Saussure and Sechehaye : Myth and Genius

Saussure and Sechehaye: Myth and Genius

Saussure and Sechehaye: Myth and Genius A Study in the History of Linguistics and the Foundations of Language By

Pieter A.M. Seuren

leiden | boston

The Library of Congress Cataloging-in-Publication Data is available online at http://catalog.loc.gov

Typeface for the Latin, Greek, and Cyrillic scripts: “Brill”. See and download: brill.com/brill-typeface. isbn 978-90-04-37814-8 (hardback) isbn 978-90-04-37815-5 (e-book) Copyright 2018 by Koninklijke Brill NV, Leiden, The Netherlands. Koninklijke Brill NV incorporates the imprints Brill, Brill Hes & De Graaf, Brill Nijhoff, Brill Rodopi, Brill Sense and Hotei Publishing. All rights reserved. No part of this publication may be reproduced, translated, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission from the publisher. Authorization to photocopy items for internal or personal use is granted by Koninklijke Brill NV provided that the appropriate fees are paid directly to The Copyright Clearance Center, 222 Rosewood Drive, Suite 910, Danvers, MA 01923, usa. Fees are subject to change. This book is printed on acid-free paper and produced in a sustainable manner.

Contents Preface  ix List of Figures  xii 1 Introduction  1 2 Who was Ferdinand de Saussure?  9 2.1 Family History and Life  9 2.2 The Cours de linguistique générale  12 2.3 Saussure’s Problem with His Intellectual Environment  17 2.4 Saussure’s Limited Intellectual Outlook and His Implicit Rationalism  24 2.5 The Saussurean Myth  29 2.5.1 The Coming about of the Saussurean Myth  29 2.5.2 Saussure the ‘Father’ of European Structuralism?  36 2.5.3 Saussure in Literature, Art and Philosophy  39 3 The Cours: A Critical Look  43 3.1 Language as a Social Phenomenon  43 3.1.1 The Social Dimension of Language  43 3.1.2 Early French Sociology  46 3.1.3 ‘Völkerpsychology’  48 3.2 Linguistics as the Science of Language, Not of Speech  51 3.2.1 The Tasks of Linguistics  51 3.2.2 The Distinction between ‘Langue’ and ‘Parole’  52 3.2.3 ‘Frequency Linguistics’ Untenable  54 3.2.4 Who Introduced the Distinction between ‘Langue’ and ‘Parole’?  56 3.2.5 The Speech Circuit  58 3.3 The Notion of Syntax and the Notion of Sentence  60 3.3.1 The Notion of Syntax  60 3.3.2 The Notion of Sentence  64 3.4 The Notion of Sign and Its History  69 3.4.1 Saussure’s Notion of Sign  69 3.4.2 The Type-Token Distinction  74 3.4.3 Some History of the Notion of Sign  78 3.4.4 The Arbitrariness of the Linguistic Sign  80 3.4.5 The Linearity of the Signifier  87

vi

Contents

3.5 Differences, Oppositions and ‘Valeurs’  88 3.5.1 Comparison with Chess  89 3.5.2 Only Differences in the Language System?  92 3.6 Synchrony versus Diachrony  94 3.7 Conclusion  101 4 Charles-Albert Sechehaye  103 4.1 Private Life  103 4.2 Scholarly Life: Preliminaries  105 4.2.1 Production and Reception  106 4.2.2 Weaknesses and Prejudices  107 4.2.3 Sechehaye and Saussure: A Paradoxical Relation  110 4.2.4 Sechehaye and Bally: At Cross Purposes  120 4.2.5 Why was Sechehaye Forgotten, or, Rather, Ignored?  125 4.3 Programme et Méthodes of 1908  127 4.3.1 Overall Survey of pmlt  129 4.3.2 Comments on Successive Chapters  133 4.4 The Essai Sur La Structure Logique De La Phrase of 1926  147 4.4.1 Overall Survey of slp  147 4.4.2 Comments on Successive Chapters  149 5 Sechehaye and the Great Subject-Predicate Debate  160 5.1 The Subject-Predicate Debate: How it Arose and Ended up in a Quagmire  160 5.2 How Did Sechehaye Deal with the Subject-Predicate Debate?  167 5.3 Why Discourse-Driven and Fact-Driven Propositions?  174 5.4 Intermezzo on the General Structure of Propositions  175 5.5 An Analytical Synthesis of the Whole Question  178 5.5.1 Definition of the Notion ‘proposition’  178 5.5.2 Anchoring and Keying  179 5.5.3 The Question-Answer Game: Underlying Cleft Constructions  181 5.5.4 Formal Aspects of tcm: The Need for ‘Parameter Theory’ in Grammar  188 5.5.5 The Collapse of Quine’s Argument of the Opacity of Modal Contexts  193

Contents

vii

6 Structuralism, Rationalism and Romanticism in Psychology and Linguistics  197 6.1 What is Structuralism?  197 6.2 Rationalism versus Romanticism: Clarifying the Terms  207 6.3 Human versus Natural Sciences  212 6.4 Reductionism  214 6.5 The Coming about of the Human Sciences  220 6.6 Early Structuralism in Psychology: The Theory of ‘Gestalts’  224 6.7 Early Structuralism in Linguistics  227 6.7.1 The Young Grammarians  227 6.7.2 Who Were, and Are, the Real Structuralists in Linguistics?  235 6.7.3 Romanticist or Nonstructuralist Grammar?  238 6.8 Summary  240 7 Conclusions  242 Bibliography  245 Index  261

Figure 0.1: Ferdinand de Saussure (1857–1913)

Figure 0.2 Charles Bally (1865–1947)

Figure 0.3 Albert Sechehaye (1870–1946)

Preface It is a common experience among historians that half-forgotten lives suddenly turn out to be of great significance.1 This is the experience I had when, a few years ago, while working on Ferdinand de Saussure in connection with the centenary of his death in 1913, I discovered the Swiss linguist Albert Sechehaye, Saussure’s junior by thirteen years, born and bred in Geneva, subordinate staff member of the linguistics department of the University of Geneva, first under Saussure, then under Charles Bally who was Saussure’s successor. When I was a student my professor told me not to read Sechehaye, as that author was guilty of the two mortal sins of ‘psychologism’ and ‘logicism’. As a result, I must confess, I did not read Sechehaye until recently, and was startled when I did. Not only did I finally realise in full that ‘psychologism’ and ‘logicism’ are not at all the mortal sins they were taken to be, but, on the contrary, necessary prerequisites for an adequate study of language. More specifically, I found a linguist with a grandiose vision, who foresaw many of the chief questions and developments of twentieth-century theoretical linguistics, in particular transformational grammar with its formal mappings, or transformations, turning propositional thought structures into sentences of specific languages. This experience was the more remarkable since, ever since my first days as a student, I had had my suspicions as regards Saussure, whose Cours de linguistique générale I have always found highly unsatisfactory in many ways. I was thus faced with a curious contradiction between the current account of ­twentieth-century European linguistics and my experience as a reader. On the one hand, there was the great and famous Saussure whose towering position in the field seemed to me to be unjustified; on the other, there was Sechehaye, who, in my view, should be counted among the most powerful theoretical linguists of the twentieth century but has, in fact, been almost totally forgotten. The questions raised by this contradiction prompted the present study, which is meant as a reappraisal of the figures of Saussure and Sechehaye, not only in relation to each other, but also in relation to the relevant intellectual, cultural and social context of the period and in relation to what is known as the foundations of language and of language study. The result is a book that has at least four threads running through it. There are, first, the well-nigh mythical status that has been conferred upon Saussure, 1 This book is a further development of my article ‘Saussure and his intellectual environment’, published in the journal History of European Ideas (Seuren 2016) as part of a special issue in honour of Hans Aarsleff, who turned 90 in 2015.

x

Preface

and, second, the denial of adequate recognition to his subordinate Sechehaye. Thirdly, there is the fascinating, colourful world of the late nineteenth and the early twentieth centuries, with its new perspectives, discoveries and insights, a world where academics were groping for clarity in areas that were only gradually giving up their secrets, if at all, in the middle of an often violent transition from status-conscious and class-ridden societies to today’s paradoxical global world. And finally, alongside all this, it turned out necessary to delve into questions of linguistic theory in their wider historical, cultural and philosophical context, not only because such questions are always present in my mind but also because discussing them rounds off, and gives structure to, the historical picture and integrates it into our own thinking about language. The present book is, therefore, somewhat unusual in that it combines the history of linguistics, including its wider cultural context, with the theory of language.

…

Of course I could not have written this book in total isolation. I owe much to the knowledge, suggestions and critique of many dear friends and colleagues. The first I must mention is my old friend Pim Levelt, who corrected me on many points and let me benefit from his vast knowledge of psychology and its history. Without his input, this book would have contained many more errors and omissions than it does now. Then there are Brigitte Bauer, Luis Miguel Rojas-Berscia, Corentin Bourdeau, Els Elffers-van Ketel, Ad Foolen, Anne Marguerite Frýba-Reber, Camiel Hamans, Dany Jaspers, Andrew Nevins, Jan Noordegraaf, Gunter Senft and Johan Siebers, who showed their loyalty by looking at various drafts and letting me benefit from their own knowledge in so far as it is relevant but does not overlap with mine. Special thanks are due also to a referee who pointed out weaknesses in the original manuscript and suggested improvements, which I have gratefully incorporated. Last, but certainly not least, I must thank the excellent library staff of the Max Planck Institute for Psycholinguistics at Nijmegen, who indefatigably provided me with the articles and books I never stopped asking for.

…

As a matter of minor importance I should mention that throughout the book, unless specified otherwise, all English translations of terms and quotes from other languages are mine. In some cases I have added the original text, or part of it, either because it has acquired fame in the original or because its complexity or controversiality may make readers wish to see the source text. Giving the

Preface

xi

original alongside the translation in all cases would have taken up too much space, as the number of quotes is very considerable. Where the original text is not given, I simply ask the reader to trust me. Pieter A.M. Seuren

Nijmegen, January 2018

Figures 1 The Château de Vufflens  10 2 De Saussure’s speech circuit (‘circuit de la parole’)  58 3 Saussure’s analysis of the linguistic sign  70 4 Saussure’s view of the relation between form (B) and meaning (A)  70 5 Ogden and Richards’ semiotic triangle  79 6 Dyadic and polyadic constituent structure of the proposition underlying ‘the boy give(s) the girl the book’  176 7 Tesnière-type stemma structure of the proposition underlying ‘the boy give(s) the girl the book’  177 8 The Kanizsa triangles  226 9 Two faces or a goblet?   227

Chapter 1

Introduction The main question dealt with in this book is the following: are the reputations of the early twentieth-century linguists Ferdinand de Saussure and Albert Sechehaye commensurate with their contributions to the field? For both, the answer to that question had to be negative, yet for opposite reasons. In Saussure’s case we must conclude that his contributions fall far short of what his reputation would make one expect, whereas for Sechehaye the opposite is the case: Sechehaye’s contributions are of the highest importance but he has no reputation at all in or outside linguistics, simply because he has been totally ignored. Older linguists only vaguely remember him as one of the editors of Saussure’s world-famous Cours de linguistique générale (henceforth Cours), ­posthumously published in 1916, and younger linguists have never heard of him. In order to argue for my answer I have to establish, first, that, despite his enormous reputation and the ubiquitous declarations that he is ‘the father of European structuralism’—declarations that are not, as a rule, supported by ­actual facts or analyses, only by vague appeals to nebulous notions—Saussure, when looked at with an uncluttered eye, did not contribute much to presentday linguistic thought or practice on either side of the Atlantic. In fact, I argue that he became the object of mythification.1 Then, I have to show that, by contrast, Sechehaye’s work was, though largely ignored, of the highest possible quality for his day and, in fact, foreshadowed theoretical developments that have played an important innovative role in linguistic thought since the 1950s on both sides of the Atlantic. As in all academic business, the conclusions reached must be based on solid arguments, based on facts or on reasonable hypotheses that are open to discussion on grounds of evidence and reason, not on grounds of prejudice one way or the other, that is, without regard for personal sensitivities. In some cases, my 1 Percival (1977: 395) speaks of a “Saussurean myth”. In Percival (2011: 237), he speaks of “the ‘Saussure cult’, that is, the wave of theoretical partisanship in Saussure’s favour that swept European linguistics from the 1930s onwards”. Hymes (quoted from Allan 2009: 262) writes (1983: 375): “Just as the comparative historical approach has its mythical founder in Sir William Jones, so the structural approach has had its mythical founder in Ferdinand de Saussure. The great respect one must have for both men does not bar inquiry into the actual part they played: in both cases, it was, most dramatically, in the symbolic use made posthumously of each”.

© koninklijke brill nv, leiden, ���8 | doi:10.1163/9789004378155_002

2

Chapter 1

arguments are based on textual critique, in the manner of good, old-fashioned philology. Their quality is commensurate with the amount of relevant and reliable information provided and the intuitive conviction carried by it. I do not consider it my task, in the present study, to trace the possible historical or personal causes of what I have called the ‘Saussurean myth’. Tracing the origins of myths is, in most cases, a question of speculation. They arise from the complex fabric of social reality in ways that mostly escape our investigative powers. This also applies to the Saussurean myth. A full analysis of the factors that contributed to it would lead into a maze of social and, perhaps embarrassing, personal details that may arouse curiosity but are less relevant to the questions that we, as linguists, are concerned with. It would, in any case, require a volume on its own. All I can, and need, do, in the present context, is establish the mere fact that, for a whole century, there has been a myth—one that has wreaked havoc. As regards the causes of Sechehaye’s oblivion, I extensively point at certain factors that probably played a role. As I see it, Sechehaye was pushed aside partly because he was snowed under by the then incipient Saussurean myth and partly also because his texts were too difficult for the majority of linguists of his day. But there was, in my view, a further and more profound reason for, or cause of, Sechehaye’s disappearance from the public scene: Sechehaye rested his entire theory on the notion of proposition as it had been presented by Wundt. And this notion of proposition was signally out of favour with the linguists of the day, who wanted their discipline to be autonomous and even independent of any other discipline, notably of logic, psychology and sociology. Due to historical accident, the notion of proposition was strongly associated, in all sorts of not always clearly defined ways, with traditional logic—which was considered sufficient reason to exorcise any linguist who worked with that notion or used that term. Sechehaye, however, did not cede to this pressure, as he was one of the very few who saw that the proposition is the central element connecting logic, psychology and linguistics. This question, which touches on wider issues of the history of ideas, permeates this entire study. The present book is thus partly concerned with the myth that has been ­woven around Saussure. But what is a myth? In a general sense, a myth is a complex of awe-inspired beliefs and attitudes, usually connected with a narrative that is, in most cases, entirely or partially fictional but accepted as true, and which has become part of a tradition within a community. Such narratives are meant either to explain phenomena in the world, as with creation myths, or to bestow greatness upon a community or a person, as with myths of heroism, where heroic figures are introduced that are either thought up or distantly

Introduction

3

related to historically real figures. Depending on circumstances, mythified ­figures are often, implicitly or explicitly, endowed with supernatural status and powers, so that they can, if appeased, ward off danger or bring on calamities if angered one way or another. Religion, myth and magic are thus of a piece, though they can and do occur separately from each other. Myths are extremely hard to dispel. Once established, their roots go deep, way into the emotional soil of the human soul. They also serve multiple purposes in a community, as they are a powerful ingredient for boosting and solidifying feelings of group identity, a factor of major importance in human communities. They are also often connected with magic, thus providing an illusory protection from calamities or capability to inflict harm on others, or they are taken as a justification of (usually false) beliefs, allowing the believer to circumvent rational argument. Normally, as long as a community has not been able to develop more reliable means of warding off or causing calamities, or has not outgrown the need for hero worship that gives its members the necessary sense of group identity, myths are an indispensable ingredient for social peace and further development. Their elimination should, therefore, be managed with extreme caution in cases where it is felt that they have become harmful or superfluous. Academic communities, however, are an exception to this rule. There, myths are to be eradicated with much less circumspection, without too much regard for personal sensitivities or group coherence—though a certain amount of tact, understanding and diplomacy will, in general, do no harm. In virtue of the very values that make academia what it is, academic communities must be strong enough to derive their identity from real qualities, based on rational grounds, not from myths or from feelings of group solidarity. This requires discipline and a willingness to cede to rational argument. Most of the time, academic myths, such as the one we are concerned with here, are devoid of religious or magical content, but they do, nevertheless, tend to create heroes taken to be gifted with special intellectual powers and thus, in an implicit and never properly explained sense, to stand above the rest of humanity. In extreme cases, such figures are even allowed by their followers to transgress the boundaries of rational thought and intellectual critique as these have become established. They then become superhuman messengers of privileged insights or revelations. That status, however, is never fully spelled out, as it would be self-destructive to do that. This is not to say that individuals with special intellectual powers do not occur. On the contrary, groundbreaking new ideas, sometimes conceptions of great generality, do, from time to time, arise in the minds of certain especially gifted and inspired intellectuals. We do not know the details of this

4

Chapter 1

­ rocess: the genesis of original ideas has so far escaped the powers of cognip tive psychology and is thus, on the whole, opaque if not mysterious—though it is known that pioneering ideas often spring from intelligent observation of hitherto unknown or refractory, and thus all too often neglected, facts or data. (It will be clear, by the way, that the as yet mysterious origin of revolutionary new thoughts or ideas is in itself, depending on circumstances, an invitation to myth-making.) In any case, inspiration does exist. It is a fact that, on rare o­ ccasions, inspired, even visionary, perspectives do arise in the minds of certain prodigies, often in the form of a first principle that works itself out in different ways over the years. Inspired ideas may, in a sense, be ‘greater’ than their authors, in that they turn out, as time passes, to be of a wider generality and greater explanatory power than the originators had been able to see in the first place. This sort of phenomenon is well-known and may be considered, in an intuitive sense, evidence of real cognitive depth or intellectual vision. Individuals that have this special gift must be respected and protected, and must, after their death, be given an adequate place in the history of their subject. Let them become legendary but they must not become the object of myth-making, although, of course, that still happens all too often. When we are faced with what we suspect of being an academic myth, the first crucial test is whether or not the ideas put forth by the mythified individual satisfy standard criteria of testability, reproducibility and empirical falsification. If they do not, there still is the possibility that something potentially great is going on, but then the ideas in question must provide a more encompassing and more explanatory framework for the rational explanation of empirical facts than previously accepted views do, whereby it is always wise to maintain a generous margin of benevolent doubt. But when these ideas merely evoke associative or suggestive networks of vague and often metaphorical ideas, and certainly when they show evidence of insufficient familiarity with the relevant subject matter, the repository of these ideas may be a great artist or ideological or religious leader of whatever kind, he or she cannot be reckoned to be a great scientist. Mere inspiration or suggestion may be intriguing in the early stages of theoretical exploration, but if that is not embedded in thorough knowledge of the subject matter and placed in the context of a new, more powerful explanatory framework, the final product will simply not do. It will be obvious that theories or persons that merely, through their pretensions or their demeanour, instill awe or admiration but fail to satisfy rational criteria of the kinds mentioned above, are prototypical foci of myth-making. Academic myth-making inevitably leads to a lowering of standards and criteria. Admirers become followers,

Introduction

5

no longer persuaded by argument but sustained by a socially shared belief and a feeling of belonging to a privileged group operating under the hero’s banner. Often also, the object of academic myth-making is identified with the institution he or she is attached to and thus becomes a standard-bearer for it. In such cases, the mythification tends to be actively supported by the institution itself. Intellectual submission, empty praise and false attribution to the mythified person of possibly important thoughts or discoveries then become rife. When an academic myth has been established, lack of agreement means lack of loyalty, a form of treason. Dissent is not countered with rational argument but met with anger, scorn, defamation and even punitive measures. When the object of the myth is criticised, refuted, attacked or, in the worst case, unmasked, then, typically, the followers, rather than feeling challenged at an intellectual level, primarily feel hurt in their feelings and often even insulted. And if it is in their power to do so, they will see to it that the dissenter becomes academically isolated or, even better, loses his or her academic post. The consequences for the discipline concerned are grave. Alternative, socalled ‘nonstandard’, paths of research are blocked by academic authorities, funding agencies and editors on grounds of reputation and status, no matter how well-founded such ‘unorthodox’ research may be. As a result, an entire field may suffer from a detrimental narrowing of focus or, worse, continue to proceed on false premises, for an extended period of time. Once the myth is there, the mythified person becomes an idol or icon one is obliged to refer and to defer to and whose views, theories or, in some cases, fantasies are only to be commented on, not to be called into doubt, let alone rejected. The Saussurean myth is a prime instance of academic myth-making. As is shown in Section 2.1, Saussure gained a reputation of being especially gifted at the early age of twenty-one, when he published an epoch-making book on the Indo-European sound system. This reputation stuck, despite a conspicuous lack of overt creative intellectual activity during the rest of his life, when he published very little but kept being seen by students and colleagues as a man who was immersed in deep theoretical thought about the nature of human language. Despite, or perhaps because of, this lack of visible signs of creative activity, colleagues and students at the University of Geneva, where he taught from 1891 till his death in 1913, began to idolise him, also, one suspects, because he appears to have been an excellent and dedicated teacher—though this reputation was based exclusively on his classes on established Indo-European linguistics, not on the teachings that underlay his posthumous book Cours de linguistique générale. Willy-nilly—but maybe just a tad more willy than nilly— Saussure thus became an icon for the Geneva school of linguistics: detracting

6

Chapter 1

from this icon came close to high treason, not only in the University of Geneva but also in the town, and soon also the country as a whole. In later years, this status spread over large parts of Europe. The threat of sanctions of whatever nature in case of dissent seems to play only a minor role in the Saussurean myth nowadays, but it is, in any case, still strong enough to make linguists, especially in Europe, feel that it is a natural obligation to refer to Saussure as the ‘father of European structuralism’, standardly attributing to him a number of basic notions and distinctions without further comment or critique, and obviously without being aware (a) that most of these attributions stem from long-established belief rather than from historical reality and (b) that even if the attributions were historically correct, the notions and distinctions involved have little to do with structuralism. Most of the views Saussure has become famous for, though taken to be original and creative by later generations, were actually taken over by him from contemporary authors, never with proper attribution or accompanied by a proper discussion of the existing literature. And to the extent that these views were original, they were vague and even confused, and, importantly, remaining without serious academic follow-up. The result has indeed been an endless series of exegetic writings,2 serving a variety of purposes not only inside but also outside linguistics, as became clear when, from the mid-twentieth century onward, Saussure was appropriated as an icon in phenomenology, art critique and other forms of less analytically minded, ‘postmodern’, semi-academic or para-academic activity. It is argued in the present book that most of the fundamental questions that were discussed by linguists in Saussure’s day and age still await satisfactory answers in our day and are thus still wide open, whereas the progress made in linguistics over the past century bears no relation to Saussure’s own thoughts in so far as they were original. To the extent that it does bear a relation to some of the ideas discussed in the Cours, it must be said that these existed independently of the Cours and were making their way into linguistic thinking anyway. Saussure’s own contribution to linguistics was, in actual fact, minimal or ­nonexistent—apart from some terminology. In sum, whatever progress was made in linguistics during the twentieth century was due to others. 2 An early, perhaps the earliest, instance of exegesis is the well-known article by the American philosopher and structuralist linguist Rulon S. Wells, in Word 1947, which starts off by saying that the Cours “needs exegesis” (which it subsequently provides in abundance). Though of very high quality in many respects and not eschewing critique, this article is totally ahistorical: no contemporary literature is mentioned and no attempt is made to situate Saussure in his historical context. It is also far too indulgent vis-à-vis the obvious defects of the Cours, which are noted but glossed over or reasoned away too easily.

Introduction

7

But not only that. The Saussurean myth has overshadowed, if not eclipsed, other contributions that contained greater promise and might have saved modern linguistics from costly mistakes and wrong turns. This point is of s­ pecial relevance, as it is argued in the present book that Albert Sechehaye, Saussure’s less attention-drawing colleague in the Geneva linguistics department, and of lesser rank than him, was developing a groundbreaking synthesis that would indeed have saved twentieth-century linguistics from many a costly wrong turn. But Sechehaye has been ignored till the present day,3 due, at least in part, to the dominant position of Saussure, who either failed to grasp ­Sechehaye’s powerful vision of language or feared it. While the Chapters 2 to 4 of the present book deal with Saussure and ­Sechehaye in a narrow sense, the subsequent chapters widen the scope of the enquiry, gradually taking the reader from these two historical individuals to the wider theoretical issues that occupied especially Sechehaye’s mind. (­Saussure no longer plays a role here, as his mind was not occupied much by these ­issues—which in itself is a further confirmation of the thesis defended in the first part of the book, namely that Saussure was, intellectually speaking, an insignificant figure, whereas Sechehaye, though largely forgotten nowadays, developed insights and perspectives that are still, or again, of the highest relevance.) This second part of the book takes the reader on a few excursions farther afield into the wider historical context that both Saussure and Sechehaye were part of. Chapter 5 discusses in detail the great subject-predicate d­ ebate—left unmentioned by Saussure but given a prominent place by S­ echehaye—which dominated linguistic theorising between roughly 1850 and 1930 and then vanished unresolved into thin air, owing in part to the terminological and notional confusion pervading the entire debate, but also in large part to the fact that linguistics, while becoming ‘structuralist’, had lost sight of the central role of the notion of proposition. At the end of that chapter, glimpses are revealed of both the complexity and the central importance of the old but now largely forgotten subject-predicate debate, in the context of a newly to be developed linguistic parameter theory, which shows that definite descriptions in natural language are often not just referring phrases, in the usual sense of that term, but denote the value of a parameter—a conclusion that promises to have farreaching consequences for the theory of language. Chapter 6 then looks at the origins of structuralism in the human sciences, in particular psychology and linguistics, in the context of the overarching opposition between (modern) rationalism and romanticism. Here it is shown 3 Levelt (2013) is an exception in that this author rightly emphasises the importance of Sechehaye’s ideas and places them in a historical perspective (Levelt 2013: 218–222).

8

Chapter 1

that linguistic structuralism is in principle a rationalist enterprise which, although the source of great advances in the study of mind and language, at the same time blocked further insight in that it parted with the notion of proposition. Linguistic structuralism is seen as a perhaps necessary but in any case transitional stage in the development of an adequate and up-to-date theory of language: necessary because linguistic utterances are structured in certain well-defined ways and these structures must be analysed, described and, if possible, explained; transitional because understanding structures is only one of the many phases required for a full understanding of that unendingly complex phenomenon we call human language.

Chapter 2

Who was Ferdinand de Saussure? 2.1

Family History and Life1

Ferdinand Mongin de Saussure (26 November 1857–22 February 1913) was born into a patrician family belonging to the upper crust of the Republic and Canton of Geneva in Switzerland. The family history takes one back to the fifteenth-century, to the noble French family de Saulxures (‘of the willow trees’), in Lorraine, the region of Nancy in north-eastern France. As the family embraced Calvinism at the earliest possible opportunity, the Saulxures were driven from Catholic Lorraine and migrated to Calvinist Geneva in 1550, where they changed their name into de Saussure (Joseph 2012: 4–6). By the time Ferdinand saw the light of this world, the family was known for a strong tradition of scientific activity, going back to Ferdinand’s famous greatgrandfather Horace-Bénédict de Saussure (1740–1799), a naturalist, explorer and inventor, practising geology, botany, chemistry and the like, especially with regard to the Alpine mountains near Geneva. According to Joseph (2012: 16), he set the intellectual standard for the family for generations to come. Saussure’s grandfather Alphonse Jean François de Saussure (1770–1853) did not distinguish himself in any way, but his grandfather’s brother Nicolas-Théodore de Saussure (1767–1845) was, though perhaps less famous than his great-grandfather, a notable chemist and plant-physiologist. His father, Henri Louis Frédéric de Saussure (1829–1905), was an adventurous but respected mineralogist and entomologist. Ferdinand’s mother, Countess Louise de Pourtalès (1837–1906), came from an immensely rich protestant banking family belonging to European, partly royal, aristocracy (Joseph 2012: 47–51). Since the seventeenth century, the Saussure family has owned the enormous Château de Vufflens (see Figure 1), built around the year 1400, with its very extensive grounds, situated not far west of Lausanne close to the northern border of Lake Geneva, summer residence of Ferdinand de Saussure’s family, the place where Saussure’s widow retired after Ferdinand’s death and the venue of the Second International Congress of Linguists in August 1931. In the city of Geneva, Saussure possessed a centrally located renaissance-style palatial mansion. 1 For Saussure’s biographical details I rely for the most part on De Mauro’s biographical notes in his edition of the Cours (Saussure 1995: 319–358) and on Joseph’s 2012 extensive, 780 pages long, biography of Saussure.

© koninklijke brill nv, leiden, ���8 | doi:10.1163/9789004378155_003

10

Figure 1

Chapter 2

The Château de Vufflens, owned by the Saussure family, who used it as a summer residence

Ferdinand was the eldest of nine children, among whom two younger ­brothers, with both of whom he was in close contact, Léopold (1866–1925), a sinologist, astronomer and French Naval Officer, and René (1868–1943), a mathematician and promoter of the international auxiliary languages E ­ speranto and Interlingua. Having finished his secondary school, where he excelled in classical languages and already started to discern etymological relations, Ferdinand, ­almost 18, entered the University of Geneva in 1875. Despite pressure from his father to focus on the natural sciences, he attended courses in a large variety of subjects covering mostly arts subjects, with an emphasis on historical-­comparative linguistics, the natural sciences making out but a tiny portion of his total ­spectrum (Joseph 2012: 168–177). During that year, Saussure decided to devote himself entirely to historical-comparative linguistics. In 1876, his father, having given in to Ferdinand’s preference, sent him to Leipzig, known worldwide as an important centre of linguistic studies. There he followed courses given by young linguists, such as August Leskien (1840–1916) and Karl Brugmann (1849–1919), who would soon stand up violently against their teachers, in particular Georg Curtius (1820–1885), professor of Greek at Leipzig University, and August Schleicher (1821–1868), professor of comparative linguistics at the university of nearby Jena. In 1879, Leskien, Brugmann and others, such as ­Hermann ­Osthoff (1847–1909), who taught at Heidelberg University, or B ­ erthold Delbrück

Who was Ferdinand de Saussure?

11

(1842–1922), who was Schleicher’s colleague at Jena, founded the new school of the Young Grammarians, which would revolutionise comparative-historical linguistics during the last few decades of the nineteenth century.2 (The views developed by the Young Grammarians are discussed in greater detail in Section 6.7.1 below.) For Saussure, the Young Grammarians’ doctrine became the basis of his linguistic thinking, though he soon began to feel that human language raised questions that were not answered by it (see his Cours de linguistique générale: 19, 117–118). At the age of twenty-one, while still a PhD student at Leipzig, he produced a book of 326 pages, his Mémoire sur le système primitif des voyelles dans les langues indo-européennes of late 1878 or early 1879 (I take 1879 as the year of publication), in which he proposed a daring and ingenious hypothesis about the original vowel system in Indo-European roots, which was dramatically confirmed by the Polish linguist Jerzy Kuryłowicz (1927), after the decipherment of Hittite and its recognition as an Indo-European language in 1915 by the Czech linguist Bedřich Hrozny´.3 This book made him famous overnight. Immediately after the Mémoire, he presented his PhD-thesis to the philosophy faculty of Leipzig University on the Sanskrit genitive absolute construction (Saussure 1881). This work, consisting of 95 pages of text, was far less epochal than the Mémoire. The Parisian linguist and Saussure’s friend Antoine Meillet (1866–1936) characterized it as follows (Meillet 1938: 177): The doctoral dissertation presented at Leipzig in February 1880 makes a distinct contrast with the Mémoire. Whereas the topic of the Mémoire impresses one by its vastness […], the dissertation On the use of the absolute genitive in Sanskrit, published in Geneva with the date of 1881, is a simple technical article. In 1881 Saussure was appointed as an instructor at the École Pratique des Hautes Études in Paris, founded in 1868 but already a highly esteemed institution for higher education, to teach Gothic and Old High-German. He stayed 2 I use the term Young Grammarians for what are commonly known as Neogrammarians. The term Neogrammarians is, though in general use, a misnomer as it fails to take into account the fact that the original German moniker Junggrammatiker, coined in 1878, was a jocular takeoff on the names used in those days for angry young nationalists, such as the Jungdeutschen, Junghellenen or, especially, Jungtürken, called Young Germans, Young Hellenes and Young Turks in English. That being so, Junggrammatiker should be translated as Young Grammarians, not as Neogrammarians. See Seuren (1998: 92–93 ), Morpurgo Davies (2004: 16) for the full story. 3 For a detailed description of the actual issues involved in Saussure’s Mémoire of 1879, see Morpurgo Davies (2004).

12

Chapter 2

there till 1891, when he accepted the new chair of Indo-European linguistics and Sanskrit at the university of his home town Geneva. In 1906 his chair was extended to cover general linguistics as well, the reason being that the incumbent professor of linguistics and philology, the Rabbi Joseph Wertheimer ­(1833–1908)—who left no trace in linguistics—had resigned and that Saussure had a reputation for being engaged in developing ideas about the general nature of language, trying to lay the foundations for a truly ‘scientific’ and ‘autonomous’ discipline of general linguistics. One should realise that, until Saussure’s appearance in 1891, the University of Geneva was just about the dullest place in the world to study linguistics and expectations were high to begin with: the appointment of the already famous Ferdinand de Saussure, who was only ­thirty-four years old at the time, would have been seen as a veritable scoop. The incumbent but merely nominal professor of general linguistics, Joseph Wertheimer, deployed no activities worth mentioning in this regard—something the University would have been painfully aware of (Joseph 2012: 172, 376, 489). As the years went by, Saussure’s health deteriorated steadily as a result, it seems, of a complex of pulmonary and respiratory problems (Joseph 2012: 625– 626). After 1910 his teaching became ever more sparse, Sechehaye standing in for general linguistics and Bally for the remaining courses of comparative linguistics, replacing the Sanskrit course with courses on comparative phonetics and morphology of Greek and Latin (Frýba-Reber 1994: 196). Saussure died in the evening of Saturday, February 22nd, 1913, at the age of 55. 2.2

The Cours de linguistique générale4

The new extension, in 1906, of his teaching duties made it necessary for Saussure to give a course on the general theory of language, which he did, though very much against his wish, as will become clear in a moment. He gave a total of three courses, in 1907 (the first half of the year), 1908–09 and 1910–11 (Godel 1957: 29), which formed the basis of the posthumous publication, in 1916, of the now world-famous Cours de linguistique générale. This book was, though published under his name, not written by him but based (besides one or two other, minor, sources) on lecture notes taken by, in total, six of his students, who never numbered more than a handful for each course.5 The text was collated by Albert Sechehaye (1870–1946) and Charles Bally (1865–1947), with the 4 I have used the fourth edition of the Cours, published by Payot, Paris, in 1949. 5 During the last course, given in 1910–11, one of the students was Sechehaye’s wife Marguerite (Godel 1957: 15; Frýba-Reber (1995/1996: 132), who made a career as a psychoanalyst.

Who was Ferdinand de Saussure?

13

help of the student Albert Riedlinger (1882–1978), who, other than Bally and Sechehaye, had actually followed the first two courses.6 Bally and Sechehaye were both, like Saussure himself, Swiss and both Saussure’s colleagues, though in subordinate positions, in the linguistics department of the University of Geneva: at the time of Saussure’s death in 1913, both had been ‘privatdozent’ for a number of years, which meant that they were expected to give courses but did not receive a salary. At that time they were already established linguists. Bally succeeded Saussure in the Geneva chair of general linguistics and comparative Indo-European studies in October 1913. Sechehaye remained privatdozent until 1929, when he was given an extraordinary professorship (see Section 4.1 for more biographical details on Sechehaye). The fact that Saussure did not himself write the text of the Cours is important. His two editors, Bally and Sechehaye, apologise profusely in their Préface to the first edition, saying that they had to try to “arrive at the thought that we only had the echoes of” (p. 8) and that the maître “would perhaps not have authorised the publication of these pages” (p. 11). Godel (1966: 493) states that Saussure was definitely averse to giving these lecture courses, as he felt depressed, uncertain and inadequate: Towards the end of 1906, he was appointed to offer a course in general linguistics at the University of Geneva, where he had been teaching Sanskrit and comparative philology for fifteen years. A friend of his told me that this new appointment simply terrified him: he did not feel up to the task, and had no desire to wrestle with the problems once more. However, he undertook what he believed to be his duty. One may infer that Saussure, had he still been alive, would indeed not have approved of the publication of the Cours. It should, therefore, be borne in mind throughout the following pages that when we speak of Saussure in the context of the Cours, what is meant is the reconstitution of his words and views as we know them through the text of the Cours, which became “a kind of ideal vulgate […] of Saussurianism” (Lepschy 1962: 69). Standards of authorship thus apply only partially. At the same time it must be realized that it was on the basis of this ‘vulgate’, not Saussure’s own words, thoughts or notes revealed to S­ echehaye himself followed Saussure’s lectures on Sanskrit and comparative philology from 1891 to 1893 (Godel 1957: 24). 6 To have an idea of the care with which Sechehaye and Bally set about the extremely difficult task of collating the students’ notes and other sources for the Cours, see Godel (1957: 95–121, 251), Frýba-Reber (1995/1996: 132–133), Engler (2004: 48–51), Sofia (2015: xxvii–xlii).

14

Chapter 2

the world at large in the second half of the twentieth century that Saussure gained his great fame and is now widely said and thought to have had a great influence on twentieth-century linguistic theory and to have been the starting point of European structuralist linguistics. Since the mid-twentieth century, an enormous amount of zeal and energy has gone into the effort to reconstruct the ‘real’ Saussure, the man behind the published version of the Cours, as well as the actual contents of the courses as they were really given, mostly on the basis of Saussure’s private notes, of which a great many have been preserved, and of notes taken by the handful of students attending his courses and their personal communications to the extent that they were still in a position to provide them. Some of the authors concerned have criticised the editors, Bally and Sechehaye, for taking too many liberties with what must have been Saussure’s real teachings, but one looks in vain for the obvious question of the relevance of those painstaking attempts at reconstruction. For even if it were to be found that the Cours, as published in 1916, differs substantially from what Saussure actually taught, it is still that book, the ‘vulgate’, not the supposed actual teaching, that forms the textual basis of the aura of genius that has arisen around Saussure’s person: the search for the ‘real’ Saussure came too late to have had any influence on the course of linguistic theory over the past century. Yet apparently, only few of the authors in question seem to have posed these questions. Present-day Saussure specialists show a strong tendency to idealise Saussure, whose greatness is taken for granted to begin with, no matter the outcome of the minute reconstructions. A striking example is Tullio De Mauro’s (1932–2017) 1972 critical edition of the Cours (Saussure 1972), a highly ­informative publication where, however, Saussure is lauded and defended at all costs, even when there is no evidence or the evidence shows the opposite. E ­ ndless lists of names are given of linguists whose works are supposedly shaped by Saussurean ideas, but hardly ever is any actual evidence given or is any precise analysis provided that would show what such influence would have consisted in or to what it would have led in present-day linguistic theory. Saussure’s greatness is axiomatic. But the evidence adduced in De Mauro’s study consists in nothing more than the names of the scores of authors who all ­mention Saussure as the ‘father of structuralism’ but never show why he should be regarded as such, the very notion of ‘structuralism’, like so many other crucial notions, most often being left in a limbo. Even if Saussure is not mentioned, his spirit is, without further clarification or illustration, taken to be present as a fruitful source of ideas that only need further elaboration. Referring to André Martinet’s (1908–1999) Éléments de ­linguistique générale of 1960, he states (De Mauro in Saussure 1972: 369): “The

Who was Ferdinand de Saussure?

15

Éléments de linguistique générale, even though Saussure is not mentioned in it, is the most radically Saussurean manual of general linguistics”. Yet, although there is little doubt that Saussure is lurking in the background throughout this book, it is at least misleading to state that it “is the most radically Saussurean manual of general linguistics”, since Martinet’s covert allusions to Saussurean ideas are above all polemical, constituting as many attempts at disengaging himself from them while promoting his own theory of “double articulation”. In fact, where Martinet does seem to have undergone some Saussurean influence, it is in regard of Saussure’s nebulous fabric of purely negative oppositions discussed in Section 3.5 (see also note 29 in Chapter 3). De Mauro even detects Saussurean influences where Saussure is not even remotely in the picture (De Mauro in Saussure 1972: 369): “One may, by way of example, cite L. Tesnière […], whose Éléments de syntaxe structurale is more inspired by Saussurean ideas than appears from explicit references”. But if one looks for evidence in Tesnière’s important and influential 1959 book, one does not find any. Tesnière (1959) is mainly about formal syntax (his dependency grammar)—a topic that was totally alien to Saussure—and also otherwise seems entirely devoid of Saussurean ideas (despite the fact that he was taught by Meillet). False attributions are made, such as the following (De Mauro in Saussure 1972: 377): “It suffices to mention the fact that Baudouin and Kruszewski themselves owe the notion of phoneme to Saussure”, which, as any phonologist will know, is false. Baudouin de Courtenay and Kruszewski may have borrowed the term ‘phoneme’ from Saussure, who used it in his Mémoire, but certainly not the notion: Saussure never had a coherent notion of ‘phoneme’—a term which he uses in inconsistent ways in the Cours—whereas Kruszewski, followed by Baudouin de Courtenay, in essence gave the term the meaning it has in modern phonology as a systematic type-level unit covering a collection of phonetically differing token-level sounds (see Dresher 2011). But let us revert to the Cours as published in 1916. There, the three courses are not distinguished from each other, as the editors decided to amalgamate the students’ notes they had at their disposal from all three courses into one continuous text. Although it is known that Saussure’s views were subject to change over the four or five years concerned—changes that are reflected in notional and terminological inconsistencies—the Saussurean tradition has never systematically distinguished between views ‘before’ or ‘after’ any given change, the way that, for example, the Wittgenstein tradition makes a strict distinction between the ‘early’ and the ‘late’ Wittgenstein, the break being placed around 1930. On the contrary, there are plenty of testimonies, not least in Saussure’s own notes and unpublished drafts, that Saussure was himself in

16

Chapter 2

a perennial state of self-doubt and insecurity about the ideas he propounded during the three lecture courses, leading to a recurring pattern of hesitation and fluctuation. That pattern remained the same throughout the period he gave his lectures. Saussure had not always been so insecure. On the contrary, as a young man he had been both cocky and intellectually precocious. At the age of eighteen he became a member of the Société de Linguistique de Paris, soon sending in six papers to be published in the Société’s Mémoires of 1877, all on single etymologies and other specific matters of comparative-historical linguistics, and becoming the Mémoires’ editor a few years later (De Mauro in Saussure 1995: 339; Joseph 2012: 199–200). Then, as we have seen, he published his Mémoire in 1879, at the age of twenty-one, while still a PhD- student. This memorable book was soon followed by his less memorable doctoral dissertation of 1881. After these achievements, however, there was a dramatic decrease in the volume and nature of his publications. Until his death in 1913 he only published a modest number of mostly very short articles, all on topics of etymological or, occasionally, dialectological detail, and a few book reviews. On the ideas expounded in the Cours he never published a word, although it appears that he had started developing these ideas from 1881 onward—probably under the influence of the then universally known and admired Parisian philosopher, psychologist, historian and journalist Hippolyte Taine (1828–1893) (see Seuren 2016). This story suggests that Saussure was a highly competent Indo-Europeanist, who had shown flashes of genius during his younger years but lost his brilliance once his youth was over. The man who gave the lectures published in the Cours is very different from the brilliant youngster: no longer the flamboyant young scholar breaching intellectual barriers in historical-comparative linguistics, but a prematurely ageing troubled soul who tried hard but ultimately felt incapable of breaching the barriers blocking a deeper insight into the nature of language (De Mauro in Saussure 1972: 357–358). In fact, such was his state of mind that, between 1906 and 1909, he seriously engaged in occult speculations about hidden messages from the subconscious or the subliminal in Greek, Vedic, Old-Germanic and Latin poetry, based on occurrences or repetitions of syllables and sounds (Starobinski 1971/1979). As this aspect of his work is irrelevant to linguistic theory, it is ignored in the present book.7 Rather than see Saussure as a prophet overwhelmed by crushing thoughts of great profundity that he found himself unable to express adequately in 7 Andrew Nevins made me aware that Roman Jakobson, whose baroque mind allowed for much that sterner academics deem beyond the pale, had great sympathy for Saussure’s ‘anagrammatic’ investigations (see Jakobson 1971c).

Who was Ferdinand de Saussure?

17

­ uman language, one should see him as a man who tried and kept trying, but h in the end discovered that he was not up to the task he had set himself. De Mauro (Saussure 1972: ii) describes him in his last years as living a solitary life: “The image of the lonely man is certainly justified by his increasing isolation, by his prolonged scientific silence, by certain features of his private life, by the sadness that shrouds his letters and his last meetings with his students”. De Mauro suggests further that this private solitude was of a piece with his aloofness with regard to contemporary linguistic thinking and to society as a whole (Saussure 1972: iii): The contrast between isolation and participation dominates not only Saussure’s private life, his human destiny. We find it again, at a deeper level, in his relations with the linguistics and the thought of his day, and of ours. The following section takes a closer look at this aspect of Saussure’s life, person and academic activity. 2.3

Saussure’s Problem with His Intellectual Environment

One disturbing feature of the Cours, and of Saussure generally, has contributed considerably to the overall lack of clarity regarding Saussure’s views and his position in the history of linguistics: the fact that he hardly ever mentions any contemporary or older authors that he either follows or opposes. According to Joseph (2003: 365), this was due to the character of the Cours, which was not “prepared […] for print”. Earlier, Joseph had justified this absence of references by saying (1996: 119): “He was after all teaching courses, not writing a book”. But surely, mentioning relevant authors is a requirement not only in academic publications but also, and equally, in serious academic teaching. Moreover, this ignoring of other authors was already evident in his Mémoire of 1879. Yet, as we know from Joseph’s (2012) Saussure monograph, he was extremely anxious that others should not purloin his own ideas without attribution. The overall evidence shows that Saussure had a problem relating to his intellectual environment. The problem is already evident in his younger years, when not mentioning relevant work by others got him into trouble with his Mémoire in 1879. Ambitious young but already established scholars, such as Karl Brugmann or Hermann Osthoff, who had been teaching Saussure in Leipzig, found many of their own ideas in the Mémoire without attribution and felt slighted to the point of privately accusing him of plagiarism ­(Morpurgo ­Davies

18

Chapter 2

2004: 23–25; Joseph 2012: 244–245). No charges were brought but the taint stuck: the Mémoire was mostly ignored in Germany, though praised to excess in France (Joseph 2012: 342–344). It does not look as if the palpable chauvinism that existed on both sides suffices by itself to explain this difference in reception. De Mauro, who, as has been shown, is patently partial to Saussure, relates (Saussure 1972: 326–327) how, early on in the piece, considerable friction arose between Saussure and his Young Grammarian teachers, notably Brugmann and Osthoff—themselves not much older than Saussure—­regarding questions of authorship, and how, as a result, Saussure’s Mémoire was met with “a conspiracy of silence” (ib.: 327). I am in no position to adjudicate in this matter, but it is clear that Saussure’s relations with his closest fellow linguists were highly problematic. In general, Saussure seems to have suffered from insufficient interaction with his intellectual environment. In the Cours, he does not enter into a dialogue with the existing linguistic literature, nor does he define his position vis-à-vis prevailing ideas. Tracing the intellectual influences he was no doubt subject to, and also spotting the strains of thought that he rejected and did not wish to be associated with, thus becomes a matter of conjecture and textual critique. The only two linguistic theorists mentioned more than once are Saussure’s fellow-Genevan Adolphe Pictet (1799–1875), a close friend and relative of his family, whom he refers to but whose ideas he rejected (Seuren 2016), and the American linguist William Dwight Whitney (1827–1894), who was likewise trying to create a more ‘scientific’ foundation for the theory of language but whose work, in actual fact, showed little if any overlap with Saussure’s.8 Saussure had met Whitney in Berlin in 1879 while preparing his PhD, and greatly admired the gracious American gentleman (Joseph 2012: 254). Apart from the handful of names in the brief historical survey of Chapter 1, Whitney is one of the very few authors actually named in the Cours, on p. 26 and on p. 110, where he is praised for his views on the social character of languages and for his notion of the arbitrariness of the linguistic sign, though also criticised for not going far enough and not seeing that (Cours: 110) “this arbitrary character radically separates language from all other institutions” (which, by the way, is not quite correct, as rituals are also often either arbitrary or based on historical 8 Sechehaye (1917: 9) justifies Saussure’s predilection for Whitney by pointing at the fact that the latter aimed at replacing the old, romanticist ways of looking at language by a more scientific, that is, rationalist, account. Other than that, however, Whitney and Saussure have little in common. Roman Jakobson’s contention (Jakobson 1971b) that Whitney had a profound influence on Saussure is not based on any analysis of either Saussure’s or Whitney’s ideas.

