such as psychotherapy interaction).20 Similarly, there is a need for a more broadly based view of the functional role of paralanguage and how it relates to non-linguistic effects. Functional definitions purely in terms of 'emotional' or 'affective' information are inadequate, as they do i0

Blasdell (personal communication, 1969) is conducting a series of pilot studies on the acoustic characteristics of paralanguage in spontaneous conversation.

PARALINGUISTICS

283

not recognize the possibility of a structural role for these phenomena,21 and a social function for paralanguage is only occasionally (and vaguely) referred to (cf. Crystal 1969b). In the latter case, the most important task is to ensure that the levels of abstraction in any social model are clearly defined and related to the linguistic variables. The situation is far more complex than appears at first sight. A person, while speaking, apart from his permanently present self-identifying vocal characteristic (his VOICE QUALITY), simultaneously produces vocal effects which identify him as a member of a number of specific communities (e.g. belonging to a particular race, or nation, or class, or occupational group). Different sets of effects identify different communities, and each set may be said to form (in the absence of a better term) a COMMUNITY-QUALITY (e.g. British, upper-class, preacher, actor).22 (The correlation between the various social abstractions and their corresponding linguistic ones, e.g. whether 'language' or 'dialect' best correlates with the concept of 'nation', need not concern us here.) The first point which has to be made is that the effects constituting a community-quality are not semantically contrastive within that community — for example, a working-class Yorkshire Englishman does not notice that any features of his voice are (a) English, (b) Yorkshire, or (c) working-class, as long as he is talking within his own community, i.e. to other working-class Yorkshire English people. The situation is complicated, however, by the fact that any of the community-qualities can be seen 'from outside', and some or all of the vocal effects from it used by speakers from outside the community in their speech, e.g. in telling a joke. All community-qualities may be affected in this way; and even an individual voice quality may be given an institutionalized role (e.g. 'he's talking like a Churchill'), though this is less common. Moreover, a community may respond collectively to another community, as in the case of people's reactions to the communityqualities of racial minority groups, or in the examples given by Stankiewicz (cf. above). A second complication is introduced by the fact that most people, not being phoneticians, get their facts wrong when they are attempting to imitate another community or individual, and instead of (or as well as) introducing vocal effects which are indeed there, produce imagined features ('stereotypes'). Often, in fact, the stereotyped features have more of an evocative effect than the real features (cf. 'You hardly sound like an American at all', and the like). Lastly, there is also the point that, while speaking, a person is using his language system with its normal range of contrasts, and this includes paralanguage. But each community may have its own paralinguistic features (analogous to regional dialect features in segmental phonology, syntax, etc.), and distinctions therefore have to be made between those paralinguistic features which belong to the language as a whole ('common-core' It is normal to talk of a grammatical function for intonation, but paralinguistic effects can expound intonational categories (e.g. whisper [cf. T r i m 1967] or creaky voice), and this has to be allowed for. " There is unfortunately no generally-agreed term for 'voice-quality of a community', hence the above coinage. Honikman's 'articulatory settings' (1964) might do, but the phenomena being discussed above should not be viewed as having a solely articulatory definition. 11

284

DAVID CRYSTAL

features, used by all speakers, whatever their class, occupation, etc.), and those features which identify specific sociolinguistic categories within the language. We can summarise the set of socio-paralinguistic possibilities in the following way: PERSONAL VOICE QUALITY

COMMUNITYQUALITIES

RACE NATION CLASS etc.

REAL or STEREOTYPED

May be introduced into the speech of another

SPEAKER RACE NATION CLASS etc.

LINGUISTIC FEATURES (including PARALANGUAGE)

It is unlikely that the above outline covers the whole range of sociolinguistic variables involved, but at least the factors mentioned have to be distinguished and assessed. I am not of course suggesting that these factors are all thoroughly understood. It is by no means clear, for example, whether the contrastive use of community-qualities by another community is legitimately or usefully classifiable as paralinguistic, or whether it ought to be called something else. And in the absence of both data and a consistent theoretical approach on the part of either linguists or sociolinguists,23 the questions can hardly be resolved at the present time. It is to be hoped, however, that the recognition of these irritating complications in paralinguistic study will act as a stimulus, and not as a deterrent, to subsequent fundamental research. 15

Some discussion of these matters is to be found in Hammarstrom (1963, and in subsequent writings, including his 1967 paper to the Xth International Congress of Linguists): he distinguishes between 'idioprosodemes* (features of particular speakers), 'socioprosodemes' (features of particular groups within a society), and 'diaprosodemes' (features of groups of speakers from a particular geographical area). See also Webster and Kramer (1968), Chreist (1964), and Cohen and Starkweather (1961).

PARALINGUISTICS

285

REFERENCES 1 9 6 7 . Elements of general phonetics. Edinburgh, University Press. . 1968. Paralanguage. Disorders of Communication 3.55-9. ADAMS, J . B . 1 9 5 7 . Culture and conflict in an Egyptian village. A M A 5 9 . 2 2 5 ABERCROMBIE, D .

35.

and K . YUMOTO. 1 9 6 6 . A study of voice identification using Japanese speech. Study of Sounds (Tokyo) 1 2 . 2 0 9 - 2 3 . ALBERT, E. M. 1964. 'Rhetoric', 'logic', and 'poetics' in Burundi: Culture patterning of speech behavior. AmA 66.35-54. ANDRÉ, E. 1965. Notes on Systems of prosodic and paralinguistic features in English. Révue de Phonétique Appliquée 1.7-13. . 1966. Le statut linguistique des traits prosodiques et paraiinguistiques. Mimeo. AUSTIN, W. M. 1965. Some social aspects of paralanguage. CJL 11.31-9. . (In press.) Non-verbal communication. Language resource information for teachers of the culturally disadvantaged, ed. by A. L. Davis. Forthcoming. BARBARA, D. A. 1963. Nonverbal communication. Journal of Communication 13.166-73. BARIK, H. C. 1968. On defining juncture pauses. A note on Boomer's 'hesitation and grammatical encoding'. L&S 11.156-9. BATESON, G. (In press.) Communication. Natural history of an interview, ed. by N. A. McQuown. New York, Grune & Stratton. BAZELL, C. E. 1954. Choice of criteria in structural linguistics. Word 10.126-35. BEIER, E. G. 1969. The accuracy of the interpretation of emotional meaning as a function of audiovisual and neutral content cues. Language Research in Progress Abstract 819C. BENYON, A. J . 1 9 6 9 . A study of the prosodic and paralinguistic features in a corpus of academic lectures in present day English. Unpublished University of London M.Phil, thesis. BERGENDAL, B . - I . , and E . SÓDERPALM TALO. 1 9 6 9 . The responses of children with reduced phonemic systems to the Seashore measures of musical talents. A preliminary study of the ability to discriminate between differences in pitch, loudness, rhythm, time, timbre and tonal memory. FPhon 2 1 . 2 0 - 3 8 . BERGER, K. W. 1964. Some factors in the recognition of timbre. JAcS 36.188891. BERNSTEIN, B . 1962. Linguistic codes, hesitation phenomena and intelligence. L&S 5.31-46. BIRDWHISTELL, R. L. 1959. The frames in the communication process. Paper presented to the American Society of Clinical Hypnosis. Mimeo. AKIYAMA, K . ,

286

DAVID CRYSTAL

. 1961. Paralanguage — 25 years after Sapir. Lectures on experimental psychiatry, ed. by H. Brosin. Pittsburgh, University Press. . 1966. Some relationships between American kinesics and spoken American English. Communication and culture, ed. by A. Smith, 182-9. New York, Holt Rinehart & Winston. BLACK, J. W . 1 9 6 1 . Relationships among fundamental frequency, vocal sound pressure, and rate of speaking. L&S 4 . 1 9 6 - 9 . BLACK, J. W., O. Tosi, S . SINGH, and Y. TAKEFUTA. 1966. A study of pauses in oral reading of one's native language and English. L&S 9.237-41. BLANKENSHIP, J . , and C . K A Y . 1964. Hesitation phenomena in English speech. Word 20.360-72. BLASDELL, R., and D . A R A M . 1 9 6 9 . Kinesic, phonetic, and prosodic communicative systems of the first-language learner. Language Research in Progress Abstract (March). BOLINGER, D. L. 1949 Intonation and analysis. Word 5.248-54. . 1951. Intonation — levels v. configurations. Word 7.199-210. BOOMER, D. S. 1 9 6 3 . Speech disturbance and body movement in interviews. Journal of Nervous and Mental Disease (Chicago) 1 3 6 . 2 6 3 - 6 . . 1965. Hesitation and grammatical encoding. L&S 8.148-58. BOOMER, D . S., and A. T. DITTMAN. 1962. Hesitation pauses and juncture pauses in speech. L&S 5.215-20. . 1964. Speech rate, filled pause, and body movement in interviews. Journal of Nervous and Mental Disease (Chicago) 139.324-7. BOWLER, N . W. 1 9 6 4 . A fundamental frequency analysis of harsh vocal quality. SpMon 3 1 . 1 2 8 - 3 4 . BROADBENT, D. E . , and M . G R E G O R Y . 1 9 6 7 . Perception of emotionally toned words. Nature (London) 2 1 5 . 5 8 1 - 4 . BRODNITZ, F. S. 1962. The holistic study of the voice. QJSp 48.280-4. BRONSTEIN, A . J . , and B . F . JACOBY. 1 9 6 7 . Your speech and voice. New York, Random House. BROOKS, R., L. BRANDT, and M. WIENER. 1969. Differential response to two communication channels: Socioeconomic class differences in response to verbal reinforcers communicated with and without tonal inflection. Child Development (London) 40.453-70. CAMMACK, F . M . , and H. VAN BUREN. 1 9 6 7 . Paralanguage across cultures: Some comparisons between Japanese and English. ELEC 2 2 . 7 - 1 0 , 4 7 . CARLSON, P., and M . ANISFELD. 1969. Some observations on the linguistic competence of a two-year-old child. Child Development (London) 40.569-75. CARRELL, J., and W. R. TIFFANY. 1960. Phonetics: Theory and application to speech improvement. New York, McGraw-Hill. CATFORD, J. C . 1964. Phonation types: The classification of some laryngeal com-

PARALINGUISTICS

287

ponents of speech production. In honour of Daniel Jones, ed. by D. Abercrombie et al., 26-37. London, Longmans. CHARLESTON, B. M . 1 9 6 0 . Studies on the emotional and affective means of expression in modern English. (Swiss Studies in English, 46.) Berne, Franke. CHOMSKY, N . A . 1 9 7 1 . Deep structure, surface structure, and semantic interpretation. Studies in general and oriental linguistics presented to Shirô Hattori on the occasion of his sixtieth birthday, ed. by Roman Jakobson and Shigeo Kawamoto, 5 9 - 9 1 . CHOMSKY, N. A., and M. HALLE. 1968. The sound pattern of English. New York, Harper & Row. CHREIST, F. M . 1 9 6 4 . Foreign accent. Englewood Cliffs, Prentice Hall. COHEN, A., J. F. SCHOUTEN, and J. 'T HART. 1962. Contribution of the time parameter to the perception of speech. Proceedings of the Fourth International Congress of Phonetic Sciences, Helsinki 1961, 555-60. The Hague, Mouton. COHEN, A., and J. STARKWEATHER. 1 9 6 1 . Vocal cues to language identification. AJPsych 7 4 . 9 0 - 3 . CONKLIN, H. C . 1959. Linguistic play in its cultural context. Lg 35.631-6. CONWELL, M. J . , and A . JUILLAND. 1 9 6 3 . Louisana French grammar, I : Phonology, morphology, and syntax. The Hague, Mouton. COOK, M . 1 9 6 9 . Transition probabilities and the incidence of filled pauses. Psychonomic Science 1 6 . 1 9 1 - 2 . COSTANZO, F. S., N. N. MARKEL, and P. R. COSTANZO. 1969. Voice quality profile and perceived emotion. Journal of Counselling Psychology 16.267-70. CRYSTAL, D. 1 9 6 3 . A perspective for paralanguage. MPhon 1 2 0 . 2 5 - 9 . . 1964. An approach to a reply. MPhon 122.23-4. . 1966. The linguistic status of prosodie and paralinguistic features in English. Proceedings of the University of Newcastle-upon-Tyne Philosophical Society, Series B (Arts), 1.93-108. . 1967. Intonation and semantic structure. Paper read to the Xth International Congress of Linguists, Bucharest. . 1969a. Prosodie systems and intonation in English. London, Cambridge University Press. . 1969b (1970). Prosodie systems and intonation in language acquisition. [Paper read to the University of Toronto Conference on Intonation, November 1969.] Prosodie feature analysis, ed. by P. Léon, 77-90. Montreal, Didier. . 1969c. Non-segmental phonology in first language acquisition: Methodological problems. Paper read to the 2nd International Congress of Applied Linguistics, Cambridge. . 1971. Prosodie and paralinguistic correlates of social categories. Social anthropology and language, ed. by E. Ardener, 185-206. (Association of Social Anthropologists Monograph, 10.) London, Travistock Press.

DAVID CRYSTAL

288 CRYSTAL, D . ,

and

D.DAVY.

1969. Investigating English style. London, Long-

mans. and R. QUIRK. 1964. Systems of prosodic and paralinguistic features in English. The Hague, Mouton. DAVITZ, J . R., ed. 1964 The communication of emotional meaning. New York, McGraw-Hill. DAVY, D . 1968. A study of intonation and analogous features as exponents of stylistic variation, with special reference to a comparison of conversation with written English read aloud. University of London unpublished M.A. thesis. DAVY, D . , and R . QUIRK. 1 9 6 9 . An acceptability experiment with spoken outCRYSTAL, D . ,

put. J L

5.109-20.

G. 1949. Mohave voice and speech mannerisms. Word 5.268-72. DIEBOLD, A. R . 1965. A survey of psycholinguistic research, 1954-64. Psycholinguistics, a survey of theory and research problems, ed. by C. E. Osgood and T. A. Sebeok, 205-91. Bloomington, Indiana University Press. DITTMAN, A . T., and L . G . LLEWELLYN. 1 9 6 7 . The phonemic clause as a unit of speech decoding. Journal of Personal and Social Psychology 6 . 3 4 1 - 9 . . 1968. Relationship between vocalizations and head nods as listener responses. Journal of Personal and Social Psychology 9.79-84. DITTMAN, A. T . , and L. C . WYNNE. 1 9 6 1 . Linguistic techniques and the analysis of emotionality in interviews. JAbSocPsych 6 3 . 2 0 1 ^ 4 . DUNCAN, S. D. 1965. Paralinguistic behaviors in client-therapist communication in psychotherapy. University of Chicago Ph.D. Dissertation. . 1969. Nonverbal communication. Psychological Bulletin 72.118-37. DUNCAN, S. D . , L. N. R I C E , and J . M . B U T L E R . 1968. Therapists' paralanguage in peak and poor psychotherapy hours. Journal of Abnormal Psychology 73.566-70. DUNCAN, S. D . , J. ROSENBERG, and J. FINKELSTEIN. 1969. The paralanguage of experimenter bias. Sociometry. DUNCAN, S. D . , and R . ROSENTHAL. 1968. Vocal emphasis in experimenters' instruction reading as unintended determinant of subjects' responses. L&S 11.20-36. EGOROV, G . G . 1 9 6 7 . Suprasegmental phonology. Moscow, University Press. EKMAN, P. 1965. Communication through nonverbal behavior: A source of information about an interpersonal relationship. Affect, cognition and personality: Empirical studies, ed. by S. S. Tomkins and C. E. Izard, 390-442. New York, Springer. FARGO, N. 1967. Prosodies and phone types in the vocalizations of young children. Mimeo. FERGUSON, C. 1964. Baby talk in six languages. AmA 66.103-14. DEVEREUX,

PARALINGUISTICS

289

C. J. 1968a. The case for case. Universals in linguistic theory, ed. by E. Bach and R. T. Harms, 1-88. New York, Holt Rinehart & Winston. . 1968b. Types of lexical information. Working Papers in Linguistics 2, ed. by C. J. Fillmore, and I. Lehiste, 65-103. Ohio State University, Computer & Information Science Research Center. FISCHER, J. L. 1966. Syntax and social structure: Truk and Ponape. Sociolinguistics, ed. by W. Bright, 168-83. The Hague, Mouton. FLOYD, W. 1964. Voice identification techniques. Washington, U . S . Dept. Comm., A.D-606 634. F6NAGY, I . and K . MAGDICS. 1 9 6 0 . Speed of utterance in phrases of different lengths. L&S 3 . 1 7 9 - 9 2 . . 1963. Emotional patterns in intonation and music. ZPhon 16.293-326. GARBELL, I. 1965. The Jewish neo-Aramaic dialect of Persian Azerbaijan: Linguistic analysis and folkloristic texts. The Hague, Mouton. GARVIN, P. L . , and P. LADEFOGED. 1 9 6 3 . Speaker identification and message identification in speech recognition. Phonetica 9 . 1 9 3 - 9 . GLEASON, H. A. 1 9 6 1 . An introduction to descriptive linguistics. New York, Holt Rinehart & Winston. GOLDMAN-EISLER, F. 1961a. A comparative study of two hesitation phenomena. L&S 4.18-26. . 1961b. The significance of changes in the rate of articulation. L&S 4.171-4. . 1961c. Continuity of speech utterance, its determinants and its significance. L&S 4.220-31. . 196Id. The distribution of pause durations in speech. L&S 4.232-7. . 1961e. Hesitation and information in speech. Proceedings of the 4th London Symposium of Information Theory, ed. by C. Cherry, 162-73. Washington & London, Butterworths. . 1964. Hesitation, information and levels of speech production. Disorders of language, ed. by A. V. S. de Reuck & M. O'Connor. London, Churchill. . 1967. Sequential temporal patterns and cognitive processes in speech. L&S 10.122-32. . 1968. Psycholinguistics: Experiments in spontaneous speech. New York, Academic Press. GRISINA, O. 1969. Hesitation phenomena in English and Russian from a phonetic, phonological and sociolinguistic point of view. Unpublished Moscow University thesis. GUMPERZ, J . J . , and D. HYMES, eds. 1 9 6 4 . The ethnography of communication. Special volume of AmA 66/6 (December). HAAS, W. 1957. The identification and description of phonetic elements. TPhS 118-59. HALL, R . A. 1 9 6 4 . Introductory linguistics. Philadelphia & New York, Chilton Books. FILLMORE,

290

DAVID CRYSTAL

HALLIDAY,

M. A. K .

1961. Categories of the theory of grammar Word 17.241-

92. and P . D . STREVENS. 1 9 6 4 . The linguistic sciences and language teaching. London, Longmans. HAMMARSTRÖM, U . G . E . 1963. Prosodeme und Kontureme. Phonetica 10.194202. . 1967. On linguistic terminology. Paper read to the Xth International Congress of Linguists, Bucharest. HAN, M. S. 1966. Acoustic-phonetic study on speech tempo. Study of Sounds (Tokyo) 12.70-83. HARGREAVES, W . A . , and J . A . STARKWEATHER. 1 9 6 3 . Recognition of speaker identity. L & S 6 . 6 3 - 7 . HAYES, A. S. 1964. Paralinguistics and kinesics. Approaches to semiotics, ed. by T. A. Sebeok, A. S. Hayes, and M. C. Bateson, 145-72. The Hague, Mouton. HENDERSON, A . , F . GOLDMAN-EISLER, and A . SKARBEK. 1965a. Temporal patterns of cognitive activity and breath-control in speech. L&S 8.236-42. . 1965b. The common value of pausing time in spontaneous speech. Quarterly Journal of Experimental Psychology 17.343-5. . 1966. Sequential temporal patterns in spontaneous speech. L&S 9.207-16. HERRIOTT, P . 1 9 7 0 . An introduction to the psychology of language. London, Methuen. HILDEBRANDT, B. 1963. Effektives Sprechtempo, reflexives Sprechtempo und Lautzahlminderung. ZPhon 16.63-76. HILL, A. A. 1958. Introduction to linguistic structures. New York, Harcourt Brace & World. HOCKETT, C. F. 1960. The origin of speech. Scientific American 203.89-96. 203.89-96. HOCKETT, C. F. (In press.) Vocal activity. Natural history of an interview, ed. by N. A. McQuown. New York, Grune & Stratton. HONIKMAN, B. 1964. Articulatory settings. In honour of Daniel Jones, ed. by D. Abercrombie et al., 73-84. London, Longmans. HYMES, D . H . 1 9 6 1 . On typology of cognitive styles in language. AnL 3 . 2 2 - 5 4 . JACOBS, M. 1956. Thoughts on methodology for comprehension of an oral literature. Men and cultures, ed. by A. F. C. Wallace, 123-30. Pittsburgh, University of Pennsylvania Press. JANOTA, P. 1 9 6 7 . Personal characteristics of speech. Transactions of the Czechoslovak Academy of Sciences, Social Science Series, 77/1. Prague, Academia. KELKAR, A. 1 9 6 4 . Marathi baby-talk. Word 2 0 . 4 0 - 5 4 . KEMPSON, R . M . 1 9 6 9 . A transcription and prosodic analysis of an approved selection of spoken English used in an interview. University of London M.A. thesis. KERSTA, L. G. 1962. Voiceprint identification. Nature (London) 196.1253-7. HALLIDAY, M . A . K . , A . MCINTOSH,

PARALINGUISTICS

291

KEY, H. 1967. Morphology of Cayuvava. JanL, series practica 53. The Hague, Mouton. KEY, M. L. 1970. Preliminary remarks on paralanguage and kinesics in human communication. La Linguistique 6.17-36. KNAPP, P., ed. 1963. Expression of the emotions in man. New York, International Universities Press. KRAMER, E. 1963. Judgment of personal characteristics and emotions from nonverbal properties of speech. Psychological Bulletin 60.408-20. . 1964. Elimination of verbal cues in judgment of emotion from voice. JAbSocPsych 68.390-6. LA BARRE, W . 1964. Paralinguistics, kinesics, and cultural anthropology. Approaches to semiotics, ed. by T. A. Sebeok, A. S. Hayes and M. C. Bateson, 1 9 1 - 2 2 0 . The Hague, Mouton. LAVER, J. D. M. 1964. The synthesis of voice quality. Edinburgh University Dept. of Phonetics PAT report. . 1967. The synthesis of components in voice quality. Proceedings of the Vlth International Congress of Phonetic Sciences, Prague, 523-5. Prague, Academia. . 1968. Voice quality and indexical information. Disorders of Communication 3.43-54. LAVER, J., and S. HUTCHESON, eds. 1972. Communication in face-to-face interLAWENDOWSKI, B. 1970. Some observations concerning emphasis. Studia Anglica Posnaniensia 2.73-83. LAWRENCE, T. Z. 1967. Regional speech of Texas: A description of certain paralinguistic features. Paper read to the Xth International Congress of Linguists, Bucharest. LAY, C. H., and B. F. BURRON. 1968. Perception of the personality of the hesitant speaker. Perception and Motor Skills 26.951-6. LERMAN, J . W., and P. H . DAMSTE. 1969. Voice pitch of homosexuals. FPhon 21.340-6. LEVIN,H., and I.SILVERMAN. L&S 8.67-85.

1965.

Hesitation phenomena in children's speech.