Who was Ferdinand de Saussure?

19

accident). There is, however, no analytical discussion of Whitney’s views and theories.9 Saussure (2002: 203–222) is a draft for an article on Whitney (which was never elaborated let alone published). In these almost twenty pages, Whitney is praised to excess but one is never told why exactly Whitney deserved that praise. But—this cannot be denied—woe upon you if you were out of favour: then Saussure’s (private) pen was dipped in bile. Poor August Schleicher, already the whipping boy of the Young Grammarians, is treated in the following way (Saussure 2002: 205; also: Godel 1954: 59): […] when this science [i.e. linguistics; pams] finally seems to overcome its torpor, the result is Schleicher’s laughable essay that collapses under its own ridiculousness. Such has been Schleicher’s prestige based on nothing but his attempt to say something of a general nature on human language that it looks as if still now he is a figure sans pareil in the history of linguistic studies, so much so that one sees linguists take on comically grave airs whenever this great figure is mentioned. […] Yet for all we can see, it is clear that one is faced with the most complete mediocrity, which does not exclude pretensions. No doubt Saussure is here referring to Schleicher’s indeed ludicrous attempt at reconstructing a short Indo-European text, his fable ‘Avis akvasas ka’ or ‘The sheep and the horses’ (Schleicher 1868), which was met with Homeric laughter by his colleagues and students (see Seuren 1998: 86). But one is struck both by the shrill tone and by the lack of argument. The same is found in his discussion of Max Müller, as one sees in the next paragraph, and also in the way he penned down his comments on Sechehaye (1908a), quoted in note 13 below. The picture sometimes painted of Saussure as a kind, gentle and urbane person does not match what is found in his private 9 Godel (1966) opens with a flourishing tribute to Whitney, which, however, contains nothing of substance. All Godel says, as regards the relation between Saussure and Whitney, is: Whitney’s ideas undoubtedly stimulated de Saussure. He never ceased to feel i­ ndebted to the American scholar and many years later, when he offered courses in general linguistics at the University of Geneva, he did not fail to mention Whitney’s name with praise and to discuss his ideas. Yet in (Godel 1957: 144) he writes, with regard to the distinction between langue and parole: “The contrast with Whitney is striking”. The same is true for De Mauro’s critical edition of the Cours (Saussure 1972), where Whitney figures prominently throughout as one of the main influences on Saussure, without it becoming clear, however, what precisely this influence consisted in. In Saussure’s own private notes, Whitney is often mentioned and praised, but his ideas are never discussed to any depth.

20

Chapter 2

notes, where he stands out as either an uncritical admirer or an abrasive and spiteful critic. In fact, there is no proper analytical discussion of any theoretical linguist in any of Saussure’s writings. In the Cours, Saussure is silent, for example, on Friedrich Max Müller (1823–1900), the famous German-born Oxford linguist, indologist and founder of comparative religious studies, who is only mentioned by Saussure in what was probably meant as a draft (but, again, never published) critique of Sechehaye (1908a), in the following words (Saussure 2002: 260): If Max Müller’s linguistically pretentious vulgarisations cannot be declared shameful, it is because all science in Max Müller’s day restricted itself more or less effectively to running from one of the thousand possible linguistic topics to another, from an anecdotal point of view, and that without any kind of serious notion of, or aspiration towards, a constitution of the scientific bases of linguistics. Like others, Müller also wished to see linguistics as a science, but in a d­ ifferent way from Whitney. While Whitney saw linguistics mostly as a historical science, Müller envisaged a scientific linguistics in which language is like, or perhaps even actually is, a biological organ that grows and evolves in its ecological environment the way biological organs do—thereby anticipating what Saussure would later call the synchronic study of language. Whitney and Müller were engaged in a bitter polemic, known all over the world of linguistics (Seuren 1998: 87–88), but the controversy is never mentioned by Saussure, even though it impinges directly on the ideas Saussure himself was developing. Nor is the German-Austrian linguist Hugo Schuchardt (1842–1927) ever mentioned in the Cours or in the private notes, though Schuchardt too strove for a naturalist scientific foundation of language studies. Schuchardt’s chief concern was with the ecology of actual language use, with the ways a language ‘lives’ in a community, with the sociological aspects of language and language use, with the actual ‘on the ground’ mechanisms of language change, with the birth and growth of Creole languages and other such earthy topics. ­Together with Johannes Schmidt (1843–1901) he developed the famous wave model (Schmidt 1872) of phonological change as a sociological process, each individual change of the local vernacular spreading over a well-defined territory.10 10

Johannes Schmidt does occur in the Cours, on the pages 277 and 287, not in the context of general linguistic theory but only in connection with his wave model of sound change, which Saussure halfheartedly, and without further comment, accepts.

Who was Ferdinand de Saussure?

21

Schuchardt (1885) was critical of the Young Grammarians as he found their view of ‘exceptionless sound laws’ too mechanical: close observation of linguistic reality shows that the famous sound laws are not like physical laws but more like regularities that allow for exceptions according to specific circumstances or preferences among the speech community (see Section 6.7.1). Since the social aspect of language was an explicit element in Saussure’s own thinking (see Section 3.1), one would have expected some reference to Schuchardt, but there is none. Nor does Saussure seem to have been affected in any way by Schuchardt’s work, no matter how well known and influential it was. The name of Wilhelm Wundt (1832–1920) does not occur in the Cours, even though Wundt’s views on language were extensively discussed by all leading linguists, many of whom were critical to different degrees.11 In Sechehaye (1908a), for example, an entire chapter is devoted to a critical discussion of Wundt’s views, apart from numerous mentions in the rest of the book (see Section 4.3). Although Saussure was thoroughly familiar with Albert Sechehaye’s 1908 book, which was written under his eyes, so to speak (see below), Sechehaye’s name, like Wundt’s, does not occur in the Cours. While Sechehaye’s view of Wundt’s linguistic ideas was, on the whole, positive though laden with critique, the German linguists of the day, such as Hermann Paul, Philipp Wegener, Berthold Delbrück, massively turned against Wundt. Saussure, however, remained silent. This is the more remarkable since Wundt came to Leipzig University in 1875 and set up his psychological laboratory in 1879, provoking great public interest, which means that Saussure must at the very least have heard and read about Wundt. He may even have met him or have been in one or two of his massive audiences. Finally, there is no mention at all, whether in the Cours or in Saussure’s private notes, of the great subject-predicate debate that was central to linguistic theorising between, say, 1850 and 1930 (Elffers-van Ketel 1991). This debate was about the fact that, in many cases, the semantic or psychological 11

Levelt writes (2013: 204): “There was no way for Wundt’s linguistic contemporaries to ignore this massive psychological ‘intrusion’ into their discipline. Wundt’s theories not only became obligatory references in just about any linguistic publication […] but were also the subject of extensive treatment and controversy”. Graffi writes in similar terms (2001: 66): “Sechehaye‘s (1908) programmatic work assumed a psychologistic perspective as fundamental: references to Wundt’s work, therefore, became compulsory”. The suggestion that the ‘compulsion’ to refer to Wundt was mainly due to Sechehaye’s 1908 book is, in my view, historically incorrect, since Wundt was being discussed and referred to all over by that time, as confirmed by Levelt. Yet the statement that “references to Wundt’s work became compulsory”, if taken as referring to social pressure within the discipline, is entirely correct, which makes it the more remarkable that Saussure did not comply, not even with a negative comment.

22

Chapter 2

s­ ubject-predicate division of a sentence does not correspond to its grammatical subject-predicate division—an extremely important and complex question (extensively discussed by Sechehaye, as is shown in Chapter 4), which has so far not found a satisfactory solution. In Chapter 5 below, this tangled question is discussed in detail and an opening to a solution is proposed. It does not look, therefore, as if Saussure was much concerned with what was going on elsewhere in linguistics or in the wider sphere of the philosophical and science-theoretic discussions of his day, despite his personal contacts with his colleague Adrien Naville, a leading authority on the philosophy and methodology of science. Yet we do, on occasion, hit on a passage that is identifiable as a reaction to existing contemporary literature, even though the author is not mentioned by name. Thus we read Cours: 33–34): One may thus think of a science that studies the life of signs as part of social life; it would be part of social psychology, and thus of general psychology; we will call it semiology […]. It is the psychologist’s task to determine the exact place of semiology [i.e., the theory of signs; see below; pams]; the task of the linguist is to define what makes a language a special system in the totality of semiological facts. The question will be taken up again below. Here we merely emphasise one thing: if, for the first time, we have been able to assign to linguistics a place among the sciences, it is because we have linked up linguistics with semiology. [italics mine; pams] Here Saussure appears to allude to Albert Sechehaye’s Programme et méthodes de la linguistique théorique of 1908, whose overarching aim was to assign to linguistics a place in the totality of sciences—a question widely discussed in the social sciences of the day, which aspired to be ‘scientific’ in the sense of the natural sciences. Saussure had, at Sechehaye’s request, read the complete manuscript before publication, as the book’s dedication to Saussure makes clear.12 The phrase “for the first time” is thus to be seen as a claim to the effect that 12

The dedication runs as follows (Sechehaye 1908a: v): To my teacher [‘maître’], Monsieur Ferdinand de Saussure. It is you who has awoken in me the interest I have in the general problems of linguistics, and it is from you that I received many of the principles that have enlightened my road in these investigations. Although my thoughts have since taken a personal direction, my ambition has been, as I wrote each of these pages, to merit your approval. You have kindly agreed to peruse the work when it was finished, and to encourage me in my enterprise by the benevolent support that I have always found in your company. I dare to hope that I have, at least to some extent, achieved my goal and it gives me pleasure to be able to dedicate this study to you as an homage and as a token of my profound gratitude. Ch.-Albert Sechehaye.

Who was Ferdinand de Saussure?

23

what Saussure presents here is a further specification of Sechehaye’s classification of theoretical linguistics as a separate and partly autonomous subdiscipline of psychology. Yet, as has been said, Sechehaye’s book is not mentioned, even though Saussure had made rather acerbic observations in a draft review of it that was never finished (Saussure 2002: 258–261; Percival 1977: 392; see also Section 4.2.3).13 By contrast, Sechehaye comments extensively on Saussure’s claim for semiology to occupy a special place in social psychology. In (Sechehaye 1917: 13) he writes: “It is through it [i.e. semiology; pams] that linguistics finds its proper place in the natural connections in the totality of the sciences”, continuing on the theme for the length of four pages. In this connection, it is relevant to mention that Saussure’s distinction between the synchronic and the diachronic study of language, put forward in his lectures, was never published by him, though he had been musing on it since at least 1894 (Saussure 2002: 11, 207). Sechehaye, however, beat him to it by publishing that distinction in his 1908 book, where it is a main theme. The first published account, therefore, was in Sechehaye (1908a)—a fact that Saussure could neither deny nor reverse.14 Saussure did introduce a new term 13

14

This is what Saussure wrote in what probably was a draft for a review of Sechehaye (1908a) (Saussure 2002: 260–261): Having correctly criticised Wundt for not having seen the grammar problem, Sechehaye himself manages not to arrive at an adequate idea of the same. For the only adequate idea of it would present the grammatical fact in its own right, showing how it differs from all other psychological or, for that matter, logical facts. The harder the author tries to tear down what to him seems to be an illegitimate barrier between the thought form and the thought itself, the further he seems to get away from his own purpose, which would be to delimit the area of expression and to conceive of its laws, not in so far as they are shared with our psychism in general, but, on the contrary, in so far as they are specific and absolutely unique in the phenomenon of language. Hardly, it would seem, the sort of thing one would expect from someone writing about a friendly and admiring younger colleague in the department, the more so because Saussure, while teaching about synchrony and diachrony, relied heavily on the relevant parts in Sechehaye (1908a) (Wunderli 1976, 1981). Saussure’s comments are, moreover, ­misguided, as Sechehaye repeatedly sets off the purely (psycho)logical aspects of sentence structure studied by Wundt from the specifically linguistic aspects studied by the grammarian-cum-theoretical linguist. The Swiss-Austrian philosopher of language Anton Marty (1847–1914), student of Franz Brentano (1838–1917), also wrote extensively on the distinction between the ‘genetic’ and the ‘descriptive’ way of studying language (Marty 1950; 1908: 52–70). His treatment of this distinction, however, is very unlike what we find in Sechehaye and Saussure, in that Marty only holds forth on meaning, not on form, and does not get beyond philosophical generalities (often opposing Wundt). It is unlikely that Marty had any influence on either Sechehaye or Saussure, which makes the fact that the same question was being discussed simultaneously in Geneva and in Prague (where Marty was a philosophy professor) significant: the issue was ‘in the air’, so to speak.

24

Chapter 2

pair: whereas Sechehaye—following Auguste Comte, who introduced the distinction ‘static’ versus ‘dynamic’ in volume 4 of his Philosophie Positive— speaks of ‘static’ and ‘evolutive’, Saussure speaks of ‘synchronic’ and ‘diachronic’, respectively. He introduces his new pair of terms in (Cours: 117), saying that he will continue to use the old pair ‘static’ and ‘evolutive’ alongside the new, which he did, but neither Sechehaye nor Comte nor any other author is mentioned. 2.4

Saussure’s Limited Intellectual Outlook and His Implicit Rationalism

What is known about Saussure is consistent with the notion that his general intellectual, including his historical, outlook, was rather limited and that his overall cultural and historical ideas largely reflected the common deposit accumulated in the French- speaking intellectual world of his day, without much of a personal input. This would mean that, though tuned to the intellectual landscape he lived in, he was relatively uninformed about the ways it had come about and about the age-old general philosophical issues involved. That said, it is clear that, in the then current competition between the rationalists and the so-called romanticists, Saussure sided with the former, not the latter (for an explicit discussion of rationalism and romanticism, see Section 6.2).15 Yet he seems to have had a greater natural affinity with specific q­ uestions of etymology than with the wider landscape of overarching philosophical and theoretical issues in the humanities. His knowledge of history was also rather limited, if we may go by the woefully inadequate and chauvinistic Chapter 1 of the Cours, “Coup d’œil sur l’histoire de la linguistique” (‘Brief survey of the history of linguistics’). In any case, if Saussure was really not, or hardly, in touch with the great philosophical and cultural issues of his day, this must have formed a serious handicap in his attempts at a general theory of language, a new theoretical linguistics. Perhaps due to the tradition in his science-oriented family, Saussure’s notion of what constitutes proper academic work was closer to the natural sciences than to the humanities. One of the things that attracted him in the new school of the Young Grammarians was the fact that these enterprising young 15

John Joseph (Joseph 2003, 2004, 2012: 150, 217–220, 647) seems inclined to let Saussure be classified, not as one who attempted to set up a formal account of language systems, but rather as a romantically inspired linguist, in a very wide sense, with an eye for less formal but ecologically important aspects of natural human language. Yet one looks in vain, in the Cours, for any original elaboration of the ecological aspects of language, whereas the allusions to and attempts at a strictly formal treatment of the language system are legion.

Who was Ferdinand de Saussure?

25

linguists had formulated a programme to study the Indo-European languages from the point of view of the regularities in their historical, more especially their phonological, development. They did pay lipservice to the general aspects of language, including the geographical and social settings in which languages are used, but they rarely said anything specific about those aspects. In accordance with prevailing views of the day, they saw a specific language as a social institution deposited as a law-driven system in the minds of its speakers, and not, or no longer, as a reified ‘organism’ (as argued for by Max Müller mentioned above), forming a disembodied separate sphere of reality, a “fourth realm of nature” (Cours: 17), next to the physical, the mental and the social. Yet, Saussure felt (Cours: 19), “no matter how great the services rendered by this school [i.e., the Young Grammarians; pams], one cannot say that it has shed light on the whole complex of questions; today the basic problems of general linguistics are still awaiting a solution”. It was precisely these “basic problems”, whatever they might be, that he aimed at addressing in the Cours in what he saw as a scientific way. It is unclear, however, how far his knowledge of the theory and general principles of science actually went. He never studied any science or philosophy of science, but, as has been said, he grew up in a strongly science-oriented family—a fact that seems to have limited rather than expanded his horizon, in that he followed Comte’s then standard idea of science as collection and classification of data rather than the more modern notion of science as hypothetico-deductive theory building as described by his friend and colleague Adrien Naville mentioned above. In any case, the philosophy and methodology of science were not in the family book, despite the enormous amount of literature, in France and elsewhere, about the nature of science, the classification of the sciences and of scientific theories, the relation between the natural and the human sciences, etc. He knew of, but apparently very little about, logic. In his private notes (Saussure 2002: 101), we find a passage about subject and predicate as a logical (Aristotelian) distinction, starting thus: In the proposition everything is reduced to subject and predicate and, thirdly, I understand, to conjunction (except in vocatives). But the subject and the predicate have nothing to do with the ‘parts of speech’, which are distinguished on different grounds: (a) The subject can be a noun or a pronoun or an adjective or a number name as immediate evidence. But also a verb (infinitive); see below why the infinitive does not change the nature of the verb. (b) All that can also be predicate. (c) Conjunction can be ‘conjunction’ or adverb.

26

Chapter 2

This passage (which sounds like an implicit reference to, in particular, the pages 121 and 122 of Sechehaye’s Programme et méthodes de la linguistique théorique of 1908, where Sechehaye is at grips with Saussure’s points a, b and c, but draws the opposite conclusion) suggests that Saussure, though clinging to a rationalist way of thinking, took no interest in logic whatsoever and had no knowledge of it, dismissing it as irrelevant to the study of language. The opposite is the case with Sechehaye, who (with Wundt) saw the subject- predicate structure, as known from traditional logic, as the schematic form of underlying propositional meaning representations and (unlike Wundt) insisted that the ‘grammar problem’ consists in the formulation of the complex rule system transforming these underlying semantic structures into well-formed surface structures. It is easy to see that Sechehaye’s attitude towards logic is a great deal more fruitful and insightful than Saussure’s ignoring it. The term logic does not occur in the Cours. It does occur, though very rarely, in his notes (Klippi 2010: 370–371), and what transpires is that Saussure doesn’t want to touch logic with a ten-foot pole. Klippi quotes from Engler’s 1968 edition of the Cours: “Static linguistics may claim many things that are normally subsumed under general linguistics. […] To this generalisation belongs even what is called general grammar [‘la grammaire générale’], which notably contains the points where linguistics borders on logic”. This is the closest Saussure ever came to mentioning logic. Saussure’s metaphorical use of the term ‘algebra’ in the Cours, may suggest that he had heard, perhaps through his brother René, of the intrusions of algebra into geometry as carried out by mathematicians like Jakob Bernoulli (1655–1705) or Leonhard Euler (1707–1783) during the eighteenth century, or of the nineteenth-century algebraic approach to mathematics and logic initiated by George Boole (1815–1864) in Ireland, though there is no evidence that he was at all familiar with Boolean algebra or its later application in classical set theory. The second wave of formalisation of logic and mathematics, initiated during the 1880s by men like Gottlob Frege (1848–1925) and Giuseppe Peano (1858–1932), and given a canonical foundation by Alfred Whitehead (1861– 1947) and Bertrand Russell (1872–1970) in their monumental Principia Mathematica (1910–1913), seems to have passed him by completely. In this respect, Saussure was a child of his immediate environment, which, on the whole, took little notice of these new developments. A closely related and conspicuous gap is the absence of any serious reference to the Grammaire générale et raisonnée, or Port-Royal grammar, published in 1660 by the Port- Royal members Claude Lancelot and Antoine Arnauld and based not only on ancient and medieval philosophy but also, to a considerable extent, on the Spanish linguist Franciscus Sanctius or Sánchez (1523–1600),

Who was Ferdinand de Saussure?

27

the last great innovator in linguistic theory before 1600 (see Breva-Claramonte 1983). The Port-Royal grammar is mentioned only once in the Cours, on p. 118, where we read (Cours: 118): On the other hand, how did those who studied language before the foundation of linguistic studies—that is, the ‘grammarians’, inspired by the traditional methods—proceed? It is curious to note that their point of view, on the question that concerns us here, is absolutely irreproachable. Their works show us clearly that they want to describe language states; their programme is strictly synchronic. Thus, the grammar of Port Royal tries to describe the state of the French language as it was under ­Louis xiv and to determine its values [‘valeurs’]. For that, it does not need the language of the Middle Ages; it faithfully follows the horizontal axis […] without ever deviating from it; this method is thus correct, which does not mean that its application is perfect. Traditional grammar fails to take into account large parts of the language, such as word formation; it is normative and believes that it has to dictate rules instead of recording the facts; it lacks overall views; often it does not even distinguish the written from the spoken word. This passage makes one suspect that Saussure never had a serious look at the Port-Royal grammar, even though he possessed a copy of it (Gambarara 1972). He appears to have been unaware of the immediately obvious fact that the Port-Royal grammar stands outside the French tradition of descriptive grammar writing initiated by the great Claude Favre de Vaugelas (1585–1650) (who is occasionally referred to for his good observations) and further developed by grammarians like Régnier Desmarais (1632–1713), Claude Buffier ­(1661–1737), César Chesnau Dumarsais (1676–1756), Gabriel Girard (1677–1748) or Nicolas Beauzée (1717–1789) and many others. Arnaud and Lancelot were not “­grammarians” but linguistic theoreticians (Allan 2010: 183–184; 194).16 Their “grammar” was not a grammar of French of whatever period but an extended argument, dressed up in a pedagogical garb (as all works on language since the sixteenth century had been) meant to show (a) that human languages, with Ancient Greek, Latin, Hebrew and French as typical examples, all follow 16

See also Robins (1967: 123): The rationalist grammars were in several ways the successors of the medieval scholastic grammars. […] were writers of universal grammars […] expounding a general theory of grammar through the medium of such languages as Latin and French. […] hey did attempt to reveal the unity of grammar underlying the separate grammars of different languages in their role of communicating thought […].

28

Chapter 2

the ontological and logical categorisations set up by the great philosophers of the past, and (b) that sentences and clauses of human languages reflect ­logico-psychological judgments—our propositions—the latter point being reactivated by Sechehaye. The Port-Royal grammar was the farewell call of the old, traditional rationalism in linguistics that had been fed by the ancient and medieval traditions but went underground until the early nineteenth century. It owes its fame and influence during that period mainly to the fact that it was the last available source for a tradition that lay dormant under the weight of enlightenment and romanticism but was nevertheless felt by some to be indispensable as a philosophical foundation for descriptive grammar writing. Beauzée’s masterful analysis of the French tense system, for example (see Seuren 1998: 73–74), would have been impossible without a philosophical foundation and shows clear traces of the chapters on ‘the verb’ in the Grammaire générale.17 17

It is a remarkable, and so far unobserved, fact that, from the late seventeenth until the early nineteenth century, little or no progress was achieved, nor was much interest shown, in linguistic circles, regarding questions to do with the relations between language, mind and world, while the relation between mind and world dominated discussions in philosophy, both on the European Continent and in Britain, culminating in Immanuel Kant’s monumental Kritik der reinen Vernunft (‘Critique of Pure Reason’) of 1781. In these eighteenth-century philosophical discussions, language played no or only a marginal role, the ‘philosophy of language’ of that period being largely restricted to vacuous speculations on the origin of language. Then, as from roughly the 1840s, language is drawn back into the mind-world-language equation, as the complex relations between these three reference points, and, more generally, the foundations of language, are again of central intellectual interest, in part through the works of Wilhelm von Humboldt and John Stuart Mill (each in their own very different ways). Meanwhile, however, the more mundane business of actual grammar writing went on with ever greater intensity (see, for example, TiekenBoon van Ostade 2008, Auer 2009, Seuren 1998: 63–74). A similar theoretical ‘calm before the storm’, combined with brisk and lively activity at a more practical level, is observed in mathematics, where the eighteenth century is characterised by huge advances in arithmetical calculus and its practical applications in mechanics. One thinks of figures like Jakob Bernoulli or Leonhard Euler, mentioned above, but also of the mathematical activities of Leibniz and Newton and their disciples, or of Christian Goldbach (1690–1764), Johann Heinrich Lambert (1728–1777), Joseph Louis Lagrange (1736–1813), Pierre-Simon Laplace (1749–1827), Adrien-Marie Legendre (1752–1833), Jean-Baptiste Joseph Fourier (1768–1830) and others, many of whom lent their names to mathematical theorems, techniques or conjectures. (I must thank Martin Mattmüller of the Bernoulli-Euler Zentrum of Basel University for initiating me into the world of eighteenth-century mathematics.) But the study of the foundations of mathematics, including the relation between mathematical reality and the human mind, did not start until after 1840, with George Boole and, again, John Stuart Mill. What, if any, will have been the common causal background to these parallel developments is a question worthy of further investigation. It looks as if the combined growth of enlightenment and romanticism during the eighteenth century

Who was Ferdinand de Saussure?

29

Given all the facts, one may reasonably surmise that the urge felt by Saussure to construct the framework for a new ‘scientific’ theory of language found its origin in his family-bred affinity with the natural sciences: his aim was to give linguistics a new status as a natural science or at least a discipline as ‘scientific’ as the work done by his great- grandfather, the famous naturalist, and other forebears. But he failed to realize that creating new and scientifically adequate foundations for the general study of language requires an adequate command of philosophical and historical issues of great complexity, overarching generality and often opaque profundity—a command he did not have. In short, Saussure was seriously out of his depth. 2.5

The Saussurean Myth

2.5.1 The Coming about of the Saussurean Myth It is difficult to say exactly when the Saussurean myth started to shape up. There had been, of course, amazement, and also admiration, when his Mémoire was published in 1879, but it does not look as if one can speak of a myth already developing at that time—certainly not in Germany, where his Mémoire had provoked mostly adverse reactions, as we have seen. The myth-making must have started during his ten years, from 1881 till 1891, at the École Pratique des Hautes Études in Paris, where he taught Gothic and Old High-German, apparently, if the reports are to be believed, with great clarity and didactic ability. Some mythification was, in any case, detectable by the time he took up his chair at the University of Geneva in 1891. It was there, in his hometown, that his reputation took on the proportions of a fully-fledged myth, based, first, on his teaching of Indo-European linguistics and then, unaccountably, shifting to his teaching of what was seen as general linguistic theory.18

18

affected philosophy and the more practical sciences in different ways, allowing them to link up only after a certain level of maturity had been achieved. Percival (1977: 392–393) points at the remarkable fact that Saussure’s reputation changed from being based on his Mémoire to being based on the Cours: At his death in 1913, Saussure was known exclusively [outside Geneva; pams] as an Indo-Europeanist. […] Barely a quarter of a century later, however, Saussure had an entirely different kind of reputation: he had come to be regarded as a brilliant theoretical innovator, as the prototypical general linguist who challenged the hegemony of historical Indo-European studies over the rest of the field—in brief, as the very founder of modern linguistics, or at least of the movement known as structuralism. He was now known chiefly as the author of the Cours. […] This bizarre reversal in Saussure’s reputation poses difficulties for the historian of 20th-century linguistics. It is clear, first of all, that the theoretical impulses which gave rise to structuralism antedate the

30

Chapter 2

A waft of myth, due, no doubt, to Saussure’s precocious Mémoire of 1879, is detectable in Antoine Meillet’s inaugural lecture to the chair of Comparative Grammar at the Collège de France in Paris of February 13th 1906. Here Meillet praises Saussure for the lectures he gave at the École Pratique from 1881 till 1891, some of which Meillet himself had attended (Meillet 1921: 2): No-one of those who had the pleasure to attend them will ever forget his well-known lectures at the École des Hautes Études, where the discreet elegance of his presentation almost made one forget the impeccable c­ ertainty and the extent of the information provided, and where the precision of a steadfastly rigorous method only allowed a glimpse of the geniality of his intuition. This praise refers to Saussure’s lectures on Sanskrit, comparative linguistics and etymology, since it was written (and presented orally) before Saussure started the Geneva lectures that would become known as the Cours de linguistique générale. Seven years later, the waft has become a definite scent. In his obituary article at Saussure’s death in 1913, Meillet wrote (Meillet 1938: 179): As for myself, there is hardly a page that I have written without feeling a remorse at claiming its merit for myself alone: Saussure’s thought was so rich that it penetrated my entire being. I would not dare to pinpoint, in what I have written, what I owe him, but I am sure that Saussure’s teaching is very much present in what benign judges have at times found praiseworthy in my own writings. This, one feels, is slightly over the top. Apparently, the myth had grown in the meantime, which made it unnecessary for Meillet to be specific about Saussure’s ‘presence’ in his writings. Yet in the first few sentences immediately following this passage Meillet does, in a subtle and extremely polite way, make it clear that he is aware of the fact that Saussure failed to present his supposedly profound thoughts in a published, or even publishable, form: Cours. But as soon as Saussure’s reputation as a theoretician was firmly established, the work began to exert a marked influence on linguistic theorizing. A curious sort of rebound effect developed: the structuralists decided AFTER THE EVENT to consider Saussure as the founder of their brand of linguistics, whereupon they made reasonable efforts to submit themselves to the influence of the Cours, as if to prove that their choice of a founding father was correct.

Who was Ferdinand de Saussure?

31

After the Mémoire, which no doubt owed its publication to the charming boldness of his youth, Saussure never felt that he had sufficiently pushed the theory of any linguistic fact forward to the point at which it could be exposed to the public. He was not one of those who hurriedly publish their ideas before they have ripened, before having fitted them into a complete and coherent system and having taken all difficulties into account. In the final sentence of the obituary, Meillet admits that Saussure, no matter how much of an authority on Indo-European matters, ultimately failed to develop the theory of language he had envisioned (Meillet 1938: 183): he had produced the most beautiful book on comparative grammar ever written [i.e., the Mémoire; pams], planted his ideas and posed solid theories, left his mark on numerous students, yet he had not completed his full destiny. As far as Meillet, with Paris as his basis, is concerned, the myth thus appears to be based more on Saussure’s old Mémoire and his lectures on Indo-European historical linguistics than on the lectures that would become the Cours. In Geneva, meanwhile, the myth had become firmly established by the time of Saussure’s death, in February 1913, at least in and around the linguistics department of the University of Geneva. Now, however, it was no longer just a myth about a scholar of Indo-European linguistics but also about a linguistic theoretician. Joseph quotes from the funeral speech held by the Swiss theologian and biblical scholar Lucien Gautier, who even brings in religious belief and “Christian experiences”, which, as we saw in Chapter 1, are closely related to myth (Joseph 2012: 629): Yes, on the scientific terrain, where he was a master and a leader, Ferdinand de Saussure could have applied to himself the words of the apostle: Now I know only in part … For now we see through a glass darkly. What is true in the intellectual domain, we hold to be true in the religious domain as well, following St. Paul, certainly better placed than any other to speak of these Christian experiences. Two pages down, Joseph comments (Joseph 2012: 631): Lucien Gautier, the successor to Jean Calvin, could not have chosen a more apt biblical reference than the passage from the letter of St. Paul to

32

Chapter 2

the Corinthians about how, in this world, we see as through a glass darkly. Saussure had perceived flashes of what lies beyond, had tried to piece them together—anguishingly, because he was not the kind of man who was satisfied with partial understanding. At last, in his lectures on general linguistics of the spring of 1911, the pieces had begun to fit. It is important to note that all this was said, whether by Gautier or by Joseph, in the absence of any concrete evidence of a clear and fruitful theoretical analysis, synthesis, or other achievement on Saussure’s part—except, of course, his precociously ingenious reconstruction of the Indo-European vowel system that went back to the late 1870s. There is no evidence at all that “the pieces had begun to fit” in the spring of 1911. Nor does Joseph give any idea of what the putative new synthesis, pieced together from the putative “flashes of what lies beyond”, amounted to. The reception of the Cours testifies to the growth of the Saussurean myth. Within Switzerland, the response was immediate and uniformly positive. ­FrýbaReber (2017) lists eight reviews, all published in 1916 or 1917, all by Swiss authors: André Oltramare (1916), Jules Ronjat (1916), Max Niedermann (1916), Léopold Gautier (1916), Jacob Wackernagel (1916), Karl Jaberg (1916), Pierre Bovet (1917) and Édouard Clarapède (1917), and all highly laudatory. Outside Switzerland, the reactions were more skeptical. Koerner writes (1999: 123): It is interesting to note that most reviewers did not come from what was still the centre of historical-comparative Indo-European linguistics, Germany, but more from the periphery: Austria (Schuchardt 1917), Denmark (Jespersen 1917), Italy (Terracini 1919), and Rumania (Bogrea 1921); from France (Bourdon 1916; Meillet 1916, 1917; Grammont 1917) and particularly from Switzerland and Swiss scholars residing in Germany (Jaberg, Niedermann, Oltramare, Ronjat, Wackernagel, all dating from 1916 […]. This diversified […] reception in itself merits study […]. It remains to note that while the reviews by the Swiss contingency were positive, if not enthusiastic, throughout, the response from France, notably Meillet, a former student of Saussure’s and life-time friend of his, was much more muted, if not lukewarm. The reviews from two other European scholars of comparable stature, Hugo Schuchardt (1842–1927), a former student of Schleicher’s and an opponent of the Neogrammarians, and Jespersen, were largely negative. In 1924, the American linguist Leonard Bloomfield (1887–1949)—who was never turned into a myth but is rightly considered the father of American

Who was Ferdinand de Saussure?

33

structuralism—published a polite but reserved review of the second (1922) edition of the Cours in the Modern Language Journal, criticising Saussure for taking the word and not the sentence as the primary unit for linguistic analysis and description.19 In 1926, the Dutch linguist Jac. van Ginneken (1877–1945) wrote, in German, an acerbic note of three lines in Indogermanisches Jahrbuch about the same second edition: “Unchanged reprint of the first edition. Of very uneven value. Dilettante extravagancies, next to deep insights into the life of languages”.20 But after roughly 1930 the myth spread and solidified in large parts of Europe, where Saussure began to be seen as the ‘father’ of European structuralism. Only Germany and Britain resisted the myth for some time, though it has meanwhile taken root there too. The German linguist Gerhard Helbig (1929–2008) gave the following account (1989: 34): In Germany in particular, there was strong resistance to the new thoughts developed by Saussure. […] It is typical for this isolation that Saussure’s work was not translated into German until 1931 (under the title ‘Grundfragen der allgemeinen Sprachwissenschaft’), that before that year the book was reviewed by only one German linguist, and that after that year it sold a mere 500 copies. It was not until after 1950 that the book was properly taken into consideration. The same holds for Britain. The so-called ‘London School’ of linguistics, headed by John Rupert Firth (1890–1960), was staunchly opposed to Saussure (see Koerner 1999: 151–166) and followed the lead of the anthropologist Bronisław Malinowski (1884–1942), but its last member and at the same time the initiator of new ways of linguistic thought, the British linguist John Lyons (b. 1932) wrote, not as a historian of linguistics but as an active participant (1968: 38): If any person is to be called the founder of modern linguistics it is the great Swiss scholar, Ferdinand de Saussure, whose lectures (reconstructed from the notes of his students after his death) were published in 1915 [sic] as Cours de linguistique générale. Many different schools of ­linguistics can

19 20

In his (1926: 154), Bloomfield, more or less gratuitously, acknowledges Saussure, along with Sapir, for having “taken steps towards a delimitation of linguistics”. “Unveränderter Abdruck der ersten Ausgabe. Sehr ungleich von Wert. Neben tiefen Einsichten in das Leben der Sprache, dilettantische Freisinnigkeiten”.

34

Chapter 2

be distinguished at the present time, but they have all been directly or indirectly influenced (in varying degrees) by de Saussure’s Cours. (Lyons failed to mention that, by the time he wrote this, all the different schools of linguistics that were in any way productive derived from the American linguist Leonard Bloomfield, who published his groundbreaking Language in 1933 in total independence from Saussure but, ultimately, inspired by Wundt, whom he greatly admired but who was never mentioned by Saussure.) Saussure’s European fame was no doubt in large part due to the fact that, by 1930, both the Paris and the Geneva school of linguistics had linked up with the incipient Prague school, largely as a result of the first International Congress of Linguists that took place in The Hague in 1928, where all the prominent figures met in a cordial atmosphere. The pooled force of the Prague and the Paris/ Geneva schools quickly became a dominant factor in large parts of Europe, looked up to by linguists in faraway places. Nowadays, it has become an automatic reflex of linguists to refer to Saussure as the ‘father of European structuralism’. Section 2.5.2 shows that this belief is not borne out by a paternity test. Nor is Saussure the true originator, as has been widely believed since World War ii, of the main or central elements in the theory of language, namely the arbitrariness of the linguistic sign, the social nature of language, and the four classic distinctions between (i) diachrony and synchrony, (ii) language as a system (‘langue’) and its use in speech or writing (‘parole’), (iii) form (‘signifiant’) and meaning (‘signifié’) and (iv) ‘associative’ and ‘syntagmatic’ relations of linguistic elements. In fact, none of these concepts or distinctions was his: they were all current in the literature that existed before Saussure taught his Cours. Keith Allan writes (2010: 261): insistence on the primacy of the spoken language is found earlier in Sweet; the arbitrariness of the sign, the differentiation of signifier from signified, the germ of the concept of valeur, and the import of the social aspect of language are all found in Whitney; the notion of associative versus syntagmatic relations and the differentiation of synchronic from diachronic linguistics are found in Paul. Allan’s answer to the question of “why Saussurean linguistics was paradigmatic in the sense that for a time it provided model problems and solutions to a community of practitioners” is (Allan 2010: 262): It was that Saussure brought together in the lectures which gave rise to the various ideas of his illustrious contemporaries and

Who was Ferdinand de Saussure?

35

i­ mmediate predecessors, fashioning them into what was thought to be a workable theory. Saussure’s insistence that linguistics is an independent science that should study language (langue) as a self- contained structured system came to characterize twentieth century linguistics. This seems to me to be essentially correct: one may indeed see Saussure’s lecture courses as an attempt to give a (selective) synthesis of modernist ideas about language that were current around 1900 and to “fashion them into what was thought to be a workable theory”, which, apparently, made for some kind of crystallisation in the minds of his students and followers. At the same time, however, this precisely shows the myth. For if this was all Saussure did in his lectures, one would, first, have expected him to refer to the real sources of the ideas concerned and, secondly, such an attempt would by itself hardly suffice to explain the fame he acquired before and after his death. It might if it had been a brilliant attempt, but, in fact, it was fragmentary, confused and generally inadequate. And as regards the “independence” of linguistics as a discipline, the last sentence of the Cours “Linguistics has for its unique and true object language, taken by itself and for itself” (see also note 37 in Chapter 3) is not Saussure’s but Charles Bally’s, who added it to the Cours on his own account as one of the editors (Koerner 1973: 222). Nor did the notion of the independence of linguistics originate with either Saussure or Bally: we find it already, for example, in Delbrück (1901), where Delbrück insists, against Wundt, that it makes no difference in linguistics what psychological theory one accepts or rejects, since language is to be studied in its own right, regardless of psychological theory. One can only conclude that there is a large gap between Saussure’s status in present-day linguistics on the one hand and his actual achievements on the other. That gap can only be explained if it is assumed that a colossal myth had come into being elevating him to his present status. As just one among countless other typical manifestations of the Saussurean myth, let me quote an anonymous contributor to Wikipedia (s.v. “Structural Semantics”), who writes: Structural semantics is that branch that marked the modern linguistics movement started by Ferdinand de Saussure at the break of the 20th century in his posthumous discourse titled “Cours de linguistique générale” (A Course in General Linguistics). He posits that language is a system of interrelated units and structures and that every unit of language is related to the others within the same system. His position later became the bedding ground for other theories such as componential analysis and

36

Chapter 2

r­elational predicates. Structuralism is a very efficient aspect of Semantics, as it explains the concordance in the meaning of certain words and utterances. The concept of sense relations as a means of semantic interpretation is an offshoot of this theory as well. Here the Saussurean myth is caught in full flight. The notion “that language is a system of interrelated units and structures and that every unit of language is related to the others within the same system” is, if anything, due to Meillet, not to Saussure, who merely subscribed to it (Cours: 124; see Section 2.5.2 below). To say that “his position later became the “bedding ground” for other theories such as componential analysis and relational predicates” is gratuitous unless backed up with evidence, which is not provided (nor is it available). The statement that “structuralism is a very efficient aspect of Semantics, as it explains the concordance in the meaning of certain words and utterances” may be true but does not imply that Saussure played any part in it. In sum, the information provided in this Wikipedia article, as in so many other publications on European structuralist linguistics, is based on hearsay, not on serious study, and it is false. 2.5.2 Saussure the ‘Father’ of European Structuralism? To what extent Saussure, brooding in his study in Geneva or at Vufflens Castle, was aware of the new structuralism we don’t know, but he must have sniffed it. All one finds in the Cours is a certain vaguely expressed preoccupation with the fact that languages are, or are defined by, systems, but his own notion of ‘system’ went no further than morphology and, to a very limited extent, the lexicon. Syntax was a closed book to him (see Section 3.3 below). In this respect, Saussure distinguished himself in no way from a mass of contemporary authors, some of whom had actually advanced further. Although it is widely, if not universally, claimed that Saussure was the ‘father’ of (European) structuralist linguistics,21 there is no evidence that he ever thought of linguistic structures. As observed by Percival (2011: 243) and Levelt 21

In his widely used textbook on semantics, Lyons wrote (1977: 231): It is the Swiss scholar, F. de Saussure, who is generally regarded as the founder of modern structural linguistics. […] t was his Cours de linguistique générale (1916) which initiated the movement now known as structural linguistics; and it is from the Cours that much of the terminology of structuralism derives. The circumstances of publication were such that Saussure’s Cours contains a number of obscurities and inconsistencies; it was not in fact written for publication by Saussure himself, and it may not faithfully represent Saussure’s ideas in every respect. It is, however, the Cours, as published, that has been of historic importance.

Who was Ferdinand de Saussure?

37

(2013: 217) (see also Section 6.1 below), the terms ‘structuralism’ and ‘structure’ do not occur in the Cours, but what is more important is the fact that the notion of structure does not occur in it either. By contrast, both the term and the notion of ‘system’ occur all over the Cours, as it does in many other linguistic writings of the time. As is pointed out in Section 6.1, many authors writing on linguistic structuralism applied the term structure not so much to the products (sentences) of the underlying system, built up of distinct parts, as to the underlying system itself, speaking of the ‘structure of a system’—although they usually did so in terms that are too vague to allow for an intelligible interpretation. Percival, for example (for once open to criticism), wishing to depict Saussure as a structuralist, refers to the following sentence occurring on page 124 of the Cours (Percival 2011: 245): “Language is a system, all parts of which can and must be considered in their synchronic solidarity”. But for Saussure, language (‘langue’) is many things. On p. 25 of the Cours, ‘langue’ is “a social product of the language faculty”, “an ensemble of necessary conventions”, “a whole in itself, a principle of classification”, and on p. 30 it is “a store (‘un trésor’) accumulated by the practice of speech (‘parole’) in the subjects belonging to any given community, a grammatical system existing virtually in each brain, or, more precisely, in the brains of a group of individuals”. One can only conclude that Saussure’s notion of what constitutes a system was so nondescript as to be vacuous. Likewise, wishing this time to depict Jakobson as a structuralist, Percival makes the cryptic assertion (Percival 2011: 247) that the term structuralism entails “a strong recommendation that languages should be described primarily as structured objects”, referring to a statement by Jakobson that a language could be regarded as “a whole where all parts hold firmly together”. But no-one at the time had any idea of what could be meant by the phrase ‘structure of a language’, or by the notion of a language as a “structured object”. Nowadays, we may perhaps speak of ‘the structure of a language system’, meaning the structure of a grammar, but that level of abstraction was, at the time, still far beyond the intellectual horizon, as it was not introduced into linguistics in anything like a precise form until Harris (1951). It is not even to be found in Bloomfield’s Language of 1933. Yet Bloomfield was, beyond any doubt, a structuralist, as he focused on the structure of the products of the system, that is, sentences, phrases, words, not on any as yet mystic notion ‘structure of the language system’.

All this is true and correct, but for the statement that the Cours “initiated the movement now known as structural linguistics” (which should, at the very least, be restricted to ­European structuralism).

38

Chapter 2

One should, moreover, keep in mind that the sentence quoted by Percival from page 124 of the Cours, “Language is a system, all parts of which can and must be considered in their synchronic solidarity”, as well as Jakobson’s description of any language as “a whole where all parts hold firmly together”, immediately remind one of Meillet’s famous dictum (Meillet 1921: 16, written in 1906): “[F]or a language forms a complex system of expressive means, a system where everything hangs together with everything” or, in the original French: “un système où tout se tient”. Yet, while these quotes no doubt implicitly refer to Meillet’s dictum, all they show is that Meillet, Saussure and Jakobson were children of their time, as whiffs of vaguely structuralist-oriented thought had been pervading the human sciences for decades (see Section 6.1 for further comment). Moreover—we may as well be frank and say it—the statement that language is a system “où tout se tient” was no more than a cryptic holistic slogan, meant to inspire awe rather than create insight.22 It is, at any rate, false, as we see all the time isolated changes in subsystems of languages that have no effect on the rest of the system. Given the material at our disposal, it is most likely that Saussure remained stuck in an as yet amorphous and nebulous notion of ‘language system’, ­without realising explicitly that this system regulates the putting together of sentences from given elements. For all we know, Saussure had no notion of sentence structure, or, if you prefer, it was so primitive that it went no further than ­one-dimensional linearity. Even prosodic structure—that is, accents and intonation—was not treated by him as consisting of distinct structural elements occurring simultaneously with the spoken sounds. As we will see in Section 3.4.5, the reason he gave for this was that intonation and accent are pronounced simultaneously with the sounds, which makes us conclude that, for Saussure, prosodic structure cannot be a distinct structural element occurring simultaneously with the spoken sounds because it is produced simultaneously with the spoken sounds. Saussure was, apparently, simply unable to imagine any other kind of structure than temporal successivity, that 22

See Koerner (1988: 62–63; 1997; 1999: 184–188) and Hewson (1990) for a full history of the phrase “système où tout se tient”. I may refer in particular to Koerner (1993), who quotes the French naturalist-zoologist Georges Cuvier (1769–1832) who wrote in 1812 (Koerner 1993: 10): Every organised being forms an ensemble, a unique closed system, all whose parts are mutually related and contribute to the same final action through a reciprocal reaction. No single part can change without the others changing likewise. Consequently, every single part, taken by itself, refers to and thus gives all the other parts. This shows that the thought expressed by linguists like Meillet, Saussure or Jakobson went back to the natural sciences of at least the early nineteenth century.