H., I. SILVERMAN, and B. L . FORD. 1967. Hesitation in children's speech during explanation and description. J V L V B 6 . 5 6 0 - 4 . LIEBERMAN, P . 1965. On the acoustic basis of the perception of intonation by linguists. Word 2 1 . 4 0 - 5 4 . LIEBERMAN, P . , and S. B. MICHAELS. 1962. Some aspects of fundamental frequency and envelope amplitude as related to the emotional content of speech. JAcS 3 4 . 9 2 2 - 7 . LIVANT, W . P . 1963. Antagonistic functions of verbal pauses: Filled and unfilled pauses in the solution of additions. L & S 6 . 1 - 4 . LOMAX, A . , ed. 1968. Folk song style and culture: A staff report on cantoLEVIN,

292

DAVID CRYSTAL

metrics. Washington, American Association for the Advancement of Science. 1 9 6 3 . On the linguistic structure of speech. Logos 6 . 1 3 - 2 3 . LYONS, J . 1 9 6 8 . An introduction to theoretical linguistics. London, Cambridge University Press. MAHL, G . F . 1963. The lexical and linguistic levels in the expression of the emotions. Expression of the emotions in man, ed. by P. Knapp, 77-105. New York, International Universities Press. MAHL, G. F., and G. SCHULZE. 1964. Psychological research in the extralinguistic area. Approaches to semiotics, ed. by T. A. Sebeok, A. S. Hayes, and M. C. Bateson, 51-124. The Hague, Mouton. MALMBERG, B . , ed. 1 9 6 8 . Manual of phonetics. Amsterdam, North-Holland Publishing Company. MARKEL, N . N . 1 9 6 5 . The reliability of coding paralanguage: Pitch, loudness, and tempo. JVLVB 4 . 3 0 6 - 8 . . 1969a. Relationship between voice-quality profiles and MMPI profiles in psychiatric patients. Journal of Abnormal Psychology 74.61-6. . 1969b. Psycholinguistics: An introduction to the study of speech and personality. Homewood, HI., Dorsey Press. MARKEL, N . N . , R . M . EISLER, and H . W . REESE. 1 9 6 7 . Judging personality from dialect. JVLVB 6 . 3 3 - 5 . MARKEL, N . N . , M . MEISELS, and J. E. HOUCK. 1 9 6 4 . Judging personality from voice quality. JAbSocPsych 6 9 . 4 5 8 - 6 3 . MARTIN, J. G . 1 9 6 7 . Hesitations in the speaker's production and listener's reproduction of sentences. JVLVB 6 . 9 0 3 - 9 . . 1970. On judging pauses in spontaneous speech. JVLVB 9.75-8. MARTIN, J. G . , and W.STRANGE. 1968a. Determinants of hesitations in spontaneous speech. Journal of Experimental Psychology 76.474-9. . 1968b. The perception of hesitation in spontaneous speech. P&P 3.427-38. MATARAZZO, J . D . , A. N . WIENS, R. G . MATARAZZO, and G . SASLOW. 1 9 6 8 . Speech and silence behavior in clinical psychotherapy and its laboratory correlates. Research in psychotherapy 3 , ed. by J . Shlien, 3 4 7 - 9 4 . Washington, American Psychological Association. M C Q U O W N , N. A . , ed. (In press.) Natural history of an interview. New York, Grune & Stratton. MIGLIAZZA, E . , and J. E . GRIMES. 1 9 6 1 . Shiriana phonology. AnL 3 . 3 1 - 4 1 . NEWMAN, S . S . 1 9 4 6 . On the stress system of English. Word 2 . 1 7 1 - 8 7 . OSMERS, M. E. 1967. A study of the present-day spoken English of church sermons. University of London M.Phil, thesis. OSTWALD, P . F . 1 9 6 3 . Soundmaking: The acoustic communication of emotion. Springfield, 111., Thomas. . 1964. How the patient communicates about disease with the doctor. ApLÖTZ, J .

PARALINGUISTICS

293

proaches to semiotics, ed. by T. A. Sebeok, A. S. Hayes, and M. C. Bateson, 11-34. The Hague, Mouton. PARRISH, W. M. 1 9 6 2 . The rhythm of oratorical prose. Studies in rhetoric and public speaking in honor of James Albert Winans, ed. by A. M. Drummond et al., 2 1 7 - 3 1 . New York, Russell & Russell. PELLOWE, J . 1 9 7 0 . Establishing some prosodic criteria for a classification of speech varieties. Mimeo. PENDERGRAFT, E. D . , and T . W . ZIEHE. 1 9 6 7 . Research in semiotic systems. Language Research in Progress Abstract 701. PENGE, H. R. 1970. A study of the distribution of pauses in a specimen of live drama. University of London M.A. thesis. PITTENGER, R. E . , and H. L. SMITH. 1957. A basis for some contributions of linguistics to psychiatry. Psychiatry (Washington) 20.61-78. PTACEK, P . H . , and E. K . SANDER. 1966. Age recognition from voice. J S H R 9.273-7. QUIRK, R., and D . CRYSTAL. 1966. On scales of contrast in English connected speech. In memory of J. R. Firth, ed. by C. E. Bazell et al., 359-69. London, Longmans. RAMSAY, R . W., and L . N. LAW. 1966. The measurement of duration of speech. L&S 9.96-102. RENSKY, M. 1966. The systematics of paralanguage. TLP 2.97-102. ROBINS, R . H . 1 9 6 4 . General linguistics: An introductory survey. London, Longmans. RUBENSTEIN, H., and M. ABORN. 1960. Psycholinguistics. Annual Review of Psychology 11.291-322. SAMARIN, W. J. 1968. The art of Gbeya insults. Paper read to the African Studies Association, California, October. SCHWARTZ, M . F . 1 9 6 8 . Identification of speaker sex from isolated, voiceless fricatives. JAcS 4 3 . 1 1 7 8 - 9 . SCHWARTZ, M . F . , and H . E . RINE. 1 9 6 8 . Identification of speaker sex from isolated, whispered vowels. JAcS 4 4 . 1 7 3 6 - 7 . SEBEOK, T. A. 1962. Coding in the evolution of signalling behavior. Behavioral Science 7.430-42. SEBEOK, T. A., A. S. HAYES, and M. C. BATESON, eds. 1964. Approaches to semiotics. The Hague, Mouton. SHAPIRO, J. G . 1 9 6 8 . Variability in the communication of affect. Journal of Social Psychology 7 6 . 1 8 1 - 8 . SIERTSEMA, B . 1 9 6 2 . Timbre, pitch and intonation. Lingua 1 1 . 3 8 8 - 9 8 . SIMKINS, L . 1 9 6 3 . Modification of pausing behavior. J V L V B 2 . 4 6 2 - 9 . SMITH, H. L. 1952. The communication situation. Washington, U.S. Dept. of State, Foreign Service Institute.

294

DAVID CRYSTAL

. 1969. Language and the total system of communication. Linguistics today, ed. by A. A. Hill, 89-102. New York, Basic Books. SOSKIN, W . F . , and P . KAUFFMAN. 1 9 6 1 . Judgment of emotion in word-free voice samples. Journal of Communication 1 1 . 7 3 - 8 0 . STANKIEWICZ, E . 1964. Problems of emotive language. Approaches to semiotics, ed. by T. A. Sebeok, A. S. Hayes, and M. C. Bateson, 239-64. The Hague, Mouton. STARKWEATHER, J. 1 9 6 1 . Vocal communication of personality and human feelings. Journal of Communication 1 1 . 6 3 - 7 2 . Suci, G. J. 1967. The validity of pause as an index of units in language. JVLVB 6.26-32. SYCHRA, A. 1 9 6 2 . Kapitola z experimentalniko vyzkumu vyrazu v hudbe a reci. Acta Universitatis Carolinae (Prague) 2 . 4 9 - 8 5 . TANNENBAUM, P . H . , F . WILLIAMS, and C. S. HILLIER. 1965. Word predictability in the environments of hesitations. JVLVB 4.134-40. TANNENBAUM, P. H., F. WILLIAMS, and B. S. WOOD. 1967. Hesitation phenomena and related encoding characteristics in speech and typewriting. L&S 10.203-15. TERANGO, L. 1966. Pitch and duration characteristics of the oral reading of males on a masculinity-femininity dimension. JSHR 9.590-5. TRAGER, G . L . 1 9 4 9 . The field of linguistics. ( S I L Occasional Papers, 1.) Norman, Okla., Battenburg Press. . 1958. Paralanguage: A first approximation. SIL 13.1-12. . 1960. Taos HI, paralanguage. AnL 2.24-30. . 1961. The typology of paralanguage. AnL 3.17-21. . 1964. Paralanguage and other things. MPhon 122.21-3. TRAGER G. L . , and H. L . SMITH. 1951. An outline of English structure. ( S I L , Occasional Papers, 3.) Norman, Okla., Battenburg Press. TRIM, J . L . M . 1 9 6 7 . Cues to the recognition of some linguistic features of whispered speech in English. Paper read to the Vlth Congress of Phonetic Sciences, Prague. VANDERSLICE, R . , and L . S . PIERSON. 1 9 6 7 . Prosodic features of Hawaiian English. QJSp 5 3 . 1 5 6 - 6 6 . VETTER, H. J . 1 9 6 9 . Language, behavior, and communication. Ithaca, HI., Peacock. VOIERS, W. D. 1964. Perceptual bases of speaker identity. JAcS 36.1065-73. WEAKLAND, J. H. 1967. Communication and behavior — an introduction. American Behavioral Scientist 10.1-4. WEBSTER, W . G., and E. KRAMER. 1968. Attitudes and evaluational reactions to accented English speech. Journal of Social Psychology 75.231-40. WEEKS, T. 1970. Speech registers in young children. Papers and Reports on Child Language Development 1.22-42. Stanford University, Committee on Linguistics.

PARALINGUISTICS

295

R. 1966. Some questions on the child's learning of phonology. The genesis of language, ed. by F. Smith and G. A. Miller, 153-68. Cambridge, M.I.T. WENDAHL, R. W. 1966. Some parameters of auditory roughness. FPhon 18.2632. WEST, J. D. 1962. Phonology of Mikasuki. SIL 16.77-91.

WEIR,

FACIAL EXPRESSION A N D BODY MOVEMENT 1

HARVEY B. SARLES

Facial expression and body movement are inseparable from linguistic activity in the communication process. Being wed to a tradition which accepts as given the 'independence' of language, however, linguists have been willing to jump to explanations of language as reflecting DEEP-SEATED, well-buffered neurological mechanisms (Chomsky 1968). There has been little attention paid to other aspects of behavior which OCCUR together with language, and are also part of the growth and development processes. Linguists have tended instead to consider gestural and other body movement phenomena either as commentary on language, per se, or as some kind of uninteresting independent communicational system.2 This field of EXPRESSION as it relates to language is a neglected area of study. This paper will show how it may be seen as an integral part of a study of language, as opposed to a language-expression dichotomous INTERFACE. Among the interested disciplines in the area of expression are anthropology, cognitive or developmental psychology, psychiatry, and ethology-behavioral biology. It appears likely that the emergent cross-disciplinary field of Human Ethology will consider this problem area (Sarles 1970a). The rallying issues are many. One is the realization that natural behavior is understudied. Psychology has been content to study a limited variety of response to stimuli in controlled experimental situations. Its picture of man as a perceptual mechanism is presently being challenged (Gibson 1966) in a movement which will have repercussions involving all of the observational disciplines which have accepted an implicitly passive, analytic picture of man as a bundle of essentially independent sensual systems. A second issue is the increasing realization that the view of man as an individual is not complete. Man — and other animals — are SOCIAL animals, and one's developing adult models are SOCIALLY successful; not simply as surviving organisms, but as people who can mate, communicate, and raise their young to live in the same cognitive sphere as they do. 1 My thanks to Martin Q. Peterson for his critical reading and suggestions during the course of writing this paper. ! Even the most empirically oriented linguists seem to have been narrowly empirical in their observations as they have been practitioners of this tradition. 'Language is language' appears to be a typical statement of linguists regardless of their positions on other issues.

298

HARVEY B. SARLES

In the context of man as a bio-social animal, it is simplistic to proceed from the assumption that language is an independent system. Linguistic phenomena are central to the maintenance of bio-social groups. Language is not merely a collection of rules for sentences or sounds, but a part of our ongoing behavior. It is thus necessary to consider verbal and other expressive behavior in many contexts to gain an understanding of how people become and maintain an APPARENTLY rational (i.e. social) existence.8 Traditional linguists must suspend their picture of language as purely verbal. A comprehensive approach to language includes recognition of larger patterns of behavior. Yet, there is no explicit recognition of some simple facts of lip and mouth movement as seen on faces; for example, on faces of parents as they recite popular children's poems. 'Baa, baa, black sheep', has a contrastive lip and mouth change on every syllable — as seen in the face-to-face (vis-à-vis) setting.4 Every bilabial and all palatal sounds (at least in American English) are characterizable not only as sound, but also as gross changes in the faces of the speaker. A child spoken to in the face-to-face sees as well as hears. And in this context, it is noteworthy that most children seem to be unable to transfer these 'language' skills to the telephone until they are in control of quite extensive grammatical and contextual 'machinery'. Seeing and hearing are part of development,5 and the face-to-face setting may be so much a part of one's interaction that only older speakers are able to speak and understand 'telephonese'. In the vis-à-vis we interact with others on a number of levels, all of which may be operating simultaneously.6 We watch skin and clothes, and see whether the * The problem of what rationality means is a very complicated one which impinges on broader questions of epistemology. From a social-behavioral point of view, it might merely mean that this picture is that one which adult humans use as a model for raising their young. Perhaps they see the world in such terms, perhaps not. Recent linguistic work has, on the contrary, apparently accepted a notion of rationality which seems overly accepting of its own truth. 4 The reader might watch himself saying this statement in a mirror, concentrating on the changes in his mouth as it might be seen by an infant or toddler, and attempt to say it much as he would, say, to a one-year-old. It is also useful to 'mouth' the words without any sound, to enhance this sense of what is seen in the vis-a-vis. 5 Blind and deaf children are of great interest in this context. How do they speak? How do the deaf handle their communication problems via such 'languages' as American Sign Language? in a conference on sign language (Center for Applied Linguistics, December 1969 - Wm. Stokoe, chairman) it was the consensus that a viable research enterprise examining communication among people (and chimps - Gardner and Gardner 1969) must take an initial strategy of considering Language and Sign Language as equivalent modes of communication, rather than considering one as the 'surrogate' or substitute of the other. • To better understand this point a musical analogy may be helpful. In a very real sense, singing or playing a musical instrument are physical, bodily processes. But the body 'serves musical ends'. The goal of playing is a musical one, and the body must adjust itself or even be forced in ways which contribute to good music - almost regardless of the contortion the body must execute. A simple example - it is musically necessary to draw the violin bow in a relatively straight line. Since the right (bow) hand moves quite a distance this can only be accomplished by a set of wrist movements which can involve both side to side and circular

FACIAL EXPRESSION AND BODY MOVEMENT

299

mouth or eyes are open or closed. We attend to fairly gross changes in the dynamics of mouths or eyes in motion. We feel and sense the underlying muscular changes of others. We smell body odors distributed differentially by skin, clothes, and hair. The problem of what expression is must also reflect a long dual definitional history. Typically, two types of sense are given to the word; one which is essentially SOCIAL (how do other people react to certain facial appearances or changes, do they reflect 'emotional' states?) and the second, a more bio-structural account (which muscles are involved in expressing 'happiness' or 'disdain'?).7 The so-called 'muscles of expression' are also referred to as the 'mimetic muscles', suggesting that anatomists believe that children mimic the movements they see, translating from parents' faces the reflections of underlying muscles. Part of the cooing, handling, and TLC which infants need to socially and physically develop is an increasing muscular or proprioceptive sense of their own bodies. Apparently, they learn much of this sense from felt contact with other people (Frank 1957). From this point of view language can be seen as muscular as well as auditory. Phonological tradition has made use of a few facial observations in constructing descriptive systems — 'open' means simply that the mouth is open further than in producing 'closed' vowels (it also means that the listener-watcher can usually see teeth and tongue movements); rounded means lip rounding (yet the English affricative consonants seem never to be described this way).8 movements - all to help insure that bow at some distance away is moving smoothly, continuously, and straight relative to the bridge (the string support), and at a particular distance from the bridge, i.e. at the point where it contacts the string. Obviously even this is compounded in moving the bow at different angles on the various strings, so the musician must 'know' a number of variables - muscular, spatial, and musical - all at the same time. Thus the musical production, like the skin of the face we see, is a by-product of muscular movements. 7 There has been a fair amount of comparative study on facial expression; much of it a belated outgrowth of Darwin's behavioral studies (Darwin 1872) - this volume is accounted by many (e.g. Sarles 1969) as the forerunner of modern ethological studies. As its title states, expression was, and to a great extent still is, considered to reflect the 'emotions' in some manner, direct or indirect. While this idea seems fairly innocuous on the surface, the history of the emotions (perhaps it should be called a 'folk history") gives it some special thrust. It appears that 'the emotions' are often given 'entity' status much like 'the body* and 'the mind' - they, like the body, are assumed to be more 'basic', to have more animal-like attributes, and ones which we humans might well share with other animals. It is in this context that a comparative study of emotions, their physiology and anatomy, has been a lively enterprise - but those attributes of the mind which have been reserved for humans, language and mind (and soul), have been assumed not to be open areas for useful comparative study (Chomsky 1966). 8 It is fairly simple to construct a personal visual phonology using a small hand mirror, and contrasting words in minimal contrast settings; e.g. 'bit', 'bet', 'bat' for relative openness of all the English vowels - the mouth is literally open wider for bat than for bet. The sounds /m, 5, J, £, z / are all 'pursed'; the sounds /t, d, s, y / have lips widened in word initial positions, but less so in word medial or final position. Contrast 'latter' and 'ladder' - the mouth is open more widely in forming 'latter'. It is still unclear as to what this might contribute to forming speech sounds, but these visual-facial factors seem to be about as ordered, patterned, and constant as do other phonological features - distinctive features or

300

HARVEY B. SARLES

On the other hand, much has remained unnoted by linguists; the vertical striae on lips are perfectly clear within the distance range of the usual vis-a-vis showing extremely slight changes in underlying circum-oral musculature. These striae indicate impending change, such as mouth opening prior to speaking (and there is a perceptible delay between these two events!).9 People's eyes are also changing contrastively at frequent intervals; it has been demonstrated (Spitz and Wolf 1946) that the smile response in infants, which is claimed to be so characteristically human, is intimately tied into reacting to HUMAN eyes. People breathe, and it is very unusual for people to speak as they inspire. We can rest assured that children are aware of this. Traditional linguistics has been willing to deal with deep, out-ofawareness phenomena of various kinds to the extent that they fit into certain linguistic structures. In this case, the phenomena of expression are comparable to linguistic phenomena, yet have not been considered to function in its structures.

EXPRESSION: TONUS A N D SET

The study of expression must deal with the physical presence of bodies as seen or otherwise sensed in face-to-face interaction.10 If this is restricted to surface phenomena, skin, clothes, and hair, changes in them represent or reflect underlying muscular and vascular changes. Any serious consideration of expression will consider both the anatomical structures which are the foundation of expression and the 'surface' of the face.11 The skin or surface of the face is organized into several areas. If we note CHANGE and CONTRAST, the areas immediately apparent are the eyes and mouth. Color or pigmentation of the eyes and skin enhance such contrasts in different persons and populations; 'steely' eyes are usually green or light blue and quite low in contrast, and changes (winks, blinks, eye shifts) may be less obvious others. It is interesting to note that essentially all professional singers have their lips under a great deal of tension, as if this tension contributes to the proper formation of their sounds. • Working with slow-motion movies and an analyzing projector we were able to note a finite and fairly 'long' (in micro time) interval; surely one which is longer than, say, a blink from start to finish - and one which is almost undoubtedly perceptible in real life/real time situations. The fact that most of us seem to be unaware of such events is probably of no more consequence for its 'reality' than, say, our unawareness of the aspiration following initial voiceless stops in English. 10 It is not unreasonable to suggest, in this context, that speech is 'expression' at a distance. One might assume that it functions in about the same ways as visual or other sensory information in communication or message transfer. In this case speech could be seen mainly as a 'garnish' in the vis-a-vis, with most information handled in other ways. Clearly this characterizes most interaction in households or other settings in which the interactors are well known to one another. The information is relegated to context - a subject which has been rarely studied, and is still considered to be unstudiable at least as a proper concern of linguists. For a counter-view see Benjamin (1969). 11 This brief paper will concentrate on the face, if only as more appealing to linguistic interests. The remainder of the body should be considered in similar ways.

FACIAL EXPRESSION AND BODY MOVEMENT

301

than in darker-eyed people; m o u t h (and eyes) a r e enhanced in relation t o d a r k e r skin. 1 2 T h e r e m a i n d e r of the facial skin seems t o b e differentially organized by 'folds' o r muscular groups which seem t o v a r y m u c h like linguistic dialects.

It appears

that one tends t o 'use' his f a c e m u c h like the people in his family, community, e t c . 1 3 Relationships between the skin and t h e deeper structures a r e n o t well known. P e r h a p s this is b e c a u s e anatomists and surgeons h a v e been interested in structure, n o t relationship, o r b e c a u s e skin lesions are amazingly self-repairing. 1 4 A n i m p o r tant aspect of this relationship is the highly flexible and plastic n a t u r e of the skin, as opposed t o t h e muscles and bones beneath it. T h e skin is not strongly attached t o bones o r muscles, e x c e p t a t relatively few points, s o it is free t o m o v e almost independently.

This enhances o r exaggerates muscle m o v e m e n t s .

T h e bony parts of the f a c e m a y be seen as a kind of foundation f o r t h e muscles. T h e s e muscles are called 'voluntary', but this t e r m implies n o necessary sense of awareness, r a t h e r suggesting that m o s t of us c a n m o v e t h e m w h e n w e w a n t t o . 1 5 Cosmetics seem to enhance such contrasts, perhaps change their nature somewhat. Eyebrows and eye liner seem to provide a contrastive 'setting' in which eye movement seems to be more obvious to the observer. A complicating issue here, of course, is the relationship between information of the sorts: 'I am available!' vs. 'I am your friend!' vs. 'I am feeling fine!' In addition we must realize that attractiveness or beauty is part of the same package. ls The question of how faces come to look like they look is an interesting one, if it is not immediately attributed to innate-genetic factors. Here, again, no systematic research has been done on facial development across human (or other) populations: in families, in the same persons over some number of years, etc. Anecdotally, the changes which can occur when people move into different socio-cultural groups, seem to be quite large. My own relatives who left eastern Europe to settle in China early in this century, appeared, to the American eye, to be quite oriental-looking a generation later. Similarly, orientals in Mexico are said to 'disappear' within a generation or two at most. The question of which children come to look like which parent, or which parent's family, also remains unstudied - although many people seem to spend a fair amount of energy making such judgments as part of their familial folklore. It is difficult to say, at present, whether children really look like one parent, or they come to look like him as they are treated as if they had originally - i.e. we might be dealing with a self-fulfilling feedback system of observation and prophecy and reinforcement. 18

14 Presently (1970) we are conducting anatomical studies of skin and its relationship to bone, muscle, fat, etc. One of our Minnesota students (Ordean Oyen), an expert dissectionistanatomist, is doing baboon facial studies in two directions - from the outside-in, and from the inside-out. Anatomical studies do the former, but the latter are necessary in order to preserve the relationship of skin, as a total tissue, to the underlying structures. 15 Much of facial (and other) muscular mobility seems to relate to 'practice' and/or to exposure in one's immediate social surround. The ability to wink or raise one or the other eyebrow are cases in point. If no one in one's reference community moves particular muscles, it appears that mobility is severely limited early in life and difficult to regain. Anecdotally, I learned to 'curl my tongue' at age 33 - urged by my children who are 'native' tongue-curlers. This occurred first during a cocktail hour, and may help give substance to what the notion 'to loosen up' might mean. It is worth mentioning that I have had many years' experience in bodily training and retraining as a violinist, and know that practice helps to overcome what appeared to be completely impossible at some earlier date. My personal bias has thus always been toward the idea of plasticity - given the right circumstances. It is clear, on the other hand, that most people are not exposed to such circumstances except quite early in life, and become apparently fixed in their muscular habits when quite young.