Who was Ferdinand de Saussure?

39

is, l­inearity. Anything like the hierarchical constituent structures proposed by Wundt and, later, in his wake, by Bloomfield (see Percival 1976; Seuren 1998: 219–227) was completely absent from his thinking, if not actually alien to it.23 Despite his solitary lucubrations regarding the nature of language, Saussure had remained a comparativist at heart and thus, like the entire school of comparativists of his day, missed the train to twentieth-century theory of language and thus to structuralism. This ironical conclusion is one good reason for denying Saussure the title of ‘father of European structuralism’. 2.5.3 Saussure in Literature, Art and Philosophy A short final comment is in order here about the curious fact that in postWorld-War-II France Saussure became the idol of a so-called ‘structuralist’, soon turned into a ‘poststructuralist’, movement in the study of literature and art in general, as well as in mainstream French philosophy, which, by then, had turned its back on rationalist thought and, suffering from an excess of phenomenological zeal, had lost itself in undisciplined effusions of a romanticist nature, largely to do with experiences of the individual (the ‘self’) vis-à-vis the external world (‘alterity’ or ‘otherness’). The connection with structuralism in any possible sense of the term has never been clarified. A typical exponent of this kind of ‘philosophy’ was the Frenchman Jacques Derrida (1930–2004), who, characteristically, went for precisely the most obscurantist element in Saussure’s thinking, his insistence on the supposed fact that everything in language is negative and oppositive (see Section 3.5.2 below). Derrida held the view that, in the study of language, expression in the form of writing takes precedence over expression in the form of speaking (Derrida 1967), which would give rise to the new science of ‘grammatology’. In fact, he proposed that the ‘sémiologie’ envisaged by Saussure should be replaced with ‘grammatology’ (Derrida 1967: 74): 23

André Martinet, who saw the distinction between segmental phonology and morphology (or ‘study of signs’) as all-pervasive in the study of human language, so much so that it even had to define both linguistics and its object ‘langue’—his well-known ‘double articulation’ (see also note 29 in Chapter 3—likewise found himself in trouble over the prosodic features of language. His solution was to eliminate them from the study of ‘langue’ and assign them to the study of ‘langage’: “But since it is the ‘langue’, rather than the ‘langage’, that forms the object of linguistics, it is justified to assert that the prosodic facts are less basically linguistic than the signs and the phonemes. […] And, since, in the last analysis, we are in search of what constantly characterises all we wish to call a ‘langue’, it is normal that we retain the double articulation and discard the prosodic facts”. (Martinet 1965: 33–34). Readers may differ as to whether this is indeed “normal”.

40

Chapter 2

One should replace sémiologie with grammatology in the programme of the Cours de linguistique générale. I will not comment on this proposal, as this book is not about Derrida. What I wish to draw attention to here is that, in this context, he introduced the term différance, as a deliberate misspelling of French différence (‘difference’) (­Derrida 1967: 91–92; italics original): On the one hand, the phonic element, the term, the fullness that one calls the sensible, would not appear as such in the difference or the opposition given to them by form. Such is the most evident importance of the appeal to the difference as reduction of the phonic substance. Here, the appearing and the functioning of the difference suppose an original synthesis not preceded by any simplicity. Such would be the original trace. Without a retention in the minimal unit of the temporal experience, without a trace retaining the other as other in the same, no difference could do its work and no sense would appear. What we have here is thus not a constituted difference but, before any determination of the content, a pure movement constituting the difference. The (pure) trace is the differance (‘différance’). It does not depend on any sensible, audible or visible (phonic or graphic) fullness. On the contrary, it is the condition for such. Although it does not exist, although it is never a being-present outside any plenitude, its possibility is anterior in law to anything one may call ‘sign’ (signified/signifier, content/expression, etc.), concept or operation, motoric or sensible. This differance is, therefore, no more sensible than intelligible and it allows for the articulation of the signs among each other inside one same abstract order—of a phonic or graphic text, for example—or between two orders of expression. (I could go on quoting this text ad infinitum, as there seems to be no coherent closure to the stream of imagery, but I stop here.) In any case, the term différance turns up again in Derrida’s Margins of Philosophy of 1982 (quoted here in an English translation by Alan Bass). Here, it appears that the term is intended to convey the composite idea of ‘difference’ and ‘deferring’, as a further development of Saussure’s ‘sémiologie’ or Derrida’s own ‘grammatology’. As an illustration, I may quote the following piece of occult writing (Derrida 1982: 4; translation by Alan Bass): Therefore, preliminarily, let me recall that this discreet graphic intervention, which neither primarily nor simply aims to shock the reader

Who was Ferdinand de Saussure?

41

or the grammarian, came to be formulated in the course of a written ­investigation of a question about writing. Now it happens, I would say in effect, that this graphic difference (a instead of e), this marked difference between two apparently vocal notations, between two vowels, remains purely graphic: it is read, or it is written, but it cannot be heard. It cannot be apprehended in speech, and we will see why it also bypasses the order of apprehension in general. It is offered by a mute mark, by a tacit monument, I would even say by a pyramid, thinking not only of the form of the letter when it is printed, as a capital, but also of the text in Hegel’s Encyclopedia in which the body of the sign is compared to the Egyptian Pyramid. The a of différance, thus, is not heard; it remains silent, secret and discreet as a tomb: oikesis. And thereby let us anticipate the delineation of a site, the familial residence and tomb of ‘the proper’ in which is produced, by différance, the economy of death. This stone (provided that one knows how to decipher its inscription) is not far from announcing the death of the tyrant. And it gets worse as the text goes on. I trust the reader will agree that it would be asking too much of me to comment on such writing in an academic context. (And, incidentally, one might also consider the fact that the style of ‘philosophising’ exemplified by these excerpts has done great harm to the already not too glorious reputation of philosophy as an academic pursuit.) Other recent authors are more explicit in depicting Saussure as a romanticist- phenomenologist thinker. An example is Gasparov (2013), whose basic lack of familiarity with historical facts, both in and outside linguistics, makes further discussion unnecessary. Another example is Stawarska (2015), who attempts to restyle Saussure as being rooted in romanticist-phenomenological thought. This author argues that the Cours does not reflect the real Saussure at all and that the editors Bally and Sechehaye wilfully misrepresented the lectures as they were really given. For this author, the real, or ‘private’, Saussure was a deep philosophical thinker, who was a phenomenologist at heart. She presents Saussure as the founder of a ‘phenomenological linguistics’, which, however, is left undefined. Unfortunately, Stawarska (2015) is marked by absence of proper academic argument: the ‘arguments’ presented are so biassed and selective that one can only speak of rhetoric and propaganda. Then, Bally and Sechehaye, the editors of the Cours, are debunked as the creators of a strawman, the ‘public’ Saussure known from the Cours. The fact that it was this ‘strawman’ that was used to put up the myth that he shaped European structuralism in whatever sense of the term, while the ‘private’ of ‘real’ Saussure, being unknown to the world until

42

Chapter 2

well after World War ii, had no influence at all, remains in limbo, which leaves it unclear why Saussure should now be promoted as the, or a, founder of ‘phenomenological’ linguistics. In any case, the accusation directed at Bally and Sechehaye has now been definitively refuted in Sofia (2015), which shows the documents as they are, in facsimile, so that everyone can now see how conscientiously, how painstakingly and how professionally Sechehaye in particular went about reconstructing Saussure’s teaching, regardless of his own opinion. Finally, even if the private Saussure were indeed a closet phenomenologist, quod non, the relevance of Stawarska’s claims is unclear—unless it is assumed that this author, who, like Gasparov (2013), is signally unfamiliar with linguistic theory and practice, aims at promoting a new paradigm in linguistics, her ‘phenomenological linguistics’, for which a restyled ‘phenomenological’ Saussure is meant to serve as an emblem.

Chapter 3

The Cours: A Critical Look 3.1

Language as a Social Phenomenon

3.1.1 The Social Dimension of Language Before we touch on the better-known features of the Cours, it is useful to have a closer look at an aspect that is usually left unmentioned but must nevertheless be highlighted for an adequate understanding ot its text and its later influence: the social nature of language systems. The social nature of language use is widely recognised and studied, especially in pragmatics, and therefore needs no special emphasis. But that languages as systems have not only psychological but also social reality is something that is hardly ever mentioned in presentday theoretical linguistics. In our day, linguists, who may be less familiar with the intellectual climate in France during the fin de siècle and only know about Saussure’s Cours from a distance, might think that the social dimension of language as put forward in the Cours is Saussure’s contribution. This, however, is not so. In fact, quite the contrary is the case. During the fin de siècle, sociological phenomena were very much in the public attention and the then new sociology was the talk of the town. Since the social, political and economic developments of the nineteenth century had a profound impact on society as a whole, especially in Europe, it was only natural that European linguists, from 1850 onward, would emphasise the social character of natural language.1 The intellectual climate of the day made it imperative for Saussure to refer, wherever appropriate, to the social 1 To give just one example, Meillet (1921: 16-17) writes, as one of many similar statements made elsewhere, before and after: [T]he reality of a language is social: it is a result of the fact that a language belongs to a definite group of speaking subjects, that it is the means of communication among the members of one given group, and that no single member of the group is in a position to bring about a change in the language. The very necessity to make oneself understood imposes on all speakers the requirement of keeping the language as uniform as possible in its daily usage. Ridicule is the immediate penalty for individual deviations. Note that this, though published in 1921, was written in 1906, before Saussure started on his Cours lectures. Meillet might have added that there are also many cases in which the speaker does not want to be understood by at least part of his audience, which, if institutionalised in a subgroup, will make for internal language variation, going against uniformity, and often also against functionality (Seuren & Hamans 2010).

© koninklijke brill nv, leiden, ���8 | doi:10.1163/9789004378155_004

44

Chapter 3

nature of language. Yet this aspect is never elaborated in any detail and no substantial consequences of this social viewpoint are detectable anywhere in the Cours. This is to be seen in the light of his insistence on the autonomy of linguistics. Saussure’s references to the social nature of language are, in fact, little more than lip service paid to the prevailing intellectual climate. He himself was, apparently, hardly interested in this aspect of language. In this regard, he did foreshadow the strictly formal varieties of twentieth-century theoretical linguistics. Yet the pressure to refer to language as a social phenomenon was strong enough for the Cours to brim over with references to the social nature of language, without, however, any further analysis of how this social nature of language interacts with languages as systems and with the development of languages. Selecting only two out of a multitude of possible quotes (see also Section 3.2.2), we read (Cours: 25): [The language ] is at the same time a social product of the language faculty and an ensemble of necessary conventions, adopted by the social body in order to make possible the use of this language faculty by individual speakers. and (Cours: 30): If we could encompass the sum of the verbal images stored in the minds of all individuals, we would touch on the social unity that constitutes each language. It is a treasure deposited by the practice of speech in the subjects belonging to a community, a grammatical system virtually existing in each brain, or, more precisely, in the brains of a totality of individuals; for a language is not complete in any single brain, it exists perfectly only in the mass of speakers. With the advent of linguistic structuralism, on both sides of the Atlantic, linguists lost interest in language as a social phenomenon and became totally engrossed in the structural properties of sounds, words, phrases and sentences. Linguistics became ‘autonomous’—a fact which, despite its great positive significance for linguistics as a discipline and a profession, also had the ­negative effect that linguists no longer wanted to have anything to do with neighbouring disciplines: the concept of autonomy was confused with that of total ­independence. Contacts were severed not only with comparative-historical liguistics—Saussure’s home ground, so to speak—but also with disciplines such as logic, psychology or sociology (see Levelt 2013: 221). By the mid-­twentieth

The Cours: A Critical Look

45

century, psycholinguistics, sociolinguistics and dialectology, flourishing and prestigious branches of linguistics before, had lost their prestige and had become marginalised, even though there were still large numbers of practitioners. Sociolinguistics and dialectology in particular were largely considered to be occupations more suitable for dilettanti than for professionals. Although psycholinguistics made a splendid re-entrance arm in arm with cognitive science, and sociolinguistics, with dialectology in its trail, did the same, largely as a result of William Labov’s powerful influence during and after the 1960s, these disciplines were, strangely enough, not integrated into the standard body of modern theoretical linguistics, which has, for a long time, remained blind to the psychological, sociological and historical aspects of language and language use. While the official narrative—incorrectly, as I argue— reckons Saussure’s distinctions between language and speech, diachrony and synchrony, and signifier and signified to be his chief contribution to modern theoretical linguistics, his repeated reference to the social dimension of language is not included in this list. We now see why: the official narrative itself has, so far, failed to take this aspect into account. Over the past couple of decades, this situation has improved somewhat, mainly because the dominant trend of formalisation in terms of monolithic rule systems has had to cede ground to more ecologically oriented approaches to language. Even so, however, modern theoretical linguistics has little or nothing to say about dialects, sociolects and interactional varieties within the same language, nor, for that matter, about the status of the standard language vis–à-vis its internal articulations. It has, in particular, never developed any clear notion of how the internal variations within each individual speaker’s (active and passive) linguistic competence are to be integrated to form the rule system of their grammar (see Seuren 1982). Nor is it mentioned anywhere in the theoretical linguistic literature that the autonomy of the language system in the speakers’ individual minds (often called the ‘language module’), and thus the autonomy of linguistics as a discipline, is in large part due to the fact that every language and language variety is part of social reality to which individual speakers have to adapt if they are to function normally in their language community. A language is imposed on its speakers from outside as a self-contained system its speakers have to master and conform to. In short, as opposed to the more or less separate branches of historical linguistics and of pragmatics, the central body of modern theoretical linguistics has still not taken into account the crucial fact that any natural language, though lodged in the individual minds of its speakers, is also the common possession of all its speakers and thus part of social reality. Nor has there been any attempt at exploring the consequences of this crucial fact with regard to the

46

Chapter 3

autonomy, or modularity, of the language system in the minds of its speakers. In this regard, Saussure does indeed conform to the main ­twentieth-century developments in linguistic theory. 3.1.2 Early French Sociology By the end of the nineteenth century, as we have seen, the opposite was the case. As a result of the many political revolutions that took place and of the often violent emergence of socialism, communism and anarchism against the  backdrop of the rapidly growing mechanised big industry during the ­nineteenth century, some academics began to take a special interest in social structures and processes, thus giving a new impetus to, or laying the foundations for, the disciplines of economics and sociology. This welter of changes and developments ultimately brought about a radical transformation of the concept and form of the modern democratic state. In the French-speaking world, the study of social structures and processes latched on to eighteenth-century French socio-political thought, a joint product of romanticism and enlightenment, though, under the ever increasing pressure of the stark realities of social and economic inequality, the early placid armchair detachment and utopianism was soon replaced with an extreme form of rationalism, called positivism—a philosophy of science that, globally speaking, reduces all reality to matter and all knowledge to sensory input. The first great name in nineteenth-century French sociology is that of ­Auguste Comte (1798–1857), born in the immediate aftermath of the French Revolution of 1789. Driven by an urge to address the pressing social problems of his day, which demanded a nonromanticist, no-nonsense approach, he sought to bring about a return to the original ideals of the French Revolution. In this spirit, he pursued what he called a ‘positivist’ approach to the phenomena of the world, which implied that societies, much as physical nature, should be seen as mechanisms regulated by laws that are discovered by means of systematic observation of relevant facts. The ‘positive’ knowledge thus obtained is the most reliable form of knowledge, as opposed to ‘theological’ or ‘metaphysical’ forms of knowledge based on irrational belief or speculative philosophy. In an attempt to close in on the notion of positive knowledge, he designed a classification of, and an overall methodology for, all the different sciences, which he divided into, first, the basic science of mathematics, then the next higher cluster of the physical sciences astronomy, physics, chemistry and biology, in that order, and, finally, the highest phase of sociology, each later, or higher, science depending on the results of the earlier or lower ones. After Comte, the problem of the classification of, and the mutual relations among, the various sciences became a standard theme in the writings of

The Cours: A Critical Look

47

s­ ociologists, psychologists and linguists. For Comte this would imply that the laws of biology should be taken as the basis for the laws in sociology—a position also held by his counterpart Herbert Spencer (1820–1903) in Britain with regard to psychology (Spencer 1855) and now called strong reductionism (see Section 6.4 below)—but he refrained from addressing this issue head-on. The question of how the psychological and social realities studied in the human sciences should be taken to relate to the physical reality studied in the physical sciences would begin to be addressed in the works of Hippolyte Taine. Hippolyte Taine (1828–1893) added psychology and the study of history to the narrow Comtean base of the ‘positivist’ human sciences, which, for Comte, consisted mainly of sociology. Taine’s aim was to turn psychology, and to a lesser extent also history, into real sciences. As he adhered, in a general sense, to Comtean positivism, this meant a reduction of psychological and historical processes to physical events. Taine’s De l’intelligence, first published in 1870, is one protracted argument for the thesis that each psychological event, including those that are shared in a community, corresponds to one or more physical events in the brain of each individual. Others before him, such as Julien Offray de la Mettrie (1709–1751), Franz Joseph Gall (1758–1828) and, as we have seen, also Auguste Comte, had said the same, but Taine differed from these predecessors in one important respect. La Mettrie, Gall and Comte held that mental and physical processes stand in a one-to-one relation to each other, forming double-faced but identical systems—which enabled the former two to deny the existence of the ‘soul’ or ‘mind’ as a separate entity and got both of them into serious trouble with religious and state authorities. Taine, by contrast, was beginning to think, in a tentative and not fully explicit way, of what is now known as token physicalism, the idea that mental processes have an autonomous systematicity of their own, not reflected in the physical, cerebral events to which, in isolation, they stand in a direct and immediate relationship. In other words, we see in Taine’s texts the first, tentative, undoing of the ruthless reductionism found in La Mettrie and Gall and still adhered to by Comte: while cognitive token events may be taken to correspond one-to-one to token events in the brain, cognitive systems are not reflected as such in the brain. If that were so, psychology would not be a field of enquiry of its own. It would, like La Mettrie‘s soul, vanish into thin air, or rather into physics. On this essential point, Taine was ahead of Comte, who still endorsed Gall’s phrenology and rejected psychology as an autonomous science (see Charlton 1959: 36, 135–136). Sociology was further developed by the Frenchman Émile Durkheim (1858–1917), now considered one of the founders of modern social science, together with Karl Marx (1818–1883) and Max Weber (1864–1920). Durkheim

48

Chapter 3

was directly influenced not only by Comte but also by Taine (see, for example, Schmaus 1994: 65). Under the influence of these two, Durkheim did for sociology what Taine had tried to do for psychology. Analogously, Saussure can be seen as a man who aimed at doing for linguistics what Taine had done for ­psychology: establish a new autonomous discipline for the complex of ­phenomena recognised as being of a linguistic nature. In pursuing this aim, Saussure simply took the social dimension of language for granted, recognising its importance but leaving it for what it was. 3.1.3 ‘Völkerpsychology’ 2 But apart from the socio-economic and political developments, there was a second, more romanticist, source for the emergence of the study of the social aspect of humanity. For this, we have to go to Germany, where one sees an increased awareness of psychological, cultural and sociological differences among nations stemming from the romantically coloured nationalist movements that arose all over Europe in the context of the ever advancing emancipation of ‘the people’ from the aristocracy and from religion.3 This led to a new interest in the social nature of humans and, after 1850, to a new discipline in Germany, under the name Völkerpsychologie, a term that has no proper equivalent in English and expresses the notion that nations are characterised, among other things, by common attitudes, beliefs, reaction patterns, value systems and, as far as possible, a common language. The best English equivalent is perhaps ethnic psychology, but this term lacks the association with ‘nation’ as an emotionally charged focus of social identity evoked by the basically romanticist term Völkerpsychologie, with its heavily loaded component noun Volk (‘people’). I will simply use the original German term, leaving it untranslated though adapted to völkerpsychology (just as the English translation of the German term Lumpenproletariat is lumpenproletariat). The new völkerpsychology aimed to discover and describe the coming about of the mental characteristics associated with each nation and their effects upon national societies as a whole. It was meant to encompass the study of language, myth, culture, religion and customs proper to each nation and would thus provide a necessary complement to the study of individual psychology. The term and the discipline were established during the 1850s, first in a short 2 I am indebted to Pim Levelt for pointing out to me the important role played by völkerpsychology in the totality of the events and developments discussed in the present study, especially with regard to one of our two protagonists, Albert Sechehaye. 3 One thinks in particular of the German philosophers Johann Gottfried Herder (1744–1803) and Johann Gottlieb Fichte (1762–1814), both powerful exponents of the new romanticistnationalist movement linking the spirit of the people (‘Volksgeist’) to language and folklore.

The Cours: A Critical Look

49

article by Moritz Lazarus (Lazarus 1851) that carried the term ­Völkerpsychologie in its title, then in a longer article jointly published in 1860 by Lazarus and his brother-in-law, the linguist Heymann Steinthal and subsequently known as the ‘manifesto’ (the mid-nineteenth century was the period of ‘manifestos’) as it declared the aims of a new journal they had just started under the name Zeitschrift für Völkerpsychologie und Sprachwissenschaft (Lazarus and Steinthal 1860). The völkerpsychology project quickly petered out, however, owing to its lack of empirical method and to its romanticist vagueness manifested, for example, by the all too frequent use of the impressionist concept ‘Volksgeist’ (‘spirit of the nation’). By the end of the century, what little remained of it had been absorbed by other disciplines, in particular sociology. Yet the project acquired a special significance by the way in which it was adopted by Wilhelm Wundt, who, over the years, developed his own concept of völkerpsychology, resulting in his massive work Völkerpsychologie, eine ­Untersuchung der Entwicklungsgesetze von Sprache, Mythus, und Sitte (Völkerpsychologie: an Examination of the Developmental Laws of Language, Myth and Tradition), a ten volume work that took the last twenty years of his life to complete, from 1900 to 1920. Wundt was inspired by the idea of a völkerpsychology as a counterpart to experimental psychology which dealt with the individual (Graffi 1995). For him, psychology had two departments, one f­ ocussing on the mechanisms of the human mind, and thus concerned with personal or private psychological processes, such as sensory perception, and open to experimental methods similar to those practised in the natural s­ ciences, the other addressing the higher functions of the mind—such as thinking, understanding, speaking—and belonging to the human sciences. This latter part of psychology is, according to Wundt, much less open to laboratory experimentation and looks more at the products of the mind, such as language, myths, customs, rites and art. Levelt writes (2013: 169–170):4 Careful historical and especially comparative analysis of such products of the mind will reveal their genesis and the complex underlying mental 4 Oswald Külpe (1862–1915), a student of Wundt’s, went to Würzburg and established his own school there around the year 1900. This school’s chief purpose was to develop experimental procedures for the testing of precisely the higher mental functions that Wundt deemed unfit for laboratory experiments. The data were obtained by the method of introspection: subjects were asked to report (in some cases extensively) on their own experiences while thinking. An early prominent member of this school was Karl Bühler (1879–1963), later professor in Vienna and author of the classic Sprachtheorie (1934), where he came to criticise the earlier introspection method (Bühler 1934: 251–255). Wundt despised the Würzburg method of introspection and made no bones about that (Boring 1950: 406; Levelt 2013: 228–232).

50

Chapter 3

processes that gave rise to them. This is the ethnic psychological [Levelt’s term for what I call ‘völkerpsychological’; pams] approach. The psychology of language can be experimental in only a limited way. It is applicable to the study of the perception of speech sounds, spoken or printed words, and of motor processes such as articulation and handwriting. But all else should come from ethnic psychology. Ethnic psychology should handle such complex processes as expressing judgments or predication by way of sentences. It will also be the preferred method for the study of language acquisition—what is it the child produces at certain stages of development? It was thus that the propositional subject-predicate structure assigned to sentences and their underlying semantic representation, though hardly of a social nature, landed in Volume 1 of the Völkerpsychologie and was, rightly or wrongly, deprived of the privilege of experimental testing. Through Wundt’s Völkerpsychologie, especially Part 2 of Volume i, which dealt with language, Wundt’s ideas entered theoretical linguistics, having been adopted by the American linguist Leonard Bloomfield, who, during the 1920s, in turn became the founder of American structuralism in linguistics. This brand of structuralism, ironically but perhaps also understandably in the context of American behaviourism, lacked, right from the start, all interest not only in the social but also in the cognitive and semantic aspects of natural human language, concentrate as it did exclusively on the formal aspects of natural language sentences, phrases and words, leaving the semantic aspect out of account. Wundt’s notion of proposition thereby went by the board. As shown below, Sechehaye was likewise deeply inspired and motivated by Wundt’s völkerpsychology, but in a totally different way, which so far has had little impact but is at least as important as Bloomfield’s ultrapositivist adaptation. Sechehaye’s point was that Wundt’s analyses of propositions (sentences) were all too exclusively semantic, not taking into account that each natural language, dialect or sociolect has its own specific ways of transforming Wundtian semantic analyses into perceptible form. But the world of E ­ uropean theoretical linguistics ignored Sechehaye and massively followed Saussure, who, apparently, wanted to have nothing to do with either Wundt or his notion of proposition. Moreover, Sechehaye, even less than Saussure or Bloomfield, was not interested in the social or ethnological aspect of Wundt’s work, only in Wundt’s structural analysis of the semantic structures underlying sentences. This way, both the nineteenth-century emphasis on the social nature of ­language and Wundt’s notion and analysis of the proposition, disappeared from the scene of theoretical linguistics. It is time to redress that situation.

The Cours: A Critical Look

3.2

51

Linguistics as the Science of Language, Not of Speech

3.2.1 The Tasks of Linguistics For Saussure, linguistics has, in a general sense, three tasks (Cours: 20): (a) to describe and trace the history of all languages it has access to, which amounts to establishing the history of the language families and define, as far as possible, the mother-languages of each family; (b) to search for the forces that are permanently and universally operative in all languages and define the general laws that all specific phenomena of history are subject to; (c) to delimit and define itself. One notes that, in this passage, which, according to Godel (1957: 77), represents the opening lecture (October 28th 1910) of Saussure’s last course given in ­1910–1911, diachronic linguistics is still given pride of place, while the synchronic aspect of language study, in fact, remains unmentioned. It might look as if that aspect is mentioned in (b), but careful reading (“all specific phenomena of history”) shows that this interpretation cannot be maintained. The task specified in (b) is still of a purely diachronic nature. There is no hint of any universal properties of, or restrictions on, the grammars of all languages, even though modern linguists might be inclined to read that into the text. Task (b) is merely an inductive generalisation over the results of task (a). Task (c) reflects the then widespread feeling that all new human sciences should be conscious of their place in the taxonomy of sciences. On p. 33, it is added that “the task of the linguist is to define what makes a language a special system in the totality of semiological facts”. This, one may take it, is meant as a partial answer to the problem set in (c): it ranks linguistics as a branch of the yet to be developed discipline of semiology (see Section 3.4.1). Apparently, Saussure (or his editors) did not notice the discrepancy between this passage and the view proposed elsewhere in the Cours, and current in twentieth-century structuralist linguistics, that “The language (‘la langue’) is a system of signs that express ideas”, and that the ‘static’ or ‘synchronic’ study of language systems is the prime task of the linguist (Cours: 128): “It is obvious that the synchronic aspect takes priority over the other, because, for the speaking mass, it is the only true reality”.5 As formulated on page 20 of the Cours, linguistics is still the historical-comparative discipline 5 Giving methodological and epistemological priority to facts of experience over facts of ­material nature is a main feature of Hippolyte Taine’s thinking, not shared by the positivist majority of academics of his day. See note 9 below.

52

Chapter 3

as it was practised by the Young Grammarians, who taught Saussure in his younger years when he was a PhD-student in Leipzig, whereas on page 128 the new, s­ynchronic point of view seems to have broken through.6 Why this new ­perspective is not given expression in his first lecture of the third course is enigmatic. If one were to go by the quote given at the outset of this section, none of Saussure’s meditations regarding the nature of ‘language’ versus ‘speech’, ‘static’ or ‘synchronic’ linguistics, the linguistic sign, the tenability of the notion of ‘synchronic’ laws, or ‘valeurs’ and ‘oppositions’, etc., in short, all the things for which Saussure has become so well-known, should be reckoned to fall under the rubric of linguistics. This can hardly have been Saussure’s intention, if only because the lectures underlying the text of the Cours were explicitly offered as lectures in general linguistics. What Saussure’s real idea was regarding the goals and tasks of linguistics can only be implicitly retrieved from the pages of the Cours, and one should not be too surprised at finding that Saussure’s ideas were far from clear on this count. 3.2.2 The Distinction between ‘Langue’ and ‘Parole’ Let us, therefore, leave Saussure’s unclear, or at any rate incomplete, description of the tasks of linguistics for what it is and concentrate on what is actually found in the Cours. A specific language (‘langue’) is, for Saussure (Cours: 30) “a grammatical system virtually existing in each brain, or, more precisely, in the brains of a totality of individuals”. By “grammatical system” Saussure means (Cours: 187) “the interpenetration of morphology, syntax and lexicology”, whereby it is understood that these three elements, though distinguishable from each other, are bound together by all sorts of mutual relations, the precise nature of which he leaves to the readers’ imagination or, more charitably, to future research. So far so good, though one should realise that Saussure’s notion of syntax, apart from being unclear, differs essentially from what it is considered to be in our day, as is explained in Section 3.3.1. By contrast, parole, or, in English, speech or language use (in the early 1960s, Chomsky introduced the unnecessary but widely used term performance) is by definition individual, even if, of course, it takes place in social interaction of whatever kind. For Saussure, parole is (Cours: 30–31): [A]n individual act of the will and of the intellect, in which we can distinguish: 6 Though perhaps not quite: see Section 3.6 below.

The Cours: A Critical Look

53

(1) the combinations by means of which speaking subjects make use of the code of the language in order to express their personal thoughts; (2) the psycho-physical mechanism allowing them to externalise these combinations. Again, on p. 38: [Parole] is the sum of what people say. It comprises (a) the individual combinations, depending on the will of those wo speak, and (b) the equally voluntary acts of phonation, necessary for the physical realisation of these combinations. Parole thus reflects the individual, not the social, aspect of language and it involves physical events (brain processes, muscular movements, sound waves plus auditory reception, and written symbols plus visual perception) that take place as a function of the language system. Any such language system is a department of ‘language in general’ or, in French, ‘le langage’ (Cours: 25, 139), along with actual speech events (‘parole’). This distinction between language/langue on the one hand and speech/parole on the other, under the umbrella notion of the French term langage—a term that has no equivalent in English—is generally attributed to Saussure, though it is also discussed extensively in Sechehaye (1908a), as is shown in Section 4.3.2 below. Saussure’s use of the term as an umbrella covering both langue and parole is somewhat idiosyncratic. Authors of the period tend to use the term somewhat differently, but for Sechehaye, whose definition and use of these terms in his (1908a: 47–54) are identical to Saussure’s. The first question is: is it correct to classify langue and parole under the more general concept of langage? This question is not of great moment, but we can say the following about it. The lexical meaning difference, in French, between langue and langage, or in Italian between lingua and linguaggio, is only partially clear. On the whole, langage and linguaggio are used in a general way and in a wide sense: ‘everything to do with language’, which corresponds to the etymology of these words. They derive from Latin linguaticum, consisting of lingua (‘tongue, language’), plus the suffix -aticum, meaning something like ‘anything to do with’. Langage or linguaggio is thus ‘anything to do with language’. Sechehaye (1917: 11) speaks of “the ill-defined ensemble of phenomena caught under the term langage”. French speakers speak of the langage des abeilles, not langue des abeilles, for what in English is called ‘bee language’, or of langage des pavillons, not langue des pavillons, for maritime flag signalling systems, or of langage du corps, not langue du corps, for ‘body language’. And

54

Chapter 3

analogously for Italian. It thus seems that langage covers more than the sum of langue and parole, as the word is used for language derivatives as well as for ways of expression that can be called ‘language’ only in a metaphorical sense. The question to be addressed next is: is it correct to distinguish between, on the one hand, langue as a system, often called competence nowadays (another Chomskyan innovation), and, on the other, parole (Chomsky’s performance), as the actual use of that system by speakers and listeners? This distinction has been common fare in linguistics at least since around 1900. It is also the main theme in Gardiner (1932). But is it correct? 3.2.3 ‘Frequency Linguistics’ Untenable This question is relevant in the present context, since there are linguists nowadays (e.g. Taylor 2012) who deny the validity of the distinction between langue/ competence as a permanent system and parole/performance as the incidental, practical use of it. For them, langue/competence is merely a (false) conjecture thought up by all too theoretical armchair linguists. What enables speakers to use language is not an underlying system of rules combined with a lexicon, but a statistically organised deposit of millions of utterances on which speakers/ listeners draw when they speak or write, or when they listen or read. What authors such as Taylor attack is less the morphology part of the grammar than what linguists now consider its syntax: ‘competence’ is tacitly equated with ‘syntax’. And syntax is said to be based exclusively on frequencies of word combinations, not on rules. That this does not eliminate the notion of competence, this time seen as a structured conglomerate of frequencies, seems to have escaped authors like Taylor, who equate competence with ‘rule system’, unlike the early proponents of the langue-parole distinction, such as Meillet, Sechehaye, Saussure and Gardiner, who, on the whole, equated competence with frequency-based habits, often confused with rules (see note 7 below). Curiously, Sechehaye, in his (1908a), dealt precisely with this question in a lengthy footnote on p. 24, choosing as his butt Franz Nikolaus Finck (1867–1910; see the discussion of this linguist in Morpurgo Davies 1975: 669–671), who, in a little book published in 1905, defended the exact viewpoint put forward by Taylor.7 From a historical point of view, therefore, modern ‘frequency’, or ‘usage’, 7 This is Sechehaye’s text (Sechehaye 1908a: 24): As an example of the way in which the problem is understood and solved by a modern linguist who also does psychology, see the booklet by Fink, entitled Aufgabe und Gliederung der Sprachwissenschaft (Halle, Hauft, 1905), where the author claims that each speaker’s language is, at any moment, a free creation, which, however, as he says, is influenced by the necessity of being understood and thus by the memory of ways of speaking heard earlier or used earlier by the one who speaks. The language in abstracto has no real

The Cours: A Critical Look

55

grammar harks back to a premodern phase of linguistic theory, in which the distinction between langue and parole was denied and the notion of syntactic rule, such as the ones responsible for the sentences quoted in (3.1) to (3.4) below in Section 3.3.1, had not yet become established. As for Albert Sechehaye himself, in his two books (1908a) and (1926), he shows a clear awareness of the socially imposed stringency of syntactic combinations but the notion of syntactic rule does not yet stand out clearly, as he uses the terms habit, rule, disposition, convention and law more or less indiscriminately. He even writes, in the footnote quoted in note 7: “Who says ‘habit’ says ‘rule’ ” (see Section 4.3.1 for further comment). Supporting ‘frequency-based or usage grammar’ is like saying that actual motorised traffic behaviour is not governed by an internalised system of traffic rules but by statistically defined mentally registered occurrences of accidentfree participation in motorised traffic—an absurdly behaviouristic notion which, if correct, would predict vast numbers of fatal accidents, far exceeding the actual numbers: imagine learning to drive by trial and error! In particular, it fails to explain how drivers avoid accidents in unique situations the likes of which have never occurred before in their personal experience. Analogously, the theory of frequency-based language use founders on the common fact that speakers experience no difficulty in producing, nor listeners in understanding, what can only be taken to be unique, or at any rate extremely rare, ­occurrences.8 So let us assume, with the vast majority of linguists, that the existence. To this we counter that there is a big difference between a memory and a habit, and it seems obvious to us that it is the latter and not the former that plays a role in the functioning of speech. Who says ‘habit’ says ‘rule’, and every rule that can be posited on empirical grounds has a real, though abstract, existence. This is how and why grammar, and thus language, exist. 8 The point is made, for example, by the following real life exchange between a father and his young son, walking about town. The boy is crying because he has just hurt himself (Seuren 2013a: 187): (i) Father: Well-educated boys don’t cry. (ii) Son: Í didn’t educate me! (with contrastive accent on “Í”) In (ii), the non-reflexive me, instead of the normally expected reflexive myself, in the meaning intended, is so rare that no amount of statistical support can possibly explain the instantaneous correct interpretation by every speaker of English, who will immediately grasp the semantic difference between (ii) and: (iii) Í didn’t educate myself! The latter means ‘Í am not a self-educator!’, which would be totally inappropriate in the given context. It is argued in Seuren (2013a: 187) that any explanation of cases such as this requires the assumption of an ‘abstract’ system of structures and rules converting semantic representations into well-formed surface structures. (ii) will then be semantically analysed as (iv) and (iii) as (v):

56

Chapter 3

­ ypothetical positing of competence (‘langue’) as a rule system directing actual h linguistic behaviour in parole is correct, because inescapable. This is not to say that frequency counts have no function in linguistics. On the contrary, they provide valuable, sometimes indispensable, information relevant to various branches of linguistics. But to claim monopoly for frequency phenomena in theoretical linguistics, and thus the wholesale rejection of the study of langue as an explanatory system of rules and restrictions, is way over the top and totally unacceptable. The actual data then consist not so much in what people actually say (as one knows, actual speech material contains a lot of gibberish), as in speakers’ implicit judgments of, or attitudes with regard to, the social appropriateness of given utterances as expressions of given meanings in given contexts or situations, just as for sociologists studying social norm systems, the data consist not so much in how people actually behave as in the community members’ implicit judgments of whether given forms of behaviour can be taken to be socially acceptable as appropriate in the given circumstances. This does not mean that the study of actual linguistic or social behaviour should be considered pointless. It only means that in the reconstruction of social norm systems, such as any given language or language variety, actual behaviour mirrors the system less faithfully than the attitudes held by individuals with regard to actual behaviour. The empirical question is: how does one reliably tap those attitudes? Often, as has been well-known for over a century now, deviant actual behaviour begins to claim or acquire legitimacy and ends up getting it, leading to a change in the language, but that is beside the point in the present discussion. 3.2.4 Who Introduced the Distinction between ‘Langue’ and ‘Parole’? Then, however, the question arises if this famous distinction between langue and parole is correctly attributed to Saussure. And here the answer must be negative. Sechehaye writes (2017: 11): “If this distinction is not absolutely new, Saussure has nevertheless had the merit of having posed it with proper insistence and of having made it the cornerstone of any linguistic edifice”. Even if it is hard to point at the exact origin of this distinction, we must at the very least allow for a close interaction between Saussure and Sechehaye in this regard. And we must acknowledge that, before Saussure did so, Sechehaye himself had (iv) not [the x such that [x educated me] be I] (v) not [the x such that [x educated x] be I] The correspondence in (v) between the two occurrences of the variable x in the same clause ‘x educated x’ will then, in principle, account for the meaning ‘I am not a self-educator’. ‘Big data’ techniques will be of no help here, or else let it be shown that they will be.

The Cours: A Critical Look

57

already made the langue/parole distinction “the cornerstone of any linguistic edifice”. In his magnum opus of 1908, Sechehaye devotes his Chapter 4 (pp. 47–54) to the distinction between langue and parole. The chapter bears the title “Theoretical linguistics belongs both to individual and to collective psychology”, echoed in Saussure’s (Cours: 30): “When we separate the langue from the parole, we separate […] what is social from what is individual […]”. We also read (Sechehaye 1908a: 53): “The study of the grammatical element in the langue thus falls within the remit of collective psychology”—a statement, by the way, we will have occasion to reject in Section 4.3.2 below. Sechehaye and Saussure differ only in that Sechehaye recognises, besides a ‘linguistics of langue’, a ‘linguistics of parole’, which studies actual linguistic behaviour, whereas Saussure (Cours: 36–39) is definitely averse to such a notion, only recognising a linguistics of langue, studying only the norm system. For Sechehaye (1908a: 48), langage is “the totality of means employed by a psycho-physical being to express his thoughts”. This is followed by an analysis of the terms used, in particular the term express. Then, after lengthy preparatory considerations, he writes (Sechehaye 1908a: 51–52): Nevertheless, no matter how important the role of the grammatical organism is in language in general and in intellectual life, one cannot deny that the conventions and rules of the grammar are not all there is to language [‘langage’]. […] First of all, it is evident that the most perfect grammatical organism could never, on its own, constitute the phenomenon of language in general [‘langage’]. The grammatical organism is at the disposal of just the speaking subject, who makes use of it by means of an act of the will and under the control of his intellect. Each instance of parole thus requires, besides the disposition that gives it its form, the free initiative of the subject, his will and his attention, which alone enable its realisation. We only have to read ‘langue’ for Sechehaye’s ‘grammatical organism’ to recognise unambiguously Saussure’s distinction between langue and parole. Who came first is hard to say. But we know (Godel 1957: 51–52) that Saussure was intimately familiar with Sechehaye’s book, the writing of which must have taken place over a fair number of years preceding its publication in 1908. We also know (see Section 2.3 above) that Saussure was less than frank when, discussing the distinction between synchrony and diachrony, he ignored Sechehaye’s book. Furthermore, Joseph (2012: 535) reveals that “Saussure’s first course included a passage on analogy […] taken directly from Sechehaye’s book”.

58

Chapter 3

Given the lack of further information, the most realistic reading of the events suggests that both Saussure and Sechehaye plucked it from the academic air, where this distinction had been circulating for some time, but that Sechehaye was the first not only to publish it but also to ”make it the cornerstone of any linguistic edifice”. 3.2.5 The Speech Circuit To explain the notion of speech event, Saussure finds a useful point of departure in what he calls the speech circuit (‘le circuit de la parole’) (Cours: 27), which consists in the circular process starting when a ‘concept’ (in Saussure’s terminology; we would say ‘propositional thought under a speech act operator’) arises in the mind of one speaker (A), who converts the ‘concept’ into sound via the acquired system of his or her specific language (‘langue’), followed by the physical transmission of the sound to the ears of the hearer (B), who recreates the corresponding ‘concept’ in his or her own mind and may call up another ‘concept’, which is then converted into sound, etc. Figure 2 ­reproduces Saussure’s schematic rendering of this circular process. The purely psychological part consists in the mutual interaction between what he calls ‘concepts’ and ‘acoustic images’ in the heads (minds) of the participants A and B, as shown in Figure 2b. One notes that Saussure makes no mention of the fact that there can be no parole, no language use, unless it is about something, either in the real or in a virtual world (for the notion of virtual reality, see Seuren 2009: 55–84). What is missing in his analysis of the speech circuit is the fact that, both in the production and in the interpretation of actual linguistic utterances, the participants in the speech circuit are in a mental relation of intentionality with regard to things in the real or an imagined but in any case external world—a relation which must be taken to be mediated by what he calls the ‘concept’, though this notion is left undefined in the Cours or anywhere in Saussure’s private notes.

a.

b.

Hearing

c A

B

i A Speaking

Figure 2

.......... c : concept i : acoustic image ..........

Saussure’s speech circuit (‘circuit de la parole’) (Cours: 27–28)

Speaking

c

i B

Hearing

The Cours: A Critical Look

59

This issue is taken up in Section 3.4.1, when Saussure’s notion of the linguistic sign is discussed. Saussure’s failure to take the outside world into account in his analysis of the notion of language is to be seen in the context of his repeated insistence that everything in language is psychological. We must assume that, for Saussure, the question of how language enables talk about and reference to elements in the, or a, world has nothing to do with language. One may perhaps surmise that Saussure would assign this property of natural language to the parole, but one cannot be certain, because he never comments on this question. What one does find, all over the Cours in all sorts of ways, is the view that, because ­everything in language is psychological, the physical properties of the sounds produced by speakers in actual speech are irrelevant to the study of the langue: the investigation of actual sounds belongs to the study of parole (Cours: 21): At bottom, everything in any language is psychological, including its material and mechanical manifestations, such as sound changes. […] The essential thing about languages, as we shall see, is alien to the phonic character of the linguistic sign. But the study of the psychological relation of reference by linguistic means is never mentioned. The exclusion of the outside world from semantics and the exclusion of actual sounds from phonology are of a piece: both reflect a solipsistic streak in Saussure’s thinking, resulting from a misinterpretation of Hippolyte Taine’s response to the Kantian paradox of knowledge.9 9 For Taine, as opposed to common opinion in his day, primary facts are facts of experience, not facts in a material world. This is clear from the following passage, which, according to Charlton (1959: 136), is repeated almost literally in two other works by the same author (Taine 1872, vol. ii: 507; translation by H. van Laun): We think that there are neither minds nor bodies, but simply groups of present or possible movements or thoughts. We believe that there are no substances, but only systems of facts. We regard the idea of substance as a psychological illusion. We consider substance, force, and all the modern metaphysical existences, as the remains of scholastic entities. This ontological agnosticism clearly derives from the philosophical dispute that has raged all over the Western world from René Descartes (1596–1650) until the present day regarding the epistemological status of the external world. The dispute peaked in Immanuel Kant’s (1724–1804) famous epistemological dilemma of the basic unknowability of the ‘thing in and for itself’ (‘das Ding an sich’). No doubt through the intermediary of Taine (see Seuren 2016), this dilemma reached Saussure and thus came to lie at the bottom of his obscure and confused, yet persistent, lucubrations regarding the ‘negative’ character of language, where all is said to be ‘negative’ and ‘oppositive’—called “domaine très obscur” by Saussure himself (Godel 1957: 83). See my critique in Section 3.5.2 below.

60 3.3

Chapter 3

The Notion of Syntax and the Notion of Sentence

3.3.1 The Notion of Syntax Like most linguists of his day, Saussure had no, or hardly any, idea of syntax in the modern sense of the term.10 The ‘grammar’ he speaks of is, in principle, a system for the formation of words, or perhaps a store of word forms—that is, morphology. Like practically all other authors of the period, Saussure places syntax in the ‘parole’, not in the ‘langue’, as appears from his definition of parole quoted above from (Cours: 30–31) and repeated here for convenience: By contrast, speech (‘la parole’) is an individual act of the will and of the intellect, in which we can distinguish: (1) the combinations by means of which speaking subjects make use of the code of the language in order to express their personal thoughts; (2) the psycho-physical mechanism allowing them to externalise these combinations. Saussure does speak of “the code of the language”, but this notion, which could have been a good starting point for the development of a theory of syntax, is not further elaborated or defined in the remainder of the Cours. What Saussure means by his use of the term syntax is stated in (Cours: 186): [Syntax] has for its object the functions attached to the linguistic units, whereas morphology only considers their form. Morphology, for example, merely states that the genitive of the Greek word phýlax (‘guardian’) is phýlakos, but the syntax informs us about the uses of these two forms.11 The closest he comes to the modern notion of syntax is in his distinction between ‘associative’ and ‘syntagmatic’ relations (Cours: 170–175). ‘Associative’ or, as they were also called later, ‘paradigmatic’, relations are manifest not only in morphological paradigms, such as the Latin case system: hortus (‘garden’) for the nominative case, horti for the genitive, horto for the dative, etc., but also in lexical derivations such as from garden to gardener. By contrast, syntagmatic 10 11

For a factually rich survey of the early history of the notion of syntax, see Graffi (2001), Chapter 2: “The rise and fall of ‘psychologistic’ syntax”. This is precisely the classical, nineteenth-century notion of syntax that my generation was taught when we started on the first principles of Latin and Greek in the 1940s. The ‘syntax’ part of our Latin and Greek grammar books consisted in the specification of the various uses of the nominal cases, such as the subjective and the objective genitive, or the ablative absolute in Latin, corresponding to the genitive absolute in Greek, etc.