302

HARVEY B. SARLES

The expression of a face reflects the underlying structure of muscle and bone, the superficial structure of the skin and the relationship between the skin and the other structures. The areas of the face in which expression occurs may be divided, for expository purposes, into an eye component and a mouth component. The set of the muscles in these areas and the movement of the skin around them defines expression. What seems to be meant by looking like oneself is some relationship between the muscle tone (TONUS) in which muscles are ordinarily held, over the bony structure underlying. Our ordinary facial expression is not passive in the sense of expression being a change from flaccid (AT REST) states of muscle tension but changes in TENSION or tonus from already 'tensioned' muscles. This is often referred to as FACIAL SET. Enlow (1968) in his elegant monograph on the growth of the face clearly shows that the development of the bony structure of the face is correlated to the use of the overlying muscles. Now a FACIAL SET is a reflection of the family in which we grow up and, equally, the group to which the family belongs. It is consistent that this facial set will be translated by the muscles to a bony set, or dynamic shaping of the skeletal face beneath the muscles.16 There are a number of aspects of BODY set which can be thought of in much the same way as facial set.17 That is, we have a range of possibilities of how to 'arrange' our shoulders (high or low — giving us the 'appearance' of having short or long necks); pelvis thrust slightly forward (adult female) or backward (adult male — both 'middle majority American'); to hold our legs such that our feet come down 'straight' as we walk, slightly in or slightly out. Woman's posture makes center of gravity adjustments to breasts. Again, these 'adjustments' seem to be culturaldialectal to a great extent: one walks like his group, holds his shoulders similarly, etc. The amount of individual variation which this allows still remains great, however, and we can identify most people at least of our 'own group' by face, and often by manner of walking at fairly great distances.

EXPRESSION A N D LANGUAGE

A linguistically sensitive rationale for the study of expression must proceed from the assumption that language relates somehow to other behavioral phenomena. This might take place within the individual at, say, the neurological level; or may involve more than one person. Whereas language seems to have lent itself to individual18

This may help to account for the anecdotal reports which have spouses growing to look more alike, masters like their dogs, etc. 17 Set may be considered as a relatively constant thing against which to measure and contrast movement. Yet we also have a sense of 'voice set' which is a similar notion, but it has not been used in this manner (Trager 1958). It is, perhaps, more useful to qualify the idea of set in an interactional context as 'the picture of the speaking-moving actor which the listener has in his mind, and against which he is measuring contrastive differences'.

FACIAL EXPRESSION AND BODY MOVEMENT

303

oriented models of language function, expression is usually seen as reflecting socialinteractional behavioral variables. Many linguists have, indeed, considered a 'functional' model of language 'use'. The most popular phrasing is of the sort: 'Language is (used) to communicate.'18 In this context, expression can also be seen as related to communication in a similar functional maimer. It is thus possible to consider several formulations of expression, facial and body, and relate them to language in terms of their similar functions. The possibility of comparing functions must relate to the kinds of 'structures' in which they are seen to operate. This requires a fairly full examination of what interaction and communication might be — yet these are neglected fields compared, for example, to the efforts of language scholars. This has resulted in a very few formulations of the nature of communication structures, and the form of expression studies thus far can be seen as reflecting one of the following: 1) communication as message or information transfer; 2) communication as a constant social process. Most models of language which have developed in the twentieth century (proceeding from Saussure) have utilized to a greater or lesser extent the first model, i.e. language as the exchange of ideas/words or thoughts/sentences. The speaker gives 'new' information to the listener. Developmentally, language acquisition has primarily meant learning words, sentences, or the rules for 'understanding' or 'creating' them. The second model of communication is that which will characterize the emergent field of Human Ethology, and towards which models of language are just beginning to develop (Sarles 1967, 1969, 1970b). This model implies, among other things, no necessary distinction between verbal and non-verbal processes as they function in communication, and truly seems to open up a potential area of comparative study. I N F O R M A T I O N E X C H A N G E MODELS

In the context of the informational model of communication, the study of expres18

There are exceptions to this. Chomsky (1968) and Hjelmslev (1961) seem to opt for a completely independent notion of language; surely independent of its function for communication. Many other linguists have taken its communication aspects for granted, but have urged that the study of language structure is to be granted priority in moving toward understanding interaction behavior. Saussure (1959), Bloomfield (1933), and others unselfconsciously assume an S-R (speaker-listener) model. See my paper, "The study of language and communication across species" (Sarles 1969) for alternative structural and procedural approaches. The notion of structure ordinarily seems to imply an analysis which looks for units and attempts to show how they are put together to make up the whole (structure). The ways of forming larger pieces out of units are often thought of as functions in relation to the structure. If, however, as appears true for notions like Language and Communication, the structure is merely defined into being, declared to be an entity by consensus and/or tradition, then it is unclear as to what notion "units' or 'functions' might mean beyond the fact that anything can be analyzed and recombined in an infinite number of ways. It is in this sense that all such definitions seem circular, or at least incapable of being disproved.

304

HARVEY B. SARLES

sion has taken a number of interesting directions — with the possibilities not yet showing signs of completeness or exhaustion. Most of these have considered units of expression to be much like words or sentences, to be examined as if they represent essentially independent structures. In the absence of any full explanation of communication, a variety of ad hoc operational definitions of structures have had to suffice. These have been rooted in the analytic traditions which consider behavior to be constituted of a number of separable (if not independent) acts or events which characterize our different modalities of operation: auditory, v i s u a l . . . olfactory. The study of 'gesture' has also capitalized on the option of a traditional dualism between EXPRESSION VS. EMOTION to establish legitimation. It has associated a movement (or set of movements isolated into a discrete segment somehow) with some emotion or 'meaning'. Heirs of the traditional dualism seem to equate the terms 'expression, rational and language structure' as opposed to 'emotional, irrational and non-verbal or verbal, non-language which had hitherto been considered to be essentially unstructured (essentially Saussure's parole). The emotions have been considered in the mind-body dualism as more 'primitive' than language; the usual equation seems to be irrational, primitive, animal-like. Biologists thus regard them as prior to man and the origin of 'true' (non-emotional) language; psychiatrically-oriented workers regard the primitive as pre-rational in each person's development; thus we see a similar set of studies emerging from these two disciplines. Excellent examples of this approach are the work of Eibl-Eibesfeldt (biology) and Ekman (psychology) which have interesting correspondences. Each selects out of the world of movement a set of operationally defined gestures, which they presume to be 'universal'. Both are engaged in testing across cultures to see if or how the 'smile' or other 'meaningful' facial expressions are handled by different cultures. The quest for universality would seem to follow from the presumption that emotions are primitive and pre-linguistic — thus they must (from the biological and psychoanalytic points of view) occur in all men. In this formulation (a 'progressive' or additive notion of evolution) man has all of the general attributes animals have — plus he has language.19 The works of both of the authors (e.g. Eibl-Eibesfeldt 1968; Ekman et al. 1969) are well worth pondering in the context of the possible relationships between language and expression. The first point of contact or interface is located in a SEMANTIC realm. Eibl-Eibesfeldt considers the 'smile' to have essentially the 'same meaning' to all people, almost regardless of experiential differences among them. He does Modern biologists have abandoned the Mendelian-progressivist view of evolution which has characterized Social Darwinist thought for the past century. A s I understand the 'modern synthesis' in genetics, the 'problem' of the origin of language would be considered to be a pseudo-problem. Their formulation is that all species are equally evolved through adaptation into different species, but none is considered 'higher' or 'more' evolved than any other. 19

FACIAL EXPRESSION AND BODY MOVEMENT

305

this apparently on the grounds that it is part of what distinguishes men from animals (just as language is presumed to do) and so all people 'must' have it. Similar or homologous movements in animals are interpreted to be non-similar in the same way as animal verbalizations are assumed to be non-language (Smith 1969: 145). Ekman et al. tested the same set of (emotion-showing) facial photographs across cultures, to see whether a variety of different people label them 'identically'. A number of facial expressions are indeed universally 'recognized'. Unfortunately Ekman does not tell us how the inter-language equivalence of terms or labels was determined, so we can only be certain that different populations are consistent in applying the same labels to particular pictures. On another level the relationships between expression (gesture) and language must lie somewhere between putative structure and experiential behavior. Language is considered to be a structure with a grammar; gesture, in order to be comparable, must also be considered in this light. Thus, it should follow that students of expression-as-acts or gestures should attempt to find a grammar of gestures. Differences between language and gesture are immediately clear in terms of linear order, and immediacy of contiguity. Nonetheless, the postulation of the notion of a GESTURAL GRAMMAR would lead to the search for similarities, probably at the level of sentences composed of gestures as units — and sequences of gestures as analogous, say to phrase structures. Another relationship between gesture and language — in the context of static studies — has to do with the understanding of individual behavior. If gesture (or any other unit of bodily position) is considered to take place on the face or on the body of the individual, then the observer can match what he says with that gesture or its immediate context. This kind of study would have relevance for the learning of behavior (gestures) possibly associated with certain forms of language. In addition, it might lend insight into both correlational and causal behavior at deeper organismic levels, such as neurological, in the context of the emerging field of Brain and Behavior (Pribram 1969).

DYNAMIC STUDIES OF COMMUNICATION

In its essential 'opposition' to more atomistically cast studies, a dynamic approach might begin from the assumption that 'new' information is a fairly rare commodity in the real world, and as such might well constitute the basis for a study of 'residual', but not 'central' phenomena. A study which concentrates on the exchange of a really new or even a fresh bit of information is thus seen as an unlikely way of proceeding to find out about communication. This is one reason why linguistic approaches which consider grammatical structure as a central human function seem (to me) to have decreasing relevance. Another consideration of the dynamic study of communication concerns the fact

306

HARVEY B. SARLES

that the notion of the truly individual-existential self seems about as rare a part of one's being as the notion of new information. Most of our actual behavior takes place in 'well-known' contexts such as the family, neighborhood, business establishment, classroom, or bureaucracy. This seems to mean that the possibility of new information about peoples' habits, likes, health . . . is not much more surprising from day to day than the appearance of the sun. New information is thus truly important, if only by contrast. To better understand this effect, see Kuhn (1962). Again there is no question that messages which embody new information do occur — hopefully even in teaching context — but it seems less useful to concentrate on essentially 'unusual' processes as the very foundation of our approach to the study of communication. It is in this context that ideal language structure may be seen as only one of a number of such structures, some remaining not well known, which enter into communication. A sample question of possible interest to linguists, which arises in the present context, is of the following form: assume, for the moment, that children do not 'have' to learn language — it is neither innate nor an incumbent process. Grant further that young children communicate very effectively with those around them (and if they did not, they simply would not survive; Blauvelt 1954) through tactile processes, 'grunts', contextually relevant eye movements — i.e. the behavior of a healthy 6-month-old. Why, then, do they 'learn' language? Immediately we are forced, on the basis of these assumptions, to search within the organism. But we are also forced to search WITHOUT — among the other people in the infant's home who begin to 'demand' speech by about 11/2 years at the latest. Models for speaking may be less in the individual than in his mother's 'working image' of him.20 Remember that many infants do not speak — we label them retarded or autistic or pathological in some way. But they may simply have made life adjustments which are different from the rest of ours (Virchow 1962). Here we may begin to see language development as part and parcel of social development. And we might even begin to ask why aphasic children are that way, rather than jumping to indeterminate diagnoses such as 'brain damage' or other such imponderables which admit of no understanding and of no possible change. The consideration of dynamic views of communication also imposes the necessary consideration of time variables, even real-time variables. Whereas the ideal structures of language, physiology, individual, or society can be seen as existing essentially out-of-time, people communicate in fairly continuous fashions. In one sense, discrete units such as sentences, phonemes, or percepts exist all by themselves; yet it might be argued that such 'essences' only emerge from the reality of interactions 20 A social-biological model suggested by primate ethology studies (Altmann 1967) suggests that 'speech' allows social interaction to occur and continue at distances. Normal biological variation ensures that some normal infants (families) will be born without an imperative for speech. A useful distinction for the reader between modern biological and non-biological approaches may be characterized by the search for differences and variation for the first, and the methodological need to arrange and analyze data for the non-biologists.

FACIAL EXPRESSION AND BODY MOVEMENT

307

in which they occur. One doesn't just work out his rules for creating sentences; he talks to someone about something; and understands what's happening to a fairly remarkable degree. People are social beings in addition to being generators of disembodied sentences. Another question which might be phrased is how do people understand one another: in what sense is language intelligible in real time? It seems to me that this question might be interpreted in two ways: first as a non-question; e.g. the real question is how does communication 'fail' or why don't people communicate better? secondly, by noting what a remarkable thing it is that there are so few communicational confusions in an otherwise messy world. The latter is clearly the appealing one to students of communication dynamics. Examples of approaches to studies which have used time include those by Miller, Galanter, and Pribram (1960) and by Birdwhistell (1968a, 1968b), Scheflen (1964), and others. Most of them have tried to work with the notion that peoples' ideas about the future often (if not always) influence their present or ongoing behavior. On the level of the sentence (Sarles 1967) this might mean that, say, the first words of otherwise similar sentences contrast as they are embedded in differentially long sentences. Or that a phrase final word might reflect in its occurrence some features of the kind of phrase in which it occurred, and something about what's coming next. (This, as opposed to 'explaining away' such actual dynamic variation by attributing it to such inexplicables as 'allophonic' or 'free' variation.) Another idea pursued by Scheflen, Birdwhistell et al. is that many such plans operate much like structures in time. Although these might be better conceptualized as being 'generative' ultimately, they have been considered to operate much like a symphony, a baseball game, or a wedding in which the outline of the form is to be 'played out'. Scheflen and Birdwhistell have approached 'The Interview' in this way, searching for patterned repetition of movement on the part of patient and therapist. But such culturally defined ritual scenes might also be unusual in the same senses discussed earlier. Communicational processes go on many hours on many days when only a more highly leveled notion of 'plan' would seem appropriate. Indeed 'chrono-biologists' are only now demonstrating that we vary in many physiological traits across the day (circadian rhythms) as well as through the week, month, and year. From the present point of view of dynamic approaches, the emergent field of 'socio-chrono-biology' will have to consider the variation in behavior as related to bio-social cycles such as the symphony, or the class-room hour. (How is it, for example, that people in varying stages of their daily cycle, can come together for an hour to listen and attend to the same lecture without disruption? Or can they?) Finally, a dynamic approach might see itself in essential opposition to the universal (language or gesture) approaches which rely on the assumption of crosscultural or pan-human similarities to be raised to the level of causation 'because'

308

HARVEY B. SARLES

they define human uniqueness. It seems to me that this rests on such an increasingly tangled mat of assumptions of the nature of human nature that they can lend little understanding to gaining new insight about it.

SUMMARY

The study of facial expression and body movement is still in its infancy, especially in regard to possible linguistic relationships. Reasons for this general neglect of such obvious aspects of man's behavior are various, but the most important are traditions of thought and conceptualization of problems to which linguists are heir. Careful description, while necessary, will not itself solve the mysteries of expression and communication. Part of the difficulty is that our philosophical traditions of thinking about man are very individual-oriented, with a primary emphasis on adult, rational, male. It is no accident that these terms are equivalent in so many senses; but this FOCUS surely does not tell us all there is to know about behavior, linguistic or other. The face, or 'parts' of it, are in constant motion while speaking. The listener's face(s) and body is quieter, but not 'at rest'; and his FEEDBACK is an important part of the face-to-face interaction where most of our ordinary conversation and communication take place. Traditions of observation have, however, called attention only to the speaker as the transmitter of 'new' information, and not to the listener. Linguists and others have acted as if such information takes place without regard to audience and context, but also as if it occurs without regard to bodies. The sentence and the word indeed appear to be independent of one's body. And a grammatical approach to facial expression through 'emotions' can also treat expressions as independent of bodies, though this requires a more creative imagination. The 'factual' basis of what humans are about is not really very strong; the traditions are, on the other hand, perhaps overly seductive. We might all agree minimally that children turn out to be essentially like their parents — except that there is variety and change upon the basic model. One way of phrasing the outline of the enterprise is, 'How does this happen?' Instead of concentrating on pathology or operationalism to define normality, we might reopen the consideration of crossspecies where children also turn out to be pretty much like their parents. Instead of assigning varying percentages to an intrinsically dualist position of nature vs. nurture, we are presently able to begin to phrase testable questions which might transcend this peculiar traditional division. Further, observation of language should include what happens in the faces and in the bodies of the interactors. It is intriguing to note that bodily relationships also SHIFT when conversation topics shift from, say, the HERE AND NOW to a narrative. There is good reason to believe that all such behavior is structured and patterned, if only because ordinary communication is so effective. There are very few 'con-

FACIAL EXPRESSION AND BODY MOVEMENT

309

fusions' in the everyday world within the relatively constant cultural contexts of family and work. REFERENCES

1967. The structure of primate social communication. Social communication among primates, ed. by S. A. Altmann, 325-62. Chicago, University of Chicago Press. BENJAMIN, GAIL R. 1969. The non-linguistic content of speech. Unpublished M.A. Thesis. University of Minnesota, Department of Anthropology. BIRDWHISTELL, R A Y L . 1968a. Communication. International Encyclopedia of the Social Sciences 3.24-9. Crowell Collier and Macmillan, Inc. . 1968b. Kinesics. International Encyclopedia of the Social Sciences 8.37985. Crowell Collier and Macmillan, Inc. BLAUVELT, HELEN. 1 9 5 4 . Dynamics of the mother-newborn relationship in goats. Conference on group processes, ed. by R. Schaffner. New York, Josiah Macy, Jr. Foundation. BLOOMFIELD, LEONARD. 1 9 3 3 . Language. New York, Holt. CHOMSKY, NOAM. 1966. Cartesian linguistics. New York, Harper and Row. . 1968. Language and mind. New York, Harcourt, Brace and World. DARWIN, CHARLES. 1872. The expression of the emotions in man and animals. London, J. Murray. EIBL-EIBESFELDT, IRENAUS. 1 9 6 8 . Zur Ethologie des menschlichen Grussverhaltens. Zeitschrift für Tierpsychologie 25.727-44. EKMAN, PAUL, E . RICHARD SORENSON, and WALLACE V. FRIESEN. 1969. Pan-cultural elements in facial displays of emotion. Science 1 6 4 . 8 6 - 8 . ENLOW, DONALD H. 1 9 6 8 . The human face. New York, Harper and Row. FRANK, LAWRENCE. 1 9 5 7 . Tactile communication. Genetic Psychology Monographs 5 6 (Nov. 1 9 5 7 ) . 2 0 9 - 5 6 . GARDNER, BEATRICE T., and R . ALLEN GARDNER. 1 9 6 9 . Two-way communication with an infant chimpanzee. Behavior of non-human primates, ed. by A. Schrier and F. Stollnetz, Vol. 3. New York, Academic Press. GIBSON, JAMES J . 1 9 6 6 . The senses considered as perceptual systems. Boston, Houghton Mifflin Company. HJELMSLEV, LOUIS. 1961. Prolegomena to a theory of language, tr. by F. J. Whitfield. Madison, University of Wisconsin Press. K U H N , THOMAS, 1 9 6 2 . The structure of scientific revolutions. Chicago, University of Chicago Press. MILLER, G . A., E. H. GALANTER, and K. H. PRIBRAM. 1960. Plans and the structure of behavior. New York, Holt, Rinehart and Winston. PRIBRAM, K . H . , ed. 1 9 6 9 . Brain and behavior 1 ; Mood, states and mind. Selected readings. Baltimore, Penguin Books, ALTMANN, STUART A .

HARVEY B. SARLES

310

B. 1967. The study of intelligibility. Linguistics 34.55-68. . 1969. The study of language and communication across species. CAnthr 10 2-3.211-21. . 1970a. Toward the study of human ethology. Anthropology and the behavioral and health sciences, ed. by O. von Mering and L. Kasdan. Pittsburgh, University of Pittsburgh Press. . 1970b. An examination of the question-response system. Semiotica 3.79101. SAUSSURE, FERDINAND DE. 1959. Course in general linguistics, tr. by W. Baskin. New York, The Philosophical Library. SCHEFLEN, ALBERT E . 1 9 6 4 . The significance of posture in communication systems. Psychiatry 274.316-31. SMITH, W. JOHN. 1969. Messages of vertebrate communication. Science, July 11, 1969.145-50. SPITZ, R . A . , and K . M. WOLF. 1 9 4 6 . The smiling response: A contribution to the ontogenesis of social relations. Genetic Psychology Monographs 34.57. TRAGER, GEORGE L. 1958. Paralanguage: A first approximation. SIL 1 3 / 1 - 2 . 1 SARLES, HARVEY

12. VIRCHOW, RUDOLPH.

Books.

1962.

Disease, life and man,

37-8.

New York, Collier

PROXEMICS

O. M I C H A E L W A T S O N

INTRODUCTION Why do North Americans typically emerge from interactions with Arabs with the feeling that Arabs are 'pushy'? What is it about North American conversational behavior that conveys an impression of distance to an Arab?

Why do students

who occupy a particular seat in a classroom during the first meeting of the class feel irritated when they arrive for the second lecture and find someone else sitting there? What is it about a French sidewalk cafe that imparts a feeling of intimacy? Why do the waiting rooms of air, rail, and bus terminals in the United States emit an air of uninvolvement? Why does it feel more crowded in a given space with a given density of people when the temperature is high than when the temperature is lower? Why do the Japanese arrange furniture in the middle of the room and Westerners along its edges?

Why is there a difference between the layout of

French cities and English cities?

These questions, which would be considered

legitimate areas of interest for the researcher of proxemic behavior, give a feeling for the content of the newly developing field of proxemics. The common thread running through the questions above have to do with human activity and space, and proxemics

is the term coined by the anthropologist

Edward T . Hall to refer to the study of man's relationship to the spatial dimension of his environment — his spatial needs and perceptions, his structuring and use of space. Before innovating the term 'proxemics' in the early 1960s (Hall 1963a and b), Hall (1968:83) referred to his work as 'social space as bio-communication' and 'micro-space in interpersonal encounters', phrases whose meanings were known to only a few specialists. In order to bring his research to the attention of a wider audience, and to distinguish it from that of scientists whose interests and activities lay in the field of outer space, Hall chose the term 'proxemics' to refer to the relationship between human behavior and space.