The Cours: A Critical Look

61

relations exist between words or morphemes as they occur in phrases or sentences. Thus, there is a syntagmatic relation between a preposition and the noun phrase commanded by it, as in for the king, and again, within the noun phrase the king, between the article the and the noun king. This causes a problem for Saussure (Cours: 172): One could make an objection here. The sentence is the prototypical kind of syntagm. But it belongs to speech (‘la parole’), not to the language (‘la langue’) […]: doesn’t it follow, then, that the syntagm is part of speech? We do not think so. What is typical of speech is the freedom of combinations; one thus has to ask whether all syntagms are equally free. He then mentions, first, the fact that every language has set expressions or ­collocations, such as English Never mind or Like father, like son, which are reckoned to belong to the language, not to speech. After that, he passes on to morphological constructions, such as unpardonable, which can be decomposed into constituent morphemes, to end up with the notion of ‘regular sentence patterns’, as in The earth rotates or What did he say to you?, which “find support in the language on account of concrete memories”, adding that (Cours: 173): One has to admit that, in the domain of syntagms, there is no clear separation between facts of language—marks of collective usage—and facts of speech, which are subject to one’s individual liberty. In many cases it is difficult to classify a combination of units, because various factors have worked together to produce it, in proportions that it is impossible to determine. Saussure is here rattling at the gates of syntax, but does not succeed in gaining access. Thus, for Saussure, as for the vast majority of linguists and linguistic d­ ilettanti of the two centuries preceding him, syntax, in the sense of how to ­combine lexical elements for the expression of any given propositional thought, was not a part of the language system (‘langue’) but, in the spirit of the romantic movement, of a free, stylistically creative and aesthetically pleasing combination of concepts and the corresponding lexical elements—that is, of ‘parole’. For Saussure, ‘grammar’ equalled ‘morphology’ and what we now mean by ‘syntax’ was unknown to him, though it was beginning to show at the horizon. A single example will show what is meant. Saussure, along with all linguists of his day, was fully aware, for example, of the fact that the vowel change ­between the English singular foot and its irregular plural feet is an isolated

62

Chapter 3

r­ emnant from Old-Gemanic, where the plural of masculine nouns involved an umlaut, in this case a change from /u/ into /i/ (Cours: 110, 122), and that the same umlaut is found in the German regular plural Füße for the singular Fuß, though the German umlaut changes /u/ into /ü/. Yet, again along with most of the linguistic world of his day, Saussure had no idea of syntactic facts, such as the fact that the markedly different word orders in the English sentence (3.1), the corresponding German sentence (3.2), the corresponding Dutch sentence (3.3), and the French equivalent (3.4) are likewise a matter of rule-bound ­constraints within the communities of English, German, Dutch and French speakers, respectively, not at all of a free, stylistically creative and aesthetically pleasing stringing together of words: (3.1) Albert has never wanted to make his pupils read the bible. (3.2) Albert hat seine Schüler niemals die Bibel lesen lassen wollen. (literally: ‘Albert has his pupils never the bible read make want’) (3.3) Albert heeft zijn leerlingen nooit de bijbel willen laten lezen. (literally: ‘Albert has his pupils never the bible want make read’) (3.4) Albert n’a jamais voulu faire lire la bible à ses élèves. (literally: ‘Albert has never wanted make read the bible to his pupils’) What we have here are clearly syntactic differences between the four languages, which, inter alia, have different complementation systems, that is, systems for the treatment of propositions embedded as arguments under a higher predicate.12 It wasn’t until the 1950s (Bech 1955, 1957) that European linguists began to be aware of the stringency of phenomena of this kind. Since then, developments in the theory of grammar—especially across the Atlantic, where Bloomfield’s Language of 1933 played a decisive role—have consolidated the notion of syntax as a partly language-specific and partly universally constrained system of rules for the expression of propositional thoughts under a speech act operator, and thus as an integral part of the grammar of any specific language. One may well find it surprising that the linguistic world was unaware of the notion ‘syntactic system’ during the whole of the nineteenth century and a good part of the twentieth. It would seem that, partly at least, the cause of this ‘blindness’ may have lain in the dominance of romanticist thought, which more or less eclipsed rationalist thought during that period. Romanticism in linguistics rejected any appeal to such rationalist notions as Aristotelian propositional structures and promoted the idea that combining words into 12

See Seuren (2003; 2017, Chs. 3–6;) for a full technical treatment of the phenomena in question.

The Cours: A Critical Look

63

phrases and sentences is a matter of natural aesthetic feeling rather than of formal rule systems. In that climate, observations such as those in (3.1) to (3.4) above simply could not be made: the sentence structures involved felt so ‘natural’ that speakers didn’t realize how complex and how rule-bound they are. But a more complete answer to this question clearly requires more detailed scrutiny. Linguists had not always been so blind to syntactic phenomena. To mention just a few highlights, the second-century CE Alexandrian Greek grammarian Apollonius Dyscolus already had some idea of syntactic constructions (Seuren 2008; Allan 2010: 103–110). The Spaniard Franciscus Sanctius or Sánchez (1523-1600), in his epochal work on linguistic theory, Minerva seu de Causis Linguae Latinae, of 1587 even showed that he had a fairly clear idea of a syntactic system transforming semantic ‘logical’ structures into well- formed surface structures (see Lakoff 1969; Breva-Claramonte 1983; Seuren 1998: 37–41; Allan 2010: 183). The Port-Royal grammar (Arnauld and Lancelot 1660/1966) continued this line of thought, using Aristotelian propositional structures as ‘input’ to the syntax, though without giving specific rules relating the propositional to the surface structures. But as the eighteenth century unfolded, romanticism gained ground and has kept some of it till the present day. Now that rationalism has made a come-back and our professional eyesight as regards syntax has been sharpened again, we know that syntax constitutes a huge part of the grammatical system of most languages, at least as large as morphology, with which it is in close interaction. But this insight only grew gradually, and it did so in the context of European and American structuralism in linguistics, with Sechehaye as an early forerunner, who was among the first to realise the immensity of the task of relating semantic representations to surface structures. Early twentieth-century structuralism re-ignited the linguists’ awareness of rules and regularities, first in morphology and then, bit by bit, also in syntax, as one began to develop an eye for syntactic phenomena. It was first by learning to see the complications and then by catching glimpses of the underlying system that the notion of syntax grew on linguists, who started by having a subliminal inkling about it but who had not been trained to descry the relevant phenomena, let alone to formulate rules or regularities for them. That bit of professional training was developed gradually during the twentieth century, first in the context of linguistic structuralism, then in generative grammar. All that linguists in Saussure’s day and age could do was talk about the ‘system’ of what we call ‘syntax’ in vague and abstract terms. Not long after Saussure’s death (and in rare cases even before, as in Ries 1894, discussed below), the notion of syntax slowly and gradually began to hatch out of its shell, to take

64

Chapter 3

shape eventually as the theory of sentence formation. Syntax is thus to be seen as belonging to the langue and not to the parole. 3.3.2 The Notion of Sentence Albert Sechehaye grappled with the notion of syntax in his masterful Essai sur la structure logique de la phrase of 1926, where he posits the primacy of the sentence over the word.13 This naturally raises the question “What is a sentence?” From Sechehaye’s texts one gathers that he considers a sentence to be a propositional structure or hierarchically ordered composition of propositional structures, explicitly so at the semantic level but transformed (and often disguised) at the surface level of grammatically formed sentences of specific languages. Sechehaye makes the incongruity between propositional structures reflecting thoughts on the one hand and well-formed surface structures on the other his central problem, complaining that he is hampered by the commonly accepted view that the sentence is a unit of ‘parole’, not of ‘langue’: “It is the opposition between the grammar and the parole that is bothering us” (1926: 120). This “opposition between the grammar and the parole” in the concept of sentence was a recurrent prominent theme in the linguistic literature of the 1920s. The first to propose that the sentence is a unit of ‘parole’, not of ‘langue’, was Hermann Paul (1846–1921), who defined the sentence as follows (Paul 1920: 121; the definition remained constant through the many editions of this book, which was first published in 1880): The sentence is the linguistic expression, or symbol, of the fact that a connection [‘Verbindung’] between various images [‘Vorstellungen’] or groups of images [‘Vorstellungsgruppen’] has taken place in the mind [‘Seele’] of the speaker, and is the means for bringing about the same connection between the same images in the mind of the hearer. We understand from this that, for Paul, the sentence is a token event, and therefore part of ‘parole’, not of ‘language’. In the 1920 edition of his book, Paul fulminates against Wundt, for whom a sentence starts not as a connection made by the speaker of two pre-existing elements but as a primarily given whole, which is subsequently split up into component elements. This, however, is not a question that concerns us here. More relevant is the fact that Paul refuses to recognise a proposition in what he calls ‘connection’, and his blindness to the fact that such connection (or proposition) events have a structure that is fed 13

See, for example, Sechehaye (1926: 100): “Our words are thus made for the sentence”, or Sechehaye (1926: 116): “Despite the priority of the sentence over the word […}”.

The Cours: A Critical Look

65

into a type-level grammatical machinery to produce the right sentence for the right language—which makes the sentence a type-level unit of language, not of speech: for the speech event resulting from the realisation of a type-level sentence we use the term utterance. (Wundt suffers from the same blindness, which was an important element in Sechehaye’s critique of Wundt’s theory.) John Ries (1857–1933) spent a lifetime on questions such as “What is syntax?” (Ries 1894) or “What is a sentence?” (Ries 1931), yet failed to come to any clear answer, despite his often sharp intuitions. Ries (1894) shows the author struggling with the problem (Ries 1894: 142): Syntax is not a semantic account of word classes and word forms, for the opposite of syntax is not the theory of forms but the theory of words [‘Wortlehre’]. The counterpart of the theory of forms is not syntax but the theory of meaning [‘Bedeutungslehre’]. Syntax […] deals with the question of how words are combined into new units […]. Its object are the word combinations—all word combinations, not just sentences, and nothing but the word combinations, not also the word classes and the word forms. [emphases original] This is not very specific, but we should not forget that we are still in 1894. Ries is more advanced here than his exact age-mate Saussure, for whom syntax “has for its object the functions attached to the linguistic units, whereas morphology only considers their form” (Cours: 186; see also note 11 above). In fact, in the passage quoted above, Ries shows up Saussure’s mistake, as Saussure does indeed confuse syntax with the semantic account of word classes and word forms. In his (1931), the counterpart to Ries (1894), Ries concludes, problematically, that sentences are units of speech, not of language, but that the constituent parts of sentences are in the remit of the grammatical system of the language at issue (Ries 1931: 62): The relation of the sentence to all earlier objects of the grammar differs from the relations the latter have with regard to each other. […] Sounds, words and word groups are artificial units of the grammar, in a sense, abstractions. The sentence is a natural unit and a linguistic reality. The former are mere products of analyses carried out for scientific purposes, elements of language that have been lifted out of their natural context and have no real life of their own. The sentence, though likewise isolated as a research object for the study of grammar, is, by contrast, not comparable to a dead laboratory preparation, but to a living organism that also

66

Chapter 3

exists outside the grammar. It is not just an element of the language but is itself language, language as real action. The sentence is living speech itself. […] Sentences are themselves always and everywhere living speech, whereas single words and word groups never and nowhere are. This means that they belong to two essentially different species of linguistic constructs. […] A sentence is a unit of living speech that does not allow for analysis into elements of the same kind, or, in short, a smallest unit of speech. How the combination of “dead laboratory preparations” can lead to “a living organism” is not explained. In the end, Ries presents his definition of what a sentence is (Ries 1931: 99): A sentence is a grammatically formed smallest unit of speech, which expresses its content with a view to this content’s relation to reality. [Ein Satz ist eine grammatisch geformte kleinste Redeeinheit, die ihren Inhalt im Hinblick auf sein Verhältnis zur Wirklichkeit zum Ausdruck bringt.] What Ries aims at defining here is not the concept of sentence but that of utterance: sentences are type-level units constructed according to the rules of a specific language and have, as such, no “relation to reality”, since for such a relation to be established one needs, besides a speech act operator, reference relations between definite terms and world entities, and such relations are not established by natural-language grammars, but by the anchoring and keying of any given sentence in an actual context of use—that is, when a sentence is realised as an actual token utterance seriously used in an actual situation (see Section 5.5.2 for further comment). Had Ries said that a sentence is (or contains) the rule-governed expression of a proposition—that is, the token mental act of assigning a property to one or more entities (see Section 5.5.1)— which can be true or false with regard to a contextually given state of affairs, he would have been closer to a solution, but both the notion and the term ‘proposition’ were distinctly out of favour at the time, as linguistics had to be ‘autonomous’, which implied ‘free from logic’. What Ries, and others at the time, failed to do is have a systematic close and detailed look at actual sentences in different languages and see how differently they express what may be reckoned to be the same proposition. Had they done so, they might well have detected systematic differences and thus have concluded that there are specific syntactic rules that hold for each language. Ries feels an inkling of remorse in this respect, yet he declines to actually do something about it, falling back on empty talk (Ries 1931: 96–97):

The Cours: A Critical Look

67

What characterizes all sentences in all languages, despite their great differences in form, is their grammatical form in and for itself. The formal property that is immanent in all of them is the very fact of being formed. The sentences of different languages have different forms, but what they have in common is that they have a form. Less obscure is the brilliant English Egyptologist and theoretical linguist Alan H. Gardiner (1879–1963), who, in his book The Theory of Speech and Language of 1932, likewise still cannot quite make up his mind about whether syntax belongs to speech (‘parole’) or to language (‘langue’). Only at the end of the book does he conclude that syntax really is more a question of language than of speech (Gardiner 1932: 184): “Thus there is such a thing as ‘sentence form’, and like all other linguistic forms, it is a fact of language, not a fact of speech”. And, after a few pages in which he still tries to uphold the notion that syntax is a matter of speech, not of language (Gardiner 1932: 199): “Be this as it may, sentence-form is indisputably the main device by which speakers ensure the right acceptance of their utterances”. And then again (p. 200): “So far as it depends upon words at all, outer sentence-form exists in the mind as a certain aptitude for putting the right words together in the right way so as to yield the appearance appropriate, as the case may be, to a statement, an exclamation, a request, or a question”. The Danish linguist Otto Jespersen (1860–1943) is another early exponent of modern syntactic theory—and practice. The theory is found especially in his widely read work The Philosophy of Grammar of 1924 (where he reinstates the notion of proposition under the name of ‘nexus’, perhaps to avoid any association with logic) and his less known Analytic Syntax of 1937. In his standard work Essentials of English Grammar of 1933 he shows his talent for detailed language description, including syntax. His monumental seven-volume work A Modern English Grammar on Historical Principles, covering four decades of work, simply testifies to the greatness, industry and solid common sense of this linguist sans pareil. Due to his penchant for detailed observation, his notion of syntax is a great deal more sophisticated and more realistic than those of Ries or Gardiner, yet he did not achieve the breakthrough that Sechehaye, who uncompromisingly reverted to the old notion of proposition, managed to achieve. As regards Saussure, he hardly ever refers to Jespersen, and on the rare occasions he does, it is in a cursory way and on points of minor importance. The early twentieth-century budding of the modern notion of syntax as the theory of sentence structure shows how linguistic thought gradually liberated itself from the influence of romanticism and idealism, and how difficult it was to shed the blinkers blocking an uncluttered, or less cluttered, view.

68

Chapter 3

That breakthrough was crucially dependent on the insight that a sentence is the finished product of an algorithmic procedure, first thought, in the school of Chomskyan linguistics, to be triggered by an ‘initial symbol S’, then, in the school of Generative Semantics, taken to be triggered by a highest predicate with its choice of argument terms—that is, by a propositional structure. In our modern age, we may say that the notional problem of the definition of the concept ‘sentence’ has been solved, even though this remarkable fact has, to my knowledge, never been mentioned explicitly in the literature. The modern concept of sentence is a formal explicitation of the ancient concept, found, for example, in Priscian’s Institutiones Grammaticae (± 500 CE): “A sentence is the systematic ordering of words expressing a complete meaning” (Oratio est ordinatio dictionum congrua sententiam perfectam demonstrans). The problem with this ancient definition (which is repeated all over the medieval and later literature on grammar) is that it is not made clear what a ‘complete meaning’ amounts to, and that it fails to specify the rule system alluded to in the adjective congrua, translated here as ‘systematic’. The Port-Royal Grammar of 1660, mentioned in Section 2.4 above, offers some relief, in that it tells the postmedieval reader (reiterating a tradional insight, which was thrown to the winds in modern linguistics) that the “complete meaning” in Priscian’s definition must be read as ‘judgment’ (‘jugement’), which stands for a propositional subject-predicate structure with a truth value (Arnault & Lancelot 1660/1966: 68): The judgment that we make of things, as when I say “The earth is round”, is called ‘proposition’. Therefore, each proposition necessarily encloses two terms, one called subject, which is the thing the affirmation is made of, such as the earth, and one called attribute, which is that which is affirmed, such as round. […] It is easy to see that these two terms belong precisely to the first operation of the mind, because that is what we conceive and what is the object of our thinking. What was still lacking, after this clarification, was a specification of the rule system implied by Priscian’s “congrua”. That rule system was precisely the object of Sechehaye’s concern, when he spoke of the ‘grammatical problem’, his ‘problème grammatical’ (see the next chapter). And since Sechehaye, anticipating twentieth-century developments in formal grammar, saw a grammar as a formal, algorithmic mapping from propositional ‘judgments’ on to the grammatically well-formed sentence structures of each specific language, we may say, in modern terms, that a sentence is the result of a language-specific

The Cours: A Critical Look

69

algorithmic procedure triggered by the highest predicate of a proposition and its choice of argument terms, and brought to a close when all argument terms have been fully elaborated. This is a so-called ‘implicit’ definition, in that it does not provide external criteria for sentencehood but simply says: a sentence is what is defined as a sentence in the grammar. 3.4

The Notion of Sign and Its History

3.4.1 Saussure’s Notion of Sign As we have seen, a language is, for Saussure (Cours: 32), the shared psychological property of a community, consisting of a lexicon and a grammar (that is, for him, morphology). The lexicon is a collection of linguistic signs. Therefore, Saussure consistently concludes, a linguistic sign is the shared psychological property of a community. A linguistic sign, moreover, is, like all signs, a psychological element sui generis, in that it links up a concept, also called the signified (‘signifié’),14 with an acoustic image, also known as the signifier (‘signifiant’), both being of a psychological nature. This union of a concept (signified) and an acoustic image (signifier) is famously visualised as in Figure 3 (Cours: 99).15 On pp. 155–156 of the Cours, Saussure presents a different take on the relation between form and meaning. There we find the well-known Figure 4 (Cours: 156), meant to illustrate the structuring power of language vis-à-vis thought (A) and sound (B).

14

15

The first technical occurrence, to my knowledge, of the term signified (French signifié) as a noun in the Saussurean sense is in Victor Egger’s (1881: 281) discussion on whether a sign precedes the signified (as clouds precede rain), or is simultaneous with (as a red face goes together with shame), or follows the signified (as destruction follows the passing of a cyclone): “No doubt the sign precedes the signified in certain cases, but in other cases it is simultaneous with it or follows it”. (‘Sans doute le signe précède le signifié dans certain cas, mais d’autres fois, il l’accompagne ou le suit’.) (italics mine; pams). The term also occurs in Taine (1878, vol. i: 331), but only as a a past participle adjectivally attached to the noun événement (‘event’). Saussure was well acquainted with Egger (1881), which he read and made notes about immediately upon its publication (Joseph 2012: 288–291). In an unfinished draft of a book, to be entitled “L’essence double du langage” and dating from about 1894, Saussure uses the term sign not for the unit of form and meaning, but for just the perceptible form (e.g. Saussure 2002: 20–21), which is more in accordance with common usage.

70

Chapter 3

concept

acoustic image

Figure 3

Saussure’s analysis of the linguistic sign (Cours: 99)

A

B

Figure 4

Saussure’s view of the relation between form (B) and meaning (A) (Cours: 156)

The accompanying text runs as follows (Cours: 155): Psychologically, abstraction being made of its expression by means of words, our thought is nothing but an amorphous and indistinct mass. Philosophers and linguists alike have always agreed that, without the help of signs, we would be incapable of distinguishing two ideas in a clear and constant way. Taken by itself, thought is like a nebula where nothing is necessarily determined. There are no pre-established ideas and nothing is distinct until the language [la langue] appears. […] Nor is the phonic substance more fixed or more rigid. It is not a mould where thought has to marry a form but a form of plastic matter dividing itself, in its turn, in distinct parts so as to provide the signifiers needed by thought.

The Cours: A Critical Look

71

It is hard, if not impossible, to find a coherent interpretation for this passage, which is in any case fundamentally wrong. There are two ways in which it can be read. One can either take the term thought (pensée) to refer to the actual process of thinking and read the term phonic substance (substance phonique) as ‘actually produced speech sounds’, in which case we are at the level of parole, or we can read these terms on the level of langue, as referring to what is now known as ‘cognition’ on the one hand and system-driven recipes for sound production on the other. On neither interpretation does the passage make sense. If read at the level of parole, the underlying thought is far from “an amorphous and indistinct mass” but is a structured composite of one or more interrelated propositions. Here the question presents itself of what we should understand by the term proposition, which has been the victim of wildly diverging definitions or lack of them Here, as elsewhere in my work, I define a proposition as: the mental act of—truly or falsely—attributing a property to one or more entities. Whereby it is to be understood (i) that the precise specification of the conditions under which a property can be assigned to entities in such a way that a true or false proposition results is, in most cases, very difficult to specify verbally (think of the famous qualia problem), and (ii) that the term entities does not apply to those entities that philosophers take to populate the actual world, in whatever (material) form, according to their ontological theories, but to the entities as they figure in, possibly naïve, human world construals on the basis of incoming sensory input plus the cognitive apparatus humans bring along when born. The entities intended here thus include all kinds of abstractions and reifications, entities we know do not exist but have somehow been thought up or imagined (as in the proposition expressed by the sentence The centaurs of Greek mythology have the lower body of a horse and the upper body and head of a man), even whatever may be taken to correspond to embedded propositions, as expressed in sentences like That John arrived this morning surprises me, where the subject is the so far mysterious entity denoted by the embedded phrase that John arrived this morning. Philosophical theories about rockbottom ‘reality’ apply to humans only in so far as they distinguish actual from virtual entities, situations and events. Humans have no choice but to work with their world construals. And these are replete with things they take to actually exist, or not to exist but which, for example, they plan to build, or wish that had

72

Chapter 3

been built, etc. It is to such cognitively shaped ‘entities’ that humans mentally assign properties thus forming propositions which together make up thought content. And such thought content is anything but a Saussurean “amorphous mass”. On the contrary, it must be taken to be highly structured, even if, so far, we do not really know much about the specific structure and organisation of such ‘thought contents’. Nor is the “phonic substance” of the stream of speech anything like “a form of plastic matter”, whatever that may mean. The best one can make of the term phonic substance on this interpretation is to let it refer to what is known as ‘white noise’, that is, an indistinct mishmash of sound where all frequencies have roughly the same amplitude, like the diffuse noise produced by distant traffic on a motorway. But a stretch of speech sound is anything but white noise: any phonetician will tell you that speech sound is highly articulated, with distinct frequency and amplitude patterns and transitions between them, marked by fractions of milliseconds of silence for the occlusives. The linguistic problem consists in recognising phonemes in these sound patterns. When taken at the level of langue—that is, with “thought” referring to cognition and “phonic substance” to system-driven recipes for sound production— the passage is equally bizarre. Cognition is a highly structured machinery, ­producing and containing inter alia concepts defined by satisfaction conditions, entity representations of many different kinds, and a network of interrelated propositions once produced and now stored in a memory or knowledge base in ways that are as yet poorly understood. Some of these stored propositions are classified as ‘true’, others as the product of imagination and thus as either ‘false’ or ‘possibly true’ (leaving necessarily false propositions out of account). And the recipes for sound production, are, of course, anything but “a form of plastic matter dividing itself, in its turn, in distinct parts so as to provide the signifiers needed by thought”—again, whatever that may mean. According to Saussure, “philosophers and linguists alike have always agreed that, without the help of signs, we would be incapable of distinguishing two ideas in a clear and constant way” (Cours: 155), but he fails to provide any references or examples. In actual fact, Saussure himself appears to be fairly unique in holding this view, which is Whorfian over the top and not held by the rank and file of Whorfians. And with good reason, for it would imply that prelinguistic humans or nonhuman animals, who had, or have, no language, were, or are, unable to make cognitive distinctions or recognise objects of a certain category. Another, much emphasised, aspect of Saussure’s notion of sign is that linguistic sign systems are but an instance of sign systems in general, such as

The Cours: A Critical Look

73

­ antomime, gestural languages, traffic, naval or military sign systems, symbolic p rituals, etc., but they are at the same time the most important among them (Cours: 33, 101). Sign systems have in common that “they express ideas” (ibid.), though—in contrast to Sechehaye (1908a: 49)—no further analysis is provided of what is meant by “express” or, for that matter, by “ideas”. The study of sign systems in general Saussure calls semiology (Cours: 33), a field of study yet to be set up but, in Saussure’s view, of immense potential importance for a proper understanding of humanity. Linguistics will thus be a branch of the more general discipline of semiology, which falls within the remit of psychology. Semiology, or the study of signs, must thus be seen as a subdiscipline of psychology. Had Saussure lived, he would have been disappointed to find that his vision of semiology as an overarching discipline has come to nothing—despite a few vain attempts—chiefly because, during the twentieth century, the notion of sign, though rightly central to his thought and that of many of his contemporaries, came under a cloud of disrespect in all branches or schools of linguistics, which wanted no truck with the vaguely esoteric and semi-mystical ‘semiotic’ movements, such as ‘General Semantics’, which arose during the 1920s partly on the basis of Saussure’s in itself not unreasonable but unelaborated notion of semiology. In my view, the almost total neglect of, and the implicit contempt for, the notion of sign in twentieth-century linguistics has done great harm to the discipline. The glaring lack of serious semantic theorising in the language sciences of the twentieth century is by far their most outstanding shortcoming.16 Saussure himself never worked out the general notion of sign any further, though, with a better knowledge of history, he could have found useful starting points in the Stoic, Augustinian and medieval notions of the sign, as shown in Section 3.4.3.17 16

17

The movement of formal semantics that developed in the wake of Richard Montague’s ‘possible-world semantics’ cannot be considered relevant to the analysis of natural language. This form of semantics may be interesting, or challenging, to some formally minded philosophers or mathematical logicians, but it fails, inter alia, to take into account the fact that human cognition, though subject to the metaphysical necessities expressed in standard mathematical logic, which is defined for the unrestricted universe of all possible situations, works with universes of situations that are restricted in a number of highly functional ways, with the result that natural human logic has a much larger number of logical relations and thus has much greater ‘logical power’ than the universally valid logic developed by mathematicians during the twentieth century. Possible-world semantics is thus not mirrored as such in human cognition and is, moreover, in conflict with wellknown observational facts of human language. See Chapter 9 in Seuren (2013a) for extensive discussion; see also note 30 in Chapter 4. I presented a compact statement of my own, causal, analysis of the notion of sign in Seuren (2009: 280–284). In my view, a sign is a perceptible (token) form F that makes the

74

Chapter 3

3.4.2 The Type-Token Distinction Of greater importance is the fact that Saussure is at pains to stress that the linguistic sign is of an entirely psychological character: it “unites not a thing and a name but a concept and an acoustic image” (Cours: 98). Saussure explicitly and repeatedly dismisses the notion of a linguistic sign as a term that corresponds to, or ‘signifies’, an object or a thing (‘une chose’) in the real or in any imagined ‘virtual’ or ‘intensional’ outside world. Above, in Section 3.2.5, I have criticised Saussure for his failure to account for the fact that the use of language necessarily involves a relation of intentionality with regard to things in the real or an imagined world, the necessary aboutness of speech. This criticism does not apply directly to Saussure’s notion of the linguistic sign, though it does indirectly. It does not apply directly since, as we have seen, Saussure maintains that a language (‘langue’) is a system of lexical signs deposited in the minds of the individuals belonging to a language community—that is, a language exists in the minds of individuals only at type level, the tokens being produced during actual speech (parole). But the criticism does apply indirectly, since any theory perceiver know, on grounds of available knowledge, that there is something else, say C, not perceptible at the time, which must have caused F. The meaning of a sign F is F’s property of making C known to the (informed) perceiver. Whoever perceives smoke and has sufficient world knowledge, knows that combustion is going on, even though the combustion itself is not perceptible at the time of perception. In this sense, the smoke is a sign of combustion (note the preposition of). This is an example of a natural sign. Conventional signs differ in that the nonperceptible cause C is a mental intent on the part of the sender, consisting of (i) a socially binding superordinate operator of assertion, command, prohibition, question or wish, and (ii) a referential propositional thought that the socially binding operator applies to. Linguistic signs—that is, linguistic token utterances (see Section 3.4.2 for the type-token distinction)—are perceptible (auditory, visual, gestural) manifestations, normally under sender’s control, of mental intents that are structured according to an underlying grammatical system, plugged in to a lexicon and combined with a phonological, graphical or gestural system and a corresponding production machinery, which together transduce the mental intent into perceptible spoken, written or gestured material. In the case of linguistic token signs, the mental intent C is intentional, that is, C is referentially focused on—the technical term is keyed to—entities in situations or events in the actual or some thought-up world. In this sense, the utterance is a sign of the mental intent C, while containing one or more (lexical) signs for world entities. At type level, a conventional sign F is not keyed to any specific, context-bound world domain, but has the potential of being keyed that way. For linguistic signs this means that they must be structured according to an underlying system in a way that will allow for successful keying in an infinite variety of possible uses. In this theory, the sentence takes primacy over the word, in that the primary signs are token-level utterances, reflecting type-level sentences. Lexical words are signs only in a secondary sense. This, however, is my analysis, not what we find in the Cours. Today’s literature about matters of meaning and reference suffers from a fatal lack of systematic analysis and hence from hopeless confusion.

The Cours: A Critical Look

75

of language that fails to take into account the fact that it is a central feature of human language, as a type-level system, that it makes token-level reference to the outside world possible, will be severely beside the point. The problem of the identification of tokens as belonging to the same type was, so to speak, ‘in the air’ by the end of the nineteenth century, even though the terms type and token did not yet exist. The identification is mentioned in Taine: “A general and abstract idea is a name, just a name, the signifying name for a series of similar facts or a class of similar individuals, normally accompanied by the perceptual but vague representation of some specimen of this class of facts or individuals”. (Taine 1878, vol. ii: 259). It is dealt with as a psychological problem in William James’s The Principles of Psychology of 1890, where, in his stately American prose, he writes, on pp. 459-460 of Volume i: This sense of sameness is the very keel and backbone of our thinking. […] This sense of identity of the knowing subject is held by some philosophers to be the only vehicle by which the world hangs together. […] Note, however, that we are in the first instance speaking of the sense of sameness from the point of view of the mind’s structure alone, and not from the point of view of the universe. We are psychologizing, not philosophizing. That is, we do not care whether there be any real sameness in things or not, or whether the mind be true or false in its assumptions of it. Our principle only lays it down that the mind makes continual use of the notion of sameness, and if deprived of it, would have a different structure from what it has. But it was the American philosopher Charles Sanders Peirce (1839–1914) who introduced the now common terms token for unique hic-et-nunc objects or occurrences, and type for the product of the mental process attributing ‘sameness’ to different tokens reckoned to belong to the same type at any level of abstraction (Peirce 1906). Tokens are by definition unique and occur in (actual or virtual) reality; types are mental constructs resulting from cognitive identification processes. Mathematically speaking, types are ‘characteristic functions’, which take concrete, individual tokens as input and deliver a type-membership judgment as value: ‘this token does/does not fall under this type’ (pace vague borderline cases) (see Section 5.5.4). In this sense, both concepts and acoustic images are types, and thus of a mental nature, and so are the Saussurean signs, which weld the two together. Saussure, though unaware of Peirce (1906) and his pair of terms, had, like Taine, a clear notion of the distinction between type and token (Cours: 151–152):

76

Chapter 3

The problem of identities is found everywhere. […] Thus we speak of identity with regard to two express trains “Geneva–Paris, 8.45 pm” leaving at intervals of 24 hours. To our eyes it is the same express train, yet, in all probability, the locomotive, the carriages, the personnel are all different. Or else, if a street is demolished and built up again, we speak of the same street, even though, materially speaking, nothing of the old street may have remained. How can one rebuild a street from the ground up without it stopping to be the same street? Because the entity it constitutes is not purely material. It is based on certain conditions that are alien to its occasional material instantiation, such as its location with regard to other streets. Likewise, what makes the express train the train it is, is the hour of its departure, its itinerary, and in general all the circumstances that distinguish it from the other express trains. […] The link between two occasions of use of the same word is not based on material identity, nor on exact sameness of meaning, but on elements that will have to be sorted out and will take us very close to the real nature of linguistic units. It is clear that Saussure placed the langue at type level and parole at token level. And the type level is indeed purely mental, part of our cognitive system that makes us interpret and categorize perceptions of token events. Thus far, therefore, the criticism of not taking the referential or intentional aspect of language into account does not apply. But it begins to apply when we consider that one, perhaps the central, function of concepts—in the modern, not any Saussurean, sense—as type-level semantic functions, is their role in the process of identifying and classifying things we take to be part of the real or of some imagined world. Concepts alone are not sufficient for this process, but they are crucial in assigning conceptual labels to any given entity of any degree of abstraction. This intentionality or potential aboutness of concepts, and of language in general, is a central part of their very raison d’être. A definite noun phrase, such as the girl, is defined in the type-level language system, but it does not refer to any given individual entity until actually used as a token in a proper context or situation so that the entity the speaker has in mind can be identified by the hearer—if all goes well. Only then do propositions in sentences acquire a truth value. A sentence like The girl was right after all is a product of the English type-level language system. We can analyse it and identify its subject, its predicate, and the adverbial adjunct after all. But we cannot say whether it is true or false, or what girl the sentence is about. For that, we need to know

The Cours: A Critical Look

77

what ­specific girl is meant and what the occasion was at which her opinion or assertion was at issue. That is, the sentence has to be keyed to a specific historical part of the (or a) world. This keying process, as badly understood as it still may be, occurs at token level. It may indeed be considered not to be part of linguistics but, rather, of the psychology of the speech process. Yet the problem of how language in general, and each specific language in particular, makes it possible to key utterances to actual situations and thus to refer to specific entities and, ultimately, to assign truth values, must be considered to be part and parcel of the discipline of theoretical linguistics. Saussure’s failure to take this intentional-referential aspect of concepts into account in his notion of language may thus be taken as a very serious point of criticism (whereby one should take into account that the same neglect of this aspect is found in the first chapter of the first volume of Hippolyte Taine’s De l’intelligence of 1870). For Ogden and Richards (1923) this is sufficient reason for an ironically damning judgment on Saussure, “a writer regarded by perhaps a majority of French and Swiss students as having for the first time placed linguistic upon a scientific basis” (p. 4). Referring to Saussure’s theory of signs, they say (Ogden & Richards 1923: 6): Unfortunately, this theory of signs, by neglecting entirely the things for which signs stand, was from the beginning cut off from any contact with scientific methods of verification. Gardiner makes a similar point (Gardiner 1932: 60): If so acute a thinker as de Saussure has failed to note the necessity of ‘things’ to every linguistic act, we may be sure that the same error is widely held. It is for this reason that I have singled him out for special criticism. On the one hand it is correct to say that, at type level, the linguistic sign is a purely cognitive element uniting a concept with an acoustic (or graphic) image, which is in turn the type-level representation of token-level sound (or script) events. It is also true to say that it is only at token level that we can speak of a linguistic sign being a perceptible form showing the perceiver that a certain concept or a certain speech act-cum-proposition has occurred in the speaker’s mind with respect to specific entities in specific situations so that reference to those entities and situations becomes possible. But it is equally true to say that it is only at type level that we can find the explanation of that token-referential function of type-level linguistic sign systems.

78

Chapter 3

3.4.3 Some History of the Notion of Sign Saussure thus defines the linguistic sign as a type-level mental unit. In this respect, he deviates from the tradition, which has always defined signs as token events functioning as signs on account of their type-level interpretation. I cannot, in this study, trace the long history of the notion of sign through the ages, but I may quote St. Augustine (354–430), bishop of Hippo Regius (present-day Annaba in Algeria) whose definition of the sign was mainly inspired by the ancient Stoic philosophers and has been extremely influential through the Middle Ages and after. In his De Doctrina Christiana, at ii.1.1, he defines the sign as follows: For a sign is a thing which, besides the form that it presents to the senses, makes something else enter into the perceiver’s thought on account of itself. (Signum est enim res praeter speciem quam ingerit sensibus aliud aliquid ex se faciens in cogitationem venire.) In his Dialectica, a few lines down the beginning of Chapter 5, one reads: A sign is something which shows not only itself to the senses but also something else to the mind. (Signum est quod et se ipsum sensui et praeter se aliquid animo ostendit.) In the wake of St. Augustine, the notion of sign acquired great significance in medieval theology, where the world was considered to be one great sign of the existence of God. But as, during subsequent centuries, theology became ever more marginalised in the development of the sciences, both the public and the academic world lost interest in the concept of sign. Yet, after a few centuries of neglect, the notion of sign re-emerged around 1870 all over the Western world of philosophy and psychology, in the context of the then rapidly expanding study of perceptual processes. By that time, the religious element in the study of signs had disappeared and all attention went to the question of how mental images, Vorstellungen in German, could be seen as signs of, or for, things in the world. In his De l’intelligence, first published in 1870, Taine devotes the first chapter to the n ­ otion of sign (Taine 1878, vol. i: 25–32), soon followed by the French psychologist-­ epistemologist Victor Egger, who devoted the last chapter of his 1881 PhD-­ thesis on inner speech to that notion. The somewhat peculiar Charles Sanders Peirce, mentioned in Section 3.4.2, followed suit: his entire œuvre is replete

79

The Cours: A Critical Look

with discussions of the sign, but Saussure may be excused for not knowing about that, as Peirce was a bit like Saussure in that he was in the habit of publishing erratically or not at all (Peirce’s complete writings were published only recently). Saussure’s Cours of 1916 is just one among the many attempts at clarifying the modern notion of sign. Somewhat later on in the twentieth century, we find Ogden and Richards’ famous 1923 book The Meaning of Meaning, which refined the traditional Augustinian notion of the sign (see Seuren 2009: 229–241 for ample discussion). These authors presented the famous triadic diagram shown in Figure 5, where “symbol” stands for what is called “sign” here (Ogden & Richards 1923: 11). Ogden and Richards’ analysis of 1923 was followed by the German psychologist Karl Bühler’s (mentioned in note 4 above) ‘Organonmodell’ in his magnum opus Sprachtheorie of 1934, and by that of the American semiotician and philosopher Charles W. Morris (1901–1979), not to mention the doctrines and writings of the theosophical and other occultist movements that flourished at the time—with the result that the very notion of sign came under a cloud of disrespect during the twentieth century (see above). All these authors developed their own notion of what constitutes a sign, though not all of them equally consistently or fruitfully. One cannot say, therefore, that Taine, or Egger, provided Saussure with a notion of sign: their notions differed too much from each other. But one can say that, as from about 1870 and until about 1940, when the interest in semantics began to fade, at least in the US, the notion of sign was central to philosophical, psychological and linguistic thinking. Whatever the merits or demerits of these notions of the sign, it is clear that the tradition defines the sign as a token occurrence of a perceptible form, whose interpretation requires type-level mental processing which reveals the existence of Augustine’s ‘something else’ in the world, Ogden and Richards’ THOUGHT OR REFERENCE CORRECT* Symbolises (a causal relation) SYMBOL

Figure 5

ADEQUATE* Refers to (other causal relations)

............................................................. Stands for REFERENT (an imputed relation) *TRUE Ogden and Richards’ semiotic triangle

80

Chapter 3

‘referent’. But, of course, Saussure was free to define the sign at type level and thus keep the whole of the language system, including the type-level signs, within the bounds of each speaker’s mind. Not so, however, for the speech circuit shown in Figure 2 above. There it cannot be denied that token-level (actual or virtual) world elements and states of affairs are essentially involved: when I ask you to hand me that book, you are addressed as a real person and the idea is that you hand me, also a real person, that particular book, again a real thing. The keying of linguistic constructs to actual or virtual situations and entities is one of the chief raisons d’être of the language system, even though we have not so far discovered what it means, in analytical terms, to say that mental content is ‘intentionally related’ to world elements or how exactly this relation of keying is established. One may, therefore, justifiably criticise Saussure for leaving the element of reference to the external world out of account in his analysis of the speech circuit, which involves the occurrence of sign tokens. This again means that one may justifiably criticise Saussure for leaving the element of reference to the external world out of account in his analysis of the language system, since the system of any given language is subject to the requirement that it must make reference to the external world possible. In Section 3.2.5 above, this flaw in Saussure’s analysis was ascribed to a deficient understanding on Saussure’s part of the metaphysical ideas held by Taine, who defended a minimalist ontology in which, counter to then current scientific notions, most of the role of physical matter is sacrificed to the complexities of psychological processing—that is, extreme nominalism. 3.4.4 The Arbitrariness of the Linguistic Sign One reason to consider linguistic sign systems to be the most central and the most important category of sign systems lies, according to Saussure (Cours: 101), in the fact that the linguistic sign is, in principle, arbitrary—a point widely belaboured in a multitude of works at the time, as, for example, in Egger (1881), whose final Chapter 6 is devoted to this notion (Egger 1881: 266): In sum, arbitrary signs have a double advantage. First, they are the only possible expression of non-perceptual ideas and of perceptual ideas that cannot be externally represented by an analogous muscular movement. […] Secondly, they alone can express general ideas as such. There is no intrinsic reason why, to mention a hackneyed example, what we know as a tree should be called tree in English but arbre in French: the

The Cours: A Critical Look

81

d­ ifference is due to mere historical and sociological accident. This may be true, but does this exclude motivated lexical forms? It clearly does not. There is, as we know, quite a bit of sound symbolism or onomatopoeia, such as cuckoo for the bird of that name, in the lexicons of languages, but, one may say, these languages would function equally well without it. Yet the extent to which natural languages have availed themselves of the possibility to conventionalise motivated, sound-symbolic forms has not been studied with great enthusiasm by the majority of theoretical linguists. There are some exceptions. Edward Sapir carried out an experiment that showed a predilection for nonce words containing the a vowel to refer to large objects and for corresponding nonce words containing the i vowel to refer to small objects of a given category (Sapir 1929). Bloomfield (1933: 245) devoted some attention to sound symbolism. Dwight Bolinger (1949) even maintained that the role of sound symbolism is so important in the lexicons of the languages of the world that the global statement that the phonological form of lexical items is arbitrary is untenable. Although this is somewhat exaggerated, he does have a point. In general, however, the issue has not played much of a role in linguistic theory, perhaps because it is felt to be too ‘romanticist’ and not ‘rationalist’ enough (see Section 6.2). In psychology, however, there is a rich tradition in studies on sound symbolism; Levelt (2013: 438–444) provides a survey. We may also mention the little known Hungarian phonetician/psychologist Ivan Fónagy, who, in Fónagy (1963), gives further references, many of them quite old, and mentions more experiments.18 Recently, a rapprochement has become visible between psycholinguists and typological linguists on the question of sound symbolism. Mark Dingemanse established (Dingemanse 2011; Dingemanse et al. 2015) that

18

One could, for example, in the sense of Ohala (1997), investigate if, in the languages of the world, there is indeed, as I suspect there is, a statistical preference for low and back vowels (and perhaps also liquids and nasals) in words that denote global totality (English all, whole, total, French tout, Greek hólos, pâs, Irish ule, Maltese kull, Swahili -ote, Hausa dukà, Malay semua, segala (with prefix se-), Dyirbal mumbay, Kilivila bobova, kumwedona, etc.), as opposed to distributive totality, where high and front vowels seem to occur more frequently, often combined with dental or labial consonants (English every or each, Malay tiap-tiap, Kilivila te-tala). If this hypothesis turns out correct, this phenomenon may well be due to a natural phonaesthetic tendency to associate the back of the mouth with the concept of totality (think of swallowing) and, perhaps, the front of the mouth with the concept of separation (think of the function of the teeth). Further research should show to what extent phenomena of this nature infringe upon, or bias, the regularity of historical sound changes.

82

Chapter 3

onomatopoeia and the related phenomenon of ideophones19 play a role in the grammars of some (African, East-Asian, American-Indian) languages. In Saussure’s day, it was widely accepted that the possibility of a linguistic sign being arbitrary has the enormous advantage of allowing for a much wider range of possible forms than can be realized in systems containing only motivated signs.20 Saussure himself subscribed to that idea, but he turned it into an absolute principle, elevating it to a status of such august importance that henceforth all talk about sound symbolism became anathema, much to the detriment of linguistic theory (Cours: 100): The principle of the arbitrariness of the sign is contested by nobody, but it is often easier to uncover a truth than to assign it the place it deserves. principle dominates the entire linguistics of langue. Its consequences are innumerable. It is true that these are not all immediately visible; it is only after many detours that they are discovered, and with them the primordial importance of the principle. What Saussure had in mind when he spoke of “innumerable consequences” is his thesis that language systems contain no positive terms, everything being oppositive and negative—a position discussed in Section 3.5.2 below and found to be incoherent (Cours: 163): As there is no vocal image that corresponds better than any other to what it is meant to express, it is obvious, even a priori, that no language fragment could ever be founded, in last analysis, on anything else than on its noncoincidence with the rest. Arbitrary and differential are two correlative qualities. Saussure here comes close to denying a priori the possibility of motivatedness in lexical forms. Research in sound symbolism has meanwhile shown him wrong. The possible arbitrariness of linguistic forms is a necessary condition for the totality of natural language systems, or else their expressive power would be severely impaired, but it is not a necessary condition for all individual lexical items: a speech community is free to assign forms to individual lexical items 19 20

That is, semi-grammaticalised, largely conventional but sound-symbolic expressions, like English bang in And then it went bang. For an excellent exposé on this question, see the little known section in Stout (1899: 471–483) on what that author calls “natural signs”—one of the best texts on the relation between motivated (that is, Stout’s ‘natural’) and arbitrary signs I have ever seen.