Several other terms

were

considered, e . g . ' . . . human topology, chaology, the study of empty space, oriology, the study of boundaries, chorology, the study of organized space' (Hall 1968:83), but the term 'proxemics was c h o s e n . . . because it suggests the subject matter to the reader' (Hall 1963b: 1022). A more specific idea of the scope of proxemics can be gained by a study of the ways in which Hall defines proxemics:

312

O. MICHAEL WATSON

1. . . the study of how man unconsciously structures microspace — the distance between men in conduct of daily transactions, the organization of space in his houses and buildings, and ultimately the layout of his towns' (1963b: 1003). 2. . . the study of ways in which man gains knowledge of the content of other men's minds through judgments of behavior patterns associated with varying degrees of proximity to them' (1964b:41). 3. . . the interrelated observations and theories of man's use of space as a specialized elaboration of culture' (1966:1). 4. ' . . . the study of man's perception and use of space. ..' (1968:83). 5. ' . . . [proxemics] deals primarily with out-of-awareness distance-setting' (1968: 83). Looking at definition 1), one sees the enormous breadth present in Hall's description of proxemics, from the manipulation of small amounts of space in face-to-face interactions, through the use of space in architecture, to urban space. Definition 2) makes explicit the communicative aspects of proxemic behavior on the interpersonal level, and definition 3) emphasizes the investigation of proxemic behavior as an aspect of culture, and thus subject to cross-cultural variation. Underlying definition 4) are notions concerning levels of proxemic behavior, 'perception' carrying connotations of man as a biological organism and 'use' carrying implications of man as a cultural organism, 'use' being the culturally influenced cognitive structuring of the raw materials gathered by 'perception'. Finally, 5) is a statement concerning levels of awareness in the maintenance of proxemic norms. One can see that proxemics, as Hall envisions it, covers a wide range of human spatial activity (from microspace to macrospace) and can be approached at different levels (biological and cultural). It is the purpose of this chapter to survey the scope of proxemic behavior, but — as befits the nature of this volume and serves ar a complement to the chapters by Professors Crystal ("Paralinguistics"), Sarles ("Facial expression and body movements"), and Stokoe ("Classification of sign languages") — to concentrate on proxemic behavior as a form of nonverbal communication on the interpersonal level: '.. . how man structures microspace, how he relates physically to other persons with whom he is interacting, and what is communicated by these physical relationships' (Watson and Graves 1966:971). THE SCOPE OF PROXEMIC BEHAVIOR

In an attempt to order the complexity which characterizes the scope of proxemics, it is useful to look first at the different organizational levels on which proxemic behavior, as well as other aspects of human behavior, can be analyzed. Stover (1970) has provided some very descriptive terms which apply to three levels, or orders, of organization: 'the ethological1 command' (responses human beings, due to the 1

Ethology, of course, is a term applied to a discipline, a field of investigation. Strictly speaking, an 'ethological command' would be a command given by an ethologist. It was clear from Professor Stover's presentation, however, that he was referring to the biological basis of behavior.

PROXEMICS

313

nature of their genetic makeup, are 'programmed'2 to make to certain stimuli), 'the universal situation' (anatomical and physiological equipment with which all men meet the needs of their lives), and 'the cultural expression' (the appropriate manifestation of behavior in the context of a learned and shared system of behavior). As a simple example of analysis of a piece of human behavior on these three levels take, let's say, walking. Bipedal locomotion is a pan-human behavior pattern with which all normal newborn humans are equipped (Eibl-Eibesfeldt 1970:401). Thus, walking may be viewed as an 'inborn skill' (1970: viii), an 'ethological command', which doesn't become fully functional until the infant is several months old. The 'universal situation', as far as walking is concerned, refers to the fact that all human beings have the same skeletal structure and musculature with which walking is facilitated, and the message 'walk!' is transmitted by the same neurophysiological mechanisms to all men. But the 'cultural expression' of walking varies somewhat from culture to culture, and La Barre (1956: 557-8) relates different ways of walking among different groups of people around the world.8 Let us turn our attention back to proxemic behavior from the viewpoint of these three levels of organization. The terms which Hall (1966:95-122) uses to refer to these three levels in his organizing model of proxemics are infracultural, precultural, and microcultural. Of the infracultural level Hall (1959:44) writes: ' . . . culture is bio-basic — rooted in biological activities. Infra-culture is the term which can be given to behavior that preceded culture but later became elaborated by man into culture as we know it today.' Thus, 'the m/racultural (level) is behavioral and rooted in man's biological past' (Hall 1966:95). Commenting on the use of the term 'infracultural' in an article by Hall (1968), Birdwhistell (1968:96) points out some inconsistencies in that Hall uses the term to convey both a diachronic (phylogenetic) sense and a synchronic (biophysiological) sense. Hall's (1966:95) reference to a precultural level, which is ' . . . physiological and very much in the present', and is the ' . . . physiological base shared by all human beings, to which culture gives structure and meaning', leaves little doubt that he has separated the diachronic (infracultural) from the synchronic (precultural). The third of Hall's analytic levels, the microcultural, deals with man's structuring and use of space as it modified by culture. Further, proxemic behavior has three aspects at the microcultural level: fixed-feature, semifixed-feature, and informal space. The fixed-feature aspect of proxemic behavior on the microcultural level deals ® I use the term 'programmed' instead of 'instinct' or 'innate' out of deference to my colleagues in anthropology who seem to have an innate releasing mechanism which responds with hostility at the mention of these words. ' I have had the distinct pleasure of seeing Professor La Barre do his impressions of walking behavior from various parts of the world.

314

O. MICHAEL WATSON

with those features of the environment which are materially fixed. The ways in which buildings are internally structured, as well as the ways in which buildings are laid out in towns and cities, are expressions of this aspect (Hall 1966:97). Semifixed-feature space, the second aspect, has to do with the ways in which people use potentially movable features of their environment: ' . . . furniture arrangement, screens, movable partitions and the like as factors in human interactions . . . ' (Hall 1963a:436). The third aspect of the microcultural level is informal (Hall 1966:105), or dynamic (Hall 1963a:429-30), space. This aspect is concerned with the ways in which humans use small amounts of space in their interactions with others. 'Dynamic' implies the active nature of the ways in which man uses microspace, and the use of the term 'informal' derives from Hall's (1959:63-91) ideas about the ways in which things are learned, the levels of awareness at which they are maintained, and the affect attached to them. The informal aspect of behavior is learned using an imitative model, 'in many cases without the knowledge that [the patterns of behavior] are being learned at all or that there are patterns of rules governing them' (Hall 1959:70). Informal awareness (paradoxically) 'describes a situation in which much of what goes on exists almost entirely out-of-awareness' (Hall 1959:73). As far as affect is concerned, there is none attached to informal behavior as long as norms are not violated, but when a violation occurs uneasiness and anxiety often follow (Hall 1959:76). The infracultural, precultural, and microcultural organizational levels provide one dimension from which to view proxemic behavior. Another dimension which would serve as an approach to proxemic behavior is the quantity of space involved in patterns of proxemic behavior. One can guage the wide range of space involved in proxemic behavior from the definitions of proxemics cited at the beginning of this chapter. At one pole stands the small amounts of space involved in face-toface encounters and at the other pole the space filled by urban areas. Hall (1966 : 95-122), in his organizing model, makes distinctions between amounts of space involved in proxemic behavior at only the microcultural level (the fixed-feature, semifixed-feature, and informal aspects of this level). It seems reasonable to assume, at least for organizational purposes, that the same distinctions can be made at the infracultural and precultural levels as well. I suggest that it is useful to divide these quantities of space into three categories, microspace, mesospace, and macrospace. The microspatial category includes part of what Sommer (1966: 60) calls man's 'proximate environment' and which he describes as 'the area immediately surrounding the individual (which) has been called personal space and has been likened to both the shell of a snail and a soap bubble': On the mesospatial level can be included the remainder of the proximate environment - \ . . everything that is physically present to the individual at a given moment' (Sommer 1966 : 60), not including the space which immediately surrounds the individual (personal space). Examples which serve to illustrate the

315

PROXEMICS

quantities of space with which the mesospatial category is concerned are architectural arrangements and movable features within these larger architectural units. Macrospace refers to the placement of these architectural units to form settlements and cities, as well as the ways in which man relates to the features of his broader physical environment.

A PARADIGM FOR THE CLASSIFICATION OF PROXEMIC BEHAVIOR

This section is concerned with the construction of a paradigm which might prove to be of some use in the organization and classification of proxemic behavior. For the vertical dimension the levels of organization (infracultural, precultural, and microcultural) discussed above will be used. The horizontal dimension will consist of the spatial quantity categories microspatial, mesospatial, and macrospatial. Microspatial 1.

Possession and Defense of Personal Space

1

•3 «

Mesospatial 2.

•a

••a eg

4. Perception of Personal Space

1^ u u U) PL,

7.

•a b* "3 5 o o

s

5.

8. Culturally Specific Systems of Proxemic Behavior in Interpersonal Relations

Possession and Defense of Movable Features; Stress from Crowding

Macrospatial 3.

Group Territory; Marking and Maintenance of Boundaries and Features Within the Territory

6. Perception of Internal Structuring of Space (Architecture) Culturally Specific Systems of Architectural Structuring and Use

Perception of Total Environment

9.

Culturally Specific Systems of Structuring and Use of Cities and Towns; Attitudes Toward Total Environment

Fig. 1. A Paradigm for the Classification of Proxemic Behavior

The infracultural level is 'behavioral and rooted in man's biological past' (Hall 1966 : 95), and is concerned with the biological necessity for man's structuring and use of space. Behavior at this level is manifested in territoriality and the reactions to the stress of overcrowding. Although Burt (1943 : 346) asserted that territoriality reaches its highest development in humans, Carpenter (1958 : 245)

316

O. MICHAEL WATSON

has pointed out that very little systematic research has been undertaken in human territoriality. Aside from anecdotal references to beggars' beats, favorite chairs, and salesmen's territories, the support for the biological bases of human spatial needs has been scanty until the recent interest in human behavior by ethologists. Cell 1 in Figure 1 (infracultural/microspatial) deals with the biological aspects of individual territoriality, the possession and defense of personal space.4 Research which suggests behavior falling into this cell includes Horowitz (1963), who, in an article which dealt with painting therapy for schizophrenics, noted that patients beginning the therapy would establish a territory and prevent intrusion by the therapist. But when rapport was established between therapist and patient, the territory was shared. Horowitz, Duff, and Stratton (1964) report that in an experiment using schizophrenic and nonschizophrenic subjects, subjects tended to display regular distances when approaching persons and objects, and that schizophrenics displayed greater individual distances than nonschizophrenics. This report receives support from Sommer (1959), who found that schizophrenics maintained greater individual distances than is considered normal. Also working with schizophrenics, Esser et al. (1964) observed territorial behavior among patients in a mental hospital ward. Cell 2 (infracultural/mesospatial) is concerned with behavior which includes the possession and defense of features within the architectural environment (furniture, etc.) and the biological limits of tolerance when people are crowded into spaces which are nonexpandable (e.g. in confinement). Sommer (1966) writes about how people lay claim and defend a place in a library against intruders, and Altman and Haythorn (1967), experimenting with volunteers, described how two men isolated in a 12' x 12' room came to display territorial behavior in their preference for items of furniture within the room. Lipman (1968), in a study involving residents of homes for the elderly in Great Britain, reports that the patients had favorite chairs which they considered their personal property and defended against encroachment. Hall (1966:161-2) reports on one of the very few studies of the consequences of stress from crowding in dwelling units. The study was conducted in Paris by Paul Chombard de Lauwe and linked social and psychological pathologies to crowding. How little we know about human spatial needs is implicit in Sommer's (1966 : 63-4) description of the confusion in individual spatial requirements in U. S. Civil Defense and military organization publications discussing shelters. Behavior falling into cell 3 (infracultural/macrospatial) is concerned with group territory and marking and maintenance of territorial boundaries. Gangs of youths in urban areas have frequently been cited as examples of group territorial behavior in humans. Block and Niederhoffer (1958 : 168), for example, point out that a gang defends its territory by means of intergang fights, or 'rumbles'. 4

Although most of these studies were not conducted with the aim of getting at the biological roots of territorial behavior, it is suggested that this behavior might belong on the intracultural level.

PROXEMICS

317

Davis (1962: 319) maintains that human gangs display typical primate behavior in their territoriality, and Kanno (1970) has shown striking evidence of marking of a gang's territorial boundaries. Lynch (1960) found that residents of three cities in the United States structured their cities in ways which recall Hediger's (1955:20) reports of the ways in which some animals structure their group territories. On a somewhat more ambitious scale, Ardrey (1966 : 238-9) contrasts the importance of territorial defense in nations which have well-integrated territorial units (as exemplified by the Battle of Britain and the Finnish thwarting of a Russian invasion in the 'Winter War') and nations which do not (as exemplified by the colllapse of France in World War II, France being a country composed of several noyaux — smaller territorial units — within the country). The precultural level in the organization of space refers to 'the physiological base shared by all human beings...' (Hall 1966 : 95), and is concerned with the ways in which man uses his senses in the perception of space. Hall (1966:39-60) divides the perception of space into two categories as functions of man's sensory apparatus: the immediate receptors (the skin and the muscles) and the distance receptors (eyes, ears, and nose).5 In terms of the paradigm presented in Figure 1, cell 4 (precultural/microspatial) deals with all the senses, for man uses all his sense modalities in the perception and structuring of small amounts of space. For the middle range of the spatial quantity dimension (cell 5), it is clear that the immediate receptors (and usually olfaction) are not involved, and in the perception of macrospace (cell 6) the eyes become the only reliable source for gathering information. Tactile communication — using the immediate receptors as modalities — is basic in man: 'Tactual sensitivity is probably the most primitive sensory process . . . ' (Frank 1960:5). Tactile modes of communication are also the most elemental, according to Frank (1957:201-2), who points out that an unborn mammal receives tactile messages by feeling the mother's heartbeat through the amniotic fluid. A child's initial orientation to space comes through exploring its environment by touch (Frank 1957:202-6, 1960 :7-9). Hall (1966:52-7) writes that one of the skin's important functions is the perception of thermal space, and that the skin has a high capacity both to transmit and receive information concerning emotional states. The blind, for example, are particularly skilled at navigation by interpreting thermal cues in the form of radiated heat, and a high skin temperature enhances olfaction in that skin odors can be smelled at a greater distance than under conditions of lower skin temperature. Olfaction also plays a role in the perception and structuring of microspace, although it is certainly not as effective in primates as it is in other mammals. Of interest here is Hall's (1966:36) point that perhaps one of the consequences of the suppression of olfaction in man gives him a greater ability to withstand crowding. !

A further discussion of the senses is included below.

318

O. MICHAEL WATSON

If humans could smell as well as some other animals they would be driven mad by the psychotic and made anxious by the smell of people with a high level of anxiety. Although vision is more important as a gatherer of information in man, Hall (1966:36) states that smell is more 'deeply emotional and sensually satisfying . . . ' Sight was the last and most specialized sensory adaptation in primates, stereoscopic vision being an important concomitant to arboreal living and was developed at the expense of olfaction (Montagu and Brace 1965: 111). In contrasting visual space and auditory space, Hall ( 1 9 6 6 : 4 0 - 3 ) points out that the eyes can detect a visual barrier at a great distance, but a sound barrier a quarter of a mile away is hardly perceptible to the ears. The chief characteristic of visual space is depth, while 'the essential feature of sound... is not its location but that it be, that it fills space' (Carpenter and McLuhan 1960:65). While visual space has a point of focus and a background, acoustic space has no boundaries and the ear favors no direction (67). The final level of organization to be considered in the paradigm is the microcultural, which deals with the structuring and use of space as it is modified by culture (Hall 1966:94-106). The proxemic behavior which is included in cell 7 of Figure 1 (microcultural/microspatial) has to do with the manipulation of small amounts of space in interpersonal relations. This chapter is primarily concerned with this category of proxemic behavior and discussion of it will be initiated in the next section. Microcultural/mesospatial (cell 8) proxemic behavior includes what Hall (1963a: 436) has called 'semifixed-feature space' and part of what Hall (1966:97) refers to as 'fixed-feature space'. Examples of research which are concerned with the proxemic behavior included in this category are descriptions of cross-cultural differences in furniture arrangements and the use of these movable features as props in interaction. Chinese have reported, for instance, that they felt like they were on trial when they sat at the opposite side of a desk during an interview (Hall 1963b: 1006), and Latin American businessmen have been known to climb across desks behind which their North American counterparts barricaded themselves (Hall 1955). Foster (1969: 3^4) reports that a Mexican Ministry of Health engineer designed and built some washtub units for a village in Vera Cruz in such a way as to stifle interaction (which was an unexpected and unwelcome result) among the women who used the tubs. Osmond ( 1 9 5 9 : 7 - 9 ) coined the terms 'sociopetaP and 'sociofugal' to refer to spatial arrangements which, respectively, brought people together in interaction or kept them apart. Sommer (1969:110-19) reports his experiments on the linkage between placement of chairs in a classroom and student participation. Also of concern in cell 8 is the internal structuring of architectural units, and cross-cultural differences in this area of proxemic behavior. In this connection we have reports by Hall (1963a: 434) of a hospital in China, built by American missionaries, which was torn down and rebuilt to meet Chinese needs when the

PROXEMICS

319

Chinese Communists arrived and expelled the missionaries. Aries (1962:390-8) points out that the familiar structuring of Western residential architecture into different rooms for different functions (bedrooms, bathrooms, etc.) is an innovation dating from the eighteenth century, before which time one room served many functions. McLuhan (1964:125) writes that before man became agricultural he lived in round houses, and after he became 'sedentary and specialized' he lived in square houses. Sommer (1969:77-97) cites examples of the nontherapeutic aspects of architectural arrangements within mental hospitals. Cell 9, the microcultural/macrospatial category, deals with the crosscultural differences in the structuring of large amounts of space. Examples in this area include the contrast in the layout of street systems in Japanese, North American, and French cultures. The Japanese, in their street system, give names not to streets, but to intersections (Hall 1959:159, 1966:99), while in Anglo-American culture lines, i.e. the streets themselves, are named (Hall 1959:159). Further, the grid system, whereby streets are laid out in a parallel and perpendicular fashion, is used in North America (Hall 1966:137). In France, however, the 'radiating star' system of streets and highways is used, where streets radiate from a center and are intersected by rings of streets concentric to the center (Hall 1966:137). Tyrwhitt (1960:90-5) points out that in the Indian city of Fatehpur Sikiri, there seems to be no fixed center and that it is hard to find the key to the city's composition in terms of Western linear perspective. The viewer of the city is presented with balanced panoramas from several vantage points, never feeling removed from the center of the city. From this attempt at the construction of a paradigm for the classification of proxemic behavior and the brief survey of examples which fit the categories in the paradigm, one gets an idea of the scope of proxemic behavior as Hall has defined it. Leaving behind those areas of proxemics discussed above, which are beyond the scope of this chapter, I will concentrate on the microcultural/microspatial aspects of proxemics — proxemic behavior on the interpersonal level.

PROXEMIC BEHAVIOR ON THE INTERPERSONAL LEVEL

This level of proxemic behavior has been described as 'how man influences his communications with others by varying the spatial features of the situation' (Hall 1963a: 430); how 'man actively uses the fixed and semifixed features given to him' (Hall 1963a: 437); and 'how man structures microspace, how he relates physically to other persons with whom he is interacting, and what is communicated by these physical relationships' (Watson and Graves 1966:971). It is clear from these descriptions that the behavior to be discussed is concerned with how man varies the content of an interaction by structuring and using small amounts of space in his orientation to other men.

320

O. MICHAEL WATSON

Hall, in his influential book The silent language, discussed the ways in which space can be used as a communicative device. He (1959:160) writes that 'the flow and shift of distance between people as they interact with each other is part and parcel of the communication process'. All people, Hall (1959:160-1) continues, have a notion of what the physical distance between interactants should be, given the definition of the interaction. People react when these rules of distance are violated by backing up when one is approached too closely or by coming closer when the distance between interactants is too great. Hall (1959:162-4) also wrote of the relationship between interaction distance and voice level, and the ways in which they were linked to the subject of a conversation. He listed eight distances, from 'very close' to 'stretching the limits of distance', and the voice levels and suitable topics of conversation associated with these distances. The distances and concomitant voice levels were, by the way, for North Americans. In later writings Hall (1964a: 45-53, 1966:109-22) elaborates on these distances and expands the areas of behavior associated with them to include not only voice level, but also visual, olfactory, and tactile cues. In these writings Hall refined the distances used earlier into four categories, each category having a close and a far phase. The first of these distances, intimate distance (Hall 1966: 110-12), is characterized by stepped-up sensory inputs, involving all the senses, and a high probability of physical involvement. The close phase of intimate distance (physical contact to 6 inches) is used for fighting and lovemaking, and the far phase (6 to 18 inches) is usually reserved for interaction with family and close friends. Personal distance (Hall 1966:112—114), which ranges from 18 inches to 30 inches at the close phase and 30 to 48 inches at the far phase, is also used for interactions with intimates, but for less intimate topics and with less sensory involvement than at intimate distance. Social distance6 (Hall 1966:114116) is regularly used for conversations with friends and business associates at the close phase (4 to 7 feet) and more formal transactions at the far phase (7 to 12 feet). Public distance (Hall 1966:116-20) is used to address an informal gathering at the close phase (12 to 25 feet) and to address a formal gathering at the far phase (more than 25 feet). This distance is usually maintained between the public and an important official. In an article Hall (1968: 92-3) presents diagrammatically how these four distance categories are linked to various sensory inputs: kinesthetic, olfactory, thermal, visual, and oral/aural. I again point out that these distances are discussed in terms of their use in North American and Northern European culture.

8

Bogardus (1933, 1959) uses the term 'social distance' to refer to a measure of the degree of acceptance given to certain persons and social groups, and is typically used to measure conflict which exists between persons or groups. The term is not used in this sense in the discussion above.

PROXEMICS

321

In summary, then, Hall has developed a set of distances used for various interactional situations and the use of the senses associated with these distances. But the basis of proxemic behavior on the interpersonal level lies not in various distances used between interactants, but in the ways these distances are perceived and maintained by use of various sensory inputs: 'Though distance is geometrically defined and measured by the conventional yardsticks of the physical sciences, the effects of distance in biology depend initially on the sensory perceptual properties of the organism' (Chappie 1970: 151). As guiding principles in his investigations of the ways in which the senses are used to structure space, Hall (1963b: 1005) asked the questions: 'By what means other than visual do people make spatial distinctions? How do they maintain such uniform distances from each other?'. By pursuing these questions in observations of interactions and interviews of the interactants, Hall (1968:94) has concluded that 'interpersonal distance is a constellation of sensory inputs that is coded in a particular way' (Hall emphasizes). Hall, thus, describes proxemic behavior as a system of interrelated variables which man uses to organize microspace. Further, he (1963b) has isolated eight variables, or 'dimensions', which provide spatial cues, and has operationally defined them. The eight factors are: 1) Postural-sex identifiers. Hall (1963b: 1007-8) uses this category to indicate the sex of the interactants and whether they are standing, sitting, squatting, or prone. 2) Sociofugal-sociopetal axis. This variable concerns the relation of the axis of one person's shoulder to that of another. Hall (1963b: 1008—9) uses nine orientations, from face-to-face to back-to-back. 3) Kinesthetic factors. This variable has to do with the 'potential to strike, hold, caress, or groom' (Hall 1963b: 1009) and has eight categories, from within body contact distance to just outside reaching distance. 4) Touch code. This variable provides for the amount of contact which takes place during an interaction, from holding and caressing to no contact. 5) Visual code. The coding of this variable provides an index of the amount of visual contact during an interaction. There are four categories, from looking a person directly in the eye to gazing off into space. 6) Thermal code. This variable provides a measure for the detection of thermal radiation from a person's body, and is coded digitally, i.e. whether or not conducted or radiated heat is detected (Hall 1963b: 1014). 7) Olfaction code. The detection of differentiated or undifferentiated body and breath odors are the concern of this variable. 8) Voice loudness. A measure is provided by this variable for the level of a person's voice during an interaction. In the section which follows, each variable in the system of proxemic behavior will be discussed separately, followed by a consideration of the ways in which the variables function together as a system.