The Cours: A Critical Look

83

that are motivated in any way whatsoever, either because of sound symbolism or phonaesthetic quality, or because of a publicly known connection with a person, a place, a historical event, or whatever. In fact, one expects language communities to maximise motivatedness of whatever kind, as motivated signs place less of a burden on memory. In this respect, therefore, Saussure, and with him the majority of language theorists, went over the top and thus stymied legitimate research. On the other hand, Saussure also stresses, correctly, that (Cours: 101) ‘arbitrary’ does not mean that each speaker can change word forms at will. On the contrary, the association of word forms with their meanings is fixed by the speech community, that is, in virtue of social acceptance.21 Arbitrariness thus allows for complex sign systems to come into being, such as natural languages, whose rules can to a large extent be defined regardless of specific forms for the signs used. The question of whether lexical forms are arbitrary or are, at least originally, naturally motivated goes back to Plato’s dialogue Cratylus, where Plato toys with the idea of a natural origin but does not commit himself. Aristotle takes a firm stand: single, non-composite linguistic signs (we now speak of ‘morphemes’) are in principle conventional and arbitrary, not natural or ­semantically motivated. In Saussure’s day, there were still authors, mostly romantically inspired dilettanti, who defended the natural-origin view, but the professionals, who predominantly belonged to the rationalist trend of thought, were largely agreed that the linguistic sign is in principle arbitrary, as long as its phonological or gestural properties stay within the bounds of possible physical realisation. According to Joseph (2003, 2004), the main source for Saussure’s notion of the arbitrariness of the linguistic sign was the Geneva-born Adolphe Pictet (1799–1875), a pre-Young-Grammarian comparative linguist and outspokenly romanticist philosopher, like Saussure belonging to the highest Geneva elite, related to the Saussure family and highly admired by Saussure in his teens ­( Joseph 2003: 367; 2012: 148).22 Such a link between Pictet and Saussure is, however, most unlikely. It is true—a fact given great prominence by Joseph—that Pictet stresses the arbitrary character of the linguistic sign, on the rare occasions he speaks of natural language in a general sense: “In language, sound 21 22

English has, or used to have, the word acceptation, French acception, for ‘commonly ­accepted meaning’. The Pictet and Saussure families have remained closely related till the present day, as appears from the intricate involvement of both families in the privately owned Banque Pictet & Cie, founded in Geneva in 1805 and now a leading asset manager for the world’s wealthy.

84

Chapter 3

has, in effect, no longer any immediate meaning; it is reduced to the role of arbitrary sign, whose meaning varies infinitely according to the diversity of languages” (Pictet 1856: 252). Yet, as rightly observed in Aarsleff (2004), the question of the arbitrariness of linguistic signs is so ancient and has been discussed so widely through the ages (also by Egger, for example, as we have just seen, in Saussure’s own age), that one does not need a Pictet to remind one of it. Moreover, Saussure explicitly rejects Pictet’s ideas on language in the Cours, first on p. 297, where the central element in Pictet’s theorising is explicitly rejected on the grounds that it has “obscured” a proper insight into the “primitive vocalism”, that is, the system Saussure himself worked out in his Mémoire of 1879, and then on pp. 306–307 of the Cours, where Saussure devotes a small part of his lectures to ‘linguistic paleontology’, Pictet’s innovation. Here he not only speaks about Pictet with reserve, but in fact explicitly rejects the man’s views. It is quite clear that Saussure was highly critical of Pictet’s pre-YoungGrammarian views and attitudes, though at the surface he always remained courteous and respectful. There is, moreover, a confusion in Saussure’s discussion of arbitrariness. He correctly distinguishes ‘absolute’ from ‘relative’ arbitrariness (Cours: 180–184). The former is what arbitrariness is really about: the fact that single morphemes, as minimal lexical units, are (free to be) arbitrary. He also correctly observes that combinations of (single) morphemes may, to different degrees, lead to what he calls ‘relative arbitrariness’. But he incorrectly seems to apply this notion to all forms of morphological combination, regardless of whether they are regular and productive or contain an element of idiosyncrasy (Cours: 181): Thus, vingt [‘twenty’] is unmotivated, but dix-neuf [‘nineteen’] is not unmotivated to the same degree, as it evokes the terms out of which it is composed and others that are associated with it, such as dix [‘ten’], neuf [‘nine’], vingt-neuf [‘twenty-nine’], dix- huit [‘eighteen’], soixante-dix [‘seventy’], etc. Here it is to be observed that what Saussure calls ‘relative’ arbitrariness manifests itself at two levels: either semantically or morphophonologically. Morphophonologically, morphemes incorporated into larger structures often keep their phonological make-up constant, as in bookbinder, but occasionally they undergo some (either idiosyncratic or rule-governed) modification, as in the word thirteen composed of three- plus -ten, but with the first morpheme modified to thir- and the second to -teen, or duchess, composed of basic duke, changed to duch-, plus the suffix -ess. Saussure misleadingly uses the term ‘relative arbitrariness’ for both unmodified cases like bookbinder and modified

The Cours: A Critical Look

85

cases like duchess. Nowadays the term allomorphy is used, which refers to the widespread phenomenon of environmentally conditioned alternative forms for what may be considered the same morpheme at a more abstract level of description. At the semantic level, the component elements of a larger structure contribute to its total meaning but do not determine it.23 When such larger structures are compositional, as the modern term goes, the meaning of the larger structure is exhaustively derivable from the meanings and the position of the component parts, as in carmaker, which simply means ‘one who makes cars’. When they are noncompositional (which they usually are in lexical compounds), they acquire a specialised meaning, usually more narrowly defined than what one would expect given the component parts (though a wider or metaphorical definition is also possible). Thus, the compound noun handyman means not just ‘man who is handy’ but, in an idiosyncratic specialised meaning, ‘man who does occasional jobs in or around the house’. While this confers a certain degree of arbitrariness to the compound structure in question, it still is, though not fully compositional, to a large extent ‘motivated’ or ‘transparent’, which helps the language user to memorise and evoke the specialised meaning and to select the compound word when speaking. This reduction of strict compositionality to mere transparency occurs with perhaps the majority of lexical compounds: the relation between tennis and elbow in tennis elbow is entirely different from what it is between tennis and court in tennis court, to mention just one further example. Semantic specialisations of this nature are extremely common, especially in compounds, but, again, the term relative arbitrariness has been given up in favour of better, more specific terms, such as noncompositionality or transparency or motivatedness. Saussure speaks of ‘relative arbitrariness’ in all these cases, allomorphic or

23

Type-level meaning must be distinguished from token interpretation, which is seldom fully determined by the component parts of the structure in question. Most linguistic composition underdetermines actual interpretation: the listener almost always needs world or context knowledge to make sense of an utterance. In principle, utterances only provide cues for interpretation: the precise actual connections are most often supplied by world or context knowledge. Grammar and semantic composition only constrain, but, except in marginal cases, do not determine, actual interpretation. To give a simple example, when I say The school is on fire, the phrase the school refers to the building, but in The school has a day off it refers to the school as a functioning institution, and in The school is unhappy about the new budget cuts, it preferably refers to the school management. Such differences are well-known and are not regulated by the grammar but by the lexical semantics of a language, in conjunction with nonlinguistic world or situational knowledge, yet within the constraints imposed by grammar and semantics.

86

Chapter 3

­ onallomorphic, compositional or noncompositional. Little wonder the need n arose for more specific terminology. Speaking of ‘relative arbitrariness’ in cases of nonallomorphic, fully compositional structures is ill-advised. These structures simply follow the g­ rammatical and semantic rules for composition. Only in allomorphic or noncompositional but still transparent structures, where there is a modification of either form or meaning, can one possibly speak of ‘relative arbitrariness’, in so far as such cases have to be stored in memory as part of, or alongside, the lexicon. The notion of ‘relative arbitrariness’ is as ancient as Plato, in the fourth century bce. Pictet picked it up and added his own thoughts. For Pictet, who looked down on ‘primitive’ languages and cultures and glorified the ‘civilised’ Western world (see Seuren 2016 note 66 and Section 4.2.2 below), the ‘arbitrariness’ of the linguistic sign applies only to the more ‘advanced’ European languages. In his view, the words of more ‘primitive’ languages may somehow have been originally ‘motivated’ in that they reflected a meaning that fitted the accompanying sound (or vice versa), but that ‘motivation’ got lost through the ages (Pictet 1877: 6–7). For Pictet, just as for one of the discussants in Plato’s Cratylus, single morphemes may originally have had a sound-symbolic origin, which would make them ‘motivated’ in a primary sense, but they may have lost their motivatedness as they became affected by diachronic laws changing either meaning or sound.24 Analogously, larger structures built up out of single morphemes were, for Pictet, likewise originally ‘motivated’ but their motivation got lost, in most cases, as the forms were modified and the meanings s­ pecialised in the course of time. In this vein, the Greek word ánthrôpos (‘man, human’), for example, is (falsely) considered in Plato’s Cratylus to be derived from the ‘motivated’ form anathrỗn hà ópôpe or ‘one who reflects on what he has seen’. Likewise, Pictet (correctly) treats a word like English ox as being derived from “Sanscrit ukshan, literally ‘he who sprays or fertilises (the cow)’, from the root uksh, conspergere, effundere (semen)” (Pictet 1877: 333). Neither Plato nor Pictet thus regarded phonologically and semantically regular morphological constructions as being ‘relatively arbitrary’, but Saussure did. In sum, if Joseph were right in maintaining that Saussure took his notion of the arbitrariness of the linguistic sign from Pictet, he would also be right in maintaining that Saussure took this notion from Plato, whom he had read in

24

Thus the English word fart, which is no longer felt by English speakers to be sounds­ ymbolically motivated, derives from the (Proto-)Indo-European root *pṛd-, which clearly is sound-symbolic and has, for that reason, resisted regular sound change in some languages, though not in English.

The Cours: A Critical Look

87

his schooldays. But neither is the case, since neither Plato nor Pictet made the mistake Saussure made when presenting his notion of relative arbitrariness. 3.4.5 The Linearity of the Signifier Saussure likewise insists, though with less force than on the arbitrariness of the linguistic sign, on what he calls “the linear character of the signifier” (Cours: 103). On the face of it, Saussure says, the fact that the acoustic signal, and the corresponding written signal in its wake, depends crucially on the dimension of linearity may seem pedestrian, but in actual fact “it is fundamental and its consequences are incalculable […]; the entire mechanism of the langue depends on it” (Cours: 103). It would have been helpful if Saussure had provided at least one single argument that would show why the consequences of the linearity of the linguistic signal are “incalculable” and why the linearity principle is so fundamental, but the reader of the Cours only has Saussure’s word for it. Nor has subsequent twentieth-century linguistic theorising, with the exception of phonology, added any basic reason why linearity should be taken to be of such central importance to the theory of language. On the contrary, what has proved to be of ‘incalculable’ importance is the fact that sentences are built up from parts that are arranged in hierarchically ordered, so- called immediate constituent structures, a notion proposed by Wundt and further developed by Bloomfield and a host of later linguists (Seuren 1998: 219–227). In these structures, the linear left-to-right dimension does play a part, but far from the only one. Saussure opposes natural languages to visual sign systems, such as the systems of visual signalling at sea, which may “create simultaneous problems in several dimensions”—problems again that are not spelled out. Acoustic signifiers, says Saussure, only have the temporal dimension at their disposal. Saussure acknowledges that the fact that linguistic form is composed of linearly ordered elements is due to the nature of the medium used, as the production and perception of sound necessarily involves the flow of time, but he neglects to mention that the nature of sound allows for simultaneously significant and mutually independent elements: pitch, loudness and sound quality may vary independently of each other, so that each may fulfill different semantic functions simultaneously with the others.25 25

That accent or intonation may make a semantic difference holds not only for phonology and morphology, but also for syntax. One example was given in note 8 above. Another is the following contrary pair: Sandra didn’t leave, because of the rain. Sandra didn’t leave because of the rain, (but because she felt ill).

88

Chapter 3

Saussure admits that there may “seem” to be a simultaneous element in the composition of linguistic sound, in particular in the distribution of accents and intonation, which creates the impression that “distinct significative elements are accumulated at the same point in time”, but, Saussure says, “this is an illusion: the syllable and its accent form but one act of phonation” (Cours: 103). The reader may well find this incomprehensible, since the phenomenon of “distinct significative elements accumulated at the same point in time” is precisely, by definition, a counterexample to the generalisation that everything in the signifiers of human languages is linear and consecutive (see Section 2.5.2). Counterexamples of that nature are plentiful in the gestural languages of the deaf, which make significantly less use of temporal linearity and much more of visual simultaneity than the sound-based languages of the hearing, a fact that in no way affects their status as natural languages. The reader of the Cours may, therefore, well be inclined to doubt that the linearity principle is so momentous and that its “consequences are incalculable”. Moreover, one wonders how “the entire mechanism of the langue” can be said to “depend on it”, since linearity is a property of actual, material sound, not of the ‘acoustic image’, whereas the langue is repeatedly and with great emphasis said to be purely psychological and to have nothing to do with the material aspects of the parole, as is highlighted in the following section. Small wonder that so many volumes have been filled with attempts at exegesis of the Cours. I will, at any rate, not add to that body of writing, leaving the tangle for what it is and calling it by its name. 3.5

Differences, Oppositions and ‘Valeurs’

By far the most enigmatic feature in Saussure’s thinking, or teaching, is his insistence on the ‘immaterial’ nature of language, closely connected, for him, with the arbitrariness of the linguistic sign discussed in Section 3.4.4 above. One consequence of this arbitrariness is, for Saussure, that the actual acoustic image of the sign, as fixed in the language system, does not matter for the system to function properly, as long as each sign has a more or less unique perceptual image assigned to it. The semantic counterpart to this is that it does not matter either what semantic content is associated with each acoustic image to form a sign, as long as all signs remain sufficiently distinct from each other. Sentence (i) entails that Sandra did not leave; sentence (ii) entails that she did. Therefore, these two sentences cannot be true at the same time, but they can be false at the same time, for example if Sandra left because of the rain.

The Cours: A Critical Look

89

3.5.1 Comparison with Chess Saussure famously tried to illustrate the arbitrariness of the acoustic half of the sign as an element in the language system by means of a comparison with the game of chess: a language is like a game of chess (Cours: 43, 125–127) in that all that matters for a move in the game is that each piece is formally different from each other piece and the position occupied by each piece on the board. As long as one chooses for each piece any old object that fits the board, one can play chess, provided no two pieces are too much alike in their perceptible form and each piece has the status defined for it by the rules of the game: it functions as a white or black rook or pawn or knight, etc. The important thing in a chess game is what, given the state of the game, each piece can do, its value, or, in Saussure’s terminology, its ‘valeur’, in the progress of a game. This comparison of a language with the chess game deserves some separate comment because of the confusion it has created. Had Saussure left it at a simple comparison whereby the chess rules are compared to the rules of grammar, as on p. 43 of the Cours, no harm would have been done, but then he draws the comparison in a different direction: on page 126 of the Cours he compares the progress of a game of chess to a change in the language: “Finally, to pass from one equilibrium to another or—in our terminology—from one synchrony to another, the move of one piece suffices”. This is confusing because on page 43 a rule application in language does not result in a language change but in an utterance—or anyway contributes to the coming about of an utterance—while according to page 126 a rule application in a game of chess results in a new system of ‘valeurs’. The synchronic interpretation is perhaps the one that modern readers would turn to in the first place, but the text on p. 126 of the Cours is clear: a move in the game is meant to correspond to a change in the language and the rules of the game should thus correspond to the ‘rules’ or regularities in language change, whereby each change affects the system of ‘valeurs’. According to Godel, Saussure himself, in his private notes, had moments when he saw the chess simile as applying to synchronic ‘values’ (Godel 1957: 73–74): Provisional conclusion. All that is synchronic can be subsumed under the term grammar (cp. grammar of the chess game, of the Stock Exchange), which implies a system in which values [‘valeurs’] are at work. There is no historical grammar: what is understood by that term is diachronic linguistics, which could never be grammatical. Indeed, Saussure repeatedly emphasises that (Cours: 134):

90

Chapter 3

iachronic facts are particular; dislocations of the system take place under the influence of events that are not only alien to it […], but that are also isolated and do not form a system among each other. But this, in turn, is hard to understand in the light of the Young Grammarian’s doctrine, narrowly followed by Saussure, which was precisely that there are diachronic sound laws affecting not just isolated phonemes but classes of phonemes according to shared phonetic features. It thus looks as if Saussure’s comparison of the langue with a game of chess brings more confusion than clarity. It is said that every comparison limps, which must be true because if a comparison does not ‘limp’ at some point, it is not a comparison but a description of the thing itself. In this case, however, the ‘limp’ is situated precisely at the point that should be clarified by the analogy. The normal use of language does not lead to a change in the language but to a change in the semantic representation of the discourse that is going on. In chess, by contrast, the ‘use’ of the game results in a different force field of the pieces with respect to each other. The purpose of language use is twofold: (i) the text-driven coherent setting up of a situational representation that the speaker intends to vouch for, or wishes to see realised or to be enlightened about, and (ii) the reconstruction and recognition by the listener of the situational representation set up by the speaker plus the speech act element of speaker’s social commitment or appeal to the listener (Seuren 2009, Ch. 4). The purpose of playing chess is to win the game. The difference is crucial and stands in the way of a simile that compares a move in the game to a change in the language.26 All one can do is compare a move in the game to an advance in discourse construction, which would be coherent and potentially enlightening. But then the analogy with grammar is lost, because discourses do not proceed along the lines set out by the grammar of the language in question but by rules of discourse construction, which are to do mainly with preservation of possible truth through a series of utterances (logico-semantic consistency) and involve presuppositional phenomena and processes of sentence-external anaphora. Saussure becomes the victim of his own metaphor when he speaks

26

This basic fault of Saussure’s comparison of the language system with a game of chess apparently escaped Godel, who devotes some space to a critical discussion of it (Godel 1957: 187). His critique concerns the fact that a chess move is instantaneous and conscious, while a language change is gradual and unperceived. If this were all there is to it, one might say that the comparison ‘limps’ (as all comparisons do) but in a harmless way. In fact, however, the comparison ‘limps’ in a way that makes it unusable for the purpose it was thought up for.

The Cours: A Critical Look

91

of the ‘grammar of the chess game’. Much better simply to forget about this entire chess business. 3.5.2 Only Differences in the Language System? Be all this as it may, Saussure concludes: “The linguistic mechanism consists entirely of identities and differences” (Cours: 151), making us tumble into another, much more serious confusion. ‘Identities’ functioning in systems are taken to be exhaustively definable in terms of differences (Cours: 168): In the langue, as in any other semiological system, what distinguishes a sign is precisely what constitutes it. It is the difference that makes the character, just as it makes the value [‘valeur’] and the unit [‘unité’]. [italics mine; pams] Two pages earlier we read (Cours: 166):27 All that precedes boils down to saying that in any given language there are only differences [italics original; pams]. What’s more: in general, a difference supposes positive terms between which it can be established. But in a language there are only differences, without any positive terms [italics original; pams]. Whether you take the signified or the signifier, the langue provides neither any ideas nor any sounds pre-existing to the 27

The notion that all is negative in language is already found in the draft text, stemming from about 1894, mentioned in note 15 above (Saussure 2002: 70): “There are in language neither signs nor significations, but only DIFFERENCES between signs and DIFFERENCES between significations”. As argued below, if any sense is to be made of this, it is that the language system can be formulated exhaustively in terms of variables, constant terms not being needed. Yet this abstract view of language, whether at the grammatical or the semantic level, would miss out grossly on the very nature of language as a means to express mental content in perceptible form. It is like doing chemistry only in terms of variables, without the actual elements. Sechehaye tries to give a Wundtian turn to this rarefied view (Sechehaye 1908a: 34): One thus sees that Wundt is interested only in the values, not in the signs of the values; he thinks that he can consider the grammatical categories in abstracto, as psychological and logical entities, regardless of the nature of the signs that carry them. But this will not do either, Sechehaye says, since the system of grammar consists precisely in the concrete ways in which the abstract ‘values’ are materialised. He continues (ib.): Yet the real grammatical problem […] consists in finding out how a system of signs must be organised materially, in concreto, for it to be the mapping of an abstract system of ideas and relations; and the importance of this problem is that the means of expression and that which is expressed condition each other mutually in absolute solidarity.

92

Chapter 3

linguistic system, but only the conceptual and phonic differences as they are defined in the system. Whatever there is in the way of ideas or phonic material in a sign matters less than what there is around it in the other signs. The proof of this lies in the fact that the value [French: valeur; italics mine; pams] of a term can be modified without touching either at its sense or at its sounds, but only because some other, neighbouring term has been changed. Yet the next paragraph presents the reader with the opposite (Cours: 166–167): But to say that all is negative in any given language is true only for the signified and the signifier taken separately. As soon as one considers the sign in its entirety, one finds oneself faced with an entity that is positive in its order. […] Although the signified and the signifier are, each taken separately, purely differential and negative, their combination is a positive fact. It is even the only kind of fact that makes up a language, since linguistic institutions are characterised specifically by the maintenance of the parallelism between these two differential orders. Taken together, these latter two consecutive paragraphs can only be taken to be mutually inconsistent. If “in a language there are only differences, without any positive terms” (p. 166), while at the same time “the sign in its entirety” is “a positive fact”, and “even the only kind of fact that makes up a language” (p. 166), one must conclude that, for Saussure, a language contains no signs, while at the same time it contains only signs. Saying that “in the langue, as in any other semiological system, what distinguishes a sign is precisely what constitutes it” and that “it is the difference that makes the character […]” (p. 168), just shows that Saussure himself was, perhaps vaguely, aware of this contradiction, but it utterly fails as an attempt to solve or neutralise it.28 The result is mystification pure and simple. Little wonder this mystification had a mesmerising effect on obscurantists like the French philosopher Derrida with his peculiar ­distinction 28

John Lyons, a staunch supporter of Saussure’s, wrote (1968: 443): As far as the empirical investigation of the structure of language is concerned, the sense of a lexical item may be defined to be, not only dependent upon, but identical with, the set of relations which hold between the item in question and other items in the same lexical system. One notes the loose way in which this author speaks about the “structure of language”, the “sense of a lexical item”, the “set of relations” and the “lexical system”. If the author had made an attempt at being more precise, he would no doubt have discovered the incoherence of this text.

The Cours: A Critical Look

93

between “différence” and “différance”, referred to in Section 2.5.3 above. In practice, Saussure operates with the notion that a language is a system of signs, which are positive elements, not a system of “only differences” (see, for example, Cours: 99, 144, 190). The entire nebulous fabric of thoughts on the purely differential and negative character of language can thus simply be dismissed.29 One can, nevertheless, understand what inspired Saussure and what he was at grips with as he kept returning to this aspect in his thoughts on language. It is probably not too far-fetched to detect the origin of these thoughts in his enormously successful hypothesis of laryngeal sounds in Indo-European dating back to his (1879). These laryngeals could not be identified in terms of their phonic substance but only in terms of their postulated effects on neighbouring sounds, and thus, in a way, in terms of the differences they made vis-à-vis other sounds. Apart from this, however, he must have been fascinated by the quintessentially rationalist (and definitely not romanticist) idea that one can, as a theoretical linguist, distill from the grammar of a language a formal system expressible in terms of variables, leaving the actual, concrete values (in the mathematical sense) of the variables out of consideration. What one then gets is known as an uninterpreted formal system. But an uninterpreted formal system is not by itself a science, only the calculus of one. It is not a science because it fails to explain the, no doubt “positive”, observed facts, which is the professed aim of science in general and of any specific science in partcular. Even theoretical physics, the most abstract of all sciences, ultimately needs a physical interpretation of its underlying mathematics. There is nothing ‘deep’ about this. Saussure only lacked the formal background to see this and thus got lost in fruitless speculations. 29

To show the corrupting infuence of this nebulous Saussurean fabric, I may quote the in other ways excellent linguist André Martinet, who, in his theoretical works, is in constant rebellious dialogue with Saussure (Martinet 1965: 6): To describe a language is to indicate in what ways it differs from any other language. Since, however, one knows neither all languages of the past nor all those of the actual world, nor, of course, any of tomorrow’s languages, one must state that, in principle, the range of variation of languages is restricted only by the implications of our definition of the term ‘langue’. This seems to imply that describing a language equals establishing a theory of ­language universals, which is more than Martinet himself can bear. His solution to this (unnecessary) quandary is quite simple: one only has to recognise that language in general, and every language in particular, is defined by a ‘double articulation’ consisting of segmental phonemes on the one hand and meaningful morphemes and morpheme combinations on the other (see note 24 in Chapter 2). I must admit that the logic of this answer is beyond my comprehension.

94

Chapter 3

It must be added, moreover, that if one looks upon a language as an uninterpreted abstract system, one misses out on the fact that the systems of natural languages are to a considerable extent themselves actually co-determined by the specific material or mental stuff forming the values of the variables. This is obviously so in phonology, where physically defined phonological features are a necessary explanatory element in the analysis and description of phonological systems. But the same applies in semantics. Facts of semantic bleaching and concomitant regrammaticalisation, such as are found in auxiliation phenomena (Seuren 1996: 80–84; Kuteva 2001) or facts of reinforcement of semantically crucial but phonologically weak elements, as in negation copying phenomena and particularly those caught under the term Jespersen’s Cycle (Jespersen 1917: 4; 1924: 335–336), would remain unexplained and without rationale. This immediately shows the nonviability of a grammatical description expressed merely in terms of uninterpreted variables. In fact, no such model for grammatical description has ever been launched with any degree of success. Saussure’s treatment of questions of this nature is, alas, highly unsatisfactory. All I can say is that what one sees here is the well-known phenomenon of a man groping for light and trying to work his way through a dense mass of indistinct and unruly thoughts and ideas which he is as yet unable to control or put in order. One can easily understand why Saussure was unwilling to teach his courses and would have been even more unwilling to see them published. He must have been, after all, a sensible man. Understandably, Saussure’s confused notion of ‘valeur’, though often presented by Saussurologists as the apex of his genius, was never picked up by the linguistic world, for the good reason that it is unworkable. No actual, wellelaborated system of Saussurean ‘valeurs’ has ever been presented and there is no good reason why one should try. 3.6

Synchrony versus Diachrony

We have repeatedly spoken of ‘diachronic’ and ‘synchronic’. It is now time to look at this pair of terms more closely, especially because it reflects one of the major distinctions Saussure has become known for. In Saussure’s day, the concept of a ‘synchronic’ or ‘static’ system regulating the forms used in actual speech or ‘parole’, as opposed to ‘diachronic’ or ‘evolutive’ changes in the language system, was beginning to be accepted among linguists, having been introduced by Wilhelm von Humboldt (1767–1835) (Jankowsky 1972: 47–49) and having become common in sociology since ­Auguste Comte (see Section 2.3 above). Early linguists who were fully aware of the distinction were, for example, Michel Bréal (1832–1915), in whose works

The Cours: A Critical Look

95

(Bréal 1893; 1891; 1897) the distinction is fundamental, as shown in Aarsleff (1981), or Sechehaye (1908a), who uses the terms statique and dynamique for synchronic and diachronic, respectively (this work had been carefully read by Saussure before its publication, as shown in Section 2.3 above). It is thus clear that the distinction between the notions synchrony and diachrony cannot be attributed to Saussure, even though he invented the terms ‘synchronic’ and ‘diachronic’. Yet the distinction, though nowadays obvious to any linguist worth their salt, was not generally known, let alone accepted, among the linguists of the day. At the time, the notion of a system behind the functioning of language was relatively alien both to the general public and to professional linguists (see Section 3.3.1). Many, even professionals, were simply unable to see that a language forms a system at any given point in time. The only perspective taken was historical: only as the product of historical developments could a language be studied scientifically. The leading linguistic authority Hermann Paul ­(1846–1921), generally considered the author who best represented the Young Grammarian doctrine, writes in his Principien der Sprachgeschichte, first published in 1880 (1920: 20):30 It has been said that there is another scientific way of looking at language than the historical one. I must deny that. What is said to be a non-­ historical yet scientific way of studying language is in fact nothing but an ­imperfect historical way […]. The moment one goes beyond the mere statement of isolated facts, the moment one tries to understand the system [‘Zusammenhang’] or to understand the phenomena, one treads on the ground of history, even if one may be unaware of that. And again on p. 28: The psychological organisms as described above are the real bearers of the historical development. That which is really spoken has no development at all. Saussure did not follow Paul in this regard. He stresses the primacy of the ­synchronic over the diachronic in the Cours on p. 128: “[I]t is evident that 30

This much quoted passage—also referred to by Anton Marty (mentioned earlier in Chapter 2, note 14) in his posthumously published (1950: 16), but written ca. 1904—shows that the distinction between synchrony and diachrony, standardly attributed to Saussure, had been a topic of discussion for at least two decades before Saussure lectured on it, or before Sechehaye wrote about it in his (1908a).

96

Chapter 3

the synchronic aspect takes priority over the other [i.e. the diachronic; pams], since for the speaking mass it is the only true reality”. However, this and other passages in the Cours starkly contrast with what Saussure intimated to Albert Riedlinger in a conversation that took place on January 19th, 1909 (Godel 1957: 29): One should start from diachronic linguistics; the synchronic must be treated on its own, but without the perpetual confrontation with the diachronic, one arrives at nothing. The ancient grammarians may have been free to do static linguistics, without any risk of confusing the two points of view, but look at what they achieved! Which shows, incidentally, not only his dismissive and abrasive way of dealing with authors and matters that fell outside the limits of his own thinking but also that he knew next to nothing about the extraordinary achievements of the ancient grammarians (see Robins 1967: 39). The best one can make of this puzzling contradiction is that Saussure, apart from being uninformed, appears to have been unclear in his own mind regarding the priority or primacy of either the diachronic or the synchronic point of view. As far as the actual issue is concerned, one sees a clear parallel here with the social sciences: what Paul says about “the psychological organisms” being “the real bearers of the historical development” must be taken to apply with equal force to sociological ‘organisms’ being ‘the real bearers of the historical development’, since, as with language, every social structure or system is a product of the past. The fact that ‘synchronic’ sociology was already well established by the end of the nineteenth century—due to the history of the new sociology, which differed radically from what had been happening in linguistics—will have been a powerful factor in the coming about of the ‘synchronic’ perspective in language studies. As regards the strictly psychological character of each individual language, Saussure does follow Paul’s classic Principien of 1880, whose first chapter is devoted to the fact that a language, though the property of a community, can only change in the minds (or brains) of single individuals, as there is no known physical mechanism that will effect changes in a number of minds simultaneously. All changes brought about more or less simultaneously in a collection of minds require individual ‘filtering’ for them to be admitted into the mind of each individual speaker. The main difference with Paul is that the latter still thinks exclusively in a historical-developmental fashion, whereas Saussure, along with Meillet, Sechehaye and others, is already well on his way towards

The Cours: A Critical Look

97

taking the, at the time still largely mysterious, synchronic ‘system’ of each language into account.31 In Section 2.3 above it was shown that Saussure had been thinking (but not publishing) about the diachrony-synchrony distinction since the early 1890s and that Sechehaye dealt with it extensively in his (1908a), saying that if one doesn’t know the system, one cannot see what exactly has changed (see Section 6.3.2 for further comment). Here we touch on the crucial link between the purely historical point of view of nineteenth-century comparative-historical linguistics on the one hand and the structuralist, system-oriented linguistics of the twentieth. The link consists in the fact that regular language change is in principle a system change—a principle that was emphasised by Albert Sechehaye, as we have seen. Regular diachronic changes in linguistic form reflect some change somewhere in the phonological or grammatical system of the language. This does not apply to semantic changes, whose causal origin must be sought in the shared cognition of the language community. But form changes reflect the language system, that is, the machinery that transduces semantic content to acoustic, graphic or gestural form. Thus, when we see that Old Latin intervocalic s became r in classical Latin (so-called rhotacism), so that, for example, original ausosa (‘dawn’) became aurora,32 just as the Old Latin genitive flosis of flos (‘flower’) changed to floris,33 and so on, we witness a change in the Latin phonological system. Likewise, the change from Early Modern English (thou) singest to (you) sing reflects a change in the morphological system of English. Or when we see that, around 1150, the Old English modal verb cunnen (‘can, be able’) ceases to occur as an infinitive or as a participle and continued life as the indeclinable finite present tense can or past tense could (Van Kemenade 1992: 159), we witness a change in the ­auxiliary system of English syntax (parallelled in a few other languages; see Seuren 2017). Or when, in Elizabethan English, the position of the English negation word not begins to move from the end of the sentence, as in She loves me not, to an inserted dummy verb do, as in She does not love me ­( Jespersen 31

32 33

Even so, the Young Grammarian Paul provides much more in the way of syntactic analysis, in the modern sense of the term, than Saussure does in his Cours. He even deals in some detail (Paul 1920: 282–286) with the great subject-predicate debate that dominated linguistic theorising between, say, 1850 and 1930 but is not mentioned at all by Saussure. Cp. Greek êốs (‘dawn’), from older êsốs. In classical Greek, intervocalic s is regularly dropped. Latin nouns in -or, genitive -oris, such as honor, honoris (‘honour’), go back to original -os, -osis (honos, honosis). Rhotacism turned honosis into honoris; the nominative honor then resulted from analogy in the Young Grammarians’ sense.

98

Chapter 3

1917: 9–14),34 this again reflects a change in the syntactic system of English. In ­sixteenth-century German, we find a clause like (3.5a) (Martin L­ uther’s ­German translation of St. John’s Gospel 6:62; see Seuren & Hamans 2010: 148), which is ungrammatical in Standard Modern German, where the correct form is (3.5b), both meaning ‘if you then (will) see the Son of man ascend’: (3.5) a. wenn ihr denn sehen werdet des Menschen Sohn auffahren (if you then see will of-the man Son ascend b. wenn ihr denn des Menschen Sohn auffahren sehen werdet (if you then of-the man Son ascend see will) Here again, we have a change from the syntactic system of standard German in Luther’s day to that of Modern Standard German.35 There are also totally idiosyncratic changes, such as the introduction or disappearance of some lexical item or idiom, or the isolated exception to an otherwise regular sound change. In such cases, we cannot speak of a change that reflects a change in the system, unless we allow the system to contain idiosyncrasies as marginal extremes, just as the system of natural numbers may be taken to contain zero as a marginal extreme instance of a natural number. A glaring example is the Portuguese word erro (‘error’, plural erros), derived from Latin third declension masculine error, with the genitive errōris. In all other Romance languages, Latin error simply followed the rule in each language for Latin third declension masculine nouns in -or. In French, they became feminine and end in -eur (douleur, erreur, pudeur, etc.), but for isolated exceptions such as the masculine honneur (‘honour’; the masculine gender possibly being due to Provençal troubadour influence) or the masculine amour (‘love’; with Provençal -ou- instead of regular -eu-) or peur (‘fear’, from Latin pavor), which lost a syllable in the process due to –v- having become a semivowel that got lost in popular medieval French. In Italian, they all remained masculine and now end in -ore (dolore, errore, pudore, onore, amore, etc.), but for paura (again ‘fear’, from Latin pavor, and again with loss of the ­semivowel—this time due to southern dialects—the feminine gender being a result of analogy with the numerous feminine nouns in -ura). And likewise in 34

35

Except for the verb know, which allowed not to occur after it until well into the nineteenth century, as in Longfellow’s famous lines: “I shot an arrow into the air. It fell to earth, I knew not where”. And, of course, in idioms like “I hope not” and “I think not”, still very much alive in present-day English. Note that the modern German verbal cluster [auffahren sehen werdet] in (3.5b) has the same left-branching structure as the verbal cluster [lesen lassen wollen] of example (3.2) discussed above in Section 3.3.

The Cours: A Critical Look

99

all other Romance languages—but for Portuguese erro, which has remained without explanation.36 In Portugese, all nouns of this class end in -or and they have remained masculine, except for the three only monosyllabic ones dor (‘pain’), cor (‘colour’) and flor (‘flower’), which have become feminine and thus form a subclass. But for no known reason Portuguese erro, otherwise a paragon of virtue, decided to take a walk on the wild side. It is a remarkable fact that the connection between change and system is well-established in phonology and morphology, but hardly or not at all in syntax, which, at the present moment at least, dominates theoretical linguistics. What one sees, right from the start of the revival of rationalist-oriented theoretical linguistics around 1850 (see Chapter 2, note 16), with the subject-­ predicate debate in the foreground, is that the general theory of language hardly interested the historical linguists, while, conversely, historical linguistics was of no concern to the theorists of language. The two brands of linguists have formed separate communities right from the start until the present day. During the second half of the nineteenth century, the historical comparativists, by and large, stood apart from those who engaged in the subject-predicate debate. After 1900, however, the landscape changed in that the synchronic point of view began to prevail, not among the subject-predicate debaters but among a faction of mostly French and Franco-Swiss (Genevan) historical linguists, such as Meillet, Sechehaye, Saussure, Bally, who had split off from the historicalcomparativist school but were, with the exception of Sechehaye, hesitant to get engaged with the subject-predicate debaters. Saussure is a clear example of this isolationism, whereas, by contrast, Sechehaye was the one most willing to cross the street and meet other brands of theorists, such as Wundt or the subject-predicate group. It may well have been the very distinction between synchrony and diachrony that kept the old comparativists and the new synchronists apart from each other, each minding their own business (and claiming their own kudos), while neither showed much interest in what was going on in psychology, the third party. As for the comparativists, I may refer to the fruitless debate between 36 Italian marmo (‘marble’) may be taken as a parallel to Portuguese erro, as it derives from Latin 3rd declension marmor, but Latin marmor has neuter, not masculine, gender and has a deviant declension (with short o in the oblique cases and hence antepenultimate accent). Moreover, neuter third declension nouns in -or are rare in Latin: the only other cases I know of are cor (‘heart’) and aequor (‘calm sea’), both again with deviant declensions. Marmor also differs from the prototypical Latin 3rd declension nouns in -or in that it has an archaic alternative form marmur, and it is derived from Greek, which has mármaros (second declension). No such idiosyncrasies apply to Portuguese erro, which is just a standard member of its class.

100

Chapter 3

Delbrück and Wundt, where the Young Grammarian Delbrück takes an agnosticist position with regard to psychological theories (Delbrück 1901). As for the new synchronists, it will suffice to point at their well-nigh rabid insistence, as from the early 1900s, on the autonomy of linguistics,37 which had to be kept free from any contact with psychology, logic, history or any other discipline. It was thus that the comparativists—that is, the Young Grammarians and their descendants—missed out on theoretical linguistics: they did not like the formal (and sometimes quasi-formal) stuff one inevitably gets involved in when studying syntax, the centrepiece of theoretical linguistics and often confused with it. Matters stand differently in phonology and morphology. Here, the general theoretical element is much less prominent and, in addition, these are the fields that the comparativists had traditionally felt most at home in. Meanwhile, an important lesson can be drawn from all this: historical linguistics, sociolinguistics and theoretical linguistics are connected by the fact that seeing a change or a dialectal or sociolectal variation should make one ask for the underlying system, while building a system should make one look for changes that have taken place in the past, or for dialectal or sociolectal variations, as these must correspond to the internal joints of the system. This provides us with the following methodological principle: When you observe a regular change or internal dialectal or sociolectal variation in a language, you may take it that this reflects a change or variation in the rule system of the language in question, even if the change or variation is not exceptionless. If you cannot locate this change or variation in the rule system, you have good reason to search for a system that reflects the change or variation in question in the simplest possible terms—which may mean an adaptation or rejection of the theory you have so far been working with. 37

The last, italicised, sentence of the Cours (p. 317) tells the tale: “Linguistics has for its unique and true object language, taken by itself and for itself”. (“La linguistique a pour unique et véritable objet la langue envisagée en elle-même et pour elle-même”.) In Chapter 4 it is shown that Sechehaye was opposed to this inordinate insistence on the necessity of keeping linguistics free from logic and psychology, which, like sociology, are to be considered adjacent disciplines interfacing in multiple ways with linguistics. John Lyons comments (Lyons 1970: 8): Linguists have tended to be somewhat insistent on the need for autonomy, because they have felt that, in the past, the study of language was usually subservient to and distorted by the standards of other studies such as logic, philosophy and literary criticism. For this reason the editors of Saussure’s posthumous Cours de linguistique générale (the publication of which is often taken to mark the beginning of ‘modern linguistics’) added to the text of the master its programmatic concluding sentence, to the effect that linguistics should study language ‘for its own sake’ or ‘as an end in itself’.

The Cours: A Critical Look

101

Language change and language variation thus become an important heuristic for systems in language, and at the same time a criterion for adequacy, in that a system that has been proposed but fails to provide the terms in which a given regular change or variation can be caught is likely to be inadequate. As regards language change, Sechehaye already saw this, when he posited that to understand a change you must know the system first. But present-day linguists, especially syntacticians, do not seem to be aware of this adequacy test, no doubt because, not having been brought up in a tradition that studies language from different points of view, they are not in the habit of linking up language change or language variation with language systems. 3.7

Conclusion

The conclusion we must draw after our careful and detailed look at Saussure’s Cours is not cheerful. If one looks at it as just an academic publication, one has to conclude that it fails to meet the standards that are normally applied. If one takes the special circumstances of its author and its coming about into account, one still wonders how it could possibly have had the career it has had. How and when did it acquire its reputation of being one of the seminal works that laid the foundations of modern linguistic science? Allan (2010: 262) gives a partial answer (quoted earlier in Chapter 2): “It was that Saussure brought together in the lectures which gave rise to the various ideas of his illustrious contemporaries and immediate predecessors, fashioning them into what was thought to be a workable theory”. This is no doubt correct. I might add that Saussure also probably brought with him an atmosphere of modernity, which may have attracted his friends, his students and the readers of the Cours. But I cannot go any further than this. In the final analysis, what we have before us, in the way of the Cours, the numerous scattered notes and other relevant documents, is a legacy that bears testimony to an intellectual poverty and impotence that could never have become the foundational source of a continent-wide dominant movement in a discipline without some extraneous factor. That factor, I maintain, was a strong shot of myth-making. This and the previous chapter have shown that, by the time of his death, Saussure had already become something of a myth among a number of French and Swiss linguists, whose influence and status were strong enough to ensure publication by a renowned Parisian publisher. The less closely connected world of European linguists would not go along with this myth for some time, as was shown in Section 2.5: it was not until around 1930 that it began to take hold outside the restricted circle of early Saussureans, especially in many parts of

102

Chapter 3

Continental Europe. We do not know of any overt attempts by Saussure himself to induce the building up of such a myth. We must, therefore, take it that there was already a cottage industry going on consciously putting up Saussure as a publicity item, as is suggested in Section 2.5.1 above, where it is mentioned, following Frýba-Reber (2017), that within a year after its publication the Cours had eight reviews, all by Swiss reviewers and published in Swiss periodicals, and all of them highly laudatory—highly unusual for a book of this nature. We just do not know for certain what caused the myth and made it spread, not with the material that is at our disposal at this moment. The important thing, in any case, is that it be clear that Saussure has been vastly overestimated during the past one hundred years, at the expense of others, such as Sechehaye, whose work was far superior to Saussure’s but was eclipsed by the Cours and thus disappeared into a void of oblivion. Only if we realise this, will we understand why European structuralism never quite made it and was, during the middle of the twentieth century, engulfed by the more serious and better founded American varieties of structuralism (which themselves soon became the victim of even more potent myths). Only then, also, will we read the forgotten authors again and will we be able to turn their insights to our advantage.

Chapter 4

Charles-Albert Sechehaye 4.1

Private Life1

We already know a great deal about Charles-Albert Sechehaye (after 1908, he left out his first name Charles, using only the second, Albert), who, as you may wish to be reminded, lived from 1870 to 1946, Saussure’s junior by almost thirteen years. If he is known at all to present-day linguists, it is as one of the editors of Saussure’s posthumous Cours de linguistque générale, but that he was a highly creative, if not prophetic, innovator of linguistic theory, rising far above Saussure and most others, is totally unknown.2 Young linguists nowadays have never heard of him. Like Saussure, he descended from a Lorraine Calvinist family that emigrated to Geneva to escape religious persecution. The Sechehayes settled in Geneva in 1689, four years after the French king Louis xiv revoked the Edict of Nantes of 1598, which had granted a certain amount of freedom to Calvinists in France. Again like Saussure, Sechehaye was born and died in Geneva, but his origins were a great deal less august: his family belonged to the bourgeois middle class (his father was active in real estate; Frýba- Reber, p.c.). On March 20th 1908 Sechehaye married Jenny Marguerite Burdet (1887–1964), likewise of French Huguenot extraction and a well-known psychoanalyst.3 Marguerite was one of the three students who followed Saussure’s last lecture course and her lecture notes were used by her husband when he compiled his version of the Cours after Saussure’s death in 1913. I found nothing about any children, other than their adopted daughter Lorraine Louisa Sechehaye-Duess (1912–2002),4 who, having first been her patient for the best

1 Most of the biographical details provided are taken from Frýba-Reber (1994, 1995/1996), the former showing us Sechehaye as an intellectual, the latter as a politically, socially and morally engaged man with a rich and active life outside academia. I am also indebted to Mme FrýbaReber for additional information regarding the lives of Sechehaye and his relatives. 2 In the opening sentence of her book, Frýba-Reber (1994: 11) regrets that the name Sechehaye “nowadays evokes hardly more than his co-editorship of the Cours”. 3 For biographical details of Marguerite Sechehaye-Burdet, see Balbuena (2014). 4 Her original name was Louisa Duess, which, upon her adoption, she changed to Lorraine Louisa Sechehaye-Duess (Frýba-Reber, p.c.).

© koninklijke brill nv, leiden, ���8 | doi:10.1163/9789004378155_005

104

Chapter 4

part of the 1930s, followed in the footsteps of her adoptive mother in becoming a psychoanalyst.5 From 1891 to 1893 Sechehaye was a student at the University of Geneva, where he followed courses in classics, philosophy, history, French literature and linguistics. His linguistics professors were Joseph Wertheimer, the then professor of general linguistics, and Saussure, at the time professor of IndoEuropean linguistics and Sanskrit (it was not until 1906 that Saussure had to take over the chair of general linguistics from Wertheimer, as mentioned in Section 2.1). In 1893 he left for Leipzig, where he stayed a mere six months, studying with the Young Grammarians Karl Brugmann (mentioned above in Section 2.1), Eduard Sievers (1850–1932) and the Indo-Europeanist and Celticist Ernst Windisch (1844–1918). Needing a livelihood, he then taught for three years at the Municipal Higher School of Commerce (Städtische Höhere Handelsschule) in Aussig an der Elbe, now an industrial town in the Czech Republic and named Ústi nad Labem. In April 1897 he found employment as a lecturer in Modern French at the Georg August University of Göttingen in Lower Saxony, where he stayed until 1901 and wrote, in German, a PhD thesis about the use of the past subjunctive mood in French counterfactuals. From then until his death in 1946 he stayed in Geneva, all the time connected to the university. His university career was slow and painstaking. Starting as an unpaid ‘privatdozent’ (which meant that he had to take on various teaching jobs outside the university for a living), he gradually moved up the ladder, becoming ‘professeur extraordinaire’ of stylistics, grammatical theory and Old French in 1929, a position without much clout and still with no salary worth mentioning.6 It was 5 There was a special link between the Saussure and the Sechehaye families in that Saussure’s son Raymond de Saussure (1894–1971), a psychoanalyst and himself analysed by Sigmund Freud, analysed Sechehaye’s wife Marguerite Burdet in 1927–1928, after which he helped her start her own practice, just as Marguerite first analysed her future adoptive daughter and then helped her set up practice. The families of Raymond de Saussure and Sechehaye also regularly organised meetings at their homes, where discussions were held and lectures were presented. What those meetings, with their well-nigh incestuous psychoanalyst-patient relations, will have been like is anybody’s guess. Nowadays, of course, such relations are strictly forbidden by the various professional psychoanalytic institutions. 6 In Frýba-Reber (1994: 194–205) one finds a complete list of all the courses Sechehaye gave at the University of Geneva from the beginning in 1903 till the end in 1946. The list is revealing in that it shows that until 1929, when he received an appointment as extraordinary professor with grammatical theory as part of his teaching task, besides stylistics and Old French, his teaching was restricted to mostly optional courses on historical French grammar, phonetics, metrical theory of French verse, stylistics (no doubt in the sense of the term adopted by Bally; see below) and language teaching methods. Only in 1912–1913, when he was standing in for the ailing Saussure, do we see him teaching on linguistic theory, restricted again in 1914 to its application to language teaching. After 1929, until his appointment to the chair of general

Charles-Albert Sechehaye

105

not until 1939, upon the retirement of Charles Bally, that, at the age of 69, he was appointed, though still only as a ‘professeur extraordinaire’ (Frýba-Reber 1994: 193), to the chair once occupied by Saussure. A year later he was elected first president of the Société Genevoise de Linguistique, founded in December 1940 by Henri Frei and Serge Karcevski, and editor of the journal Cahiers Ferdinand de Saussure, the official organ of what is known as the ‘Geneva school of linguistics’, still in publication today. In his private life, Sechehaye was very much socially engaged, active in all kinds of local, national and international initiatives and always driven by a strong sense of duty and morality that manifested itself mainly in political writings and in an active interest in education. As a lifelong convinced but rebellious Christian, he strove for a better society and for world peace. For ten years, from 1933 to 1943, he was editor-in-chief of the Geneva journal L’Essor (translatable as ‘The Leap Forward’), an independent, critical journal, founded in 1905 and still running, devoted to world peace, social solidarity and respect for life. In the articles he wrote for L’Essor, he manifested himself as a severe critic of the political establishment, a socialist, at times even a revolutionary, and a fearless antifascist, always invoking moral values. Besides these political activities, he was active in education, devising new methods for language teaching and regularly taking a teaching part in summer courses for foreign students organised by the famous Institut Jean-Jacques Rousseau, founded in 1912 and dedicated to the advancement of educational methods that focused on the child, rather than on the teacher. 4.2

Scholarly Life: Preliminaries

As a scholar, Sechehaye was the opposite of ‘trendy’, prepared to go against current attitudes and opinions rather than give up his convictions. From his two books, Programme et méthodes de la linguistique théorique (henceforth pmlt) of 1908 and his Essai sur la structure logique de la phrase (henceforth slp) of 1926, one gets the impression that he was at bottom a conservative, holding on to traditional values and not giving up traditional insights if he felt that these were solid, but at the same time a great innovator, exploring novel insights if he felt that was necessary to find answers to new questions. In exploring linguistics in 1939, we see his teaching tasks extended with topics of a more general and theoretical nature (once even, in 1931, when he was briefly standing in for Bally, with “general linguistics”). But it was not until after his appointment, in 1939, to the chair of general linguistics that he had a free hand in teaching general linguistics courses.

106

Chapter 4

uncharted territories, he was independent and intellectually fearless, of a clear mind, and, as we will see, a visionary, anticipating not only many modern insights (though of course not in the professional terminology that has established itself in the meantime)7 but also the questions and discussions that come with them. It is my definite feeling that, had he been given the attention and appreciation he deserved, European linguistic structuralism would have flourished and perhaps taken a leading and beneficial role, and would not have atrophied the way it has since the 1950s. 4.2.1 Production and Reception Although mostly known for his part in the editing of the Cours, Sechehaye had a modest production of his own.8 First, there are his two books, Programme et méthodes de la linguistique théorique of 1908 (pmlt) and Essai sur la structure logique de la phrase of 1926 (slp). These are the publications I will mostly fall back on in the discussions that follow. Besides these, there is his Göttingen PhD-thesis of 1902 about the subjunctive imperfect in French counterfactuals, written in German and published as a long article in a French translation in Romanische Forschungen of 1905. Then, besides a booklet on French historical grammar published in 1909, there are some thirty-odd articles, book reviews and occasional papers, only few of which, in particular his inaugural lecture of 1929, published as (Sechehaye 1930), and his contribution to the first volume of the Cahiers Ferdinand de Saussure of 1941 (Sechehaye 1941), are to do with the general theory of language. Not an overwhelming lot, but distinctly more than Saussure’s œuvres complètes. The two books pmlt and slp are the main sources for his thoughts on grammar in the context of the general theory of language. Neither of these two books was ever translated into any other language. Although his pmlt of 1908 won the Prix Amiel, a local Geneva prize, it never had a second printing or a new revised edition (though it was brought out as a facsimile paperback in 2011 by Nabu Press, Charleston, South Carolina). According to Frýba-Reber (1994: 12), pmlt had seven reviews, which were, on the whole, positive except that Sechehaye was (unjustly) criticised for being too much centred on psychology and too little on linguistics proper. One review was a tongue-in-cheek one by Meillet, in the Bulletin de la Société linguistique 7 This is not to say that Sechehaye’s technical terminology is totally ad hoc or, worse, inept. On the contrary, there is a great deal of Western philosophical and grammatical lore to be found in the terms adopted by Sechehaye—a source of inspiration that is badly missed in much of our modern terminology. 8 A complete chronological bibliography of Sechehaye’s academic and semi-academic publications and extant unpublished writings is given in Frýba-Reber (1994: 183–188).

Charles-Albert Sechehaye

107

de Paris (Meillet 1908), where Sechehaye is called “a man who knows linguistics, who has thought about all its problems”, but where nothing is said about the quality of his thoughts—the same Meillet, by the way, who, two years later, wrote the grossly unfair and dismissive comment (Meillet 1910: 790): Mr Sechehaye restricts himself to pure theory. His only object is to draw the plan for a theoretical linguistics while specifying its links with general psychology. He hardly touches on specific linguistic facts. He aims mainly at highlighting the need for a theory of general morphology from a static point of view, a theory that is indeed badly needed. But even where he poses the entirely historical problem of language development, he fails to escape from purely psychological considerations. It is true that Meillet was writing here for sociologists, not for linguists, but the implication is still that Sechehaye’s book is not really worth while, not even for linguists. Meanwhile it is clear that Meillet, though himself a linguist, had failed to see that pmlt was a prelude to an enormous programme of structuralist and strictly linguistic (not psychological) research into the form and nature of grammar and of the structures generated by it. Then there are the sour notes written down by Saussure in reaction to pmlt and probably forming the draft of a review which, characteristically, was never published (Godel 1957: 52; Saussure 2002: 258–261; see note 13 in Chapter 2).9 Sechehaye’s slp of 1926 was likewise relatively successful during the first few years after its appearance. It was awarded the Prix Volney by the Institut de France in 1927. A reprint was brought out in 1950 by the same publisher (Champion in Paris). It had eight reviews, all antedating 1930, all positive, and almost all commenting on the difficulty of the book (Frýba-Reber 1994: 44–46), which, it is true, went way beyond what the critics themselves could cope with. 4.2.2 Weaknesses and Prejudices I have, in the preceding pages, repeatedly expressed my great appreciation for Sechehaye as a theoretical linguist. This does not mean, however, that he never slips up or never falls into vague phrases or gratuitous speculation, 9 A similar insensitive reaction to Sechehaye’s way of looking at language is found in the review written by the French classical scholar Louis Havet (1849–1925) of the Saussure festschrift Mélanges de Saussure published in 1908, where Havet dismissively comments on Sechehaye’s contribution (Sechehaye 1908b) (Havet 1909: xxix; quoted from Frýba-Reber 2001a: 127): “Sechehaye […] takes language for what it is [Havet uses the cryptic phrase ‘dans son être’; pams], contenting himself with seeing it as a system expressing ideas”. Havet just failed to see what inspired Sechehaye.

108

Chapter 4

sometimes even sliding into what we now consider to be nonsense. Let me give one example from the very few that could be given. Speaking about phonological systems—although he was one of the first linguists to have a clear idea of the distinction between phonetics and phonology, the study of speech sounds was clearly not his forte—he writes, having ventured to posit a relation between a speech community’s phoneme inventory-cum-phonological system on the one hand and its physical type on the other (pmlt: 158, repeated in similar terms on pp. 253–254): Can we also establish a certain cause-effect relation between the phonological system and the psychological tendencies of the subject? If there is one, it will naturally be more complex and more subtle. Yet we think that one may rightfully suppose there is. The system can be more or less rich, according to whether the individuals are more, or less, capable of differentiating a large number of elements, or whether their tendencies lead them to variety or to monotony. […] It is not psychologically indifferent that one ethnic group adopts an intensity accent while another goes for a tone accent, that one language has rules regulating the quantity of its vowels while another has not. The quality of the articulations used may also be determined by psychological factors. One language has a predilection for clear vowels, as Italian does, another, such as German, multiplies its consonants and reduces the vocalic element in its syllables. Why? Is it an illusion to think that this corresponds to the psychological differences one knows that exist between these two peoples: the one being extravert and loving colours and everything that strikes the senses, the other attaching greater importance to the intellectual and subjective aspect of things? I propose to ignore such occasional lapses—the more so because, later in life, Sechehaye explicitly rejected such ideas (see, for example, Sechehaye 1933: 59). They have no bearing on the central issues of Sechehaye’s theoretical work and are the result of prejudices that were widespread at the time and cannot in general be held against individuals who grew up with them. In his later work, no such passages are to be found. The closely related issue of the Eurocentric racial and sociocultural prejudices, if not worse, found in the linguistic literature of the nineteenth and early twentieth centuries, and also, to some extent, in Sechehaye’s writings, deserves attention but should not lead to radical rejection: a modicum of historical relativity is required. It would be worthwhile to make a detailed study of this phenomenon, but all I can do here is point at it, reminding the reader that one

Charles-Albert Sechehaye

109

should see such utterances in the light of the climate of opinion that prevailed at the time and which the authors in question did not always subject to the kind of critique one might have wished. The same, I am afraid, may well apply to us when we are read by future generations. An outstanding representative of deeply engrained racism was A ­ dolphe Pictet (1799– 1875), much admired by the young Saussure (see Section 2.3). Pictet was a hard-core exponent of the racism of his day, as appears from Pictet (1856: 117), where he draws an ascending scale from the ape to the “negro” to the European, saying that “the negro tends more towards the animal than the European type does”, or from the swollen opening of his (1859), where the “European race” is said to be “destined by Providence to rule one day the entire globe” and to be “privileged among all other races by the beauty of its blood, the gift of its intelligence”, etc., etc. (see Seuren 2016, note 66). Still to be qualified as racist, though perhaps less crassly so, were Wilhelm von Humboldt (1767–1835), who, incredibly, thought the Chinese were an inferior race (Seuren 1998: 111–112), and William Dwight Whitney, much admired by Saussure, who wrote the following (Whitney 1867: 181): As, here in America, a single cultivated nation, of homogeneous speech, is taking the place of a congeries of wild tribes, with their host of discordant tongues, so, on a smaller scale, is it everywhere else: civilization and the conditions it makes are gaining upon barbarism and its isolating influences. Such utterances, though of course to be utterly condemned, should not lead to a wholesale rejection of their authors, who were as much the victims of their climate of opinion as many of us are of ours. In Saussure’s Cours and his personal writings I found no overt signs of racism or of classism. But then, Saussure was a prototypical Eurocentric who hardly ever touched on ‘exotic’ languages, always restricting himself to European languages of relatively well- studied language families. On the rare occasions that he refers to speakers of languages spoken in other parts of the world, he refers to them as ‘savages’ [‘sauvages’, e.g. Cours: 261]. Saussure never touches on questions to do with class differences among speakers, though dialectal differences are frequently mentioned. Nor did I find any open racism or classism in Sechehaye’s works, but there are clear signs of cultural prejudice in his pmlt, where I found a few passages like the following, in which Sechehaye claims a superiority of Western vis-à-vis non-Western languages and culture: “[…] we, who are in possession of languages that are much more perfect from a logical point of view ” (pmlt: 35), and “[…] savage peoples who are not yet in possession of a somehat sophisticated grammar [‘une grammaire un peu perfectionnée’]” (pmlt: 89).10 In his slp I found no such passages, though slp contains more references to ‘exotic’ languages than pmlt. Apparently, Sechehaye had educated himself in this regard during the eighteen years that passed between the two books. 4.2.3 Sechehaye and Saussure: A Paradoxical Relation A third preliminary feature that needs discussion is the paradoxical relation between Saussure and Sechehaye, first colleagues in the same department, then the one deceased but still ruling from his grave, the other a minor and unpaid member of staff. While alive, Saussure, though intellectually speaking no match for Sechehaye, held sway over the latter on account of their different status in the academic (and perhaps also the social) world. While dead, he cast a spell of such magic that Sechehaye kept being fettered by it until the end of his days. Or so it seems. One might have expected that Sechehaye, otherwise not afraid to criticise governments and fight against established but unjust social structures (FrýbaReber 1995/1996), would have pursued his own course in the theory of language. And so he did, though, apparently, not freely. Instead of coming out squarely against Saussure in those multiple respects in which he disagreed, he kept deferring to Saussure till the end of his days in what strikes one as an almost servile way. Given Sechehaye’s character as it transpires from his biography, it seems unlikely that his admiration for Saussure was a cover, kept up to the outside world for reasons of career or social standing. Yet the available evidence, though mostly of an indirect nature, suggests that he was under heavy pressure in this regard. One should not forget that he badly needed promotion in his university, which he did not get until 1929, when he was finally promoted to the still modest rank of ‘professeur extraordinaire’, a rank he kept till his death in 1946, as has been said above. As an example of Sechehaye’s show of extreme reverence for Saussure I may quote from his lecture of February 28, 1913, when he was temporarily standing in for Saussure. Since Saussure had died just six days before, this lecture took the form of a memorial speech. Referring to Saussure’s lectures which he 10

By contrast, American anthropological linguistics has always, from its late-eighteenthcentury beginnings, fostered an ideologically driven egalitarianism (Andresen 1990), inspired by the French philosophy of the Enlightenment—a philosophical movement that crucially influenced the founding fathers of the United States of America Benjamin Franklin and Thomas Jefferson, who took a firm hand in launching and directing anthropological and linguistic research among the American Indian peoples.

Charles-Albert Sechehaye

111

himself had followed twenty years earlier, from 1891 till 1893, Sechehaye commemorates his old teacher as follows (I quote, and translate, from Frýba-Reber 1994: 179): The extraordinary quality of this maître was his art of organically linking up each new topic to what had been explained before. He put things into our heads in such a solid order that, for that reason alone, they stayed fixed in our minds. This highly developed didactic gift he never deployed with greater happiness than in his lectures on Sanskrit grammar […]. He knew the art, not of stripping the grammar of its complications, which would have amounted to a falsification, but of methodically inculcating it, in a continuous and steadfast way, without ever leaving anything behind that had not been perfectly established and assimilated. He brought to his lectures the very qualities that made his knowledge stand out so much. One felt his command of the subject matter not only in his absolutely infallible memory but also, which is much rarer, in his marvellous power of synthesis and simplification. With him, the light shone on the essentials. In every subject he traced architectural lines that impressed us as definitive. The perhaps somewhat overdone nature of this praise may be attributed to the fact that he was paying tribute to a man who had just passed away—hardly a proper occasion for critical or relativising remarks. Yet other passages, such as the excessively adulatory dedication to Saussure of pmlt (see note 12 in Chapter 2) as well as frequent other references to Saussure, give the impression, or are intended to do so, that Sechehaye idolised his ‘maître’, despite the fact that their ways had parted intellectually. That they had parted intellectually is very clear.11 As we saw in note 12 of Chapter 2, Sechehaye himself explicitly says so in the dedication of his pmlt of 1908: “Although my thoughts have since taken a personal direction, […]”. 11

Amacker, who, like me, places Sechehaye far above Bally intellectually, writes: “After the publication of the Cours, Bally used Saussurean ideas rather superficially, whereas Sechehaye ended up interpreting them in a manner very close to our present understanding of the Cours” (Amacker 2000: 263). Since, however, Amacker fails to say what “our present understanding of the Cours” amounts to, it remains unclear to what extent, according to this author, Sechehaye did or did not remain faithful to Saussurean ‘orthodoxy’. The body of Amacker’s article, however, makes it clear that Sechehaye was, right from the start, critical of Saussure’s theoretical notions, despite his declarations of loyalty, which, in Amacker’s words, were “a way to manifest ostensibly […] his attachment to the Geneva School” (Amacker 2000: 214).

112

Chapter 4

In particular, he rejects, though without mentioning Saussure by name, Saussure’s continuous but fruitless brooding over the ‘negative’ and ‘oppositive’ character of language systems, discussed in Section 3.5.2 above. Here are some quotes that show this. The first is (pmlt: 34): If the language did nothing but express logical relations, one could without doubt study these relations in themselves, without caring about what can correspond to them in the grammar. But if, as Wundt asserts repeatedly and with good reason, everything in the language has a psychological coefficient, this coefficient is neither determinable nor thinkable without the association of the idea with certain signs. These signs confer their psychological form upon the idea, and it is through their mediation that the idea associates itself in turn to other ideas of the same form. No grammatical value exists without having found a material support, a corresponding sign in the language. The value exists in the sign and in virtue of the sign. To consider it detached from what incarnates it, to attribute to it a psychological quality without considering the actual quality the grammatical process may have, is to substitute for reality an arbitrary, a priori representation of what one thinks that is, instead of taking the things as they are. And (pmlt: 144): We already have compared this abstract and deductive science [i.e., the grammar; pams] to algebra. Through symbols it renders equivalents of thoughts, which makes it a logical problem, but one of a practical logic […], which depends on forms and conditions of the totality of mental life. […] This science, however, should not be compared to pure mathematics, speculating about perfectly abstract relations, but rather to celestial or physical mechanics, which shows how nature obeys these abstract laws. The interaction of forces that make matter move produces effects that the mathematician can analyse. The same applies here: each effort by human intelligence to express its thoughts by means of signs creates something that the logician must give an account of. The terminology is still uncertain—Sechehaye’s use of the terms logic, logical and logician, for example, does not stand out for consistency—but even so it is clear that he takes distance here from Saussure’s musings about his favourite but opaque tenet that everything in language is negative and oppositive.

Charles-Albert Sechehaye

113

The implicit rejection of Saussure’s ‘differential’ and ‘negative’ point of view is repeated, for example, in (slp: 94–95): Whether the characteristics of a class are internal to a word, as in a special system of flection, or external (particles, word order) is immaterial, as long as there are positive facts and the class exists in the reality of the language. [italics mine; pams]. I may also refer to a passage in a lecture he gave in 1930, (re)published in Cahiers Ferdinand de Saussure of 1944 (Sechehaye 1916: 46; quoted and translated from Frýba- Reber 1994: 176): Amongst other penetrating and forceful points, the present lectures have the merit of developing rigorously some of the consequences of a principle that dominates all of linguistics: a language consists exclusively in the links between conceptual values (ideas) and material values (ideas of given sounds). Such links, for the rest, exist only in so far as they form a unit and are determined by the system they belong to. Sechehaye clearly had no sympathy for Saussure’s attempts to abstract away, in the study of language, from actual meanings and actual phonological forms and to reduce everything to an abstract system of variables, his reason being that one cannot engage in the study of grammar without taking into account the actual formal means for the expression of meanings, as well as the actual meanings themselves. That Sechehaye was right, in this regard, is clear: semantic ‘bleaching’, for example, often goes hand in hand with categorial weakening, as in the case of auxiliary verbs (Seuren and Hamans 2009), and cannot be considered ‘in the abstract’, without taking the actual meanings and forms into account. It is difficult to get a coherent picture of Sechehaye’s attitude versus Saussure’s, who, himself, seems to have felt compelled privately to express a low opinion of Sechehaye, if we are to go by his unpublished note about Sechehaye’s pmlt, quoted in note 13 of Chapter 2. Even if Sechehaye did not know, or not realise, that his boss did not appreciate his qualities—which is highly unlikely—his repeated public professions of allegiance to Saussure are hard to interpret. I see, in principle, three possible answers. One is that Sechehaye was a real and sincere admirer of Saussure who, for him, remained the great teacher he had had the privilege of being a student of in his younger years, a man whom he had come to idealise and even to idolise. In this interpretation, Sechehaye’s hero worship must have been so great

114

Chapter 4

that he simply ignored Saussure’s obvious weaknesses and deficiencies, which he must have seen but will have put down to the unavoidable side effects of genius. His hero worship would then have been such that the fact that he deviated from several of Saussure’s viewpoints was immaterial to him. If so, he must also have accepted Saussure’s low opinion of him as correct, thinking of himself as an unworthy member of the linguistic community. What speaks in favour of this interpretation is that Sechehaye appears to us as a man of great moral integrity, who would not easily, against his own conscience, yield to pressure to help propping up a façade of unity and solidarity, masking a reality of conflicting ambitions. Moreover, as is made clear in note 5 above, there were, through his wife’s psychoanalytic activities and her friendship with Ferdinand de Saussure’s son, the psychoanalyst Raymond de Saussure, frequent and friendly social contacts between the two families. Many scholars I have spoken to do indeed believe that this interpretation is the correct one. What speaks against it, however, is that it does not fit in well with what we distill from the relevant texts. Frýba-Reber, discussing Sechehaye’s virtual disappearance from history— which, remarkably, she describes as some kind of atonement (“cette entrée au purgatoire” )—is one of those who opt for this interpretation. She seems inclined to seek the answer to the paradox of Sechehaye’s relation to Saussure in the former’s total effacement with regard to the man he allegedly considered his great master. Having stressed the fact that linguists all over Europe thought very highly of the Geneva school of linguistics, she writes (Frýba-Reber 1994: 13): Nevertheless, after his death Sechehaye’s name will only be mentioned in connection with Saussure’s, and that chiefly in the context of controversies to do with the Cours. It is not easy to explain this entrance into purgatory, but the oblivion his work fell into is no doubt related to the attitude of effacement of the author himself in front of the man who was, in his eyes, the uncontested master, and in whom he saw, with heartfelt sincerity, the model of a true scholar. I have some difficulty with this interpretation. Apart from the fact that Sechehaye’s ‘effacement’ with regard to Saussure hardly explains the oblivion his own work fell into, it fails to resolve the conflict between, on the one hand, Sechehaye’s intellectual and moral independence as illustrated above and, on the other, his ever obsequious attitude with regard to Saussure. Sechehaye must have seen the depressions, the bouts of despair, and, given his own advances in the theory of language, the lack of substance and the lack of progress

Charles-Albert Sechehaye

115

in Saussure’s thinking and simply cannot have looked upon Saussure as “the model of a true scholar”—unless, of course, he was so bewitched by the Saussurean myth that he could no longer see the things for what they were. The second interpretation is that Sechehaye’s apparent admiration was part of a façade which the Genevan linguistic establishment forced him to keep up so as to save the appearance of a united front with Saussure as its figurehead. This would explain a great deal but not everything. It seems to fit in with a large part of the evidence we draw from the relevant texts, Sechehaye’s own and those written by others, but it shows Sechehaye as a man who deliberately played a double role, one for the outside world and one for himself. This is not the Sechehaye we know from his biography and the decriptions we have of him. Sechehaye was far too honest a man to live such a life. The third answer, the one I have eventually come to lean to, stands midway between the two. In this interpretation, Sechehaye combined honesty with a certain amount of self-deception, a not uncommon phenomenon in people who try to make the best of a difficult situation. Support for this interpretation comes from the many occasions on which he expresses his admiration for his great master or his support for Saussure’s ideas. Telling, in this respect, is Sechehaye’s speech held at the publication ceremony of the Cours in 1916, published as (Sechehaye 1917). Knowing what he wrote in his pmlt of 1908 and in his slp of 1926, we see that, in this speech, Sechehaye does his level best to present Saussure’s Cours as the epitome of linguistic wisdom, without, however, forsaking his own principles and viewpoints. The result is a highly abstract, nonspecific and somewhat far-fetched construction in which he tries to make his own ideas fit in with Saussure’s and where the differences remain hidden under a mantle of abstract terminology and associations. Nothing, however, indicates that Sechehaye is insincere or hypocritical. On the contrary, this text shows him precisely as a man who is trying to make the best of a difficult situation, without compromising too much either the truth or his own honesty. But a certain tension must have been there, noticed by his immediate environment. The situation was indeed difficult and became more difficult as time passed, as a closer look at slp will show. At the end of his slp of 1926, there is a fourand-a-half- page Appendix (slp: 219–223) which, in my view, it is impossible not to see as a public declaration of loyalty to Saussurean orthodoxy, an auto da fé if you like. Everything suggests that this Appendix was forced on Sechehaye as a result of objections that more orthodox, and more powerful, Saussureans had apparently raised in regard of his book manuscript before it was published. An immediate indication that this was indeed the case consists in the profusion of references to the Cours in this Appendix, always meant to

116

Chapter 4

show that he was a true Saussurean, despite appearances to the contrary. The Appendix starts as follows (slp: 219): In 1908 we published a work entitled Programme et méthodes de la linguistique théorique, which dealt with the classification of the disciplines that deal with language. In that study, we insisted on the necessity to embed these disciplines in one another according to a logical principle. We distinguished the study of the facts of language taken in their static aspect from the study of the evolutions and we posited the necessity to establish a science of the pregrammatical, that is, of free and spontaneous expression preceding all conventional organisation. These ideas still seem to us to be correct and are in perfect agreement with the distinctions so solidly established by F. de Saussure in his Cours de linguistique générale. In particular, the science of the pregrammatical is necessarily implied in all study of organised speech and serves as its basis together with static linguistics. An essential difference between the Saussurean and our doctrine is that the Cours de linguistique générale does not draw from its distinctions any principle for rigorous classification and focuses more on the reciprocal relations existing among the different aspects of the linguistic fact. Thus, for Ferdinand de Saussure, language exists for speech, but it also arises out of speech; it emanates from it and makes it possible, and nothing forces us to place the one before or above the other. It is a complex that can only be analysed by abstraction. We, by contrast, perceive a principle of subordination and of classification. We place speech [‘la parole’], in its pregrammatical form, before language [‘la langue’]. We see here how Sechehaye defends himself against accusations to the effect that he had deviated from Saussurean orthodoxy: “These ideas still seem to us to be correct and are in perfect agreement with the distinctions so solidly established by F. de Saussure in his Cours de linguistique générale”. He begins by saying that he stands by what he wrote in his pmlt, especially as regards his theory of pregrammatical speech and his insistence on a ‘logical’ classification of the language sciences. What he says about the relation between language and speech results, one may assume, from criticisms regarding his theory of the nesting, or embedding, of the language sciences. For Sechehaye, the study of language is ‘embedded’ in the study of speech, whereas for Saussure—who for once, in my view, is right here—the two coexist in mutual dependence. Just as a social code of behaviour for living together peacefully in a community arises out of actual peaceful social interaction and at the same

Charles-Albert Sechehaye

117

time improves it by some sort of bootstrapping process, a language, as a codified system, arises out of speech while at the same time improving it, again by some sort of bootstrapping process. Sechehaye’s fixation on the hierarchical classification of the sciences got him into a similar predicament in Chapter 6 of his pmlt, with regard to his embedding of collective in individual psychology, as is shown below in Section 4.3.2. But the bulk of his defence consists of merely terminological issues.12 Points of critique that had been raised are thus made light of, being reduced to anodyne questions of terminology. For example, he had, apparently, been criticised for not making any use of the Saussurean distinction between associative and syntagmatic as made in the Cours (pp. 170–175). Defending himself, he links this issue to his use of the term symbol, a use rejected by Saussure (Cours: 101), which reduces this criticism to a mere terminological detail. The result is a passage that trivialises a serious theoretical issue and thus strikes one as contrived and defensive. The relevant text is quite interesting (slp Appendix: 220–221): The first object that the science of language [‘langue’], taken as an institution, has to study is the simple and autonomous conventional sign, which is by itself, besides other, equally autonomous signs, the carrier of a meaning. These are, roughly speaking, the unanalysable words of our dictionaries, such as cat, house, two, often, etc. In a scientific treatment, one must consider everything that concerns their constitution, both from the point of view of their sounds and from that of their ideational value [‘valeur d’idée’]. This first discipline we have called ‘symbolics’ [‘symbolique’],13 because we used the term ‘symbol’ for the linguistic sign. But this term has been criticised for not being appropriate to designating an arbitrary sign.14 To name the linguistic sign one has used the words sème or sémantème or other similar ones. But to designate a science with the help of these terms would lead to a fatal confusion with disciplines that have already been given a name: sémantiques, sémasiologie, sémiologie. To find an exact and convenient name we resort to the correct and fruitful distinction made by F. de Saussure between the associative 12 13 14

Cp. Amacker (2000: 247): “All this illustrates well how careful one must be not to let oneself be fooled, in Sechehaye’s texts, by the use of Saussurean terms; almost always he uses them in a sense that is his own”. Here a footnote incorrectly refers to pmlt: 136. On p. 137 Sechehaye uses the term symbol a few times, but symbolique occurs nowhere in pmlt. Here a footnote refers to (Cours: 101).

118

Chapter 4

and the syntagmatic relation15 and we will speak of associative grammar. In fact, the arbitrary and autonomous sign is the sign that is significant [‘significatif’] only in virtue of differences in meaning and in form found between it and the other autonomous signs of the language.16 Since these signs do not have any necessary contact with each other in the sentence, these differences are only manifest in the association of ideas. The value [‘valeur’] of two, for example, is based on a kind of implicit comparison that we make spontaneously with one, three, four and the signs for other ideas that are closely connected with that of two. [italics original; pams] A first comment that comes to mind is that Sechehaye does not make it clear how a resort to the distinction between syntagmatic and associative grammar could possibly solve the terminological issue of sign versus symbol. Then, one wonders what Sechehaye means when he says “we will speak of associative grammar”, since he does not work with, or speak of, that notion at all. The term associative grammar occurs extremely rarely in slp. It occurs immediately after the quote just given (slp Appendix: 221): “To this associative grammar another, syntagmatic, grammar must be added” and on the very last page of the Appendix, that is, of the book (slp: 223): Thus, who says “associative grammar” means by that term what in the description of a language state belongs to pure association, whereas by “syntagmatic grammar” one should mean all the combination-related facts or procedures which, in the sentence, graft themselves onto the associative facts. In the body of the text, the term associative grammar occurs only on page 4, in a paragraph which clearly is a post hoc addition I assume Sechehaye had to put in, like the Appendix, so as to make it appear to the outside world that he was a loyal Saussurean. This particular paragraph is in turn specifically referred to in the Appendix (slp: 220): “But the grammar, which, as was said on p. 4, we take in the widest possible sense, itself contains divisions that one must know

15 16

Here a footnote refers to (Cours: 170). Here a footnote refers to (Cours: 159, 163, 166). These references strike one as disingenuous, given Sechehaye’s repeated statements (see above) to the effect that a linguistics based only on differences and oppositions alone leads nowhere. In making these latter statements, he referred to (Cours: 99, 144, 190), as we have seen. The Cours thus becomes a source of quotations supporting contrary points of view—a common feature of texts that are considered sacred.

Charles-Albert Sechehaye

119

in order to give each question its proper place”. Here follows the paragraph in question (slp: 4): With F. de Saussure, we use the term grammar in its most general sense. Grammar is for us everything that concerns the organisation of the language: sounds, lexicon, syntax. In grammar, we deal with the part that concerns the combinations of signs. For reasons to be given in the Appendix, we will call this the syntagmatic grammar, as opposed to the associative grammar, which deals with signs of ideas taken in isolation. One could equally well speak of a phonological grammar”. [italics original; pams] This is, again, a curious passage. It must be noted first that it interrupts the flow of Sechehaye’s exposé: the paragraph before and the one after link up naturally with each other, as the reader who looks at the original will be able to ascertain. The paragraph just quoted, occurring between the two, strikes one as incongruous, a fremdkörper. Then, one notes how Sechehaye, having first assured the reader that he will follow Saussure’s use of the term grammar (as set out in Cours: 185–187), subsequently tells the reader that he himself will deal only “with the part that concerns the combinations of signs”, which means that he will not “use the term grammar in its most general sense” and thus will not follow Saussure in this regard—as indeed he does not. The term associative, in the Saussurean sense, does not occur anywhere else in the text of slp except in the Appendix, as we have just seen, while the term syntagmatic is found all over. These are things a sleuth cannot let go by unnoticed. The entire book is about the proposition with its overall subject-predicate structure and about the fact that the grammars of specific languages turn this pattern into surface structures consisting of hierarchically ordered constituents and subconstituents all the way down to the phonemes—a topic not touched upon at all by Saussure. This leaves open the possibility that Sechehaye’s apology may have been à contrecœur and that he did not think much of the objections made. His critics, one suspects, may well have found Sechehaye’s show of submission unconvincing but let it pass nevertheless, preferring to avoid an open conflict. Of course, all this is conjecture based on textual critique and close reading, not on direct explicit evidence, but, given the lack of such direct evidence, how else can one explain the anomalies and inconsistencies observed? I cannot escape from the impression that there existed a covert but perceivable hostility with regard to Sechehaye on the part, first, of Saussure and then of Saussure’s successor Charles Bally, and that Sechehaye was coerced to add this Appendix,

120

Chapter 4

together with the contentious paragraph on p. 4, after he had completed his manuscript. On the other hand, human nature is full of surprises and the possibility that Sechehaye was indeed blind to Saussure’s weaknesses, or that he preferred not to scrutinise his own attitudes in this matter too closely, cannot be excluded. Be all this as it may, one thing seems fairly certain: slp did not go to press without a nihil obstat from authoritative experts, probably Antoine Meillet and/or Charles Bally,17 the latter, like Saussure, not being an intellectual match but likewise in a position to cut him off from academic life. As regards the personal relation between Saussure and Sechehaye, perhaps further source-digging or an even more incisive textual critique will settle the issue. For the time being, I have no choice but to leave the issue open. 4.2.4 Sechehaye and Bally: At Cross Purposes A further look at the question of Sechehaye’s disappearance from history takes us to his colleague and superior Charles Bally, Sechehaye’s senior by five years, who, having first, in 1912, unsuccessfully tried to obtain a chair in ‘stylistics’ (Amacker 2000: 222; Joseph 1912: 611–615), took over Saussure’s chair of general linguistics and comparative Indo- European studies in 1913, after Saussure’s death. Sechehaye, by contrast, had to wait for another twenty-six years, till 1939, to succeed Bally at the age of 69, though, as has been said, still only as a ‘professeur extraordinaire’, the rank to which he had been promoted in 1929 to teach stylistics, grammatical theory and Old French. Until 1929, Sechehaye never had the opportunity to teach on the things dealt with in his two books— apart from a brief stint as a replacement for Saussure in 1913, when the latter was too ill to teach and finally died (see note 6 above for details on Sechehaye’s teaching commitments). It is thus clear that, for most or perhaps all of his active life, Sechehaye occupied a subordinate position in the hierarchy of the Geneva linguistics department, overshadowed first by the mythical Saussure, then by the essentially vapid Bally. Bally seems to have treated his subordinate Sechehaye even worse than Saussure did.18 This appears, for example, from Bally (1944; first published in 17 18

Both are thanked and praised profusely in footnote 1 on p. 7 of slp, along with an acknowledgement of a subsidy from the Genevan Société Auxiliaire des Sciences et des Arts, which made the publication of the book possible. As a telling detail I may refer to Sofia (2015: xxxi–xxxiii), where the author quotes from a letter written by Bally to Sechehaye on September 18th, 1913, not long after the decision had been taken to publish Saussure’s lectures as a book, which later became the Cours, and also shortly after Bally had been designated as Saussure’s successor in the chair of general linguistics (Saussure himself had died on February 22nd, 1913). On the basis of a conversation Sechehaye had had with Saussure’s widow, he was under the impression

Charles-Albert Sechehaye

121

1932), where the second chapter (pp. 53–75) in particular closely paraphrases large sections of Sechehaye’s slp, with only one casual and badly inadequate reference (on p. 55) to the true originator of the ideas in question. This chapter, which is in fact a rambling and undigested hodge-podge of heterogeneous notions, contains expatiations not only on monoremes and diremes and on the transition from coordination to subordination, but also on the distinction between topic and comment (see Section 5.2), which Bally calls ‘thème’ and ‘propos’, all of which are extensively discussed in Sechehaye’s slp (see Section 4.4.2 below). Throughout his academic life, Bally concentrated on what he called ‘stylistics’, which is not stylistics but the informal and impressionistic study of ‘affective’ elements in speech of whatever kind. Stylistics, for him, was part of the study of ‘parole’, the actual use of language, not of the underlying system, the

that he had been entrusted with the task of collating the students’ notes, as he in fact did, turning them into a publishable book. He had meanwhile received copies of these notes. Suffering from bad health, he had been spending the summer in the countryside, where he had barely started on this difficult and time-consuming job. While still out in the countryside, he received Bally’s letter of September 18th, 1913, in which Bally wrote: “Dear friend, For some time I have been waiting impatiently for news regarding the notes taken by Dégallier and Joseph [students whose notes were being used; pams]. I hope you don’t mind if I tell you that I need them very much and you frighten me a little by telling me that you have started working on them personally and that you won’t have finished until the New Year. […] It is difficult for me to agree with your views, since your way of collating does not correspond at all to mine. I had thought that your much appreciated collaboration would consist mostly in your rapid marking of the variants contained in the notes taken by Madame Sechehaye [who attended Saussure’s last course; pams]. […] So please return these notebooks to me by registered post in the last week of September at the latest, and, if possible, straight away. After your return, we will have occasion to see how you and I can deal with the manuscripts. The part not yet collated by you, on langue, is precisely the part I need most. […] Time is pressing. At the request of the Saussure family, and of several of my colleagues, I have morally committed myself to the undertaking of this collating job to see if it can lead to a publication, which cannot be kept in the docks for an unlimited amount of time. I am waiting for them, as I have spoken about them at conferences, in my lectures and in one of my books. I must be in a position to refer my students to them. In my inaugural lecture [held on October 27th, 1913; pams] I am planning to give a sketch of Saussure’s work and method, for which reason the notes are necessary to me.” The peremptory, if not imperious, and arrogant tone adopted by Bally will have hit Sechehaye unpleasantly, especially because, until a few months earlier, they had been colleagues of equal (humble) status, both being unpaid ‘privatdozent’.

122

Chapter 4

‘langue’, and linguistics should, in his view, focus on this kind of ‘stylistics’— clearly at cross-purposes with Sechehaye.19 Despite all this, Sechehaye wrote, in the privacy of his study, producing his two masterpieces, first, the Programme et méthodes of 1908, then the Essai of 1926, discussed in detail in the Sections 4.3 and 4.4, respectively. Both books show an author who is capable of clear, daring and penetrating thought, but neither book is easy to read. Frýba-Reber uses the term cryptanalytical when she writes about her own reading (1994: 55):20 Yet thought is not always easily accessible and we will, over the last pages of this first part, share with the reader the difficulties we encountered in the course of our cryptanalytical reading of the text. Bally, reviewing Sechehaye’s slp in an 82-lines local newspaper column, steers away from the difficult theoretical aspects, focusing on Sechehaye as a champion of new methods in language teaching. He describes Sechehaye as a man who saw the study of the abstract and remote aspects of the mechanics of language as potentially useful for application in the classroom (Bally 1926: 1; taken and translated from Frýba-Reber 1994: 45): It is important to note that, while writing these pages, Mr Sechehaye has not addressed himself only to the specialists. He belongs to those who estimate that language, as a social institution, must be of interest to the 19

20

In his (1908b), published the same year his pmlt appeared, Sechehaye had criticised, in extremely polite but nevertheless clear terms, Bally’s notion of stylistics (Sechehaye 1908b: 167): “This conception of a stylistic science, distinct from all others, is based on an idea to which we, for our part, would not be able to subscribe”. Sechehaye was not alone in this. In a critical article on Bally’s ‘stylistics’, Otto Müller wrote, quoting a review of Bally (1905) by the respected Belgian Romance scholar Maurice Wilmotte (Müller 1922: 9): I will only quarrel about the title and the definition he gives of stylistics, which upsets established usage without any advantage to anyone. A treatise on French composition or elocution is not a stylistics and calling by that name a disquisition that lacks any concern with literary style amounts to introducing a tiresome and annoying […] terminology. By contrast, Amacker writes, comparing Sechehaye’s writings with Bally’s (Amacker 2000: 227–228): It is rather different with Sechehaye, whose scientific productions are easier to analyse […] than those of Bally, in particular because his ideas, though expressed in a period where one had to respect the rules of the art of good writing, are nevertheless expressed with a clarity that is in general remarkable. Moreover, as he writes a great deal more frequently than Bally on theoretical questions, there is hardly any text of his that one can leave out of consideration.

Charles-Albert Sechehaye

123

whole of society, and his pedagogical experience has made it possible for him to make problems that count as the most thorny of general linguistics accessible to every thoughtful mind. […] One knows how many attempts have already been made to establish the correspondence between the grammatical forms and the frames of intelligence: an enterprise so daring, so daunting even, that the practical study of grammar has so far remained almost totally formalist. The reader may well be puzzled by what might be perceived as an inconsistency between the first and the second part of the passage quoted: if a proper formulation of the correspondence “between the grammatical forms and the frames of intelligence” is such a daunting task, then how could the formulation of this correspondence be useful in the classroom? It seems to me that this puzzling passage must be seen in the context of Sechehaye’s and Bally’s common didactic engagement, resulting no doubt from the fact that, in those days, Switzerland, and Geneva in particular, was an internationally renowned centre of educational innovation. In this context, the question of how to teach both foreign and native grammar in the schools had become a topic of wide interest and controversy ever since the late nineteenth century (as to some extent it still is). By the 1920s, traditional grammar had become the whipping boy of educationalists but no clear alternative was proposed (Elffers-van Ketel 2016). Sechehaye made an attempt, just as Bally did. But whereas Bally insisted on the practice of nonliterary, common usage, Sechehaye propagated a ‘constructive method’ for the teaching of syntax, which should replace the existing teaching methods and at the same time satisfy strict scientific requirements. FrýbaReber (1995/1996: 130) quotes the following passage, where, as a true Swiss, Sechehaye describes his ‘constructive method’ by comparing it to a timepiece, with its system of interdependent cogs and other pieces (Sechehaye 1916: 47–48): Take each piece, one after another with its form and its function, and consider its proper place in the clockwork between the spring that gives the movement, regulated by the balance-wheel, and the hands that mark the time. This cannot be done in any arbitrary order. There are the essential constitutive elements one must start with; then there are other, secondary ones, which exist only for and in virtue of the former. As the mind follows this natural procedure, which is like the logical genesis of the whole ensemble, it will see and understand each element, form and function, in its proper nature and in its relation to the whole. The intelligible piece has become an organ.

124

Chapter 4

From our modern vantage point we see that what Sechehaye had in mind was what we now recognise as the immediate constituent (IC-)analysis introduced in Bloomfield (1933) as a result of Wundt’s representation of sentence meanings in terms of subject and predicate and further subdivisions. Sechehaye proposes that in analysing a sentence, one should first take its main constituents, then the subconstituents, and so on until the minimum building blocks, Bloomfield’s ‘ultimate constituents’, have been identified: Sechehaye was a true structuralist, both in his theory and in his didactic proposals. That Bally, in the last sentence of the passage quoted above from his 1926 review, implies that all attempts made in the past were fruitless because they had been cast in “almost totally formalist” terms, would seem to show that, in this review, he preferred to present Sechehaye’s ‘constructive method’ as ‘antiformalist’ or ‘nonformalist’, which is puzzling, because Sechehaye himself explicitly and repeatedly speaks of grammatical mechanisms, which, by their very nature, are, if not ‘formalist’, certainly ‘formal’. This is no doubt because, for Bally, a proper, didactically fruitful, analysis and description of sentence structures does not lie in ‘formalisms’ but in nonformal, intuitive and approximative descriptions of how linguistic forms relate to their meanings. By contrast, Sechehaye was much more of a formalist and cannot have agreed with Bally in this respect. This leaves us with the question of how Sechehaye himself managed to reconcile his own expectation that a proper theory of grammar would indeed imply the setting up of formal systems of daunting complexity on the one hand with, on the other, his idea that the results of his work in this regard would be of benefit to the practical teaching of native and foreign languages. The opening page of his slp of 1926, along with the passage just quoted from his (1916), bears testimony to this idealistic belief in a union of theoretical analysis and teaching method. How realistic this is, is a question I will leave open, as the present book is not about how to teach grammar to young students. But it does seem to me that Sechehaye’s value as a theorist of language lies less in any possible didactic applications of his theoretical work than in the theoretical synthesis he envisaged and partly developed. The relative inaccessibility of Sechehaye’s texts—Frýba-Reber, as we have seen, characterises her reading of Sechehaye’s texts as “cryptanalytical”— might give rise to the thought that their very difficulty and abstractness may have contributed to Sechehaye’s oblivion after the first flares of interest immediately upon the publication of his books. Perhaps it is only with the hindsight of a century that, retroactively, an adequate assessment appears to be within reach: Sechehaye may have been too far ahead of his time. Possibly, but, on the other hand, the fame of Saussure’s Cours kept growing and growing, even though its readers failed to grasp what exactly it was intended to impart.

Charles-Albert Sechehaye

125

Perhaps this is because, in the case of the Cours, readers thought they had understood, reading their own pre-existing ideas into Saussure’s words, whereas more disciplined texts do not so easily lend themselves to this kind of selfdeception. Such texts tend to gather dust until they happen to be rediscovered. After 1950, slp seems to have vanished into thin air, along with the older pmlt. By that time, European structuralism had begun its decline, collapsing on account of its own weakness and successfully challenged by the new theory of Generative Grammar that rapidly spread across Europe after 1957. Yet despite this decline, Saussure’s Cours kept being looked up to as a shining specimen of linguistic theory, not only by those who stuck by European structuralism but also, though to a lesser extent, by the new generativists. 4.2.5 Why was Sechehaye Forgotten, or, Rather, Ignored? Why did Sechehaye never make headlines? This is a complicated question. To begin with, it seems certain that he was eclipsed by Saussure’s undeserved glory.21 Then, if I read the situation correctly, it may have been felt hat he should be kept in the background because he was suspected of deviating from his master’s ideas, which will have made him a social outcast. An additional cause, as has been said, may well have been the fact that Sechehaye’s theoretical writings were not only inspired, which his colleagues apparently failed to see, but also too difficult for them to understand, which they will have resented as it dented Saussure’s monopoly, their trade mark, and also because it made them look and feel inferior. It seems anyway that he came under a cloud of suspicion, never mentioned openly but none the less real for it. His inferior position in the department made it all the easier for his colleagues and superiors to push him aside as being irrelevant. After that, the damage was hard to repair. But apart from all this, there seems to have been a deeper and wider background to Sechehaye’s eclipse from twentieth-century linguistics: Sechehaye did not adhere to the view, not only propagated by Meillet and Saussure but 21

Amacker suggests—a suggestion I find hard to comprehend—that it was especially because of the (re-)interpretation of the Cours after the publication of Saussure’s private notes and of the manuscripts on the basis of which the Cours was published, that is, after 1957, that Sechehaye was pushed out of the arena of theoretical linguistics (Amacker 2000: 254): It seems to me […] that the importance of Sechehaye as a theoretician has been eclipsed too long by the prestige of the later interpretations of the Cours (cp. Engler 1983: 542). Since Godel’s Sources manuscrites of 1957, and Rudolf Engler’s publication, as from 1968, of the critical edition of the Cours [see Saussure 1968, 1974; pams], our view of the first half-century of discussions on the Saussurean theses has been partially distorted by the retrospective projection of these interpretations, notably, as far as I am concerned, those by Godel and De Mauro (1972), onto those of the earlier commentators.