322

O. MICHAEL WATSON

THE VARIABLES IN THE SYSTEM OF PROXEMIC BEHAVIOR

Postural-Sex

Identifiers

Although Hall (1963b) treats this variable as belonging in the proxemic syndrome, it is my contention that it should not be considered as a part of the system. This is not to say that an interactant will not respond proxemically to the sex and posture of another person, but I maintain that these responses will be made, in this case, to variables outside the system of proxemic. The posture and sex of interactants is more usefully treated as a part of the context in which proxemic behavior takes place, and these and other situational factors will be treated in a later part of this chapter.

Sociofugal-Sociopetal

Axis

As was mentioned above, Osmond (1959) coined the terms 'sociofugal' and 'sociopetal' to refer to spatial arrangements which brought people together or kept them apart, and various researchers have investigated the orientation of bodies toward one another. Winick and Holt (1961), for example, found information in the ways in which group therapy members positioned themselves in relation to other group members during therapy sessions. In an article written by Sommer (1959), he tells how he noticed that in the cafeteria of a mental hospital, which was used by patients and employees alike, people who sat across the corner of a table from each other were more likely to engage in a conversation than were those in other positions at the table. In the experiment wherein subjects were asked to go to tables and discuss a topic with each other, Sommer discovered that employees, nurses, and nonschizophrenic patients most frequently sat across the corner from each other, while schizophrenics were more likely to sit at distant places from each other or across the table from each other. Further, Sommer found that when a decoy was placed at a table and subjects were told to go discuss a topic individually, schizophrenics were more likely to sit alongside the decoy, and 'normals' were more likely to sit across the corner of the table from him. In another experiment, Sommer (1961) used employees of a mental hospital as subjects. Subjects were divided into groups and each group chose a leader. The groups were then told to sit at a table, and the group leader most often chose the end positions at the table. When the leader did not occupy the end position, the other members of the group sat opposite, rather than alongside the leader. In still another study, Sommer (1962:115) concluded that 'people prefer to sit across from one another rather than side-by-side. We have now learned two exceptions to this — when the distance is too far for comfortable conversation, and secondly, when the distance across exceeds the distance side-by-side.'

PROXEMICS

323

Steinzor (1950) concluded that the location a person occupies in a seated group plays an important role in interaction, and Hare and Bales (1963) confirmed this in a task oriented small group experiment. Hearn (1957) found that in a small group with a leader who sat passively by, group members tended to interact with people opposite them and in their line of vision rather than those next to them, but when the leader became active in coaching and criticism just the opposite effect was observed. One can see from these experiments that components of the situation play a part in the orientation of one person to another. Hall (1963b: 1009) writes that 'which components of the sociofugal-sociopetal axis are favored and for what transactions, is largely culturally determined. These components are also linked with the social setting and the age, status, and sex of the [parties involved in the interaction].' Kinesthetic Factors Distance between individuals has already been mentioned in connection with Hall's classification system for interaction distances in humans. This variable has been called 'kinesthetic' by Hall (1963b : 1009) because the concern here is with distance between interactants as determined by the potential to 'strike, hold, caress, or groom' one another. Hall (1955, 1959, 1960, 1961, 1963a and b, 1964b, 1966, 1968; Hall and Whyte 1960) has noted in his publications the cultural attitudes toward different distances in given situations. In an experiment which dealt with the placement of cutout figures, Kuethe (1962a) noted that a high commonality existed in placing a child's figure closer to a woman's figure, a dog's figure closer to the figure of a man, and to place human rather than nonhuman figures together. Kuethe (1962b) also found that the tendency to place man and woman figures together was greater than to place two woman figures together, and figures of men facing each other were placed closer together than those facing away from each other. Little (1965), using similar techniques, had subjects place figures on a grid after the experimenters had supplied information about the degree of acquaintance and the setting. It was found that the closer the acquaintance, the closer the figures were placed together; similarly, the distance between figures decreased progressively in interactions involving the office, the home, and the street corner. In the same experiment, Little had subjects place live 'actresses' after describing the conditions under which the interaction was taking place, with similar results as with the cutout figures. The obvious conclusion was that interaction distances are strongly influenced by degree of acquaintance and the setting in which the dyad interacts. Felipe (1966) observed that, in a small group experiment, co-operating individuals sat significantly closer together than did those individuals who were competing or coacting, and concluded that distance is a variable which is important in interaction. Finally,

324

O. MICHAEL WATSON

Feshback and Feshback (1963: 499), in a study of fear in children, parenthetically report that telling ghost stories to the children decreased the circle in which they were sitting from 11 to 3 feet. These studies, even the ones using cutout figures, suggest that distance between interactants is varied according to various factors in the context of the interacting.

Touching The tactile experience of a person growing up in Anglo-American culture is divided into three phases, according to Frank (1957 : 202-6, 1960: 7-9). The first phase begins with birth and continues through a child's finding contact surrogates for its mother by handling blankets, toys, and its own body. The second phase begins around five or six years old for boys and somewhat later for girls, when contact with the child's mother is less frequent. Around puberty the third phase begins with the seeking to give and receive contact. Boys walk with their arms around the shoulders of other boys, and girls with their arms around the waists of other girls. Then heterosexual contact is sought and is manifested by kissing, caressing, and sexual intercourse. But the amount and kind of touching which occurs between people in interaction is culturally defined (Frank 1957:207, 1960 : 10; Hall 1963b: 1011, 1966 : 147). Further, Frank (1960 : 11) asserts that much of a culture's social relations are learned and maintained through touch codes: hand shaking, hand holding, kissing, etc. Although the study of the amount and kind of touching which takes place during an interaction would seem to be an important contribution to the understanding of the total communication process, no model for the analysis of touching behavior has been constructed for any culturally specific system (Kauffman 1969 : 1). Also of importance is the study of the context in which touching takes place (Hall 1968 : 95). Visual Code It has been estimated that the eyes are about a thousand times more effective than the ears in picking up information. Using acoustic cues a blind person can gather information from the center of a 100-foot circle, but with sight he could see the stars (Hall 1966 : 61). Murphy (1964 : 1265), in an article concerning the Tuareg of North Africa, notes the importance of using the eyes to gather information during an interaction: 'The Tuareg is not a mouth-watcher but rather an eye-watcher and . . . during interaction his eyes are fixed by the steady stare of his respondent. On one occasion, I countered this by wearing dark glasses, but my Tuareg friends retaliated by the same technique and succeeded in totally effacing themselves.'

PROXEMICS

325

Argyle and Dean (1965) assert that of the many functions of eye contact in dyadic interaction one of the most important is the feedback provided by the gathering of information concerning the feelings of the other person. Hutt and Oimsted (1966 : 346) affirm the belief that eye contact is important in interaction and report that gaze avoidance is 'a persistent and characteristic feature in children with the syndrome of 'early infantile autism'. Further, Hutt and Ounsted (1966 : 355) feel that adults normally avoid the gaze of others only when in a state of high arousal (e.g. embarrassment or fear). Chance (1962 : 84) feels that gaze aversion serves to temporarily cut off sensory receptors from a potential conflictprovoking situation. Kendon (1965), in a study of gaze and eye contact, found that during an interaction the listener looks at the speaker for long periods of time, while the speaker tends to look away part of the time. I hasten to point out that the studies mentioned above used North American and English subjects. Cross-cultural evidence indicates that eye behavior varies from culture to culture. Navajos, for example are taught not to gaze directly at another person during a conversation, and Greeks have been noted to use their eyes more in interaction than Anglo-Americans (Hall 1963b: 1012). Crosscultural data also confound statements that eye contact serves to assert dominance. Arab males, for instance, have been found to employ mutual direct eye contact during most of an interaction (Watson and Graves 1966, Watson 1970), compared to Kendon's (1965) British subjects' 30 to 60 per cent rate of eye contact during an interaction. Are we to assume that Arabs are more 'dominant' than Britishers, or that Arabs are vying for dominance in their interactions while the British are not? Obviously much more work needs to be done on the comparison of culturally specific patterns of eye behavior in communication.

Thermal Factors Not much is known about the importance of conducted and radiated body heat in the structuring of space in interpersonal relations, but anecdotal evidence suggests that it does play a role in interaction (Hall 1963b: 1014). An indication that at least some thermal cues in at least one culture are taken into consideration is evident in a quote from an English book of etiquette: 'When a gentleman offers a lady his chair he should engage her in conversation for a few moments, thus giving the seat time to cool' (Playboy Magazine 1967 : 975).

Olfaction Almost 2000 years ago, Ovid recommended: 'See that your teeth are clean . . ., Let your breath be sweet, and your body free from rank odors . . . ' (1960 : 121). An

326

O. MICHAEL WATSON

evening's monitoring of television in the United States would convince one that success in business, love, and interpersonal relations are dependent on using the proper perfume, cologne, mouthwash, toothpaste, and deodorant. There is little doubt that olfaction plays a communicative role in interpersonal relations. Physicians rely heavily on olfaction in the diagnosis of disease: 'Doctors must be able to differentiate the stink of uremia, diabetes, alcoholism, lung abscess, and other serious illnesses. Even psychiatric diagnosis may be aided by a good nose, since the sweat of chronically schizophrenic patients carries a peculiar odor that rats and perfume experts are able to detect' (Ostwald 1964:16). Olfactory (and thermal) factors are functions of the immediate receptors, so these factors cannot come into play except at very close distances between individuals. If we knew the cultural framework of the ways in which olfaction and detection of thermal radiation were used in interaction, these factors could be inferred from the distance at which the interaction takes place. Unfortunately we have almost no culturally specific information concerning these variables.

Voice Loudness 'The loudness of the voice is modified to conform to culturally prescribed norms for (a) distance, (b) relationship between the parties involved, and (c) the situation or subject being discussed' (Hall 1963b : 1016). Voice loudness is one way to maintain spacing between individuals, but use of voice loudness as a spacing mechanism varies from culture to culture. The level of the American voice is 'suspiciously' low to an Arab, and Arabs seem 'loud-mouthed' by American standards (Hall 1963b : 1005). Further, Arabs have mentioned explicitly in interviews that raising the level of the voice is one way to span distance (Watson 1970 : 105). There are apparently stylistic differences in voice loudness also. A Puerto Rican gave an indication of the differences in Latin American and North American voice level styles when he commented on the loudness of fraternity parties on college campuses in the United States: 'This is not personal involvement, but just a lot of people screaming' (Watson 1970 : 107). Voice loudness is, then, another variable in the proxemic system about which very little is known in culturally specific contexts.

THE SYSTEM OF PROXEMIC BEHAVIOR

In the section above each proxemic variable was discussed separately and evidence was given to support the notion that each of these variables play a communicative role in interaction. But the study of proxemic behavior, strictly speaking, is concerned with these variables as they interrelate to form a system of

PROXEMICS

327

behavior. Thus, the ways in which humans perceive and structure microspace is a 'synthesis of many sensory inputs: visual, auditory, kinesthetic, olfactory, and thermal' (Hall 1966 : 172). Chappie (1970:152) refers to this system of variables as being 'somewhat overcomplicated by a concern with the number of senses — and sensory inputs — involved which, except in closeness, reduce to two: sight and hearing'. So the question is raised: why consider proxemic behavior as a system of interrelated variables? I wish I had massive empirical evidence to inject at this point, to lend support to the notion of a system of variables, but the fact is that no such empirical evidence exists. There does exist, however, enough anecdotal evidence to justify the pursuit of knowledge concerning the interrelationship of the proxemic variables. Hall's (1963b: 1005) research with foreigners studying in the United States indicates that Arabs, when interacting with Anglo-Americans, were made uncomfortable by the fact that Americans were reluctant to interact within the zone of olfaction. Further, 'Arabs also experienced alienation traceable to a "suspiciously" low level of the voice, the directing of the breath away from the face, and a much reduced visual contact' (Hall 1963b: 1005). The Americans, on the other hand, were made nervous by 'the Arab look, touch, voice level, the warm moisture of his breath, the penetrating stare of his e y e s . . . ' (Hall 1963b : 1005). The Arab behavior was considered by the Americans to be inappropriately intimate, and such behavior is 'only permissible on a non-public basis with a member of the opposite sex' (Hall 1963b : 1006). Sommer (1969 :147) writes: 'Visiting Synanon or Kerista one is struck by the freedom taboos against body contact. There is also a directness of gaze, a process of continual confrontation, that makes the visitor uncomfortable... Hippies complain that 'squares' do not look at one another, they treat each other as nonpersons.' The implication of Sommer's statement is that even when people are physically close they can, by the manipulation of the sense modalities, create the fiction that they are alone. Often people find themselves crowded together for a period of time — on an elevator, in a subway, etc. — but involvement with other people can be prevented by avoiding eye contact, keeping the voice at a low level, by not facing other people directly, etc. It has been a source of insight for me — as well as a source of amusement — to ride in crowded elevators with a friend and manipulate proxemic variables. A high voice level is met with surprise and backing away. The same result occurs from making eye contact or facing another person more directly than the side-by-side stance commonly used in elevators in the United States. I feel that as proxemic research progresses, it will become increasingly clear that these variables do comprise a system of behavior for the perception and structuring of space. One inconvenience I sometimes experienced iii so small a house, is the difficulty of getting to a sufficient distance from my guest when we began to utter the big thoughts in big words. You want room for your thoughts to get into sailing trim and run a course

328

O. MICHAEL WATSON

or two before they make their port... Individuals, like nations, must have suitable broad and natural boundaries, even a considerable neutral ground, between them. I have found it a singular luxury to talk across the pond to a companion on the opposite side. In my house we were so near that we could not begin to hear — we could not speak low enough to be heard.... If we are merely loquacious and loud talkers, then we can afford to stand very near together, cheek by jowl, and feel each other's breath; but if we speak reservedly and thoughtfully we want to be farther apart, that all animal heat and moisture may have a chance to evaporate.... As the conversation began to assume a loftier and grander tone, we gradually shoved our chairs farther apart till they touched the wall in opposite corners, and then commonly there was not room enough. Henry D. Thoreau, Walden (1854)

PROXEMICS AND SEMIOTICS '

Semiosis has been defined by Charles Morris (1938 : 3) as 'the process in which something functions as a sign . . . ' , and the study of sign systems is termed semiotics. Semiosis, according to Morris (1938 : 3), has three principal components: the sign vehicle ('that which acts as a sign'), the designatum ('that which the sign refers to'), and the interprétant ('that effect on some interpreter in virtue of which the thing in question is a sign to that interpreter'). Further, Morris (1938 : 6-7) abstracts the relationships between the principal components of semiosis into three different aspects: the semantic, the pragmatic, and the syntactic. The semantic dimension of semiosis deals with the relationship of signs to the things signified, the pragmatic dimension with the relationship of signs to the behavioral responses which they elicit, and the syntactic dimension with the relationship of one sign to another sign within the same system of signs. The purpose of this section is to discuss proxemic behavior on the interpersonal level in a semiotic frame of reference. The term 'proxemic sign' as used in this paper is defined to mean that element of behavior which is subsumed within each of the analytically distinct subdivisions of the five categories which operationally define the system of proxemic behavior. Using the category 'eye contact' as an illustration, looking another person directly in the eye is one proxemic sign, looking at his face is another, looking in his general direction is another, and directing the gaze away from him is yet another. The Semantic Aspect of Proxemic Behavior The semantic aspect of proxemic behavior deals with the relationship of a proxemic sign to the meaning attached to the sign. In terms of the semantic dimension, proxemic norms, once internalized, are maintained largely 'outside awareness' (Hall 1964a: 41, 1966:109), i.e. people are not conscious of the importance of these norms until they have been violated. Thus, the meanings of proxemic signs become 7

This section is based on Watson (1969).

PROXEMICS

329

clearer when they are violated. Violations of proxemic norms most typically occur in a cross-cultural interaction in which two different patterns of proxemic behavior clash and produce interference. Since foreign students in the United States are likely to be exposed, to a greater or lesser degree, to violations of their proxemic norms every day, they seemed to be a particularly suitable sample to question about the meanings attached to various proxemic signs. The subjects were asked about differences in proxemic norms between North Americans and people in their own countries, and what violations of these norms meant in terms of their own systems of proxemic behavior. Each category of proxemic behavior will be discussed in turn. 1) Directness of Facing (Sociofugal-Sociopetal Axis) Empirical evidence8 indicates that Arabs, Latin Americans, and Southern Europeans face each other more directly during an interaction than do North Americans. Most Arab subjects indicated that to face another Arab with the indirectness which characterized North American interactions would mean that the person is not paying attention. Most Latin American and Southern European subjects answered similarly, saying that this indirectness would signal lack of interest. Subjects from both the Arab and Latin American groups mentioned that directness of facing is generally not as important as eye contact. East Asians, Indians and Pakistanis, and Northern Europeans fall into the same group as North Americans in directness of facing, so less specific indications of the meaning of the violation of this proxemic norm were obtained. None of the East Asians, interestingly enough, were able to think of any meaning associated with different degrees of directness of facing. Also interesting was the fact that several Northern European subjects mentioned that directness of facing and eye contact go hand in hand, in contrast to the remarks by Arabs and Latin Americans. 2)

Distance Although the Arabs interacted the closest of any group observed, several Arab subjects said that distance is not very important. An Arab, they continued, compensates for distance by shouting and eye contact. Latin Americans felt that North Americans used more distance in their conversations and that to use such a distance in Latin American would seem 'cold'. One student from Ecuador said that standing in line for tickets in the United States made him nervous because of all the space between people. Southern Europeans felt that using the greater North American distance in a conversation would appear to be too impersonal, or to impart a feeling of 'lack of harmony', as one French subject put it. The East Asians indicated that standing closer than is normal would imply superiority. Several subjects from the Indian-Pakistani group felt that North Ameri8

The empirical evidence discussed in this section is based on Watson and Graves (1966) and Watson (1970).

330

O. MICHAEL WATSON

cans were a little more distant in their interactions and that not to stand close to a person in India or Pakistan is not to be close to that person. Northern Europeans used about the same distance in interaction as North Americans, and most of them said that they got nervous when approached too closely. 3)

Touching Almost half of the Arab, Latin American, and Southern European subjects touched each other during an interaction, while only two East Asians and no Indian-Pakistani, Northern European, or North American subjects touched. The Arabs reported that it just wouldn't seem right not to touch another man during an interaction. The Latin Americans considered it extremely unusual not to touch during an interaction, and not to do so would appear to be 'cold'. Southern Europeans felt that they touched each other much more than North Americans, and not do so would be unfriendly. To the East Asians, touching another person during a conversation would indicate anger, loss of self-control, or over-friendliness, and Northern Europeans couldn't imagine a situation where one would touch another man a great deal. Although none of the subjects from the Indian-Pakistani group touched during an interaction, most of them felt that men in their countries touched each other more than North Americans and not to touch would impart a feeling of distance. 4) Eye contact The Arabs, Latins, and Southern Europeans put more stress on the importance of direct eye contact than did subjects from the other groups. One Arab said that conversation was impossible without the use of the eyes. Many Arab, Latin American, and Southern European subjects mentioned that to employ less eye contact is taken to mean impatience to break off a conversation or a lack of interest in the conversation. Subjects from the other groups frequently mentioned that too much eye contact signals anger, hostility, or prying. 5) Voice loudness Arabs mentioned that distance is spanned by raising the voice and several subjects from the Arab, Latin American, and Southern European groups indicated that to employ a voice loudness as low as that typically employed in North America would indicate shyness or embarrassment. Most East Asians mentioned that raising the voice indicates anger, threat behavior, or loss of self-control. Many IndianPakistani subjects said that it is impolite to speak too loudly. Among Northern Europeans raising the voice seems characteristically to convey anger. The brief examples cited above to illustrate the semantic dimension of proxemic behavior serve to emphasize the fact that different meanings are attached to the same proxemic signs in different culturally specific systems of proxemic behavior.

PROXEMICS

331

The Pragmatic Aspect of Proxemic Behavior The pragmatic aspect of proxemic behavior is concerned with the relationship of a proxemic sign to the response it elicits from an interpreter. Again, it is more illuminating to talk about violations of proxemic norms than their maintenance, and to employ the distinction which Hall (1964a) makes between adumbrations and cues. Adumbrations are precedents and accompaniments to the 'formal' topic of a transaction and serve as a feedback mechanism in adjusting or maintaining behavior. An adumbration, in Hall's (1964a : 157) words, 'is a perceivable manifestation of A's feelings of which he may not even be aware', while a 'cue is a short message of minimal redundancy in full awareness from A to B that indicates what A wants B to do'. One can see that the principal difference between adumbrations and cues is the level of awareness at which each is employed. Proxemic signs can be used both as adumbrations and as cues in communication within a culture, but in a cross-cultural situation proxemic signs are most frequently cues, or, more appropriately, m/scues. When two people from the same culture engage in a conversation, proxemic signs can be used as adumbrations in accordance with the subjective expectations of the interaction. A person is engaged in a conversation with his friend. Both share the same system of proxemic norms. Both of them have defined the situation in the same way and the behavior elicited is reinforced and maintained in part by proxemic signs used as adumbrations. The conversants use a distance, a degree of eye contact, an amount of touching, and so on, which is appropriate to the situation. Things go smoothly and the interaction is broken off. If questioned, the interactants would probably be hard put to specify the rules which govern the proper distance, amount of touching, and so forth, which went on during the interaction. Proxemic signs can also be used, at a higher level of awareness, as cues in culturally prescribed ways: you're angry so you raise your voice, you're hostile so you stare, you want to make a point so you jab your friend in the chest with your finger. You use, in other words, a proxemic sign as a cue to elicit a response from your conversational partner. In a culturally specific context proxemic signs can be used to help guage the smoothness of an interaction or to dare a person to punch you in the mouth, but the signs are usually interpreted correctly and elicit the expected response. Correct interpretation of a proxemic sign is often not the case in a cross-cultural interaction where the proxemic patterns of the interactants clash. For example, Arab, Latin American, and Southern European subjects gave a composite description of North Americans which characterized them as shy, uninterested, embarrassed, and generally 'cold'. These impressions are due, in large part, to a misinterpretation of proxemic signs, and such misinterpretations are understandable in light of the earlier discussion of the meanings of proxemic signs. An Arab and an American engaged in a conversation might define the situation in the same way, but the proxe-

332

O. MICHAEL WATSON

mic adumbrations used by each could be noticeably different and therefore serve as cues. The American's failure to maintain direct eye contact is normal to the American, but the Arab is very likely to read it as a cue which he interprets as meaning lack of interest. The behavior elicited from these misinterpretations is most often, to use Goffman's (1957) phrase, 'alienation from interaction', i.e. the interactants begin to direct more attention to what's going wrong than to the topic of conversation, and such responses are obviously not conducive to good communication. The Syntactic Aspect of Proxemic

Behavior

The syntactic aspect of proxemic behavior concerns the relationship of a proxemic sign to other proxemic signs. The task in investigating the syntactic aspect of proxemic behavior is to attempt to construct a network of proxemic signs. Using the operationally defined, phenomenally distinct units defined as proxemic signs, correlation matrices can be constructed which give some idea of what the interrelationships between these units are (Watson and Graves 1966, Watson 1970). Matrices correlating proxemic signs in different culturally specific systems turn out to be quite different, suggesting that proxemic signs within different systems have different relationships to each other. Thus, it is the analysis of proxemic behavior at the syntactic level of semiosis which serves to demonstrate how little is known about proxemic behavior. The way in which 'proxemic sign' has been used in this section is a convenient analytical device which may or may not have relevance within a culturally specific system of proxemic behavior. The isolation of proxemes — contrastive units of proxemic behavior which are relevant within culturally specific systems — seems to be a task which must be undertaken if a meaningful analysis of proxemic behavior as semiosis is to be made. But we must know what the culturally meaningful signs are before we can determine the ways in which they are used.9 PROXEMICS IN THE CONTEXT OF NONVERBAL COMMUNICATION The discussion above of the semantic aspects of proxemic behavior should have made it clear that it functions as a form of nonverbal communication. As a system of communication, Hall ( 1 9 6 3 b : 1021) sees proxemics as ' . . . a transaction between two or more parties, or one or more parties, and the environment'. Other researchers, primarily in the field of environmental (or ecological) psychology, have expressed an interest in the transactions between man and his environment (Barker 1968). Altman and Lett (1967 : 1 - 3 ) define the interests of ecological psychology as 'the nature of the mutual interaction between man and his environment.. ., how people use their bodies and manipulate objects in their environment...'. Altman • See the section on 'Levels of Analysis: Proxemics and Proxetics', below, for a more complete discussion of proxemics.