126

Chapter 4

well-nigh universally adopted throughout twentieth-century linguistics, that linguistics, like psychology, should be entirely autonomous and independent of any other discipline, no matter how closely connected. One recalls the last sentence of Saussure’s Cours (see note 35 in Chapter 3): “La linguistique a pour unique et véritable objet la langue envisagée en elle-même et pour elle-même” (‘Linguistics has for its unique and true object language, taken by itself and for itself’). Instead, as shown below, Sechehaye supported a much less strict view of the autonomy of linguistics, whereby room is left for organic ties with adjacent disciplines such as psychology, logic, history, sociology, the physics of sound and the physiology of sound perception. In this respect, too, Sechehaye’s viewpoint was prophetic, in that, over the past twenty years or so, linguistics has developed intensive contacts with adjacent disciplines such as psychology, biology and, in particular, genetics (though the ties with logic have so far remained problematic). It seems to me that Sechehaye generated a considerable amount of hostility, in particular by his free use, following Wundt, of the notion, if not the term, proposition, against the overwhelmingly dominant trend among the linguists of his day to stay away from anything remotely reeking of logic. This testifies to Sechehaye’s independence in this respect, but he was made to pay for it. Since Aristotle, both the notion and the term ‘proposition’, in an academic context, have always been closely associated with logic. But as from the late nineteenth century, psychology and linguistics were struggling to assert their autonomy, especially with regard to logic, with the result that, after 1900, their practitioners refused to have anything to do with logic, which, if you like, was seen as the old ‘colonial power’. In addition, logic itself became mathematical during the late nineteenth century, severing all ties with the mind and maintaining only ambiguous ties with language. Aristotelian logic, which had traditionally been seen as a faculty of the human mind, was, from now on, considered faulty and inferior and thus simply dismissed, despite its great intrinsic interest not only for logic as such but also for linguistic semantics (see Seuren 2010; 2013a: 257–339; 2014a). This way, logic became the bogeyman of both psychology and linguistics, and nobody in these fields dared speak of ‘proposition’ anymore: the propositional baby was thrown out with the logical bathwater. For Wundt himself this was not yet the case: his (1880) is called Logik, though it is, in fact, a book on thinking. But in post-Wundtian psychology, the aversion to logic, and with it to both the term and the notion of proposition, became rampant. This would explain at least in part why, for example, Wundt was never even mentioned by Saussure and why Sechehaye, who placed the proposition at the centre of his theorising, was refused recognition.

Charles-Albert Sechehaye

127

In actual fact, however, the notion of proposition is central not only to logic but to an equal degree also to the study of cognition and of language. The close association with just logic is historical accident, due to Aristotle’s using the term mainly in the context of his logic. But the notion of proposition is not the exclusive property of logic. It is, in fact, the one central unifying factor providing a common point of departure for logic, cognitive psychology and theoretical linguistics. Rejecting this common role of the proposition means rejecting harmonious connections among related disciplines, such as exist, for example, among theoretical physics, chemistry and biology. The proposition connects the three disciplines concerned, logic, cognitive psychology and theoretical linguistics, and forms their common basis. Logic specifies formally what propositions are necessarily true given one or more true propositions. In psychology, one should realise that all declarative mental processing, including memory and the storage of knowledge, takes place or is organised in terms of, what is or is not taken to be the case—that is, of propositions. And as regards linguistics, given the fact that natural language utterances per se express propositions (under a speech act operator establishing some specific form of socially binding commitment; Seuren 2009, Ch. 4), it is only natural to assume that their meanings are optimally representable in the format of propositional structures. The grammar is then a set of procedures or rules transforming propositional into sentential structures (always interpretable, in token usage, in the context of shared world and situational knowledge and shared knowledge of preceding discourse). Yet as it is, modern psychology is largely given to experiments and cannot be said to have a generally accepted philosophical and theoretical basis. Theoretical linguistics has, for most of the twentieth century, either focused on the combinatorics of linguistic forms (morphology, syntax) or on associations and frequencies (behaviourism, cognitivist linguistics, frequency-based or ‘big data’ linguistics), while the linguistic subdisciplines, in particular the study of language acquisition, have been equally indifferent to the notion of proposition. (Sechehaye, as is shown below, took a vivid interest in language acquisition processes, but always from the propositional point of view.) All in all, the proposition simply disappeared from the script, and Sechehaye with it. 4.3

Programme et Méthodes of 1908

For a proper understanding of Sechehaye’s programme in the theory of grammar, and of language in general, it is necessary to understand that he was driven by one overarching and all-pervading goal: the reduction of the surface

128

Chapter 4

structures of sentences to hierarchical compositions of propositional structures, each consisting of a predicate on the one hand and one or more argument terms on the other—his grammatical problem (‘problème grammatical’). This meant, among other things, that all lexical items of all different classes must be reduced to predicates: not only verbs and adjectives, but also nouns, quantifiers (all, some, most, the numerals), prepositions and case markers, conjunctions (because, while, although), connectives (and, or, but), and much more, will have predicate-status at the level of semantic analysis. Sechehaye, using the terms idea and notion for what we now prefer to call predicate, expressed this by saying (1908b: 167–168): What do our words, like all conventional elements of our language, such as particles, prefixes, suffixes and syntactic combinations of these diverse elements, represent? Ideas, and nothing but ideas by themselves. Their raison d’être consists in being linguistic substitutes of certain notions. […] All conventional means of expression invariably represent ideas of any order. Flection, for example, […] translates, by means of its case forms, relational ideas but it may also correspond to modal, that is, subjective, ideas when, in the conjugation, it marks the difference between an indicative, a subjunctive or an imperative. In modern terms we would say that, in a sentence like He was dancing on the table, the surface-structure preposition on is taken to represent the highest predicate in the semantic analysis. As a predicate, ‘on’ takes two argument terms, the nominal object term the table and the propositional subject term he danced, as if the sentence was “his dancing on-ned the table”, with the quasi-verb ‘on’.22 The grammar of each specific language L thus has the task of converting semantic analyses of sentences—that is, hierarchical compositions of predicate-argument structures—into semantically corresponding, well- formed surface structures of L. It is essential for a proper understanding of Sechehaye’s work that the reader keep this overall viewpoint in mind when reading the following pages, even though, of course, Sechehaye himself was as

22

In fact, this is how this sentence reads in Korean. Note also that the negation of this sentence, He didn’t dance on the table, is naturally taken to deny the prepositional phrase ‘on the table’, not the assertion of his dancing, just as the negation in, for example, His dancing didn’t disturb the neighbours denies the disturbance of the neighbours, leaving the fact of his dancing unaffected.

Charles-Albert Sechehaye

129

yet unable to present a fully fledged machinery actually achieving this transformation process.23 4.3.1 Overall Survey of pmlt The outer framework of Sechehaye’s pmlt of 1908 is formed by the question of what place should be assigned to linguistics in the totality of sciences—a question that preoccupied all practitioners of the human sciences of the period and should be seen in the light of the widespread efforts to create an autonomous psychology and an autonomous linguistics. These efforts made it necessary to assign to these new disciplines a place in the overall landscape of science, so that they can delimit themselves and stake off their own exclusive territory. From this viewpoint, Sechehaye’s preoccupation with the place of linguistics among the other, equally autonomous, disciplines is perfectly understandable, whereby it is to be noted that Sechehaye differed from Saussure, who, as has been shown, addressed the same question, in that Sechehaye attempted to create a framework in which linguistics, psychology and sociology entertain good neighbourly relations, whereas Saussure, like Meillet, was concerned to maximise the differences and to pull up impenetrable barriers between them. I will take this encompassing framework into account only in so far as it has an impact on the methodological questions discussed by Sechehaye. The main relevance of pmlt, and, at the same time, Sechehaye’s main concern, does not lie in the framework but in the central theme of both pmlt and slp, which is the theory of grammar. Unlike Saussure, who was wrestling with the overarching parameters of human language, Sechehaye wrestled with what he called the grammatical problem (‘le problème grammatical’), which consisted, for him, in the technical problem of how to transform, by means of a systematic set of rules (his ‘mechanism’ or ‘organism’), the semantic representations of sentences, cast into the form of logico-semantic propositions, into semantically corresponding well-formed surface structures of any specific language in question, in which the ‘original’ logico-semantic form of the sentence structure often is not preserved or recognisable only with great difficulty. For Sechehaye, the grammatical problem “is, if not the only then certainly the most important, object of linguistic theory” (pmlt: 23).

23

Sechehaye realised that establishing formal systems mapping thoughts onto corresponding surface structures is a task of great magnitude (Sechehaye pmlt: 45): “The following pages have no other pretention than to sketch a programme, the execution of which is a long-term project that goes beyond our powers”.

130

Chapter 4

The gist of Sechehaye’s 1908 study is captured in the following budding insights: (i) Sentence meanings are representable in terms of subject-predicate structures as in the propositions of traditional Aristotelian and Medieval predicate logic—a rationalist insight shared with Wundt and with Lancelot and Arnauld’s Port Royal grammar of 1660, and revived independently, and in a more modern form, in McCawley (1967, 1970, 1972). (ii) The grammatical subject-predicate structure of natural language sentences often does not reflect the subject-predicate structure of the semantically underlying propositions—an insight shared with the participants in the great subject- predicate debate of Sechehaye’s own day. (iii) Most importantly, between the subject-predicate structures known from predicate logic on the one hand and the surface structure of the corresponding sentences on the other, there must be taken to exist, for each specific language, a complex system of rules/habits/conventions/ dispositions/laws (Sechehaye does not yet distinguish these notions; see Section 3.2.3) transforming the former into the latter—an insight shared implicitly with the entire, centuries-old tradition of grammar writing and thinking about grammar, and explicitly (that is, supported by precise formal analyses and descriptions) with the school of ‘Generative Semantics’ which, as a branch of generative grammar, flourished between 1965 and 1980.24 The importance of the book lies in the fact that it begins to lay the foundations for, and thus announces, the detailed study of formal grammar as a universally constrained system transforming the propositional structures that represent linguistic sentence meaning as a product of contextually integrated thought into well-formed sentences of natural languages. Sechehaye actually uses the term syntactic transformation on page 22 of pmlt, though not as a technical term but more or less casually: Searching for a truly rational method for the study of the syntactic transformations, we opened Wundt’s volume in the hope of finding a solution, 24

The generative semantics movement of the 1970s would have found inspiration in Sechehaye’s work if it had known about it. The leading generative semanticist Jim McCawley (1938–1999) wrote (McCawley 1973: 244): “It should be evident by now that I am proposing a system of semantic representation that is along the lines of the notational system used in symbolic logic”. One page later, McCawley mentions the fact that this point of view was already proposed in Wundt (1900), thanking Keith Percival in a footnote for bringing this to his attention. Had Percival and McCawley dug a little deeper, they would have come across Albert Sechehaye as well.

Charles-Albert Sechehaye

131

or at least the elements of a solution, to the problem that we had posed ourselves. [italics mine; pams] This is, to my knowledge, the first occurrence in the linguistic literature of the term syntactic transformation in the sense in which it is used today. But a clear notion of (transformational) rule is still absent. Interestingly, we encounter the term generative principle in Sechehaye (slp: 30): “ subject and predicate has become the generative principle [‘le principe générateur’}, the central organ of the entire grammatical mechanism”. In this sense, both Sechehaye’s pmlt and his slp were prophetic. Sechehaye considers it to be the task of theoretical linguistics to follow an inductive process leading to a deductive system from which the observed facts can be formally derived—that is, to what we now know as an algorithmically organised grammar of any specific language (pmlt: 47): In the first chapters of this essay we have justified in principle the existence of theoretical linguistics […] brings together what is similar and looks for constant relations so as to elevate itself by induction from particular facts to the last principles, then to redescend from these to the particular facts by deduction, thus explaining them as necessary manifestations of these principles under the given conditions. Again interestingly, very much the same thought is expressed by the American linguist Zellig Harris, whose highly original work continues the lines set out by Bloomfield and who thus became the inventor of generative grammar (Harris 1951: 372–373): The work of analysis leads right up to the statements which enable anyone to synthesize or predict utterances in the language. These statements form a deductive system with axiomatically defined initial elements and with theorems concerning the relations among them. The final theorems would indicate the structure of the utterances of the language in terms of the preceding parts of the system. See also the following passage, which shows Sechehaye as a true structuralist who actually presaged the notion of a formal algorithmical grammar (Sechehaye 1916: 76): The spirit of syntax, so to speak, must be constructive and architectural. What this science requires, by its very nature, is not an enormous learned

132

Chapter 4

compilation of superficially classified facts but a substantial and insightful résumé of them. Sechehaye thus anticipated the notion of a transformational grammar organised in the terms of a propositional subject-predicate format. Albert Sechehaye was much inspired by Wundt’s tree diagrams representing underlying sentence meanings in the propositional format of subject-predicate structures (Wundt 1880: 53–54, 57; 1900: 329). His problem with Wundt was, as we have seen, that Wundt does nothing about the relation between these semantic structures on the one hand and the surface structures of the various languages on the other. Still lacking an explicit notion of syntax and not having much of an interest in phonology, Sechehaye could do little but restrict his notion of structure to morphology and a small bit of shallow syntax. He speaks, following Humboldt (see Jankowsky 1972: 48–49) and the Young Grammarians, of the organisation of the sentence (‘organisation de la phrase’) for what we call its structure, and of organism or mechanism for what we call the underlying grammatical system. As one form of ‘organisation’ he mentions linear ordering rules (‘règles d’ordonnance’) (pmlt: 116) that vary from language to language, but these are not sufficient to define a grammar: “We have mere word order, as in many primitive [sic; pams] languages, we have composition (as in German Kirchturm [‘church tower’]), flection (Charles’s hat), prepositions (le chapeau de Charles [‘the hat of Charles’])” (pmlt: 117–118). We even have different word class selections, he continues, in that the same semantic content can be expressed by very different grammatical means. (An example would be the grammatical difference between the semantically equivalent English and German sentences I like swimming and Ich schwimme gerne, respectively). Moreover, he distinguishes consistently between “an abstract form of organised language, which is the very form of our thought” on the one hand and “the conventional sounds by means of which this abstract form is realised” on the other (pmlt: 122), thereby defining the grammar of a language as a system mapping semantic representations onto surface structures—not the other way round, as for him, a sentence expresses a thought. One of the things that make Sechehaye’s nascent notion of structure different from what is found decades later in fully fledged structuralism is the fact that, for Sechehaye, at least some structures, or constructions, reflect psychological processes. If, for example, the genitive relation is expressed by mere word order, as in any equivalent of chapeau Charles, meaning ‘Charles’s hat’, “it is the hat that makes one think of Charles, and not the other way round” (pmlt: 118). Such (quasi-)psychological accounts of sentence processing as a surrogate for grammatical analysis and description are frequently found in

Charles-Albert Sechehaye

133

both pmlt and slp, and Sechehaye was also frequently criticised for it. There are, however, also many passages where Sechehaye emphasises that grammatical structure is a matter of language-specific rules of grammar, which have nothing to do with the psychology of sentence processing. This just shows how nebulous and unstable his notion of a grammatical system still was. One should not forget that we are in early twentieth-century Europe, where all human actions, including the production of utterances, were considered to be guided directly by psychological forces reflected in these actions and open to awareness and reporting. 4.3.2 Comments on Successive Chapters But let us look at pmlt as it presents itself in its successive chapters, only some of which can be discussed in detail. Thus the sections on phonetics and phonology will, on the whole, be left aside, as they are of lesser relevance in the present context, despite their intrinsic interest and despite the fact that Sechehaye has highly interesting things to say on these subjects. Nor will the chapters or sections in which Sechehaye discusses questions of language change be commented on, no matter how interesting his texts are in this regard. This way I hope to be able to transmit to the reader what I experienced as I was working my way through the book. The reader will then be able to reconstruct, bit by bit, the theoretical edifice in so far as it was created and erected by its author back in the earliest years of the twentieth century, while at the same time being made privy to my, often extensive, comments. The numerous and sometimes lengthy quotes from pmlt should give the reader a direct taste of Sechehaye’s ideas and of his style of writing (as far as it transpires through the translation). After a first chapter (pp. 1–10) in which Sechehaye distinguishes between classifying and law-driven sciences,25 historical linguistics falling under the former and thus needing a counterpart in the latter, he further elaborates this thesis in the second chapter (pp. 11– 22). In the third chapter (pp. 23–46), Sechehaye launches into a critique of Wundt, the main point, repeated over and over again, being that Wundt fails to see what Sechehaye calls the grammar problem (‘le problème grammatical’), that is, the problem of formulating precisely and in all details the mapping of the propositionally structured semantic representations of sentences onto the corresponding well-formed surface sentences 25

Here Sechehaye followed Adrien Naville (1845–1930), philosophy professor at the University of Geneva, for whom the sciences fall into two complementary main classes, the classifying (i.e. taxonomic) and the law-driven (i.e. explanatory) ones, (‘sciences des faits’ versus ‘sciences des lois’) (Naville 1901).

134

Chapter 4

in any given natural language. This implies that Wundt’s analysis of sentence meanings as propositional subject-predicate (or predicate-argument) structures needs as a complementary counterpart a structural-transformational grammar converting these semantic structures into language-specific surface structures. Sechehaye’s Chapter 3 starts off as follows (pmlt: 23): The essential critique we think we must formulate with regard to Wundt can be summarised in a few words: he has addressed all linguistic problems that seemed to him to be of a psychological nature, but he has neglected everything that he thought to belong more especially to the domain of grammar. In a word, he has not understood the importance of the grammar problem. […] This is, if not the only then certainly the most important, object of linguistic theory. (italics mine; pams) He thus implies the autonomy of linguistics with regard to psychology and other human sciences, posited more explicitly on p. 37, where he also implies the necessity of distinguishing between language as a regulating system (called ‘organism’ here) and the use made of it in speech (pmlt: 37): True, gives us a few examples of syntactic constructions of different types, discussing them from a psychological point of view. But in doing so, the author appears to be planing above the grammar, looking at it from above or from outside. What is lacking is an exposé—no matter how elementary provided it is as general and systematic as possible— of how actual linguistic expressions come about. If he has failed to provide that, it is because he always has the speaking subject in mind and does not believe that grammar itself is an object of psychological study distinct from the speaking subject. And yet the linguist has to do with linguistic organisms whose laws impose themselves on the speaking subjects, who are themselves unable to change them other than within strict limits. These are organisms that must be analysed in their constitutive parts and explicated as to the way they function. There is thus room here for a whole rational and general grammar, of which Wundt gives us only a few insufficiently coordinated principles. In a way, Sechehaye’s critique of Wundt is ambiguous. He blames Wundt for not having seen the importance and the complexity of this mental transformational system and thus not having done anything about it—that is, for not having dealt with the grammar problem. Yet at the same time he sets the grammar problem apart from psychology, making it a problem for linguists, not for

Charles-Albert Sechehaye

135

psychologists. It is as if he reproaches Wundt for not having become a theoretical linguist, besides being a psychologist. A few quotes will illustrate this. First, we read (pmlt: 32–33): This chapter is among the richest and most interesting ones. Yet, here again, we have grave objections to raise regarding the way in which the author has addressed and dealt with his subject matter. […] Where we expected a chapter on the theory of language, we find scattered fragments of historical grammar provided with psychological commentary. (italics mine; pams) And (pmlt: 37): Here we do not blame Wundt for anything except that he stopped short of crossing the threshold of a linguistic discipline, having first provided us with its first elements. A few pages down, however, Wundt is forgiven for not going into the grammar problem, as the psychology of the speaker and the underlying system of grammar belong to two distinct disciplines, psychology and theoretical grammar, the latter being embedded in the former (pmlt: 41): If this [i.e. the conversion of logico-semantic sentence structures into surface sentences; pams] is the role of language and its grammar in our mental life, are we wrong when we estimate that the theoretical discipline that occupies itself with this should explain the mechanism, the functioning and the life of this marvellous instrument, and that the psychology of the speaking subject and his creative functions can only serve as an introduction to a higher [i.e. embedded; pams] science? And (pmlt: 43): If Wundt did not squarely address the linguistic problem, if he failed to see that it cannot be solved unless the grammar problem is addressed as well, it is because, being a psychologist rather than a linguist, he was more concerned to practise psychology with respect to language than to practise the psychology of language properly speaking. But apart from Sechehaye’s critique of Wundt, what were his own ideas about such issues as logic, grammar, structure? These we find in the chapters

136

Chapter 4

following his critique of Wundt. Language is defined in Chapter 4 (pp. 47–54) in the following way (pmlt: 48): Language [‘le langage’] […] is the totality of means which a psychophysical being has at its disposal to express its thoughts. This is followed by a few pages in which Sechehaye tries to explicate the notion ‘express’. He finds that difficult because the notion is too wide to delimit the notion of ‘language’. His final conclusion is that the border between language and nonlanguage in the use of conventionalised signs is vague (pmlt: 49): “We have to do here with a difference of degree, not a qualitative difference”. Ulti­ mately, the decision between what is language and what is not depends on our actual familiarity with the phenomenon we prefer to call ‘language’. In the end, he thus abandons any attempt at precise definition, leaving it to the linguist to decide whether or not anything qualifies for the predicate ‘language’ (pmlt: 50): “To show under what conditions the rational study of language presents itself to us one must know language itself as it is, and not just the idea one may have of it. With this definition we will be able to find out in nature what deserves to be called language”. In this vein, he finds that any language (langue) must have a grammar, that is, a system (an ‘organism’ or ‘mechanism’ in his words) of habits or rules (he does not distinguish between these notions) telling speakers how to compose their utterances given the thoughts they wish to express (pmlt: 51): “[…] the grammatical organism is like a channel through which our intellectual activity must pass for it to get materialised”. But besides the grammar of the language, there are also extralinguistic elements in any utterance, reflecting momentary psychological states. Accordingly, he draws a distinction between the language system, which is the object of theoretical linguistics and falls under collective psychology, and extralinguistic (we would say pragmatic) elements in speech, which fall under individual psychology (pmlt: 52–54): Furthermore, it is not hard to observe that besides those expressive elements whose value rests on a grammatical convention, language contains many more such elements that seem to depend directly on the free spontaneity of the subject. Some accents, intonational nuances, in many cases also the word order, and a large number of similar phenomena in our sentences are dictated by a kind of psychological necessity and, though outside all fixed convention, they help to comprehend the discourse and are directly interpreted by the intuition of their psychological causes. […] The study of the grammatical element in the language thus falls under collective psychology.

Charles-Albert Sechehaye

137

As for the extragrammatical elements, which are not subject to any conventional rule, these depend directly on the psychophysiological activity of the speaker […] must be explained entirely by the laws of simple or individual physiological psychology. Our theoretical linguistics thus depends at the same time on individual psychology and on collective psychology. This makes us predict that there is good reason for subdividing this science into two complementary disciplines, one being incorporated into individual, the other into collective psychology. (That contrastive accents may make for truth-conditional differences, as shown in note 8 of Chapter 3 and in note 18 of Chapter 5 below, was something Sechehaye was not aware of. In this respect, he does not differ from the entire linguistic world of today, where such facts as are mentioned in these two notes are likewise unknown or in any case not taken into account. Awareness of such facts is bound to cause a radical shake-up in current ideas about semantics and grammar, and about the role of pragmatics.) Chapter 5 (pp. 55–65) is of a philosophical and epistemological nature, touching, inter alia, on the great question of reductionism. Each science is embedded (‘emboîtée’) in a lower, more comprehensive science. Like Hippolyte Taine, mentioned in Section 3.1.2, Sechehaye defends the CartesianKantian position that the first and most certain evidence comes from human thought and experience, which is why, he says, mathematics is the first of all sciences.26 In his hierarchy of sciences, psychology is embedded in biology (pmlt: 57–58): [T]his is the reason for the primacy of mathematics. From there, one descends stepwise towards ever more complex realities. Knowledge of an organised and living being is thus obtained by the solution of a series of problems of distinct and naturally subordinated orders: mathematical problem, mechanical problem, physical problem, chemical problem, biological problem. […] According to this principle, the biological sciences […] comprehend the physiology of the nervous system, in which psychology, especially human psychology, is in turn embedded.

26

In this, Sechehaye unwittingly followed the Dutch mathematician-philosopher Lutzen E.J. Brouwer’s (1881–1966) intuitionism in mathematics, which develops a mathematics exclusively from formalised mathematical introspection. The interested reader may consult, for example, Brouwer (1907, 1967); Iemhoff (2013). See also Chapter 5.

138

Chapter 4

Having distinguished individual from collective psychology, the latter being embedded in the former, he then classifies collective psychology as the last of the natural sciences (pmlt: 59): “This collective psychology is the last of the sciences of nature and serves as the basis for the moral sciences, which build upon it”. The “moral sciences” are (pmlt: 60) “for example, history and its philosophy, the social and political sciences and everything concerning philosophy, art and religion”. Sechehaye had some idea of token physicalism, but only in nucleo, not explicitly. His notion of token physicalism is closely bound up with his ideas about the nested hierarchy of sciences. There are, he says, some unmistakable criteria for nesting, or embedding (pmlt: 61–64): 1° It must be possible to think […] of the facts dealt with in the e­ mbedding science independently of the facts of the embedded science. Thus, I can think of my body without the life that animates it. […] 2° An unmistakable sign of embedding is the circumstance that nature offers us facts that are attributable to a certain order independently of any mixture with a different order, while, by contrast, this different order appears only in combination with the facts of the former. […] 3° In the third and last place, it must be observed that the embedded ­science always studies more complex and, mostly, also more concrete phenomena than the embedding science. […] [A]n unmistakable sign of embedding is the circumstance that the input of a new science is required for a proper understanding of how a phenomenon whose essential principle, its principal factor, is already known in a rational manner is practically manifested in concrete reality. In Chapter 6 of pmlt (pp. 67–73), Sechehaye discusses a tangle he has worked himself into: his theory of hierarchical nesting of the sciences turns out not to fit in well with his thesis that collective psychology is embedded in individual psychology. According to his theory, it should be possible to talk about extragrammatical facts, which belong to individual psychology without any mention of the language system, which falls under collective psychology. But that is not possible, since extragrammatical facts are only that in virtue of the fact that there is a grammar, a language system (pmlt: 68): “One is faced here with a kind of question begging [‘pétition de principe’] from which it is difficult to escape”. His answer is that one can imagine what he calls a ‘prelanguage’ (‘langage prégrammatical’), that is, a “language without any element of organisation, of habit or any other convention” (pmlt: 69), as when one has no other means

Charles-Albert Sechehaye

139

of communication than to improvise with sounds and gestures. This, however, turns the whole problem into a question of terminology: it all depends on how you wish to define ‘language’. True, as we have seen, Sechehaye defines ‘language’, in the general sense of the French term langage, as follows (pmlt: 48): “Language [‘le langage’] […] is the totality of means which a psychophysical being has at its disposal to express its thoughts”. But he also says (pmlt: 51): What strikes us above all when we observe language [‘le langage’] spoken by humans is […] its organic character. There is a grammatical mechanism consisting of fixed conventions, by means of which ideas are associated with certain signs and with certain grammatical rules specifying how the signs are to be combined. He finds a way out by reformulating the concept of ‘langage’ (pmlt: 78): Language [‘le langage’], as we have defined it and as we generally observe it, first appeared to us as something organised. We have needed a closer examination to recognise that the rules of language [‘les règles du langage’] do not make up all of language. This proves that we have to do with a phenomenon in which the grammatical factor is predominant: it is to this factor that it owes its most essential character. It is up to the reader, of course, to judge the merits of this answer. For me it is not convincing, the reason being that the tangle does not arise when one realises that the study of language does not belong to collective but to individual psychology. One cannot, in my view, maintain that the study of grammar belongs to collective, or social, psychology. Social psychology studies the specific effects of groups as opposed to individuals. That is, it studies the phenomena that only arise when humans are placed together in groups. Linguistic interaction does not require a group: it mostly occurs between individuals or with just the speaker or reader involved. The study of language change or language settling (see Seuren 2013a, Ch. 1) may be reckoned to fall under social psychology as a special subdiscipline but the study of language systems may not, the reason being that there is no language change or language settling unless it is accepted by the speech community, which is a social process and requires a group. Individual psychology, by contrast, studies processes, structures and systems in the individual mind, whereby it is irrelevant whether or not these are, knowingly or unknowingly, shared with other individuals. From this perspective, the study of grammar must be seen as belonging to individual

140

Chapter 4

psychology, in which case no tangle arises.27 Sechehaye has become the victim of his passion for classifying the sciences in terms of one great hierarchical synthesis. Chapter 8 (pp. 95–101) expands on the nature and method of collective psychology and is of less direct relevance in the present context. The longer Chapter 9 (pp. 103–125) deals more directly with the notion of grammar and is, for that reason, highly relevant, and also highly interesting, as one sees budding notions and insights that have proved extremely useful and fruitful alongside withered remnants of the past. I will focus on the exciting new ideas rather than on the withered remnants. The chapter starts off as follows (pmlt: 103): The essential problem of theoretical linguistics dealing with spoken language is the grammar problem […]. This problem must be resolved deductively under the control of the facts, starting from an exact knowledge of what is given. This is refreshingly modern. The phrase “deductively under the control of the facts” presages Karl Popper’s famous method of hypothetico-deductive theory building and of falsification (Popper 1959/1935). Moreover, as we saw in note 24, it announces deductively organised transformational grammar, which is sufficient reason to give us pause. The instrument in the use of articulated language is intelligence, Sechehaye’s term for what we prefer to call the rational mind or cognition (pmlt: 104): What is the instrument used by this agent [i.e. the collectivity; pams], what is the form of the mental activity especially put to work in the realisation of this aim [i.e. the transmission of thought; pams]? We have said it before: it is human intelligence. […] constant forms corresponding to the various operations of the mind, the capacity to formulate a judgment, to express logical relations. The supreme rule for a grammar that is systematic and evolves is logic. [italics original; pams] The terminology is a little antiquated, but if we read ‘regular form-meaning correspondences’ for ‘constant forms corresponding to the various operations

27

That Sechehaye classifies the study of grammar as a subdiscipline of social or collective psychology is no doubt related to the fact that Wundt classifies the study of language as a subdiscipline of his ‘völkerpsychology’, as shown in Section 3.1.3 above.

Charles-Albert Sechehaye

141

of the mind’, ‘proposition’ for ‘judgment’ and ‘predicate-argument relations’ for ‘logical relations’, we are very close to a rationalist theory of grammar in the spirit of Sanctius and the old Port Royal grammar mentioned in Section 2.4 above. We also have a reference, albeit a vague one, to the role of logic. Let us have a closer look, since the relation of Sechehaye’s theory of language to logic is of central importance to him, as appears from the title of his 1926 book Essai sur la structure logique de la phrase.28 He defines the term logic as follows (pmlt: 104): However, this term is open to equivocation. In general, and with good reason, one understands by the term logic an entirely theoretical science which aims to capture the basic relations of thought in their purest and most abstract form. Logic is an exact science, sister to mathematics. We could say, following the familiar analogy between the natural and the language sciences, that the logic of grammar is to philosophical logic what the forms and movements we see around us are to the forms and movements of geometry and mechanics. Grammar is a practical and applied logic. That the term is “open to equivocation” is very true: it has for a long time been used in a bewildering variety of senses. I will spare the reader a survey of the ways the term logic has been used over the centuries and limit myself to saying that such a survey would lead to the conclusion that, in common usage, the meaning of the terms logic and logical has been whittled down to ‘(something) that makes sense’. Sechehaye does not like that: he wants to reduce the term to manageable proportions, with the result that he sees logic, if I interpret him correctly, as the analysis and description of the laws of sound reasoning—a common view during the nineteenth century. If this analysis and description is fully formalised, as was attempted by Augustus De Morgan in his (1847), we have indeed “an exact science, sister to mathematics”. Sechehaye had not heard, however, of the modern, fully formalised propositional and predicate logic developed by mathematical philosophers during the late nineteenth and early twentieth centuries, in particular Frege, Peano,

28

Frýba-Reber (1994: 34–35) tells us that, having used as a working title for his 1926 book Le problème grammatical, he subsequently considered: Un essai sur la structure de la phrase, Un essai sur la structure de la phrase dans ses éléments logiques et psychologiques, La structure de la phrase. Premier essai: les idées, and La structure de la phrase grammaticale. Premier essai: les idées, until he finally settled for the title it has.

142

Chapter 4

Russell. Or if he had, he did not study it, since that logic is emphatically presented as not being a formalisation of correct ways of thinking but as a system of metaphysically necessary relations among well-defined sets of propoitional formulae, expressed in an analytically precise formal language. Nor did he have to study it, since standard modern logic is a metaphysically necessary mathematical construct successfully applied to physical nature, but not reflecting the logical operations of the human mind, which makes an ingeniously selective use of the possibilities afforded by it (Seuren 2010; Seuren & Jaspers 2014).29 It sounds as if, back in the 1940s, Sechehaye had an inkling of this when he wrote (Sechehaye 1941: 81; quoted and translated from Frýba-Reber 1994: 162): ≤Logic> is neither the determining cause nor its goal. Language is there to serve our practical needs, it was constructed from materials that were already available, and for that reason it is imaginative and, as a consequence, limited in its possibilities of abstraction. What may have escaped Sechehaye, however, as it still escapes even the most advanced logicians and cognitive and linguistic theorists of our day, is that the classic propositional subject-predicate (or predicate-arguments) structure has at least the following three distinct but mutually related functions (for an analysis of the notion of proposition, see Section 5.5.1 below):30

29

30

This new formal logic (or ‘logistic’, as it was often called in the early days) is applicable to the full extent to physical matter, just the way arithmetic is, but its creators did not, on the whole, express themselves clearly on the extent to which it could possibly be applied to the logical powers of the human mind and to the semantics of natural language—that is, as part of a theory of truth for natural language sentences. The Polish-American logician Alfred Tarski (1901–1983) was an exception in that he was clear on this issue: mathematical logic is not, as such, applicable to human language and the human mind. Logicians should beware of trying to apply mathematical logic to natural language and creating a formalised truth-conditional semantics for it in the terms of standard modern predicate logic, since natural language is intrinsically vague and very unlike formal logical languages (Tarski 1946: 27–28). (The reason given by Tarski is false: natural language is a great deal less vague than Tarski thought and it can to a much larger extent be caught in terms of a formal logical system, but this system differs considerably from standard modern logic; see note 16 in Chapter 3.) Tarski’s advice was not followed by American logicians and philosophers such as Donald Davidson (1917–2003), Saul Kripke (b. 1940), David Lewis (1941–2001) or Richard Montague (1930–1971), who unsuccessfully used the concept of ‘possible worlds’ for the analysis of the semantics of natural language (for a detailed critique see Seuren 2013a, Ch. 9). This insight was still present in Arnault & Lancelot’s Port-Royal Grammar, as was shown in Section 3.3.2 above, but it got lost during subsequent centuries. Sechehaye revived it, though it is not known whether he was inspired by the Port-Royal Grammar.

Charles-Albert Sechehaye

143

(i) It provides the format for the cognitive specification of actual and possible states of affairs, and thus for the acquisiton and storage of knowledge. (ii) It forms the basis for logical entailment relations, as well as for processes of practical or theoretical induction. (iii) It is, for these reasons, the most appropriate format for the semantic representations of natural language sentences to be cast into. The linguist is not directly interested in (i) or (ii) but all the more in (iii): if the semantic representations of sentences are cast in the format of propositions, then this format must be retrievable from the surface structure of sentences, which means that propositional structure permeates the grammars of natural languages. The use of the propositional subject-predicate (or predicate-arguments) format in semantics or in grammar (syntax) is, therefore, neither a ‘logicism’ nor a ‘psychologism’, but fully justified from a strictly linguistic point of view. The criticism poured on Sechehaye after the appearance of his pmlt to the effect that his approach to language was too much infected by logic (‘logicism’) and by psychology (‘psychologism’) is, therefore, unjustified in so far as it was based on insufficient insight into the fact that the propositional format fulfills at least the three distinct though mutually dependent functions just mentioned. On the other hand, it was justified in so far as Sechehaye, like Wundt before him, was insufficiently clear in his own mind about the status and multifunctionality of the propositional format and all too often still fell back on (romanticist or idealist) psychological notions and theories that are incompatible with the subject-predicate structure of thought content. Yet the direction his thoughts were taking is clear: he was gradually detaching himself from the romanticist or idealist notion of how sentences come into being and explored with increasing fervour the notion that the underlying structural principle of sentences is the subject-predicate (or predicatearguments) format—the format that forms the basis of all logical deduction and empirical induction. It was not for nothing that he eventually called his second book ‘Essay on the logical structure of the sentence’ (‘Essai sur la structure logique de la phrase’)—even though, as argued below, the word propositional would have been more appropriate than the word logical. Chapter 10 (pp. 127–130) is again on the nesting of the sciences and thus interests us less directly—but for two passages in which the author shows that the synchronic study is prior to the diachronic study of language, stressing that one has to know the system before one can understand the change (pmlt: 127–129):31

31

We find Sechehaye’s saying echoed in Roman Jakobson’s statement with regard to language disorders (Jakobson & Halle 1956: 55):

144

Chapter 4

[T]he results of the static disciplines, no matter how imperfect still, are a necessary prerequisite for the evolutive sciences. The state of a language is the environment that must be known before the evolution of the language can be understood. […] We can show this by a simple appeal to common sense: to understand an evolution one must first know what it is that has evolved. […] One cannot show what has happened to the various elements of a language in the course of an evolution without first defining them and having intimate knowledge of their nature. These definitions are provided by the static sciences. […] How could I understand, and make others understand, the modification that an engineer has carried out on a machine if I have not first formed a clear idea of the way the machine functions and of the aim the engineer wished to achieve when modifying certain pieces? Any wellinformed person listening to my so-called explanations would say that I am uninformed and am saying the wrong things. That is what happens in linguistics when one does not embed the science of evolutions in the science of states. And again, on p. 129: We have just said it: evolutions cannot be thought of without the states whose idea is implied, whereas, by contrast, a language state can very well be thought of without any knowledge of the evolution it is a part of. Whether this was what Saussure thought is unclear. As was noted in Section 2.4, Saussure taught, speaking about traditional grammarians who could not possibly take the historical development of languages into account, that their method was “absolument irréprochable” and “juste” (Cours: 118). Yet, as was said in Section 2.4 above, Godel (1957: 29) quotes Albert Riedlinger’s account of a conversation held with Saussure on January 19th, 1909. where, according to Riedlinger, Saussure said: One should start from diachronic linguistics; the synchronic must be treated on its own, but without the perpetual confrontation with the To study adequately any breakdown in communication [due to some language disorder; pams] we must first understand the nature and structure of the particular mode of communication.

Charles-Albert Sechehaye

145

diachronic, one arrives at nothing. The ancient grammarians may have been free to do static linguistics, without any risk of confusing the two points of view, but look at what they achieved! Whatever one may make of this, it does not look as if Saussure held a firm view on this subject. The gibe at the ancient grammarians, at any rate, was totally unjustified. It not only testifies to Saussure’s lack of knowledge of the ancient grammarians’ monumental work, it also makes the extreme poverty of his own contribution to the theory of grammar and the practice of grammar writing stand out the more painfully. There can be no doubt that Sechehaye, and the Saussure of the Cours, were right and that the Saussure conversing with Riedlinger was wrong. The argument that one can perfectly well analyse and describe a language without any recourse to its past development, cuts wood and is, in fact, confirmed by the great achievements of the ancient grammarians, who did without any comparative historical grammar. The distinction between the synchronic and the diachronic was the same for both, but they differed, apparently, in that Sechehaye did, but Saussure did not, draw clear methodological consequences from it. After the short Chapter 11 (pp. 131–134), in which Sechehaye deals with the embedding of phonology in static ‘morphology’, a term which, for Sechehaye, is taken in “a very wide sense, including everything to do with the abstract form of language: lexicon, flections and syntax” (pmlt: 128), we come to Chapter 12 (pp. 135–160), entitled “Programme for the science of organised language in its spoken form”. Organised language is a term used by Sechehaye to distinguish system-driven language as we know it from ‘pregrammatical language’, which precedes ‘organised language’, both in a phylogenetic and in an ontogenetic sense: pregrammatical language is the not yet fully developed language of either prehistoric societies or very young children. After some musings of a general nature, Sechehaye comes to a concrete point on p. 139, where he discusses the question of whether it is the word or the sentence that holds primacy in the theory of language. Against Saussure, who defended the primacy of the word, Sechehaye maintains that (pmlt: 139): “The unit in language is the sentence; the elements of the sentence have a value only in their quality of being parts of this unit”. He had said it earlier, in his critique of Wundt in his Chapter 3 (pmlt: 36–37): [O]ne cannot possibly deny the logical priority of the sentence over the word. As a consequence, knowledge of this superior unit, in which everything is virtually contained, is the primordial condition of all rational knowledge of the grammatical mechanism. […]

146

Chapter 4

[T]he linguist has to do with linguistic organisms whose laws impose themselves on the speaking subjects, who are themselves unable to change them other than within strict limits. It is these organisms that must be analysed into their constitutive parts and explained as to the way they function. There is thus room for an entire rational and general grammar [‘toute une grammaire rationnelle et générale’], of which Wundt gives us only a few insufficiently coordinated principles. Yet the two views are not simply opposed to each other. They meet in oneword sentences, in particular those produced by very young children (pmlt: 139–140): In these, the modern theory that wants the sentence to have primacy over the word and the old theory that wants the sentence to be the product of stringing words together are reconciled with each other. […] It is perfectly true that the word, with its grammatical value, is the product of analysis. But it is no less true that every sentence constitutes a synthesis. The reconciliation of these two truths, which seem to be contrary to each other, is quite easy once one allows for the existence of isolated symbols [i.e., one-word sentences; pams], which are both idea and thought, both symbol and sentence. Just as the single word papa may indicate all sorts of thoughts in which the idea of the child’s father plays the main role, the word dodo will represent all sorts of thoughts in which ‘sleep’ is the dominant element. Bring these two sentence-symbols together in a synthesis in the order that corresponds to that of the ideas, and say, for example, papa dodo to express the thought that daddy is asleep, and you have created the first proper sentence. Now analyse this ensemble and you will find in the word papa the substance, the subject of the declaration, and in the word dodo the attribute or the predicate, the corresponding idea being conceived as a quality or as an action. […] The group consisting of these two juxtaposed words has become a construction, a syntax, according to the etymological meaning of these terms. An absolutely similar operation occurs in a more advanced stage of language evolution, when, as a result of an operation of synthesis and one of analysis, two coordinated sentences are transformed into a single sentence consisting of two propositions, one being the main and the other the subordinated one. At this point, a footnote is attached (pmlt: 140): Here is an example, taken from real life, of how a sentence results from the juxtaposition of two one-word sentences. A 21-months-old child, who

Charles-Albert Sechehaye

147

still mainly produces one-word utterances, hears the sound of a carriage passing outside. Immediately he cries out: Coco, coco, coco (that is, ‘look, a horse!’). Then he listens attentively and, as the noise diminishes, he adds pati (that is, ‘it is gone’ [French: ‘il est parti’; pams]), and after a short pause he says again, in a softer tone, as if to complete his thought, coco (i.e., ‘the horse’). This last word is either another one-word utterance: ‘I am talking about the horse’, or the subject of the sentence whose predicate has just been pronounced: (it is gone), the horse; the horse (is gone). The remainder of this chapter is devoted to discussions regarding the sounds of languages, a topic that is not our main focus of interest. The long Chapter 13 (pp. 161–214) deals with the relation between historical phonology—his “phonétique”—and historical grammar, again a subject that is of lesser interst in the present context, though, of course, highly topical in Sechehaye’s day. Chapter 14 (pp. 215–261) continues in this vein, dealing with historical phonology and historical grammar. Here we find many interesting and relevant observations regarding reanalysis, idiomaticisation, frequency, and a robust critique of the Young Grammarians’ doctrine of the unexceptionability of sound laws (see Section 6.7.1), but nothing that strikes us as visionary the way his insights about semantic form, grammar and propositional sentence structure do. The book closes with a short chapter summing up the author’s conclusions. 4.4 The Essai Sur La Structure Logique De La Phrase of 1926 4.4.1 Overall Survey of slp What Sechehaye is concerned with, in his slp even more than in his pmlt, is the thesis that the underlying semantic structure of natural language sentences consists in a hierarchical composition of propositional structures, each proposition consisting, for him, in a binary division of subject and predicate.32 In carrying out this programme of reducing sentences to compositions of propositional structures, Sechehaye enters realms of vertiginous abstraction that will have struck his linguistic contemporaries as too far removed from the mundane realities of language and too much caught up with logic and psychology. (In this respect, the world of linguists has not changed too much in 32

To those present-day semanticists and formal grammarians who have not totally distanced themselves from the notion of propositional structure, a polyadic division into a predicate and its one, two or three argument terms, as set out in Section 5.5 below, seems more appropriate, but this point is of secondary relevance in the present context.

148

Chapter 4

the meantime.) slp is essentially a continuation of the view on grammar developed in pmlt, which inevitably pushes the intrepid investigator towards the overarching parameters of the relations between the world as processed and interpreted by the cognitive mind, the mind itself with its creative powers of thinking up imagined or hypothetical states of affairs, and language as an instrument for the reciprocal transfer of the results of such mental activities to fellow humans. As we saw in note 28 above, Sechehaye was in doubt what to call his slp for a long time but finally settled on the title it has. Yet from our modern vantage point we may well feel that this title is not the best choice either and that the word logique had better be replaced with propositionnelle, as the book’s paramount theme is the propositional nature of both semantic representations and the surface structures derived from them. But then, by 1926 the notions of proposition, logic and the cognitive mind had still not crystallised out. In fact, they did not do so after 1926 either, as up to the present day the mainstream developments in both logic and the cognitive sciences would keep the notion of proposition at arm’s length. Modern logic, having convinced itself that logic is a matter of mathematics, nothing to do with cognition, lost much or most of the relevance it used to have for the study of language. The cognitive sciences (including linguistics), on their part, took up a deep aversion to the notion of proposition, which was considered either too ‘logical’ or too ‘semantic‘, in any case having no place in the dynamics of the human mind, no matter whether the mind was seen as a ‘spiritual’ entity or, as in behaviourism, as a physical system of cerebral stimulus-reaction patterns, or as anything else between or outside these two extremes. How wrong can one be and how right Sechehaye was! For him (slp: 30), the notion of proposition, composed of subject and predicate (‘la phrase à sujet et à prédicat’), is “the generative principle (‘le principe générateur’), the central organ, of the entire grammatical mechanism” (italics mine). We may add that it is also “the generative principle, the central organ”, of all cognitive functioning. Of the three roles played by the proposition—its role (a) in the cognitive specification of rational world knowledge, (b) in logic and induction, and (c) in the representation and specification of sentence meanings, as set out above in Section 4.3.2—only the third is relevant to Sechehaye’s programme, but the other two roles are crucially relevant, each in its own way, to inductive and deductive logic and to the cognitive sciences in general. The neglect, or avoidance, of the notion of proposition has, in a major way, stifled progress both in logic and in the cognitive sciences—notably in linguistics, which has so far failed to integrate its pragmatic, semantic and grammatical theorising into a unified whole, owing in great part to its failure to realise that the notion of proposition is the unifying element.