PROXEM1CS

333

and Lett present a classification model for dealing with the use of the environment in interpersonal relations. It is useful to view proxemic behavior in terms of the model, a simplified representation of which appears in Figure 2, and which is concerned with dyadic interactions.

Fig. 2. A General Classification Model (after Altman and Lett 1967)

The column of elements at the far left of Figure 2 are referred to by Altman and Lett (1967 : 12) as 'antecedent factors', i.e. antecedent to an interaction and by means of which an interactant 'sizes up' the interaction. Antecedent factors include personal characteristics of the interactants (age, sex, personality traits, etc.); physical-physiological characteristics (height, weight, deformities, and emotional state); interpersonal characteristics (degree of acquaintance, compatibility of the interactants, etc.); and environmental characteristics (temperature, light, noise level, etc.). These antecedent factors serve the interactants in providing a situation definition, the next step in the model. In defining the situation in which the interactants find

334

O. MICHAEL WATSON

themselves, they adjust their behavior accordingly, using environmental props and self markers. Environmental props are the chairs, tables, etc., and the ways in which the interactants use them. Self markers refers to the ways in which people use their own bodies in an interaction. Environmental props and self markers are integrated into an over-all gestalt profile. The gestalt profile conveyed over a period of time is a social interaction. The evaluation and assessment part of the model concerns the judgments of interactants as to the appropriateness of their behavior. If an interactant judges his behavior to be inappropriate in terms of his evaluation about his expectations of the situation, he can either define a new situation or make a readjustment within the context of the expectations originally defined at the beginning of the interaction. In summary: 'when individuals are in one another's immediate presence, a multitude of words, gestures, acts, and minor events become available, whether desired or not, through which one who is present can intentionally or unintentionally symbolize his character and his attitudes' (Goffman 1957 : 104). When proxemic behavior is considered in the context of Altman and Lett's model, it becomes apparent that it can be included in the 'environmental props' and 'self markers' categories of the model. In terms of environmental props an interactant can use chairs, tables, partitions, and other moveable features of the environment to orient himself in relation to other interactants. An interactant can also arrange his own body and manipulate his sensory inputs to structure microspace, such behavior falling into 'self marker' category. But proxemic behavior is only one element in the composition of the 'self marker' part of the model, as other forms of somatic communication play a role in interaction. Darwin (1872 : 365), for example, described the importance of facial expression and body position as a communicative device: 'The movements of expression in the face and body . . . are in themselves of much importance for our welfare.' More recently Ekman (1964) has shown that people have the ability to correctly indicate the relationships between photographs of interactants and verbal behavior using the head and body as cues. La Barre (1956) also has described the ways in which people use their bodies to display their feelings.10 Birdwhistell's (1952, 1960, 1966) work demonstrates that kinesics, which he defines as ' . . . the study of body motion as related to the non-verbal aspects of interpersonal communication' (1952 : 3), also belongs in the 'self marker' category of the model. From the discussion in the section concerned with proxemics as semiosis, it should be clear that proxemic behavior communicates. In some cultures, for instance, raising one's voice indicates excitement, while in others it may indicate anger or loss of self-control. Direct eye contact in one culture might indicate interest while in another it might be construed as a threat. Now, in terms of Altman and Lett's (1967) model, consider what takes place when two people from the same 10 See the essay in this volume by Professor Sarles for a more complete discussion of body motion and facial expressions.

PROXEMICS

335

cultural group interact. First of all, when two persons sharing the same cultural system interact, they have a common basis for defining the situation, and if each of them has the same subjective expectations of the interaction — 'sized up' the situation properly — things will go smoothly until the interaction is broken off. Secondly, even if the interactants do not define the situation in the same way, sharing the same cultural system they have a common knowledge for the interpretation of cues: a readjustment can be made or a new situation can be defined. Suppose that A and B, two persons sharing the same culture, enter into an interaction. A determines his relationship to B by means of antecedent factors. B's clothing, his emotional state, his degree of acquaintance to A , the temperature, the time of day, and the setting all help A to define the situation. If A sums up the antecedent factors and judges B to be his social subordinate, he will act accordingly. If B determines from the antecedent factors that A is his social superordinate, then the subjective expectations of both interactants will correspond and chances are the interaction will proceed according to plan. Let's suppose, however, that A judges B to be his subordinate, but B, by some oversight, fails to recognize A as his superordinate and fails to act toward A in the ways which the culture specifies one should act toward a person of higher status. B may violate cultural norms in interacting with A , and A may cue him to readjust his behavior and redefine his relationship with A. 1 1 If B doesn't behave correctly toward A proxemically, A may cue him by standing closer or farther away, he may raise his voice or lower it, he might look B in the eye or look away, according to cultural prescriptions. One can imagine a situation in which B could intentionally ignore A's cues in order to demonstrate his contempt, or to provoke conflict, or for any number of reasons. The point is that people who interact with a common basis for interpreting the multitude of cues which go into an interaction usually know what is taking place during the interaction. But consider two people from different cultures who engage in an interaction.

PROXEMIC

INTERFERENCE

Imagine two well-meaning persons from cultures with different patterns of proxemic behavior entering into an interaction. Let's assume that neither has the culturally specific information with which to define the situation in terms of the other's expectations. Both interactants might define the setting as, say, friendly conversation between two males in a public place, but the distance, amount of eye contact, etc., specified by each culturally specific system of proxemic behavior might not correspond. Both interactants would hypothetically be constantly making adjustments 11 Superordinate-subordinate confusion occurs frequently in an academic setting. Administrative staff sometimes mistake professors (who are usually perceived as superordinates) for students (whom they perceive as subordinates). The younger professors, with their longer hair, beards, and lack of coat and tie, cause a confusion of cues which is both enlightening and delightful to witness.

336

O. MICHAEL WATSON

to fit their altered expectations, but probably to no avail. Altman and Lett (1967:17) summarize this dilemma. An inconsistency in A's and B's personal definition of the situation is important as a potential initial determinant of interpersonal conflict and stress. To the extent that there is a discrepancy in their expectations, there will be a reverberation throughout die remainder of the model. Each person would begin experiencing unanticipated behaviors on the other's part, which eventually could be the basis for an incompatible relationship. This misinterpretation of proxemic behavior has been oalled 'interference' by Hall (1963b: 1005), who defines it as a clash between the two patterns, or a perceived absence of patterning. 'When people from cultures with different proxemic patterns interact, they cannot be relied upon to place the same meaning on the same element of proxemic behavior' (Watson 1970:109); they obviously have no choice but to interpret the proxemic behavior of the other interactant in terms of the proxemic patterns of their own culture. Thus, proxemic interference has to do with misinterpretation of proxemic cues. Another factor which plays a part in proxemic interference, and which further confounds cross-cultural interaction, is the level of awareness at which proxemic patterns are maintained. Hall (1966:121) has used the metaphor of a person being enclosed in a 'bubble' in referring to the zones which surround an individual and are thought by that individual to be his inviolate personal space. The dimensions of these zones vary from culture to culture: Arabs, Latin Americans, and Southern Europeans carrying around with them smaller 'bubbles' than Northern Europeans, North Americans, or Asians (Hall 1966, Watson 1970). Thus, when W. H. Auden (1965:4) wrote: 'Some thirty inches from my nose, the frontier of my Person goes', he was describing the dimensions of his personal space. Further, these 'bubbles' are transported and maintained 'informally', i.e. 'outside awareness' (Hall 1955: 84-5, 1959:73-4, 1964b: 41, 1966:109, 1968:83). People are not conscious of the informal rules which govern proxemic patterns, and when violations of proxemic norms occur misinterpretation, anxiety, and alienation from interaction often follow. Alienation from interaction has been described by Goffman (1957:108-11) as being represented by four types: external preoccupation, self-consciousness, interaction-consciousness, and other-consciousness. External preoccupation can be described as occurring when interactants focus their attention on something which is unconnected to the interaction. When proxemic interference occurs, external preoccupation does not actually have to occur for behavior to be interpreted as such. Thus, foreign student interviewees frequently mentioned that the lack of directness in facing and lack of eye contact which North Americans display gave the impression that there was a lack of interest in the conversation. Self-consciousness occurs when the interactants focus their attention on themselves in order to locate the source of the interference; or they may turn to the behavior of the other interactant as the cause of the disturbance, which is other-consciousness. It is also possible

337

PROXEMICS

that 'a participant in talk may become consciously concerned to an improper degree with the way in which the interaction, qua interaction, is proceeding, instead of becoming spontaneously involved in the official topic of the conversation' (Goffman 1957:107), which is defined as interaction-consciousness. All of these forms of alienation from interaction are likely to result from proxemic interference, at the expense, obviously, of the interaction. When proxemic interference occurs, the interactants can usually detect something has gone wrong, without finding out precisely what it was. Hall (1963b: 1005) has provided an excellent description of proxemic interference: W h e n approached too closely, Americans removed themselves to a position which turned to be outside the olfactory zone (to be inside was much too intimate for the Americans). Arabs also experienced alienation traceable to a 'suspiciously' low level of the voice, the directing of the breath away from the face, and a much reduced visual contact. T w o common forms of alienation reported by American subjects were self-consciousness at the cost of involvement and other-consciousness. Americans were not only aware of uncomfortable feelings, but the intensity and the intimacy of the encounter with Arabs was likely to be anxiety provoking. The A r a b look, touch, voice level, the warm moisture of his breath, the penetrating stare of his eyes, all proved to be disturbing. The reason for these feelings lay in part in the fact that the relationship way not defined as intimate, and the behavior was such that in the American culture is only permissible on a non-public basis with a person of the opposite sex (Hall 1963b : 1005).

LEVELS OF A N A L Y S I S : PROXEMICS A N D

PROXETICS 12

Kenneth Pike (1954:8) coined the terms 'etic' and 'emic' and applied them to two different ways of studying systems of human behavior, using as an analogy the analytical distinctions made by linguists between phonetic and phonemic approaches to sound systems. The distinctions between etic and emic approaches have been treated in detail elsewhere (e.g. Harris 1968:568-604, Pike 1966:152-63), so a summary of these distinctions will suffice here. The etic point of view approaches a system of behavior from outside the system, using criteria which are external to the system. The etic approach provides an initial base from which the observer can begin his analysis of the system. Statements made about the system in etic terms, in Harris's (1968:575) words, 'depend upon phenomenal distinctions judged appropriate by the community of scientific observers'. The emic approach, on the other hand, is concerned with studying behavior from inside a single, culturally specific system. Criteria used in an emic description are drawn from the contrasts made within the system itself and are relevant in terms of the internal functioning of the system. Emic distinctions are those which are recognized as meaningful to the interactants who 'use' the system. 12

This section is based on Watson (1%8).

338

O. MICHAEL WATSON

So, a behavioral scientist interested in a particular area of human behavior might search for universals in this area, and use etic units and categories in describing this behavior as it occurs in cultures around the world. If he were interested in describing how this behavior is structured in one, culturally specific system, the etic categories previously uncovered would prove useful in providing an entré into that particular system and would serve as an analytical base until he had gained the knowledge necessary for an emic description of the system. Proxemic behavior can be fruitfully explored using the emic and etic approaches. From Hall's (1963b) operationally defined proxemic variables and the system of notation which he provides for the recording of observations of proxemic behavior, it is clear from the level of analysis used that the area of behavior under investigation is not proxemics, but proxetics. Hall (1963b : 1021) was aware of this when he wrote: . . this presentation is concerned more with proxetics than proxemics, and is therefore only the first of a series of steps in a long, complex process.' So far, proxemic research has been almost entirely in an etic framework. We have Hall's etic categories, we can observe, measure, and quantify the proxemic behavior of all the peoples of the world, so, etically speaking, all is well. But these things don't tell us much in emic terms, i.e. what the relationships are between proxemic variables in different culturally specific systems. Even given the differences between, say, Arab and American systems of proxemic behavior, all we can really talk about is the contrast between the two systems, not the contrasts made within each system. Hall (1968:94) has emphasized that there is no known universal mechanism for setting distances, i.e. different culturally specific systems of proxemic behavior make greater or lesser use of different senses. This amounts to different systems of proxemic behavior: 'This is not a matter . . . of generalizing about Latinos standing closer than North Americans, or moving each space zone up a notch...' That different culturally specific systems of proxemic behavior are structured differently is supported by some empirical evidence. If it were true that all systems of proxemic behavior were similar in interval structuring (i.e. each variable in any system bore the same relationship to all other variables in the system), and culturally specific systems of proxemic behavior differed from each other only in terms of quantity (i.e. interactants in one system would face one another more directly, display more eye contact, etc., than interactants in another such system), then we would expect these consistencies to be demonstrated by a simple correlation matrix. If all proxemic systems were structured the same way, the values on a correlation matrix would remain about the same for any proxemic. This appears not to be the case. Analysis of empirical data (Watson and Graves 1966:983, Watson 1970 : 97-9) demonstrates that matrices correlating the different proxemic variables within different culturally specific systems are not similar. This suggests that the variables within different systems bear different relationships to each other. Internal differences between proxemic systems were also supported by interviews

PROXEMICS

339

with subjects (Watson 1970). Several Arab subjects, for example, told me that directness in facing didn't really matter as long as eye contact was maintained. Northern Europeans, on the other hand, insisted that shoulder axis and eye contact went hand in hand. North Americans, when approaching friends on the street, typically avoid eye contact and speaking until the other person is close enough to exchange greetings or begin interaction, i.e. they use physical proximity to establish contact. Arab subjects, however, indicated that there is no such hesitancy among them to begin interaction: upon seeing a friend approaching in the distance they typically use voice loudness and eye contact to span that distance, beginning interaction from afar. Such behavior would be embarrassing and offensive to most North Americans. Emic analysis of culturally specific systems of proxemic behavior would be necessary in order to isolate the proxemes — contrastive units of proxemic behavior — in a system. The etic categories isolated and used as a basis for measurement of proxemic behavior include, as previously mentioned, such things as the directness with which interactants face one another, the amount of touching which goes on during an interaction, the physical distance between the interactants, the voice loudness with which interactants converse, and the eye contact which takes place during an interaction. These categories are broken down further into smaller units of analysis which are observable and measurable. The category 'eye contact', as was mentioned in a previous section, has four analytical subdivisions, which amount to: looking another person directly in the eye, looking at him in the area of his face, etc. In etic terms these are four separate units, easily observed. But, in emic terms, the boundaries between some of these units might not be significant in a culturally specific system. It could be, for instance, that in a culturally specific system of proxemic behavior looking a person directly in the eye and looking at him in the area of his face might not be contrasting units, i.e. might be alloproxes of the same proxeme and have the same meaning. The same thing would apply, of course, to the subdivisions of other proxemic categories. Further investigations along emic lines might confirm some things which were suggested by my own research {Watson 1970). It appears, for example, that certain variables in a proxemic system are influenced and altered by certain factors, thus changing their relationships to other variables in the system. A specific example of this phenomenon regards the amount of touching in 'contact' cultures (i.e. Arabs, Latin Americans, and Mediterranean Europeans). People from these cultures, after they have been in the United States for a while, come to realize that the amount of touching which goes on between males in their own cultures is looked at askance in this country. Arab subjects, who had been in this country for a while, told of going to meet their newly arrived Arab friends and extricating themselves from an embrace long enough to tell their friends that this sort of behavior is frowned upon in this culture. One Italian subject put it nicely when he told me: 'When I came to this country I touched less consciously, but now I touch

340

O. MICHAEL WATSON

less unconsciously' (Watson 1970:95). These and other responses by subjects were supported by data analysis, which demonstrated that the longer a member from a contact culture had been in this country, the less he touched during an interaction. Finally, a word about 'contact' and 'non-contact' cultures. Hall has borrowed the terms 'contact' and 'non-contact' from the ethologist Hediger (1950: 111, 1950: 66) and applies them to human groups in terms of proxemic behavior. Hall (1963b: 1023, 1964b: 44-45) suggests that all cultures fall into one or the other global types. The contact group is composed of members who face one another more directly, interact at closer distances, touch one another more, look one another in the eye more, and speak at higher voice levels than do members of the non-contact group. But it would appear that as more culturally specific systems of proxemic behavior are described and analyzed, the more difficult it will become to utilize the global contact and non-contact typologies. Data analysis (Watson 1970) suggests that instead of the entire proxemic syndrome being used to classify a culture as contact or non-contact, the variables closeness (kinesthetic factors) and touching might be used for this purpose. Similarly, eye contact and shoulder axis could perhaps be used to classify a culture along a 'direct-indirect' dimension. If these relationships do indeed exist, it might then be possible to find a culturally specific system of proxemic behavior which could be classified as 'contact/indirect', or 'non-contact/direct'.

PROXEMIC RESEARCH: WHERE DO WE GO FROM HERE?

The emphasis in this chapter has been on a demonstration of how little is known about proxemic behavior. In the case of proxemics I feel that the traditional closing statement calling for more research is amply justified. The question which must be asked, however, is: what kind of research? Hall (1968:94-5) has listed 15 problem areas to be pursued in proxemic research. They are rather general and include distance differentiation and maintenance and the effect attached to these distances; the study of fixed-feature, semifixed-feature, sociofugal, and sociopetal space; the investigation of boundaries and hierarchies of space; the nature of touching, looking, smelling, and listening behavior; the social dimensions in the use of space; and more investigation of spatial needs. I can only endorse these suggestions, as they are all important in the pursuit of knowledge of proxemic behavior. I would particularly like to see some work done in urban areas, in an attempt to uncover the effects of crowding in ghettos and other high density areas of the city. Further, I feel that anthropologists, while maintaining the traditional notions of the importance of culture in human activity, should temper their traditional prejudices against biological explanations of behavior with

PROXEMICS

341

a deeper understanding of the growing body of literature in ethology. It seems obvious that arguments of 'instinct' vs. 'learning' are not quite so simple as they used to be, and problems lie in the understanding of the interaction between biological man and cultural man. More specifically, I advocate the study of culturally specific systems of proxemic behavior as it occurs in culturally defined settings. Proxemic behavior does not occur as a set of independent variables in a natural setting, but are dependent on a multitude of factors in the environment of the interaction — social, emotional, physiological, etc. Finally, from a strictly personal point of view, I would urge students of anthropology and, indeed, the other behavioral sciences, to consider doing research in proxemic behavior because it is — to me — one of the most interesting and rewarding areas of investigation in the behavioral sciences. Proxemic research is not only just plain fun to do, but also plays an important role in the understanding of the total communicative process. REFERENCES ALTMAN, I . ,

and

W.

E. HAYTHORN.

1967.

The ecology of isolated groups.

BS

12.169-82.

I., and E. LETT. 1967. The ecology of interpersonal relatons: A classification system and conceptual model. Paper presented at the conference on social-psychological factors in stress. University of Illinois, April. AUDREY, R. 1966. The territorial imperative. New York, Atheneum. ARGYLE, M., and J. DEAN. 1965. Eye-contact, distance, and affiliation. Socio-

ALTMAN,

metry 28.289-304.

Centuries of childhood. New York, Knopf. About the house. New York, Random House. BARKER, R. 1 9 6 8 . Ecological psychology. Stanford, Stanford University Press. BIRDWHISTELL, R. L . 1 9 5 2 . Introduction to kinesics. Washington, D.C., U.S. Department of State. . 1960. Kinesics and communication. Explorations in communication, ed. by E. Carpenter and M. McLuhan, 54-64. Boston, Beacon. . 1966. Some relations between American kinesics and spoken American English. Communication and culture, ed. by A. G. Smith, 182-9. New York, Holt, Rinehart and Winston. . 1968. Comment on Proxemics, by E. T. Hall. CAnthr 9.95-6. BLOCK, H. A . , and A . NIEDERHOFFER. 1 9 5 8 . The gang. New York, Philosophical Library. BOGARDUS, E . S. 1 9 3 3 . A social distance scale. Sociology and Social Research ARIES,

P.

AUDEN, W .

1962.

H.

1965.

17.265-71.

. 1959. Social distance. Yellow Springs, Ohio, Antioch Press.

342

O. MICHAEL WATSON

W. H. 1943. Territoriality and home range concept as applied to mammals. Journal of Mammalogy 24.346-52. CARPENTER, C . R. 1958. Territoriality: A review of concepts and problems. Behavior and evolution, ed. by A. Roe and G. G. Simpson, 224-50. New Haven, Yale University Press. CARPENTER, E., and M . M C L U H A N . 1960. Acoustic space. Explorations in communications, ed. by E. Carpenter and M. McLuhan, 65-70. Boston, Beacon. CHANCE, M. R. A. 1962. An interpretation of some agonistic posture. Symposia of the Zoological Society of London 8.71-89. CHAPPLE, E. D. 1970. Culture and biological man: Explorations in behavioral anthropology. New York, Holt, Rinehart, and Winston. DARWIN, C. 1872. The expression of emotions in man and animals. London, Murray. DAVIS, D . E. 1962. An inquiry into the phylogeny of gangs. Roots of behavior, ed. by E. L. Bliss, 316-20. New York, Harper and Row. EIBL-EIBESFELDT, I. 1970. Ethology: The biology of behavior. New York, Holt, Rinehart, and Winston. EKMAN, P. 1 9 6 4 . Body position, facial expression, and verbal behavior during interviews. JAbSocPsych 6 8 . 2 9 5 - 3 0 1 . BURT,

ESSER, A . H . ,

S . AMPARO, R . N . CHAMBERLAIN, E . D . CHAPPLE, a n d

N . S . KLINE.

1964. Territoriality of patients on a research ward. Recent Advances in Biological Psychiatry 7.37-44. FELIPE, N . Interpersonal distance and small group interaction. Cornell Journal of Social Relations 1.59-64. Ithaca, N.Y. FESHBACK, S., and N. FESHBACK. 1963. Influence of the stimulus object upon the complementary and supplementary projection of fear. JAbSocPsyc 66.498502. FOSTER, G. 1969. Applied anthropology. Boston, Little, Brown. FRANK, L . K . 1957. Tactile communication. Genetic Psychology Monographs 56.209-55. Reprinted 1966, Communication and culture, ed. by A. G. Smith, 199-209. New York, Holt, Rinehart, and Winston. . 1960. Tactile communication. Explorations in communications, ed. by E. Carpenter and M. McLuhan, 4-11. Boston, Beacon. GOFFMAN, E . 1957. Alienation from interaction. Human Relations 10.47-60. Reprinted 1966, Communication and culture, ed. by A. G. Smith, 103-18. New York, Holt, Rinehart, and Winston. HALL, E. T. 1 9 5 5 . The anthropology of manners. SA 1 6 2 (April). 8 5 - 9 0 . . 1959. The silent language. Greenwich, Conn., Fawcett. . 1963 a. Proxemics — the study of man's spatial relations. Man's image in medicine and anthropology, ed. by Iago Galdston, 422-45. New York, International Universities Press.

PROXEMICS

343

. 1963b. A system for the notation of proxemic behavior. AmA 65.100326.

. 1964a. Adumbration as a feature in intercultural communication. AmA 66:6/2.154-63. . 1964b. Silent assumptions in social communication. Disorders of Communication 42.41-55. . 1966. The hidden dimension. New York, Random House. . 1968. Proxemics. CAnthr 9.83-108. HALL, E . T . , and W . F . WHYTE. 1 9 6 0 . Intercultural communication: A guide for men of action. H O 1 9 . 5 - 1 2 . HARE, A. P . , and R . F. BALES. 1963. Seating position and small group interaction. Sociometry 26.480-6. HARRIS, M . 1 9 6 8 . The rise of anthropological theory. New York, Crowell. HEARN, G. 1 9 5 7 . Leadership and the spatial factor in small groups. JAbSocPsych 5 4 . 2 6 9 - 7 2 . HEDIGER, H . 1950. Wild animals in captivity. London, Butterworths. . 1955. Studies of the psychology and behavior of captive animals in zoos and circuses. London, Butterworths. HOROWITZ, M . J. 1 9 6 3 . Graphic communication: A study of interaction painting with schizophrenics. American Journal of Psychotherapy 1 7 . 2 3 0 - 9 . HOROWITZ, M . J . , D . F . DUFF, and L . O . STRATTON. 1 9 6 4 . Body-buffer zone. Archives of General Psychiatry 1 1 . 6 5 1 - 6 . HUTT, C., and C. OUNSTED. 1966. The biological significance of gaze aversion with particular reference of the syndrome of infantile autism. BS 11.346-56. KANNO, H . 1 9 7 0 . Group territoriality: An ethological analysis of youth gangs in Chicago. Symposium on Biology, Behavior, and Aggression, Purdue University, April. KAUFFMAN, L. E. 1969. Tacesics: The study of touch. Unpublished ms. KENDON, A. 1 9 6 5 . Some functions of gaze direction in social interaction. Acta Psychologica 2 6 . 2 2 - 6 3 . KUETHE, J. L. 1962a. Social schemas. JAbSocPsych 64.31-8. . 1962b. Social schemas and the reconstruction of social object displays from memory. JAbSocPsych 65.71-4. LA BARRE, W. 1956. The cultural basis of emotions and gestures. Personal character and cultural milieu, ed. by D. G. Haring, 547-63. Syracuse, N. Y., Syracuse University Press. LIPMAN, A. 1 9 6 8 . Building design and social interaction. The Architects Journal 147.23-30.

K. B. 1965. Personal space. Journal of Experimental Social Psychology 1.237-47. LYNCH, K . 1960. The image of the city. Cambridge, M.I.T. Press. MCLUHAN, M. 1 9 6 4 . Understanding media. New York, McGraw-Hill. LITTLE,

344

O. MICHAEL WATSON

F. A., and C. L. BRACE. 1965. Man's evolution: An introduction to physical anthropology. New York, Macmillan. MORRIS, C. W. 1938. Foundations of the theory of signs (International Encyclopedia of Unified Science, 2). Chicago, University of Chicago Press. MURPHY, R. F. 1964. Social distance and the veil. A M A 6 6 . 1 2 5 7 - 7 4 . OSMOND, H . 1959. The history and social development of mental hospitals. Psychiatric architecture, ed. by C. Goshen, 1-9. Washington, D.C., American Psychiatric Association. OSTWALD, P . F . 1 9 6 4 . How the patient communicates about disease with the doctor. Approaches to semiotics, ed. by T. A. Sebeok, A. S. Hayes, and M. C. Bateson, 1 1 - 3 4 . The Hague, Mouton. OVID. 1957. The art of love. Translated by R. Humphries. Bloomington, Indiana University Press. PIKE, K . L. 1954. Language in relation to a unified theory of the structure of human behavior. Summer Institute of Linguistics. . 1966. Etic and emic standpoints for the description of behavior. Communication and culture, ed. by A. G. Smith, 152-63. New York, Holt, Rinehart and Winston. SOMMER, R. 1959. Studies in personal space. Sociometry 22.247-60. . 1961. Leadership and group geography. Sociometry 24.99-110. . 1962. The distance for a comfortable conversation: A further study. Sociometry 25.111-16. . 1966. Man's proximate environment. JSocI 22.59-70. . 1969. Personal space. Englewood Cliffs, N. J., Prentice-Hall. STEINZOR, B. 1950. The spatial factor in face-to-face discussion groups. JAbSocPsych 45.552-5. STOVER, E. L. 1970. The relation of aggression in individuals to conflict and war between groups and nations. Symposium on biology, behavior, and aggression. Purdue University, April. TYRWHITT, J. 1960. The moving eye. Explorations in communication, ed. by E. Carpenter and M. McLuhan, 90-5. Boston, Beacon. WATSON, O . M . 1968. Proxemics and proxetics. Paper presented to the American Anthropological Association. Seattle, November. . 1969. Proxemics and semiotics. Paper presented to the American Anthropological Association. New Orleans, November. . 1970. Proxemic behavior: A cross-cultural study. The Hague, Mouton. WATSON, O . M . , and T . D . GRAVES. 1966. Quantitative research in proxemic behavior. AmA 6 8 . 9 7 1 - 8 5 . WINICKC., and H . H O L T . 1 9 6 1 . Seating positions as non-verbal communication in group analysis. Psychiatry 2 4 . 1 7 1 - 8 2 .

MONTAGU, M .

C L A S S I F I C A T I O N A N D DESCRIPTION OF SIGN LANGUAGES

WILLIAM C. STOKOE, JR.

Among the languages of the world are many which have but rarely been noticed, or, when notice has been taken of them, which have been supposed a single system and called 'sign language' or 'the sign language'. This treatment is like that accorded language by primitive societies in which the same word used to name their language also means 'language in general' or 'talking'. But the cases are different. Primitive groups remain primitive because of isolation. Lacking contact with other groups speaking other languages, they have had no occasion to reflect on the diversity of tongues nor to separate the action of talking from the action of talking in a particular language. In the case of a sign language, however, most observers have worked from outside the sign-linguistic community, and to them the fact that there is communication in gestures without speech seems sufficient to characterize all such activity as sign language. Sign languages differ in several respects, and if an attempt is made to identify and classify them, their considerable number, their differences in kind from other languages, and their structural diversity all become apparent. One way to begin is to take the most limited kind of denotatum that a sign of a sign language can have. This minimum is a letter of an alphabet. There are a number of standard alphabets in use and perhaps as many sign languages to denote them. These are all codes consisting of pairs of a gestural sign with an alphabetic symbol conventionally denoted by it, and use of such a system is commonly called 'fingerspelling' by its American users. There are other kinds of writing systems which do not use letters nor the alphabetic method of representing language elements, and the classification of sign language must also take into account systems in which a sign denotes an element of a non-alphabetic writing system, whether syllabic, morphemic, or ideographic. The question of what precisely distinguishes an ideogram from some other kind of graphic symbolization need not be oversubtle. For the purposes of this portion of sign language taxonomy an ideogram is taken to mean any representation of a morpheme, a one-syllable word, or a synonym of that word in a single 'character' of the writing system (Chao 1968). In the kinds of sign languages now being considered signs are composed in three

346

WILLIAM C. STOKOE, JR.

ways. In some, signs are made with one hand only; others require the use of both hands; in still a third class the signs are made with one hand or both in conjunction with other parts of the body. These three ways of making signs and the two kinds of script, alphabetic and non-alphabetic, distinguish six classes of sign language. Each of the six is a class of semiotic systems in which a kind of sign recognized by its composition, a kind of denotatum recognized by its graphic nature, and a conventional relation between sign and denotatum define the class. All this is shown graphically in Figure 1 in which the numbered arrows indicate the six classes. Alphabetic symbols

Ideographic symbols

1 Monocheric signs

Amphicheric signs

Mixed signs

Fig. 1

There a number of members of class 1, among them the American Manual Alphabet, which is pictured in most dictionaries of English published in the United States (Figure 2). Other sign languages in this class are in use in France, Russia, Scandinavia, Finland, Eire, Canada, Japan, and Oceania. Differences between members of this class are determined in part by the alphabets used; e.g. the American manual alphabet has signs for the 26 letters, the Russian signs for the 32 letters of the Cyrillic alphabet (Gejl'man 1957). But these ngn languages also differ when there is no alphabetic necessity for it. The signs for 'f', 'g\ and't' are made differently by American and European signers who use the same alphabets for printing. The American 'hand' for't' is a classic form of the manus fica of antiquity, which still has an obscene or insulting significance in many parts of the world (Barakat 1969); but the French and Irish manual alphabets, perhaps for this reason, denote't' by a different sign. Irish hand signs also differ from American in other particulars. The Irish hand for 'h' is the manus cornuta (Ibid.), index and little finger upraised from the closed fist. Although this sign is a clear icon of a capital H, its connotation may have kept it out of the French manual alphabet, direct source of the American. A still more curious difference is to be seen within Irish fingerspelling. Until they are in their middle teens, boys and girls in Cabra, Dublin, in the schools

CLASSIFICATION AND DESCRIPTION OF SIGN LANGUAGES

R

W

The AMERICAN MANUAL ALPHABET CH Fig. 2

347

348

WILLIAM C. STOKOE, JR.

founded respectively by the Christian Brothers in 1846 and Dominican nuns in 1849 have different signs for's'. The girls hold the hand still, as is done for the 's' in American fingerspelling, but the boys 'throw it', i.e. they begin with the same hand configuration but rapidly open the hand as it is thrown forward. When the pupils are in the fifth form, about a year from the time they will leave school, their teachers say that they somehow become conscious of this difference which has been visible all along and adopt the s-sign of the other sex. Apparently as adults they have free variation, the two manual s's being allosigns in their system. Lest all this detail of minor variation in a sign for one letter in one sign language of many seem trifling, it may be kept in mind that, in round figures, about one person in one thousand of the general population is deaf, and a great many persons use sign languages of this kind. Moreover, it is just this kind of detailed description of communication differences that will make possible solid information about the cultural systems and their transmission among a very little known group of people. Class 2 of alphabetic sign languages has but a single extant member known to the writer. It is used in Great Britain and in Northern Ireland and was used in the school for the deaf in Nova Scotia some years ago. It is used as are the languages in class 1 for spelling words but in it the signs are made with both hands instead of one. As a means of rapid and accurate communication between deaf persons or between the deaf and hearing both monocheric and amphicheric alphabetic codes have great utility. Just as conditions may sometimes make use of radiotelephony impossible but allow contact by Mouse code, so these substitution codes with signs for letters can facilitate communication. The writer has carried on a conversation with a deaf-blind poet in the United States, using the American one-hand alphabet. His hand lightly touching the hand of the sender was his means of receiving messages and his replies were in speech and sign language of another kind, sometimes delivered before the sender's halting fingerspelling had completed a question. Again with a deaf-blind man met in Edinburgh the two-hand alphabet proved an indispensable medium. His left hand and the writer's right combined for messages to him; to reply he spelled in the normal English way. Early accounts treat both class 1 and class 2 sign languages together, giving not only examples of the ones in present use but also examples of extinct sign languages of these two classes. George Dalgarno (1680) in "Didascalocophus, or the deaf and dumb man's tutor" shows a cut of a hand with positions for the letters on the left hand and the directions to make a cross touch for the vowels with (doubtless the right index) finger and to point to the consonants with the right thumb. Use of a glove with the positions (Figure 3) marked reduced the load on the memory, and references to Dalgarno's glove appear in later treatises on the education of the deaf (Siger 1968). A late use of this kind of class 2 sign language is attested by the tiny left-hand kid glove with letters marked on it which is preserved in the library of the Alexander Graham Bell Association in Washington, D. C. Robert Graves

CLASSIFICATION AND DESCRIPTION OF SIGN LANGUAGES

349

I

Z after Hodgson 1953 : 92 E 1 apparently in error forF U* apparently in error for J

Fig. 3

(1952:114) in The white goddess shows (Figure 4) what he believes to have been the arrangement on the left hand of the twenty letter Ogham alphabet. While his is an imaginative reconstruction of the history of the transmission of alphabetic lore and its connection with Druidic calendar magic and computations, this conjectural class 2 alphabet is the only link between a class 2 sign language and class 3, of which more below. Much of the interest since the sixteenth century in class 1 and class 2 sign languages has been centered in their utility for teaching the deaf; but there can be no doubt that the earlier interest in them was connected with two other uses: as technical aids to memory, and as means of secret communication when the vast majority of men were illiterate. There is moreover in treatises on sign languages of these classes as early as the tenth century (MSS attributed to Bede, De computo vel loquela digitorum, Vatican Library, and Biblioteca Nazionale, Turin; Siger 1968:28) a special class of sign languages. These have as denotata numbers or numerals, some of them using one-hand signs, some amphicheric signs, and some both hands and the signer's body. By the

350

WILLIAM C. STOKOE, JR.

r\ n r F /-v

S

NG

/

Z

U

after Graves 1952 :114 Fig. 4

time that these treatises were compiled the writers identified the signs, shown in drawings, later in woodcuts, by conventional Roman numeration, so that one might take the position that the signs denoted the numerals and those the number in computation. However historians of mathematics are convinced that manual numeration long antedated systems of written numerals. Thus Levi Leonard Conant at the end of the last century writes: A most inviting field for research is . . . furnished by the primitive methods of counting and of giving visible expression to the idea of number. Our starting point must, of course, be the sign language, which always precedes intelligible speech; and which is so convenient and so expressive a method of communication that the human family, even in its most developed branches, never wholly lays it aside. (From Newman 1956.) The correct classification of sign languages that denote numbers instead of elements of script may not be possible in the scheme adopted here. The Arabic numerals in current use are ideograms, so that monocheric, amphicheric, and mixed signs to denote them are clearly non-alphabetic; but certainly to mathematicians the denotata are not elements of language but are numbers independent of the words used to denote them in any language. In terms of Figure 1, then, these sign languages may be called classes '4', '5', and '6', but their denotata are not strictly the elements of an ideographic script of any language as language is here being defined.

CLASSIFICATION AND DESCRIPTION OF SIGN LANGUAGES

351

ALFABETO DE SORDO-MUDOS DE ESPANA

Class 3 sign languages seem to be extinct, unless Spanish deaf persons continue to make the sign for 'h' as it is shown in Figure 5, after Pedro Ponce de Leon (1520-1584). Enough of these signs are identical or very similar to those in a modern alphabet like that shown in Figure 2 to support the tradition that the Spanish manual alphabet was that adopted in France in the eighteenth century. Ponce de Leon's sign language has only one sign that requires the use of the signer's face and so is but minimally a class 3 system. Truly class 3 sign languages involve in every sign something other than a hand. Two examples are mentioned in The book of Ballymote, 'Cos-ogham ('leg-ogham') in which the signaller, while seated, used his fingers to imitate inscriptional Ogham with his shin bone serving as the edge against which the nicks were cut. In Srottogham ('nose-ogham') the nose was used in much the same way.' (Graves 1952:113). The denotata in these were the twenty letters of the alphabet Graves gives as: B L F S N H D T C Q M G NG Z R A O U E I. But there is another stratum in sign

352

WILLIAM C. STOKOE, JR.

languages of this class. The list above gives the phonetic values of the Ogham alphabet, but these were denoted by nicks or fingers, one to five on one side of the edge for the first five, on the other side for the second five, slantwise across the edge for M through R, and perpendicularly across for the vowels. While this could conceivably be the direct representation of the morphophones of a Q-Celtic language, its orderliness - five digital elements in four manners of presentation - suggests a substitution code like Morse for the letters of a conventional alphabet. Complete sign languages of classes 4, 5, and 6 may exist, but the only examples the writer has been able to find are subsystems of a sign language used by deaf persons in the Republic of China. When one hand is used alone (class 4) its index finger draws the Chinese written symbol in the air, using the same order of strokes as in conventional calligraphy. This kind of sign presents to the addressee the mirror image of the character as it would appear on a surface normally used for writing, but this rotational transformation does not seem to interfere with understanding. It is not unusual for hearing speakers of Chinese to resort to an indicated drawing of a character when a word as pronounced is not understood by the hearer; but in these cases the speaker traces the character in the palm of his hand so that both he and the hearer see it in normal orientation. Class 5 signs in the same sign language utilize both hands as icons of the 'word' or written character (Chao 1968:102f). When the element of script is a symmetrical figure, the configuration of both hands is identical; e.g. ^ ^ p e i 'north' is signed with the hands back outward, touching at the little finger joints, the thumbs and forefingers showing the main outlines. The hands need not be identical in configuration for other symbols. The first three fingers of the right hand held across the upraised lift index, palmar surfaces together, makes an iconic sign for -^r'rule'. In this sign as in the one made by pressing the right index tip against the second joint of the left index, parts of the hand having no iconic function are ignored. The denotatum of this last sign is itself another kind of sign, of course, and a sentence from Chao makes a pertinent reminder that there are strata of denotation involved in class 4, 5, and 6 sign languages: 'The written symbol X represents as much the spoken word jen as the meaning 'man'; the written form man represents as much the meaning 'human being' as the sound [maen]' (Chao 1 9 6 8 : 1 0 5 ) . Class 6 signs in this Chinese sign language use the face or neck as the central element of the written symbol with the hands representing the sides, e.g. 'south' signed with the hands at the cheeks. The three classes of sign-denotatum relation so far considered in this Chinese sign language do not constitute the whole repertoire of its signs; but instead of considering others used by deaf persons of Taipei, we will move to the next branch of the classification. The first six classes of sign language, it will be remembered, are characterized by having as denotata either alphabetic or non-alphabetic symbols in the script of a particular language. If the join of these two classes is called ' + script', it obviously

353

CLASSIFICATION AND DESCRIPTION OF SIGN LANGUAGES

belongs to a higher class which may be called specific language'. On the same level with ' +script' is '-script', and since ' + specific language' dominates '-script', it is obvious that sign languages of this class must have two sets of denotata, the words (or morphemes) of some language, and the meanings of the words. The signs of this class, like the written symbols of Chinese which Chao discusses, represent as much spoken words as they represent their meanings. The classification so far is shown schematically in Figure 6. +

specific language

+

-script

script

+ alphabetic

monocheric

- alphabetic

amphicheric

mixed

Fig. 6

In the sign languages used in Trappist monasteries (Barakat 1969) may be found signs with - script, + specific language denotata. It would be difficult to determine whether some of these signs had meanings or sounds as denotata were it not for one example Barakat gives. Monks in a Trappist monastery in the United States sign 'milk'-'walk'-'key' to denote Milwaukee. Such puns or canting signs make clear that between sign and denotatum there are the signer's and the addressee's association of the sounds and meanings of English words. In other countries other languages with their own syllables and their own meanings enter into the complex relationship in this kind of sign language. Still more complexity is involved in a second example also from Barakat. For Cincinnati the monks repeat the sign for 'sin' and follow it with signs for 'a' and't'. They use their own amphicheric spelling code, but in this instance the operational denotatum is the letter-name's sound. These few examples are sufficient to show that in this portion of sign language taxonomy the feature + specific language is dominant. Used to make puns and nearpuns, the signs lose some of their utility as fixed code symbols and become counters in an ingenious but sometimes capricious game played by speakers of a specific language.

354

WILLIAM C. STOKOE, JR.

LaMont West, after his doctoral study of American Indian sign language (1960) made field studies of the language of aborigines of Australia and has told the writer in conversation of another sign language which belongs in the + specific, -script class. Taboo prevents aborigine widows from speaking in the first year of widowhood; so they learn the tribal sign language and carry on with signs instead of speech. The old men of the tribe, West says, complain that the widows do not sign right. They must have a basis for judging, and it would be a reasonable inference that the signs are taken as representing the sounds and meanings of the tribe's language and are judged as signs against recollected experience of seeing other signs made somewhat differently. West also adds, however, that the chief complaint is that the signers do not strictly observe silence. A different class of sign language from all those so far considered is one which is anthroposemiotic (Sebeok 1969) but not related to a particular language, i.e. -specific language. The example par excellence of this class is the sign language or languages used by Indians of North America and formerly by the scouts, trappers, and other non-Indians in contact with them. The most interesting feature of this class is precisely its lack of direct use of one language as its integrating infrastructure. The denotata of signs in this class of sign languages are as much the words of several languages as the meanings in which the words converge. The signs of this Indian sign language are monocheric, amphicheric, and mixed in composition. As signs some of them are indexes, pointing to or otherwise contiguous with whatever is denoted, some are icons, showing some kind of similarity to the object, action, or relation denoted; and some may be, with iconic features obscured or disregarded, symbols, arbitrarily associated with a denotatum which common nouns of other languages denote. Moreover, because the denotata of signs in this -specific language class are the meanings of the words of whatever languages its users speak, the signs may be treated as words and assigned to grammatical categories. It would seem that a search for language universals could have been most profitably carried on when it might have been possible to find speakers of an Algonkin-Wakashan language, of a Hokan-Siouan language, and an IndoEuropean language all communicating in a -specific sign language. The grammar of this sign language might have preserved structures that are transformed into widely divergent surface structures in the speakers' native languages. (For a similar search for deep structure n a sign language see Woodward, 1972.) Some such notion as this of a fundamental syntactic structure in Indian sign language may have been behind the common nineteenth century belief, still persisting in some quarters, that sign language is a single semiotic system, that it is pancultural, and that it is a species-specific human attribute. Another fact that may have contributed to the view that there was but one human sign language, universally intelligible, is the similarity seen when comparison is made of parts of American Indian sign language with parts of sign languages used by deaf persons in America and in Europe. But selection of data allows proof of

CLASSIFICATION AND DESCRIPTION OF SIGN LANGUAGES

355

almost anything, and semiotic systems like other systems can be proved congruent only by complete system-wide comparison. In the scheme being presented here Indian sign languages and certain sign languages used by deaf populations are not even contiguous. The node joining sign languages which do not use a specific language as their range of denotata and those which have the feature + specific language may be labeled 0 can be written for suns, but how does one write the case form sun's? When problems like this became pressing, some new inventions were needed. The most important development at this stage is the idea of phonological equivalence of parts of utterances. This is inherent in the structure of language, and many persons are aware of it; but to come upon this idea in developing writing seems not to have been easy. However, the realization that the phonological analysis of words can lead to ways to write parts of them did come about, and apparently came about repeatedly in various parts of the world. So, if one says suns for the plural, and has a symbol to write that plural, one can also write the possessive sun's in the same way. Or if one writes the numeral 2 as two lines, say ||, then one can write to and too in the same way. Such writing is now no longer directly sememographic, but is still entirely logographic. The words being written here are, however, phonemic entities, and so the writing is PHONEMOGRAPHICALLY logographic; perhaps the term PHONEMOLOGOGRAPHIC can be used. Once this step is reached, another discovery can be made. The terms suns and sun's are morphemically different but phonemically the same; by writing them the same we are writing morphemes phonemically; let us call this PHONEMOMORPHEMOGRAPHIC writing. If these developments and discoveries and inventions take place with a language that has mostly short words, and most of these of one syllable, and especially if the syllable structure is a simple one, say predominantly CV and CVC, then the writing system may rather quickly become SYLLABOGRAPHIC. This kind of syllabographic writing is apt to be rather consistent in its representation of phonemes: that is, each sequence of CV or CVC is usually written in the same way every time, regardless of its morphophonic or morphemic structures. Once phonemosyllabographic writing has been achieved, the step to a PHONEMOGRAPHIC system, in which an individual phoneme is represented by a single symbol — at least in theory — is easy. The four main streams of writing that have been mentioned have all produced various kinds of systems. The Sumerian system originally was sememographic and logographic. It developed some phonemomorphemographic aspects, and gave rise to systems that were syllabographic (Akkadian), and eventually to phonemosyllabographic (Hittite, Old Persian) and phonemographic (Ras Shamra) systems. The Chinese system started as a sememologographic system, and remained such, with only a few phonemomorphemographic aspects; it gave rise to phonemosyllabographic systems (Japanese), and also to phonemographic ones (Korean, modern Chinese

384

GEORGE L. TRAGER

'phonetics'). The Egyptian system again was sememologographic, developed some phonemomorphemographic aspects, and very early gave rise to phonemosyllabographic systems (the Egyptian syllabic writings); from it followed the basic Semitic systems, which were phonemographic in a special way, and from these followed a host of phonemographic systems of various kinds. The Mayan system was logographic, with the sememographic aspect less emphasized than in the Old World systems; from it developed some phonemomorphemographic systems, and possibly even some phonemographic practices. Of the aspects of language that could be written (see the beginning of this section), a combination of morphemes and sememes, in words, was the first to be represented by the main known writing systems. Later, representation extended to syllables and separate phonemes; but all the writing systems kept some of the earlier types of representation, and have even in some cases expanded them. It was not until modern times, with the invention of phonetic alphabets, that representation of sounds as such came to be practiced. Some linguists have also devised symbol systems for articulations, but these are not really writing, at least not in the form that most of them take. Much phonetic writing, however, when it uses diacritic marks, is representing articulations. The phonemographic systems that are called ALPHABETS (see below, and 2.2) all represent phonemes, though sometimes only partially, and often not consistently. Moreover, there are many situations in which allophones are represented, or in which the phonemic analysis is mistaken and the symbols are, accordingly, misleading or confusing. Established alphabetic systems usually represent some of the morphophones of a language, and morphemic writing is common. Syntactic operators are sometimes shown, by hyphens, lack of spacing, or the like. Sememes are often represented by certain words, and by other devices. Punctuation, which is mostly modern, attempts to indicate some sememes (capitals for proper nouns, for instance), and marks off units of occurrence and discourse (capitals for the beginnings of sentences, indentation for paragraphs, and the regular punctuation marks). The symbols of logographic writing systems are called by such names as HIEROGLYPHICS (Egyptian, Mayan), GLYPHS (Mayan), CHARACTERS (Chinese); some call many of these kinds of symbols pictographs, and do not believe they are really writing. Systems whose symbols represent mostly syllables are called SYLLABARIES (Akkadian, Japanese kana, Cherokee). Systems whose symbols represent phonemes are called ALPHABETS, and the symbols are called LETTERS. All the known writing systems actually mix several of the basic principles, and there is no purely alphabetic or purely syllabic or purely logographic system in existence. Details will be found in sections 2 to 10 and their subsections. Here we may mention the following characteristics of our own writing system: The numerals we use are logographic, and also sememographic (by position):

WRITING AND WRITING SYSTEMS

385

7 'seven'; 7[3] 'seventy [...]', 7[23] 'seven hundred [...]', etc. When the spelling of strange words or names uses such devices as 'Saigon is pronounced "SIGH-gone"', it is phonemologographic. The capitalization of the first syllable is actually a late phonemographic device, to show the phoneme of primary stress. The spelling of meat 'flesh' to distinguish it from meet 'to encounter' is phonemologographic and morphemographic at the same time, and our phonemographic principles are being strained by our use of two different graphs for the same phonemic entity — ea and ee for /iy/. When we write 'He X-ed out that clause', reading 'X-ed' as /ekst/, we are writing syllabographically; and this particular instance is also phonemosyllabographic, because X is read only as /eks/. In books that are printed in special type, with decorative initials and other devices of that kind, we are representing metalinguistic considerations of various types. Similar considerations also pertain to cases such as the comic-strip 'Pogo', where some of the characters have their speeches written in black-letter type, suggesting the paralanguage — pompous, orotund tones; and even the ordinary remarks have some words or syllables emphasized by heavier lines than others to indicate prosodic and paralinguistic characteristics. Analogies with these comic-strip devices, as well as the balloon coming out of the characters' mouths to contain their speeches, are found in Aztec manuscripts, having been independently invented by the Aztecs long before our modern use of them. Even pictographs are present in our everyday writing. It is customary to indicate profanity by a collection of unpronounceable symbols: 'He went away swearing loudly: "I'll be &§V7oD if I ever talk to that ? ! # * again".' Many readers will say 'blankety-blanked' for the first set of symbols, and 'so-and-so' for the second, but since those readings are not predictable, and are not conventionally stylized, the sequences of symbols can only be of the nature of pictographs. They present a situation, but each observer furnishes his own linguistic material to interpret the symbols. The next section treats the spread of writing in the world. 1.3 The oldest known writing systems, those of the Indus valley and of Sumer, have not survived. The spread and influence of the Indian one are not known; it is probable that the system died out for lack of proper cultural environments, and left no traces, unless it gave rise elsewhere to the idea of writing. The Sumerian system and its descendents flourished in Mesopotamia and the adjacent parts of the Near East — ancient Palestine and Syria, parts of Arabia, and ancient Persia — for some 1500 years or more; it almost certainly influenced Egyptian and Egyptian-derived writing in one way or another. Egyptian writing spread to the Semitic world after 2000 B.C. The Semitic alphabets gave rise to the alphabets of the Mediterranean world after 1000 B.C., and also to the alphabets of India about the same time. The Semitic-derived Greek alphabets gave rise to the Italic alphabets between 800 and 500 B.C.; and in Christian times Greek produced Coptic around 200 A.D., Gothic after 300 A.D., Armenian and Georgian

386

GEORGE L. TRAGER

around 500 A.D., and the Slavic alphabets after 800 A.D. Latin writing conquered the whole of western and central Europe during the first half-millennium of the Christian era, and after the age of discovery — 1500 and on — it spread to America and other parts of the world. It is now the leading alphabetic system, with Russian Cyrillic in second place but far behind it. Mayan writing spread to the Aztecs and other peoples of the valley of Mexico, but was wiped out by the Spanish after 1500. At the present time, the efficient functioning of Western culture is highly dependent on literacy — the knowledge of reading and writing. Statistics on literacy are not very accurate, and in any case are somewhat misleading. In the United States of today, everybody's life is governed by the necessity of being literate : the infant's food and health are taken care of by written prescriptions and directions; and by the time a child learns to talk he is already aware that the marks on paper, on jars of food, or on public buildings are symbols of language and are to be read; long before a child can read himself he asks, 'What does it say?' Note the term 'say' ; he does not ask 'What is written there?', but 'What does it sayT This is functional literacy. Despite their misleading nature, a few statistics on literacy are in order. Usually they are concerned with the ability to read and write of persons 10 years of age and over, and record the percentage of such persons in a nation or some other political entity. The figures do not, and cannot, tell us how well people read and write. Literacy of 95 per cent or more of the population is found in most of the United States (all the states except Arizona, New Mexico, Texas, Arkansas, Louisiana, Mississippi, Alabama, Tennessee, Kentucky, Virginia, North Carolina, South Carolina, Georgia, Florida, Alaska); in Canada except Yukon Territory, the Northwest Territories, Quebec, Labrador and Newfoundland, New Brunswick, Nova Scotia, Prince Edward Island; in all of Australia except the Northern Territory; in New Zealand; and in Europe in Iceland, Great Britain, Ireland, Norway, Sweden, Finland, Denmark, Germany, Belgium, northern and northeastern France, and Switzerland. Literacy from 90 per cent to 95 per cent is found in Texas, Arkansas, Tennessee, Kentucky, Florida, Georgia, Virginia, Quebec, Labrador and Newfoundland, Nova Scotia, New Brunswick, Prince Edward Island; in Europe in Estonia, Latvia, Austria, Hungary, the Netherlands, and parts of France; and in Japan. In the range from 80 per cent to 90 per cent literacy, we find Arizona, New Mexico, Louisiana, Mississippi, Alabama, North Carolina, South Carolina, Northern Baja California; northern Argentina and also in the southern pampas of Argentina, in northern Chile and the area around Santiago, in the area around Lima in Peru; in most of European Russia and in Siberia; in northcentral Spain. From 70 per cent to 80 per cent of the population is literate in Alaska, Yukon, the state of Nuevo León in Mexico, central Argentina, Uruguay, some parts of Chile, Colombia around Bogotá, some parts of northern Spain, eastern White Russia,

WRITING AND WRITING SYSTEMS

387

Rumania, central Italy, central Asian U.S.S.R. (except the Kirghiz and Tajik republics). From 60 per cent to 70 per cent literacy is found in northern Mexico, Southern Baja California, Guyana, the state of Sao Paulo in Brazil, Poland, Lithuania, Greece, Kirghizia, Tajikistan, central Italy, southern Spain. From 50 per cent to 60 per cent literacy is found in Yucatan in Mexico, the Republic of South Africa, Portugal. Between 40 per cent and 50 per cent literacy is found in the Northwest Territories of Canada, in central Mexico, in interior central Brazil, in Siam, and in the Philippines. Most of Africa, the interior of China, and the interior of India are counted as having less than 10 per cent literacy. It will be noted that high literacy is found in a few languages only — English, French, the Scandinavian languages, Finnish, German, Dutch, Estonian, Latvian, Hungarian, Japanese, Spanish, Russian. Somewhat lower is literacy in Italian and Portuguese. The lowest literacy rates are in the new countries of Africa, in South America and south Asia, and in China. Except for China, these are countries in which until recently formal schooling was largely in nonnative languages, and no attempt was made to achieve literacy in the native languages. It is said that learning to read and write Chinese is harder than to achieve literacy in a language written in an alphabet. But in Japan, where the Chinese writing system is complicated by the special adaptations to Japanese, there is nonetheless high literacy. It may be true that some writing systems are harder than others, but it is the educational system and the rest of the culture that really determine the extent of literacy. One final thing must be said: regardless of the percentages of literates, all the countries of the world that are participants in Western culture are, or are becoming, highly literate. Populations with low literacy are mostly outside Western culture — Indians in South America, Africans of all kinds, Arabs in the Near East, and nonMoslem and non-Christian Asians. It is likely that in the next few decades literacy in all the Western-culture countries will be over 90 per cent — in the Latin or the Cyrillic alphabets mostly — and that in Africa and Asia there will be growing literacy in various native languages, but also in the established world languages, especially English. 2. WRITING IN THE ANCIENT NEAR EAST

2.0 The oldest writing system in the Near East seems to be the Sumerian. It is described as CUNEIFORM, meaning 'wedge-shaped' (Latin cuneus 'wedge'), from the shape of the strokes which form the symbols. This shape was a secondary result of the writing materials used; the surface for writing was smoothed soft clay, and the writing instrument was a stylus that had a narrow triangular or rectangular cross section; one end was pressed into the clay to produce the marks. The original pictures from which

388

GEORGE L. TRAGER

the system was elaborated were replaced by combinations of wedge-shaped marks, with the curves replaced by lines at angles to each other. Most of the symbols came to have little or no resemblance to any possible original picture. It should be added that the wedge shape and the finished system were the result of specific developments and inventions long after the idea of writing had begun among the Sumerians. Two examples must suffice: an old symbol for the word for 'ox' is & ; this eventually became 'fish' was at first x > , then It is not certain what the date of the earliest Sumerian writing was. Some place it around 4000 B.C., others as much as nearly a thousand years later (3100 B.C.). Considering the many changes that had to take place before the system reached the stage found in the numerous Sumerian documents after 2500 B.C., the longer period of development seems reasonable. The earliest inscriptions seem to involve labels indicating ownership of objects, and quantities: the symbols were word signs, representing numbers, names, artifacts. The idea of using the signs which derived from pictures as representations of phonemic sequences arose fairly early among the Sumerians, and after 3000 B.C. the writing system was largely syllabic. The symbols represented words, mostly of one syllable (that being the nature of the language), and then syllables of the forms CV, CVC, VC; the device of combining a CV and VC to spell out a CVC syllable was used, and it thus became unnecessary to devise too many CVC signs. There were about 600 word signs in Sumerian, and about 150 syllabic signs. Around 2500 B.C. the Akkadians, speakers of a Semitic language (see References and Notes), came into contact with the Sumerians, and adopted Sumerian writing. They adapted it by turning it into a real syllabary, retaining only a few word signs for names of divinities, persons, and places. Often, Sumerian word signs were spelled out in the syllabary, and most of our knowledge of Sumerian comes from these phonemic indications. At about the same time as the Akkadians, another people, the Elamites, adopted and adapted Sumerian writing. The Elamite language is not known to be related to any other, and we do not have the kinds of clues to reading it that we have in Akkadian because of the latter's relationship to other Semitic languages which are well known. After 2000 B.C., the Hurrians adopted writing from the Akkadians, and by about 1700 B.C., the Hittites, speakers of an Anatolian (Indo-Hittite) language, had learned to write, using Akkadian and Hurrian models. Almost a thousand years later, around 800 B.C., the Urarteans, a people in present-day Armenia, began to use cuneiform writing. About 1400 B.C. there was devised in what is now Syria an alphabet of 32 letters, in cuneiform shape, for writing a Semitic language of the time, Ugaritic. It was also used to write Hurrian. It is possible that this was simply a Semitic alphabet transformed into a cuneiform style, since none of its symbols seem traceable directly to Akkadian ones, but some of them resemble Canaanite letters. This system is known as the Ras Shamra alphabet, having been discovered in a Syrian village of that name.

389

WRITING AND WRITING SYSTEMS

'a M m

¥

TTY

x i

'i

>»-

n

r

i YY

T s

m

%

g v *

M d

g

X Y %

h

w

11 S* ? ?

*Y

z h• wh ~ i

r

S

>*
,o

A.D

g d

(b) c

(a)

(d)

a

S, g d, d

e

3-3

e

3

e, e

w

Í-3

V

3

V,

i

z

I

z

B.H

A-

B

A

B

m

th

0.0

th

—

\ > K

i h i

1

(»)

[i] 0

S-P /V/Á ? G]

Q

h, A

H

1

k; c,g

K

>1

I, /

L

m

m, m

M

n, n

N

M

M

(ffl)

(?)

—

—

—

0

(0)

(*)

—

—

O

p

i n

¿

n

1 r

(?)

q-q

u X ph

(?) r s

s t

P á

T t W u (X) w 0

hh T - v

—

-8

—

—

AA /

n

P. P

1

P> P

—

*

T v-v

P

Q

—

r s; s, z

r,

t,

a

r, r

R

*

S,

5

S

t;

t, d

T

t

V

u; v=u V,

—

—

X

—

—

—

—

—

—

6 —

l

—

—

a

M

f,

/

8

d

Osean and U m b r i a n Alphabets

f, f è

—

—

tu, W],

Y

WRITING A N D WRITING SYSTEMS

419

left-to-right direction may indicate some direct late West Greek influence. The differences of these alphabets from one another are perhaps not surprising in view of the relatively small use of writing at the periods involved. That there are these differences suggests that writing was introduced to the various peoples of ancient Italy by individuals and at different times, and that the introducers were not too expert in the use of writing and were possibly fairly ignorant of the language being written. 3.9 The Greek alphabet was first extended to non-Greek languages in ancient times. In Asia Minor Lycian and Lydian (Anatolian languages — see References and Notes) were written in modified Greek alphabets in the last few centuries before the Christian era. And, of course, the extension of West Greek writing was what brought about the Latin alphabet. After the introduction of Christianity, Greek missionaries began to extend the alphabet in the eastern Mediterranean area and the Near East. In the third century A.D. the Coptic alphabet (see 4.1) was formed for writing the Egyptian of that time — Coptic. It contains the 24 letters of Greek, with seven additional letters taken supposedly from demotic writing: s,f, h /x/, h, j, c, and a letter for the syllable ti. In Coptic, Greek theta, phi, khi, represent /th, ph, kh/. The supposed demotic origin of the seven last letters is quite possibly a myth; the letter forms are such that quite different origins are possible (see below, 4.1). In the fourth century, the Gothic alphabet was based on Greek (see 4.1). In the fifth century, Greek missionaries brought Christianity to the Caucasus, and the Armenian and Georgian alphabets were devised. See 4.2 for the Armenian and Georgian alphabets (see References and Notes for material on the Armenian and Georgian languages). In the eighth and ninth centuries the Slavic alphabets were perfected. See 4.3 for the alphabets. In the early eighteenth century Russian Cyrillic was simplified and modified to look more like Latin letters. In the nineteenth century the several other languages using Cyrillic adopted the Russian forms of the letters, with each language omitting unneeded letters and adding a few new ones. Early in the nineteenth century this activity brought about the Serbocroatian elimination of the Cyrillic preiotacized vowel letters (denoting /ya, yu/), and the introduction of the symbol j from the Latin alphabet. The Russians extended their alphabet to Aleutian, Yakut, and Ossetic during the nineteenth century, with new letters added and others omitted (see References and Notes for these languages). After the Russian revolution in 1917, for about ten years efforts were made to introduce Latin-letter orthographies for all the non-Russian languages of the Soviet Union; then, after 1930, all these new orthographies were replaced by Cyrillic equivalents — Kirghiz, Uzbek, Tajik, Buryat Mongol, and others; in these Cyrillic variants, all the Russian letters are used, so that Russian loanwords remain unchanged except for grammatical endings, but new letters or letters with diacritics are introduced as needed. There are also, however, instances of the use of arbitrary digraphs, such as of (i.e. ob) for ¡aj. In these new practices,

420

GEORGE L. TRAGER

Russian Cyrillic is being used in about the same way as the Latin alphabet has been extended, rather than the way Greek was originally modified (see 4.3, end).

4.

CHRISTIAN ALPHABETS

4.0 As has already been indicated, the spread of Christianity carried with it the spread of writing. Where the missionaries used the Latin alphabet for their own language, they merely adapted it slightly to the sounds of the new language, added an occasional diacritic or digraph, and let it go at that. The world is full of numerous Latin-letter orthographies which are more or less unsuited to the languages for which they are used. But where the missionaries used the Greek alphabet, they usually modified some letters and introduced new ones. The modified Greek alphabets that have survived are Coptic, Gothic, Armenian, Georgian, and the various forms of Cyrillic. Coptic and Gothic are presented in 4.1, Armenian and Georgian in 4.2, and the Slavic alphabets in 4.3. The old Germanic alphabet known as the futhark, though pre-Christian, is at least in part an offshoot of the Greek alphabet, and is given in 4.4. 4.1 When Greek Christianity came to Egypt in the third century, the Greek alphabet was adapted for writing various Egyptian dialects then in use. Collectively these dialects are known as Coptic, and a single modified Greek alphabet is used for them. Since the sixteenth century, the use of Coptic has survived only in the Ethiopic Christian church and the few Coptic Christian churches left in Egypt. The alphabet is the Greek alphabet in majuscule and minuscule form, with seven added letters that are commonly said to be derived from demotic writing (cf. 3.9). The letter forms of the Gothic alphabet (see References and Notes for a discussion of the invention of the alphabet and the origin of the letters) exhibit some difference in style from those of Coptic, but both alphabets are based on the same kind of Greek writing of about the same period. Greek letter forms in the early centuries of the Christian era were written in much the same styles as were those of Latin (capitals, uncial rounded forms, various scripts); Coptic and Gothic are both based on uncial forms. See 4.2 for Armenian and Georgian, based probably on cursive forms of somewhat later date with very different letter shapes, and 4.3 for the two Slavic alphabets — Glagolitic, based on very much elaborated cursive forms, and Cyrillic, based on post-uncial styles and showing many characteristics of letter shapes in common with Coptic and Gothic. The table on p. 421 lists the Coptic and Gothic alphabets, with a column of Greek (or other) letters that are the originals of the other two; numerical values of Gothic are given. A glance at the table shows that the Coptic alphabet is evidently composed of the Greek letters, with some additions at the end. The use of majuscule and minuscule

WRITING AND WRITING SYSTEMS

Coptic

Greek or other originals

•A a Bp r

r

AS Be [see text]

VK, A .

a

BII

b

T P

g d

6 €

e z ê

Z Ç

H h

0 0 [see text]

0 II K

11

Kx AX M (t Nv S Ç [see text] 0 o see text]

Demotic

•-M

th i

K

.u

H h

Oo Iln

k

y m Ml) 9 m Mi)

o

II

[see text] P p [see text] S CT,ç [see text] T T Y u [see text] ©

i [see text]

a. B

r

g d

6

e

P

C

c

m

M

n

H

n

X

G n n

y

•1

{?]

r s t

4> q> X x f

qr

III

Wti

\

«o

g ,

O

f

-

O

O

C n

ï>>

N

4)

((

+1

i>>

N

bî

\

O N

O M

O Q O O © ^ m vo N

M

«Û

«

i l

»n

.H

H

X

ï ÎO>

X

o

ho

3

Sj

' o ' > « Î

M

hSÎ

V

tH A

V

ci

S

o .5P "S 8 «»

M

§ i

¿

^

iP

m

6

a

4)

Ci rQ

S

§

^

«i

£ •8 •S £ ft,

S

f -

s «

a

»•-S >o

Î O C "

«»

>2 wi

l

(O

s

F

«

^

jî

®

e /

s «i

¡s

M fc

^

»

*

U > •

'

=

s K

à P

rae

o

zhani

o c\

pari

o oo

S

o o o o o o o o m \o oo

H J . ^

fJ e ->—. Ë

F » .

o vo

0 0

lazi

Old Name numerical value of letter

nari

Transliteration

GEORGIAN

o m

5?) (modified

o o o i-i CM

S?)

00 ON o o

Nv (modified

o 0 1 i" O nn £ >y £¡

S. 0

(replacing

U

mani

g tP

M (i.

o

AX ?

(modified £)

Letter Transcription TranslitProbable eration Modern Modern Numeror possible ical West East (iOld Greek or value Modern Old Armen. Armen. other originals Letter Armen)

|

o to

IUO

ARMENIAN

GEORGE L. TRAGER

g o

? r -s

T -5

T5

I

o o o o o T-H

>o S

IV.

•! éz

427

WRITING AND WRITING SYSTEMS

8

CUM

¥