Charles-Albert Sechehaye

149

4.4.2 Comments on Successive Chapters In the Introduction of slp (pp. 1–7) Sechehaye tells the reader how much he owes to a variety of contemporary and older scholars. The first mentioned is Saussure (slp: 2–3): Over the past twenty or thirty years, the psychology of language has vindicated its rightful place besides or, we would rather say, above the historical science. More recently, the Cours de linguistique générale by Ferdinand de Saussure has reinstated [‘remis en honneur’] the badly neglected study of language states, which, by a different and perhaps safer path, likewise brings us back to the study of grammatical forms. [‘[…] ce qui, par une autre voie—peut-être plus sûre—nous ramène également à l’étude des formes de la grammaire.’] One notes how Sechehaye makes it clear, in an extremely circumspect way, that it was not through Saussure that he came to study the forms of grammar and that the “path” he himself followed—that is, via Wundt—was “perhaps safer”. One also notes that he does not say that the distinction between the evolutive (diachronic) and the static (synchronic) study of language was introduced by Saussure, correctly reminding the reader that the static study of language had existed long before (for over two millennia, in fact) the evolutive study came into being. Nor does he mention the fact that he, Sechehaye, was the first to explicitly publish that distinction, in his pmlt, a story told in Section 2.3 above. Having thus referred to Saussure, Sechehaye then mentions (slp: 3), in that order, Wundt and Schuchardt (in Germany), Meillet, Vendryes and Bally (as students or followers of Saussure), Sapir (in America), and Noreen and Jespersen (in Scandinavia) as authors that have inspired him. (One notes the striking contrast with Saussure, who hardly mentions any authors at all, as we saw in Section 2.3 above.) Reminding his readers that slp is his way of making good on a promise made in his earlier pmlt, he adds that he can only partly deliver, since he does not provide a complete theory of grammar, that being too tall an order, but can only deal with questions regarding sentence meaning (slp: 3): Thus far we have redeemed only a part of our commitment, as we address only questions regarding the ideas, that is, the logic and the psychology of the sentence, leaving aside the formal aspect of the grammar problem. We are not trying to set out the precise combinations of signs in their material conditions but rather what they express in the way of general determinations. It would be interesting to situate the problem thus delimited in the whole of the linguistic disciplines, but this would mean a return to what

150

Chapter 4

we have already said in an earlier work [that is, his pmlt; pams]. Yet, so as not to overburden this introduction, we refer the readers who are interested in this question to an Appendix [that is, the Appendix discussed in Section 4.2.3 above; pams], where they will find our actual views expressed on the organisation of linguistics. Here we will limit ourselves to insisting on two or three essential points of principle, of terminology and of method. For Sechehaye, grammar “has been created and exists for the purpose of providing a form to an element of thought and it is in living speech that these grammatical orms33 are born” (slp: 5)—an insight that was largely lost in dominant varieties of linguistic theorising during the second half of the twentieth century. To see how grammar actually shapes up in the minds of speakers one best looks at the early stages of language acquisition in very young children, whose ‘stammerings’ (‘balbutiements’) shed light on the essential principles of grammar. In fact, much of the text of slp is taken up with extensive discussions of early utterances actually produced by children during the first stages of their acquisition of the mother tongue as registered by a variety of authors. But the actual purpose of a static study of language is a theory of language based on both reason and facts (slp: 5–6): What we aim for is a reasoned enumeration and an exact definition of the notions that allow for a proper analysis and classification of all the facts of syntagmatic grammar. These notions, let us be frank about that, are not exempt from a certain amount of a priori. […] But this does not mean that the facts have no role to play in our enquiry. On the contrary, it is always because of them that problems arise, and the comparison of as large a number of different languages as possible will always be the necessary point of departure of all research and the indispensable instrument of control with regard to the results obtained. In this respect, we have to admit that we are not sufficiently equipped to be able to pretend that we can do much more than present a simple attempt. The basis on which we build our construction is somewhat narrow, since we are familiar only with some of the languages of the

33

The French text has “normes” (‘norms’), which appears to be a misprint for “formes” (‘forms’).

Charles-Albert Sechehaye

151

Indo-European family. What we have been able to see beyond those […] is too little to speak of. Within the Indo- European language family, we have focused especially on our native language—the only language one can have an entirely intimate knowledge of. This statement of the method to be followed could have been be made by a modern student starting on linguistic theory—though an average modern student would have a far more restricted command of other languages, IndoEuropean or other, than Sechehaye gives evidence of. It also goes far beyond what is found in Saussure’s Cours. In the chapters that follow, Sechehaye systematically expands on the notion that sentences reflect a propositional subject-predicate structure representing their meaning—a topic already central to his earlier pmlt but elaborated here in greater detail. Taking up his earlier discussion in Chapter 12 of his pmlt in the context of the question of the primacy of the sentence or of the word, he starts off with one-word sentences or monoremes (Chapter 1, pp. 9–17), meaning sentences (or utterances) that consist of one single nominal expression. (The question of monoremes had been broached earlier in Paul 1920: 129.) Here he looks at monoremes from the point of view both of the origin of language and of early language development in children, the latter taking pride of place. Giving many examples of actually observed early language use, Sechehaye sketches how the young child proceeds from affection-laden monoremes to representational utterances, in which the child expresses a proposition, even though there are no separate words to denote subject and predicate. He concludes (slp: 17): We say, therefore: in the normal monoreme act of communication, the given circumstances represent the paramount SUBJECT of the sentence, and the word, the conventional linguistic sign, is its PREDICATE.34

34

This may be true for some monoreme utterances, but in many others, especially those where the uttered word w has a deictic function, w is typically the subject, the missing predicate being left to the hearer’s powers of inference. For example, when a child utters “Daddy”, pointing at daddy, who has put on a funny hat, then the monoreme stands for the subject, or some other argument term, of an incomplete proposition (such as “look at daddy”), whose predicate has to be inferred by the hearer because the child’s vocabulary is still too restricted. One may justifiably surmise that in the prelinguistic stage, when the child (or the ape) only points and uses no language at all, the object pointed at is always the subject of an implicit proposition whose predicate is to be inferred for lack of vocabulary—a thought that may be relevant to theories on the origin of language.

152

Chapter 4

The much longer Chapter 2 (pp. 19–41) deals with two-word sentences, or, as he sometimes calls them, ‘diremes’. Here Sechehaye tries to explain, in terms of home-bred psychology, how two separate, juxtaposed words in a child’s utterance can begin to form a meaningful construction. Leaving that explanation for what it is, we focus on his distinction, based inductively on a discussion of actually observed utterances, between two main construction classes in early child language, head-complement and subject-predicate sentences. The former, which are also called idea-sentences (‘phrases-idées’), do not contain an overt subject and predicate term but merely a two-word noun phrase either referring to an object or entity (see note 34) or expressing a predicate. The latter, by contrast, also called thought-sentences (‘phrases-pensées’), overtly express a propositional thought, that is, the assignment of a property to an object or other sort of entity. An instance of an early learner’s two-word ‘idea-sentence’ (head-complement sentence) is, for Sechehaye, provided by the German utterance Bub bein (‘Bub leg’) in answer to the father’s question “What is this?” (the father is pointing at Bub’s leg). An instance of a ‘thought-sentence’ (subjectpredicate sentence) in early speech is (pmlt: 28, 30) the German Kuh dall (‘the cows stable’), in answer to the question “Where are the cows?”, or also (pmlt: 33) the French Coco pati (‘the horse gone’), an example discussed earlier in (pmlt: 140), as we saw in Section 4.3.2 above. Subject-predicate or ‘thought’ sentences in early learners’ speech are considered of greater importance than monoreme or ‘idea’ sentences. Prophetically, Sechehaye even speaks, as we have seen, of the subject-predicate sentence type forming the “generative principle” dominating the whole of grammar (slp: 30): n the syntactic development of the language as an instrument of thinking and of communication, the idea-sentence type only plays a secondary role, whereas the thought-sentence type, that is, the subjectpredicate sentence type, has become the generative principle [‘le principe générateur’], the central organ, of the entire grammatical mechanism. (italics mine; pams) When there are two consecutive subject-predicate structures, they are first connected by juxtaposition (slp: 39–41): ‘I know—you are at home’. But then, as the language acquisition process proceeds (or, alternatively, as the language faculty develops in the history of mankind), the second sentence is made subordinate to the first and begins to function as its object term: ‘I know that you are at home’. This is not only a stage in language acquisition by individuals but also the origin of complementation systems in languages,

Charles-Albert Sechehaye

153

though some languages may not reach that stage and do without sentential complementation.35 Sechehaye, incidentally, uses immediate-constituent analysis as introduced later in Bloomfield (1933) to clarify this point (slp: 40–41): the sentence is first divided into a subject and a predicate, then the predicate may be subdivided further into a verb and an object term, which may itself again be a subjectpredicate structure, etc. In fact, Sechehaye thus envisages a process of infinite recursion of (embedded) propositional structures. Like McCawley after him, Sechehaye was aware that it is in the interest of the unity of the theory to reduce all lexical word classes in surface sentences to predicates, their surface class membership being due to the grammar of each specific language. Verbs and common nouns are obviously predicates, and so are adjectives. But the other lexical word classes are likewise reducible to predicates at the semantic level of description. Thus, prepositions can be taken to function as transitive predicates in semantic structure (slp: 84, 150). A sentence like Jack is in Paris can be analysed semantically as a propositional structure with in as the main predicate, Jack as the subject term and Paris as the object term, yielding something like ‘Jack in-s Paris’, with ‘in’ as a quasiverb. Jack lives in the wood will then be ‘Jack’s living in-s the wood’. If the object term the wood is incorporated into the predicate—a common phenomenon in all languages: think of English work wonders, wreak havoc, take care and similar collocations—then the predicate is in the wood, as Sechehaye says (slp: 150): “In the wood […] is thus a real predicate of what precedes [i.e. the semantic subject; pams]”. This is only one step removed from treating sentential conjunctions as ‘abstract’ binary predicates. A coordinated sequence such as ‘I am leaving—I don’t like it here’ may become ‘I am leaving because I don’t like it here’, where because has been inserted as an abstract higher predicate over two propositional argument terms, meaning something like ‘be-caused/explained-by-thefact-that’: ‘my leaving [is-caused/explained-by-the-fact- that] I don’t like it 35

One is reminded of the recent sensational and slightly comical commotion around Daniel Everett’s finding (Everett 2005) that the Amazonian language Pirahã lacks complementation and thus recursion, which he presented as a counterexample to Chomsky’s claim, made over the past fifteen years, that the crucial defining characteristic of human language is recursion (which, for reasons that are unclear from an academic point of view, he prefers to call merge). Apart from the fact that any sensible linguist will see that this claim by Chomsky is baseless if not bizarre, Everett’s observation merely means that Pirahã, as a socially accepted language system, has remained stuck in a stage of stunted development, a fact that is borne out by the living conditions of the community that speaks the language, as Everett makes clear.

154

Chapter 4

here’. This step is taken in Chapter 12 of slp, which deals with subordinating conjunctions. The chapter opens thus (slp: 205):36 Conjunctions introducing a circumstantial subordinated clause really are ‘propositional prepositions’. One can, in effect, assimilate completely the class of subordinating conjunctions to that of the prepositions. Whereupon Sechehaye likens the ‘neutral’ preposition of (French de) to the neutral complementiser that (French que).37 In the chapters 4 (pp. 43–90) and 5 (pp. 91–118) of slp, Sechehaye discusses the extremely complicated question of how the well-known word classes and sentence parts of surface sentences correspond to the lexical and structural elements of the underlying propositional structures. In the latter, nominal (or substantival) expressions are needed to denote the entities referred to by the subject and other (non-sentential) argument terms of the predicate, while the predicate, no matter its surface category, is needed to denote what is said of the entities referred to. Sechehaye’s text is not optimally clear in these pages, but he seems to be saying that nominal, verbal and adjectival expressions in surface sentences are the main categories of grammar, because they reflect most directly the division into a subject and a (verbal, adjectival or nominal) predicate—or rather into a predicate and its argument terms—in propositional structures. Commenting on a long but fruitless tradition, he devotes an inordinate amount of space to showing that these expression classes do not necessarily correspond to ontological or general cognitive categories. Nor do they necessarily correspond to lexical word classes (noun, verb, adjective, adverb, preposition, etc.). He recognises that nominal expressions do not always denote concrete entities, that verbal expressions do not always denote processes and that adjectival expressions do not always denote qualities, stressing that the three classes are interchangeable. Qualities are sometimes expressed verbally, as in the English verb squint, which can also be expressed as be cross-eyed, or 36

37

Sechehaye might have referred here to Jespersen (1924: 89): Nor is there any reason for making conjunctions a separate word-class. Compare such instances as “after his arrival” and “after he had arrived”, “before his breakfast” and “before he had breakfasted”. […] The only difference is that the complement in one case is a substantive, and in the other a sentence (or a clause). Interestingly, in colloquial spoken Dutch, the neutral complementiser dat (‘that’) is often replaced with the neutral preposition van (‘of’): Ik dacht van het ging regenen (‘I thought of it was going to rain’) instead of standard Ik dacht dat het ging regenen (‘I thought that it was going to rain’).

Charles-Albert Sechehaye

155

English limp, which is not only a verb but also an adjective and a noun. Both processes and qualities are sometimes expressed nominally, as in the destruction of the city or the height of the building. Predicates expressing properties can be taken from the class of nouns, as in Mauritian Creole (the French-based Creole of the Indian Ocean island Mauritius), where one says Li ti trè sagrẽ, literally transferred to French ‘il était très chagrin’ (‘he was very sadness’), and meaning ‘he was very sad’. Likewise the nonstandard but very widespread use by (uneducated) Dutch speakers of the noun paranoia for the adjective paranoïde (‘paranoid’): Hij was helemaal paranoia (‘he was totally paranoia’), is said instead of the standard Hij was helemaal paranoïde (‘he was totally paranoid’). We can even interchange grammatical subject and predicate while expressing the same propositional content. An example would be the Andean language Aymara,38 where the normal way of saying ‘I sing’ is to say (the equivalent of) ‘the singing is mine’. Sechehaye alludes to this kind of phenomena in his earlier pmlt, on p. 117, but there he refrains from discussing such examples, saying that they “have the inconvenience of falling outside our habitual ways of speaking”. In slp he comes again close to this topic when he names a section of Chapter 5 (slp: 98): “On the arbitrary attribution of an idea to a category in the language”. On the next page we read (slp: 99): What is important for us is to note that the ideas of the language are very far from classifying themselves […] naturally into categories. On the contrary, at first sight arbitrariness seems to reign supreme here. One observes that the most diverse ideas dress up with equal ease in any word class whatsoever without anything changing except the character of the category in question. Nowadays, we are in a position to say that nominal argument positions in a proposition require a grammatical form that reflects reification, a mental process whereby a thought complex is cognitively packaged as an entity of whatever kind, very much the way computer scientists can package any amount of information as an ‘object’, leading to what is known as object-oriented programming.39 This way, a property can be assigned to a given entity to form a whole 38 39

The example was provided to me by Luis Miguel Rojas-Berscia (p.c.). Sechehaye expressed this by saying (slp: 103): “Any idea, of whatever natural category from a logical point of view, can be reduced to the category of entity”. And again (Sechehaye 1941: 88; quoted and translated from Frýba-Reber 1994: 162):

156

Chapter 4

proposition, whereby the ‘entity’ need not have the ontological status of an entity (in whatever ontological theory) but is in principle the result of a cognitive process of reification. By contrast, predicate expressions manifest cognitive packaging in terms of properties seen as sets of entities to which reified entities can be said to belong or not to belong, which is precisely what propositions do, as they say of entities that they do or do not belong to the class denoted by the predicate. Logicians distinguish between different ‘orders’ or ‘types’: what is expressed by a predicate is per se of a higher order (or type) than what is expressed by a nominal expression in a referring term position. Individual entities are zero-order elements; first- order predicates denote sets of zero-order elements, second-order predicates take terms that denote first-order elements (that is, sets of entities), etc. What Sechehaye was trying to convey is, at least in part, that lexical meaning is independent of set-theoretic order (or type) taken in this sense. One cannot, of course, blame Sechehaye for not being familiar with notions of this nature or with the corresponding modern terms. He made the best of the notions and terms that were available to him. In a way, one can even say that his lack of expertise in modern formal logic gave him the advantage of greater freedom in the exercise of his creative intellectual powers, allowing him to disregard the all too stifling conceptual frame of set theory. The phenomenon of reification, in particular, cannot be adequately described in terms of set-theoretic ‘order’, as it requires a vastly richer and more flexible system of mental reification procedures than can be provided in the terms of set theory. When carried through to its inevitable conclusion, this means that propositional structures as underlying meaning representations only have as their parts the two categories of nominal and predicative expressions, defined on grounds of their functional role in propositions, just as McCawley had it (McCawley 1972). All surface-structure word classes, such as verbs, adjectives, nouns, adverbs, prepositions, conjunctions, etc., are due to the grammatical machinery of the specific languages, which specifies a surface- structure category for each abstract predicate. They are thus in principle semantically and ontologically indifferent and individual words may change class over time.40 This leaves open the possibility of there being universal restrictions holding for the

40

We, humans, are in effect bold enough to turn every concept, no matter which, into an entity, to assign to it qualities, to give it a place in processes and relations. It is thus that we manage to manipulate all those ideas the way we manipulate those that come to us from the perceptible world and to fit what we think in their regard into the discursive unfolding of our sentences. For example, the word for the complementiser ‘that’ in Sranan (the dominating, largely English-based, Creole of Surinam) is taki, derived as a semantically bleached form from

Charles-Albert Sechehaye

157

grammars of specific languages as to how they can manipulate underlying nominal and predicative expressions—a question that is still largely undecided. Nor does it exclude the possibility of there being semantic or ontological preferences in the languages of the world for the assignment of a grammatical or lexical category to certain classes of notions. The fact that questions such as these are still largely open in present-day linguistic theory is largely due to its failure to take note of Sechehaye’s ideas. The sixth chapter of slp (pp. 119–142) is about what I have called the great subject- predicate debate. I am leaving it undiscussed here, as Chapter 5 of the present book is entirely devoted to this great question. Chapter 7 of slp (pp. 143–166) deals with sentences, phrases and clauses that do not exhibit a clear subject-predicate structure and are taken to have an ‘implicit logic’. Here, Sechehaye deals in particular with impersonal sentences (It is raining), existential sentences (There are cows in the stable), predicative adjuncts (She went home, satisfied that all was well), relative clauses, absolute adjuncts (Arms akimbo, she frowned at the man), resultative phrases (He painted the door blue), adverbials with implicit, context-bound presuppositions, such as only or even (Only/Even Harry laughed),41 and similar categories. In Chapter 8 of slp (pp. 167–172), Sechehaye attempts to fit infinitival and participial forms of verbs into the subject-predicate format. The chapters 9 and 10 (pp. 173–199) deal again, but in greater detail, with the transition from coordinated to adverbial subordinated propositions, whereby the adverbial subordinating conjunction (because, although, while)—with or without the clause commanded by it—represents a higher (‘abstract’) predicate (slp: 182): Often, perhaps even most often, and in any case in a fully objective sentence, a subordinated clause will be the psychological predicate even if it occurs after the main clause: John is not going out (subject)—because it is raining (predicate). In itself this does not pose any problem. But the following problem now presents itself: what difference is there between such a sentence and another where two coordinated propositions ­occur in sequence expressing the same psychological relation but with a ­ conjunction marking the relation: John is not going out,

41

English talk used as a serial verb: English He believes that I am ill has become Sranan A bribi taki mi siki. Words such as only or even were known in late medieval philosophy as exponibilia, or exponibles, as their semantics requires discourse-dependent propositional completion in the sense that they “need further analysis in order to lay bare their underlying logical form and to make clear under what conditions can be said to be true or false” (Ashworth 1973: 137).

158

Chapter 4

for it is raining. Unless propositional coordination and subordination are assigned basic roles, the distinction between these two kinds of grammatical link seems to become vague.42 In Chapter 11 (pp. 201–204), Sechehaye discusses the question of what role to assign to subordinate clauses in special, difficult cases, such as She was twelve, when her mother died, as distinct from When her mother died she was twelve, meaning simply ‘at the time of her mother’s death she was twelve’. In the latter, she was twelve can be regarded as the semantic subject to the semantic predicate at the time of her mother’s death, but is this also the case with She was twelve, when her mother died? Other such examples are discussed, such as No sooner had she left than the men started fighting, or nonrestrictive relative clauses, such as I found your brother, who was working in the library, wasn’t he? Chapter 12 (pp. 205–209) deals again with subordinating conjunctions. As we have seen, it starts off as follows (slp: 205): “Conjunctions introducing a circumstantial subordinated clause really are ‘propositional prepositions’ [‘prépositions de proposition’]”. Sechehaye then wonders why the class of conjunctions is divided into subordinating and coordinating conjunctions, whereas no such terminological distinction exists for prepositions. He concludes that the question is of a purely terminological nature (slp: 208): One thus sees that usage unites rather disparate and badly classified things under the term conjunction. The great fault of this traditional terminology is that it fails to take into account the difference between coordination and subordination—a difference that is too deep for it to be neglected without troubling consequences. No matter what one calls things, one needs one term to designate everything that coordinates […] and two terms for the designation of the domain of the particles, one for the nominal domain, which will be preposition, and one for the complete

42

The difference seems to be accounted for at least in part by an appeal to speech act theory. The coordinating conjunction for (French car) gives the reason for asserting the preceding clause: John is not going out it is raining, whereas the subordinating conjunctions because, since, as are primarily interpreted as giving the reason for, or cause of, what is described in the main clause: It is because it is raining that John is not going out, but they can also be used, with subtle differences in acceptability, in the sense or coordinating for. The same goes for the conditional conjunction if, which can be used both at propositional and at speech-act level: If you go, I will stay at home (propositional level) versus If you’re hungry, there’s food in the fridge (speech-act level).

Charles-Albert Sechehaye

159

or elliptical substantive propositional domain, which will be, if you like, subordinating.43 The book ends with Conclusions (pp. 211–217), followed by the Appendix (pp. 219–223) discussed above. 43

Sechehaye distinguishes correctly (slp: 215) between substantive subordinate clauses, which occupy nominal positions in the overarching proposition, and adjectival subordinate clauses, that is, relative clauses. He might have added the class of adverbial subordinate clauses, which occupy adverbial positions in the sentence and normally require a conjunction (because, when, etc.).

Chapter 5

Sechehaye and the Great Subject-Predicate Debate Having placed the proposition at the centre of his theory of grammar, Sechehaye could not possibly, as Saussure had done, stay out of the fray of what I call ‘the great subject-predicate debate’ that had been raging since the 1850s (Elffers-van Ketel 1991). It is now time to explain in detail how this debate arose and what it was about. The upshot will be that we have to do here with a central section of grammar and semantics that has, so far, hardly been explored but offers the prospect of a much more profound and thus more adequate insight into basic features of natural language. It is highly regrettable that the entire issue has been put aside in theoretical linguistics since the 1930s and has de facto been handed over to pragmatics, which is not the right place for it. Since this handover, the real nineteenth-century background of the subjectpredicate debate has been forgotten: the pragmaticists who now study ‘functional sentence perspective’, ‘topic and comment’, ‘topic and focus’, ‘theme and rheme’, ‘information structure’ or what not tend to be unaware of their nineteenth-century predecessors.1 5.1

The Subject-Predicate Debate: How it Arose and Ended up in a Quagmire

During the nineteenth century, the comparative-historical linguists were not the only linguists around. There were also theorists, quite a few of them— mostly German—reviving the rationalist tradition that had stagnated since the mid-seventeenth century (see note 15 in Chapter 2). But most of these lived in a separate world, hardly interacting with the comparative linguists in théir separate world. One thinks, for example, of the German scholars Heymann Steinthal (1823–1899), Georg von der Gabelentz (1840–1893), Philipp Wegener (1848–1916), Theodor Lipps (1851–1914), and also of the Cambridge philosopher and psychologist George Stout (1860–1944), mentioned, in different contexts, 1 A rare exception, and at the same time a confirmation of what has just been said, is Gundel & Fretheim (2004) (brought to my attention by Corentin Bourdeau). In this, otherwise very sensible and well-informed, article, out of the many nineteenth- or early twentieth-century predecessors only Gabelentz (1869) is mentioned (erroneously dated 1868), but merely as a pro forma reference, without any comment or analysis. Nothing is said about the fact that the subject-predicate debate dominated linguistic theorising from, roughly, 1850 to 1930, or about why it did. © koninklijke brill nv, leiden, ���8 | doi:10.1163/9789004378155_006

Sechehaye and the Great Subject-predicate Debate

161

in note 20 of Chapter 3 and in note 18 of Chapter 6. Linguists who combined comparative-historical with foundational work were few and far between. An example is Wilhelm Meyer-Lübke (1861–1936), a tried and trusted historical linguist who also thought and wrote about the deeper aspects of human language. Wundt, though very much concerned with the theory of language, stood apart from these. He knew about their work, at least in a global sense, but he was not part of their community. In his Völkerpsychologie of 1900, Wundt addressed both the comparativists (see Section 3.6) and the theorists but the intersection of his own work with that of either of these two groups was too small for him to be drawn into either camp. The linguistic theorists just mentioned all focused their attention on one big question, which has remained unsolved till the present day. This question arose from the discovery, around 1850, that: If the old Aristotelian concept of the proposition as the mental act of assigning a property to one or more entities is taken seriously, and if this act is taken to be linguistically expressible, in any given language, as an utterance in which the property is expressed by the predicate and the entities are referred to by the argument terms, then one often finds an incongruity between the mentally assigned property with its mentally selected entities on the one hand and the grammatical predicate with its grammatical subject and possibe other argument terms on the other. Aristotle divided the proposition corresponding to a sentence (or utterance) like ‘Socrates is wise’ into two components, the subject term ‘Socrates’ and the predicate term ‘(is) wise’. The semantics he attached to this division is what it still is today: a proposition is the act of mentally assigning the property expressed by the predicate, in this case ‘(be) wise’, to the referent of the subject term, in this case the person called ‘Socrates’. The assignment of the property denoted or expressed by the predicate to the mental representations of the referents of the argument terms (for Aristotle just the subject term) may be (i) singular or uniquely determined, as in ’Socrates is wise’, or it may be (ii) distributive or, as we now say, quantified, that is, extending over a class, in which case it may be (iia) universal, as in ‘all Athenians are wise’, where the property is assigned to (the representation of) any member of the class of all Athenians, or (iib) particular (we now say existential), as in ‘Some Athenians are wise’, where it is assigned to (the representation of) any member of an indeterminate subset of the class of Athenians.2 2 Aristotle’s predicate logic, the well-known Square of Opposition, is based on propositions of type (iia) and (iib) plus external and internal negation. Perhaps surprisingly, the simplest

162

Chapter 5

During all the centuries from Aristotle till about 1850, it was not seen that the division between subject and predicate is applicable at two distinct levels of analysis, the psychologically realistic and mostly discourse-dependent level on the one hand and the surface grammatical level, driven by grammatical markers for argument term functions and predicates on the other. Nor was it seen that it is often, perhaps standardly, the case that what is subject and what is predicate at one level is not what is subject and what is predicate at the other, despite an overall (but not absolute) identity of truth conditions. The Ancient grammarians who adopted and applied the Aristotelian subject-predicate distinction to the grammatical structure of sentences unwittingly missed out on this frequently occurring incongruity. This went undetected mainly because the division appeared to work out well, both in logic and in grammar, as long as the wider context of actually made utterances was not taken into account. That this unintentional oversight was exposed precisely during the middle of the nineteenth century, does not look like historical accident. In all probability, it was due to the fact that (a) the theoretical linguists of those days had all benefited from a good old classical education, which meant that they were familiar with the Aristotelian notion of proposition, and (b) in the wake of the great movements of both Enlightenment and Romanticism during the preceding one hundred years, they had begun to develop an eye for the daily realities of language and speech, beyond the isolated examples that had been used until then. Against the essentially rationalist background of their classical education, they simply hád to get startled when faced with the complex facts of ordinary language and its use. What, then, is that incongruity between subject-predicate structures in contextually bound propositions on the one hand and their grammatical manifestations on the other? Given the notional and technical complexity of this question, the best I can do is illustrate the point by means of an example, such as the one provided by Steinthal in the following passage from Steinthal (1860: 101): One should not be misled by the similarity of the terms. Both logic and grammar speak of subject and predicate, but only rarely do the logician and the grammarian speak of the same word as either the subject or the predicate. […] Consider the sentence Coffee grows in Africa. There can be no doubt where the grammarian will locate subject and predicate. But the logician? I do not think the logician could say anything but that form of a proposition, classified as the ‘singular’ or ‘uniquely determined’ type (i), plays no role in Aristotle’s logic.

Sechehaye and the Great Subject-predicate Debate

163

‘Africa’ contains the concept that should be connected with ‘coffee grows’. Logically one should say, therefore, ‘the growth of coffee is in Africa’. Translated into more up-to-date terminology, this means that a sentence like Coffee grows in Africa has as grammatical subject the word coffee and as grammatical predicate the phrase grows in Africa, but in a psychologically realistic sense—Steinthal says ‘logically’, for lack of a better term and because his notion of logic was fuzzy—the sentence should be read as ‘the growth of coffee takes place in Africa’, that being the normal reading, with the property of being located in Africa assigned to the (abstract or reified) entity denoted by the phrase ‘the growth of coffee’. In fact, in what may be considered a normal reading of the negation of this sentence, Coffee does not grow in Africa, what is denied is that it is in Africa that the growth of coffee takes place, suggesting that it takes place elsewhere. Steinthal clearly has a point here, since the reading he proposes will without doubt frequently occur in ordinary discourse. But there are three immediate difficulties. The first is of an empirical nature: how do we verify the semantic, or psychologically realistic, division into subject and predicate? Gabelentz was aware of this. He observes (1901: 370): But if one wants to give the inductive proof for all this, one has to be careful with examples. For the phenomena to do with positions in the sentences of different languages are not unambiguous or equivalent. The second, related, problem consists in the fact that other semantic subjectpredicate divisions are also possible. When I say COFFEE grows in Africa, with emphatic accent on COFFEE, the semantic subject is ‘what grows in Africa’ and the semantic predicate is ‘COFFEE’, as the property of being coffee is assigned to what is designated by the phrase ‘what grows in Africa’. (In general and globally speaking, main sentence accent falls on the underlying semanticopsychological predicate, which, as we are in the process of demonstrating, does not necessarily have to coincide with the grammatical predicate.) Or, to take a different example, the simple sentence Scott wrote Ivanhoe has a clear grammatical structure: Scott is the subject term and wrote Ivanhoe is the (composite) predicate. But how about the psychologically realistic semantics? We may say, of course, that this English sentence expresses the proposition by which the property of ‘writing Ivanhoe’ is assigned to the person called ‘Scott’. This is what we normally do in semantic and/or grammatical analysis. But what if we start assigning contrastive or emphatic accents? In SCOTT wrote Ivanhoe the roles are reversed: now the subject is ‘the one who wrote Ivanhoe’

164

Chapter 5

and the predicate is ‘Scott’, so that the sentence now means ‘the one who wrote Ivanhoe is Scott’, and in Scott wrote IVANHOE the semantico-psychological subject is ‘what Scott wrote’ with the predicate ‘Ivanhoe’: ‘what Scott wrote was Ivanhoe’.3 The third immediate problem was—and still is—of a terminological and technical nature. The terms grammatical subject and grammatical predicate are clear enough for the present purpose. But what about the terms s­ emantic, psychologically real, semantico-psychological or logical? At the time (as still nowadays owing to a still prevailing lack of clarity in the areas concerned), these terms lacked clear and generally accepted definitions. The result was a Babylonian confusion of tongues—a problem that plagued our nineteenthcentury predecessors as it still plagues us at the start of the twenty-first. The discussions that began taking place everywhere were thus weighed down by a stifling notional and terminological confusion, due to the fact that those involved were venturing into a terra incognita they were not equipped for. The restricted insights and technical tools at their disposal were insufficient to satisfy their ambition to unravel the realities of language and language use, just as the much further developed theoretical insights and formal equipment of present-day researchers are still unequal to the tasks at issue. Gabelentz, for example, proposed a division between a ‘grammatical’ and a ‘psychological’ subject-predicate division. The ‘grammatical’ subject takes nominative case, etc., while the ‘psychological’ subject is the representation that presents itself to the mind first (Gabelentz 1869: 378): What does one wish to achieve when one speaks to another person? The answer is that one wants to arouse a thought in him. In my view, this implies two aspects: first, one has to direct the interlocutor’s attention (his thinking) to something, and secondly, one makes him think this or that about it. I call that of or about which I want my addressee to think the psychological subject, and that which he should think about it the psychological predicate. In the sequel it will become clear how much these categories often deviate from their grammatical counterparts. Those familiar with (traditional) logic, like Wundt or Jespersen (1924: 147), further spoke of a ‘logical’ subject, representing the thing ‘spoken about’ in the 3 Interestingly, French grammar forces speakers of French to stay closer to the semanticopsychological division, in that it requires the forms C’est Scott qui a écrit Ivanhoe (‘it is Scott who wrote Ivanhoe’) versus C’est Ivanhoe que Scott a écrit (‘it is Ivanhoe that Scott wrote’). Simple contrastive accent, as in English, is, in general, not allowed in French.

Sechehaye and the Great Subject-predicate Debate

165

Aristotelian sense. Confusion all over. In the end, Theodor Kalepky exclaimed (Kalepky 1928: 20): ‘Such a confusion simply cries out for relief’ (‘Eine derartige Wirrnis schreit förmlich nach Abhilfe’). In fact, Kalepky, like a few others such as Carl Svedelius (1897) and, as a last comer, Manfred Sandmann (1954: 78–81; 105–109), called—without success—for a grammatical theory without subject and predicate at all (see Section 6.7.3 below). The only ray of light came from a few authors who sought a connection with discourse. Theodor Lipps, Philipp Wegener and, in their wake, the visionary British philosopher-psychologist George Stout, all mentioned above, and, in addition, that other Brit Alan H. Gardiner (1879–1963), as well as the great Danish linguist Otto Jespersen (1860–1943), all realised that the semanticopsychological subject-predicate assignments that differ from surface subjectpredicate divisions correspond to the dynamics of ongoing discourse: coherent discourse gives rise to (explicit or implicit) questions, which are then answered by a following sentence. Jespersen wrote (1924: 145): The subject is sometimes said to be the relatively familiar element, to which the predicate is added as something new. […] This may be true of most sentences, but not of all, for if in answer to the question ‘Who said that?’ we say ‘Peter said it’, Peter is the new element, and yet it is undoubtedly the subject. What Jesperen left unmentioned but clearly intended to imply is that this ‘new element’ systematically carries a prominent accent, whereas the ‘relatively familiar element’ is systematically presented without much accent and under a flat intonational contour. Subsequently, the problem became a central issue in the Prague School of linguistics, founded in 1926 by Vilém Mathesius and Roman Jakobson. In the Prague School, the accentual and intonational patterns of sentences (utterances) were given a prominent place, which led to the notion of what I prefer to call topic-comment modulation (tcm) but is referred to in the literature under a variety of names, as has been said. From there it was picked up by scholars all over the world studying what is now widely referred to as information structure, considered to be more of a pragmatic than of a semantic nature. Yet despite the important new insights gained in the study of information structure, this more pragmatics-oriented approach has so far failed to provide a clear idea of how surface sentence structure reflects propositional structure as moulded in running discourse. In actual fact, it does not seem totally unjustified to state that the failure of the modern information structure approach to cast tcm in the terms of propositional structure is ultimately responsible for

166

Chapter 5

its lack of theoretical clout, in respect of both grammar and semantics. This issue, too, is thus still wide open. Yet it is, at the same time, also closed, since the fact that, during the entire period from the 1850s till the 1930s, the subject-predicate debate was the one most central issue of theoretical linguistics, seems to have been erased completely from the collective memory of present-day linguists, including the ‘information structure’ community. In fact, the entire subject-predicate debate petered out without having been brought to anything like a satisfactory conclusion, the reason being, no doubt, that an adequate solution requires insights and analytical tools that were not available at the time. If those who participated in the debate had continued to think in terms of propositional structure, their chances of reaching a solution would have been significantly better. The whole, now largely forgotten, debate was by definition rationalist, not romanticist, as it was principally concerned with the old Aristotelian division of propositions into a subject and a predicate. It was, moreover, structuralist— at least in spirit—in that the subject-predicate division is a structural division assigning structure to propositions. The fact that this complex of questions is now being studied by rather nonrationalist and nonstructuralist students of what is nonetheless called ‘information structure’ testifies to the still prevailing unclarity and confusion surrounding the general questions of human language. The position taken here and argued for below is that Lipps, Wegener, Stout, Jespersen, Gardiner and also Sechehaye, as we will see in the following section, were on the right track, even though they were, in the end, unable to carry their arguments through and thus reach a satisfactory solution. The track taken by these scholars and followed here implies that one must, for the bulk of natural language sentences, distinguish between two distinct proposition types that are largely but not entirely (see the Sections 5.5.3 and 5.5.4 below) equivalent as regards their factual truth-conditions. We distinguish between, on the one hand, a discourse-driven proposition type (called ‘modulated proposition’ or ‘modprop’ in Seuren 2010: 390; see also Seuren 1999a), reflecting the genesis of the proposition in the dynamics of discourse and displaying the structural features of what are known as cleft constructions, and, on the other, a fact-driven proposition type (called ‘flat proposition’ or ‘fprop’ in Seuren 2010: 390), which may, on the whole, be taken to be a more direct mapping of the way humans see and analyse situations in the actual or any virtual world.4 4 Aristotle himself was obviously unaware of the difference between discourse-driven propositions on the one hand and more factually oriented predicate-argument surface structures on

Sechehaye and the Great Subject-predicate Debate

5.2

167

How Did Sechehaye Deal with the Subject-Predicate Debate?

As was observed in Section 2.3 above, Saussure seems to have remained totally unaffected by, or uninterested in, the great subject-predicate debate, as not a trace of it is found in either the text of the Cours or Saussure’s private notes—no doubt because he was thoroughly averse to anything to do with the notion of proposition. Sechehaye, by contrast, saw the importance of this issue and discussed it extensively, especially in his slp of 1926, clearly veering in the direction taken by Lipps, Wegener and Stout during the fin-de-siècle. Although Sechehaye may well have known about Lipps’, Wegener’s and Jespersen’s ideas in this respect, it is less likely—though, of course, still possible—that he knew about Stout’s work, the Anglophone world still being relatively marginal in the Continental linguists’ scope of interest. Yet his discussion in slp clearly echoes the ideas developed by George Stout, whose terminology, however, is perhaps more opaque than that of Sechehaye, who wrote (slp: 17): We say, therefore: in the normal monoreme act of communication, the given circumstances represent the paramount SUBJECT of the sentence, and the word, the conventional linguistic sign, is its PREDICATE. Stout, by contrast, does not speak about a child’s earliest utterances but, more generally (and more daringly), about the way in which each new utterance in a discourse is an addition to an already existing mental representation of a situation described in preceding discourse (Stout 1918, vol. 2: 212–214): This relatively indefinite schema, which becomes articulate in the process of thinking, is what logicians call a universe of discourse, and what in ordinary language is called a subject or topic. […] The essential mark of the subject-predicate relation is that it constitutes a sentence. […] To explain the nature of the subject-predicate relation is at the same time to explain why discourse is broken up into distinct sentences. […] The predicate of a sentence is the determination of what was previously indeterminate. The subject is the previous qualification of the general topic or universe of discourse to which the new qualification is attached. The subject is that product of previous thinking which forms the immediate basis and starting-point of further development. The further development is the predicate. Sentences are in the the other. Taken literally, the Aristotelian definition of the proposition as the mental assignment of a property to one or more entities is open to both interpretations (see Seuren 1999b).

168

Chapter 5

process of thinking what steps are in the pr0cess of walking. The foot on which the weight of the body rests corresponds to the subject. The foot which is moved forward in order to occupy new ground corresponds to the predicate. […] All answers to questions are, as such, predicates, and all predicates may be regarded as answers to possible questions. If the statement, ‘I am hungry’ be a reply to a question, ‘Who is hungry?’ then ‘I’ is the predicate. If it be the answer to the question, ‘Is there anything amiss with you?’, then ‘hungry’ is the predicate. If the question is, ‘Are you really hungry?’ then ‘am’ is the predicate. Every fresh step in a train of thought may be regarded as an aswer to a question. The subject is, so to speak, the formulation of the question; the predicate is the answer. […] The ultimate subject is always the universe of discourse. In some cases this ultimate subject is also the proximate subject of a sentence. When this happens, it may not be represented by an expressive sign and it may even be incapable of being adequately expressed. Under this head come impersonal sentences such as ‘It rains’, ‘It grows late’, etc., and exclamations such as ‘Fire!’ ‘Murder!’ This means that in monoreme utterances the subject is formed by Stout’s implicit ‘universe of discourse’, while the uttered word is the predicate, just as Sechehaye has it. Modern linguistic theory, suffering as it does from a fatal narrowing of its range of vision, has failed to take up this strand of thought and develop it further. Had it done so, a serious theory of discourse incrementation, presupposition, topic-comment modulation and natural logic, all based on the notion of universe restriction (Seuren 2014a, 2014b), would have become an integral part of linguistic theory a long time ago. Elaborating on the subject-predicate distinction, Sechehaye begins by distinguishing between, on the one hand, utterances produced ‘in real life’ [‘la vie’] and, on the other, those expressing a simple fact, regardless of context (slp: 30–31): In the sentence Kuh dall, the word Kuh was subject because it took up the idea of the question [the question being: ‘Where are the cows?’; pams]. But, taken by itself, it was a simple objective idea like that of the word dall, which served as predicate. In fact, all that counted in all this was the specification of a fact: ‘the cows are in the stable’. But now suppose that we have to do with the same thought but placed in a different context, or without any context, so that the idea of Kuh, ‘the cows’, is no longer connected with a question and thus no longer has the same psychological character. Now it has become arbitrary whether we assign to it a different

Sechehaye and the Great Subject-predicate Debate

169

intonation and importance from the predicate dall, ‘are in the stable’. In fact, if we look at these two notions, ‘cows’ and ‘stable’, in all their logical nudity, the same sentence may mean not only ‘the cows (subject) are in the stable’ but also and equally well ‘in the stable (subject) there are cows (predicate)’. These two terms may, if you like, be each other’s subject and predicate. That is what logic teaches us. But we must come down from pure abstraction to real life, and there we must recognise that as long as there is communication, that is, activity, there is necessarily also a movement of thought, which moves from the term taken as point of departure, as given or presupposed, to the term of successful conclusion [‘aboutissement’], that is, subject and predicate. […] Complete parity of the two terms is a marginal case of an entirely theoretical nature. The distinction between subject and predicate is thus a postulate of life, and in ordinary living language [‘langage’], permeated as it ordinarily is by affectivity, it is easily observed. [italics original; pams] Sechehaye thus heads straight for the great subject-predicate debate. In this debate, he takes—perhaps unknowingly, and thus independently—the position formulated by scholars like Lipps, Wegener, Stout, Gardiner and Jespersen, all of whom distinguished a discourse-determined subject-predicate structure, reflecting a question-answer game, from a grammatical subjectpredicate structure in surface sentences, the former being recoverable from the latter either by grammatical markings or by intonational or word-order phenomena. Let us assume that, in principle at least, an intelligible coherent discourse is driven by a, mostly implicit, question-answer game, started off by a so-called ‘feeder’ (Van Kuppevelt 1991). Let the feeder be a ‘hot-news’ sentence: The president has died! This feeder gives rise to a number of questions, for example, When did the president die?, or What did the president die of?, etc. Anticipating such questions, the speaker will proceed by answering them, saying, for example, He died last night at 10.35 pm. Now the part he died is the topic, expressing what the (possibly implicit) question is about, the given element known from the feeder, expressible nominally as his death. The part last night at 10.35 pm answers the question When did the president die?, or rather its underlying form ‘the president’s death occurred AT WHAT TIME?’, which makes ‘last night at 10.35 pm’ the comment, answering the question by specifying the time of the president’s death. In the given context, the sentence He died last night at 10.35 pm is now paraphrasable as the propositional structure ‘the president’s death was LAST NIGHT AT 10.35 PM’, or ‘it was LAST NIGHT AT 10.35 PM that the

170

Chapter 5

president died’. The discourse-determined topic has now become the subject of the discourse-determined subject-predicate structure and the newly added comment has become the predicate that specifies the topic/subject. In this type of structure, the subject term typically designates a parameter, in this case the parameter of the time of the president’s death, while the predicate assigns a value to the parameter. One may say that the copula be here assigns a unique value to a parameter. This analysis implies that the topic-comment structure of a sentence is NOT a pragmatic addition to the surface sentence already formed by the grammar, a final ‘packaging’ of the information supplied before it is dispatched to the listener, as is standardly assumed nowadays in linguistics, pragmatics and in formal semantics, but, on the contrary, part and parcel of the meaning of the sentence in question (see Section 5.3 below for relevant evidence), and thus necessarily also a matter of grammar. Sechehaye, however, did not reach full clarity in this respect. Let us see how he proceeds. On the first page of his Chapter 6, Sechehaye, much in the way Jespersen did, explains the topic-comment structure as follows (slp: 119): When one poses the question Who is sick?, one assigns to the pronoun the role of grammatical subject. Yet in actual fact this sentence amounts to The sick one is who? where the idea that the question relates to can only be the predicate. This is borne out by the answer It’s John. And if I say John is sick, then, although ‘John’ is, in my utterance, the grammatical subject, in reality it is the psychological predicate. Further down in the same chapter we read (slp: 132–133): It is obvious that, when speech [‘la parole’] neglects the properly psychological values [‘les valeurs proprement psychologiques’] of the constructions, it still makes use of the logical values [‘valeurs logiques’]. The grammatical groups Subject : Predicate and Principal : complement remain the indispensable vehicle for mutual comprehension. They no longer represent the movement of the thought but rather the relations in which the various [nominal; pams] ideas stand to each other in the totality of the sentence. The grammatical forms thus always keep at least a part of their function: their logical role. This logical role is derived from their original and basic psychological role. It is in virtue of the linguistic knowledge of the relations that normally unite the grammatical terms that language provides a solid substrate for the constructive elements of its complex sentences.

Sechehaye and the Great Subject-predicate Debate

171

And again (slp: 136–138): We have seen the effects of the grammatical automatism in moderately affective discourse, and we have seen how a higher degree of affectivity puts this automatism aside to revert to the expressive grammar. There are things to say about the cases where […] the discourse turns […] to a more intellectual, more objective form than what the speaking subject’s emotions would perhaps prefer. […] If I give an order in a natural way, I will say It is John who will do this job [‘C’est Jean qui fera ce travail’]. But if I wish to assume greater dignitiy, I will say, in a slightly more formal way, John will do this job [‘Jean fera ce travail’], making John the subject of my sentence, but in reality it is the predicate. […] All the facts that we have mentioned so far, and probably others as well that are different in nature but similar in their effects, take place on the stage of speech [‘ont pour théâtre la parole’]. It is because of them that grammatical forms sometimes break through the psychological frames they were made for. But what belongs to speech may become usual and thus pass into the language. It is thus entirely natural that these discrepancies between form and background [probably meaning: ‘comment and topic’; pams] are reflected in certain parts of the grammar. Sechehaye is thus seen to be in doubt about whether to assign such phenomena to the dynamics of speech or to the grammar system, wavering between two opposite views. On the one hand, he proposes that, in actual speech, certain elements are ‘added’ to the grammatically defined form of the sentence. On the other, however, he recognises that an analysis in terms of speech processes alone is not enough because these ‘additions’ are themselves subject to the regime of grammar and vary from language to language, which inevitably makes the incorporation of topic-comment structures into surface structures a necessary part of the ‘grammar problem’.5 To give a simple example, the French equivalent of English John sold the CAR, NOT the MOTORBIKE is: C’est la VOITURE que Jean a vendue, PAS la MOTOCYCLETTE (‘it is the CAR that Jean sold, NOT the MOTORBIKE’). That is, where English can do with mere emphatic/contrastive accent, French 5 As was observed in Section 4.2.4 above, Bally (1944, first published in 1932) liberally borrows from Sechehaye in this respect, without, however, properly citing Sechehaye as his source, or at the very least as relevant earlier literature—a fact not noted in any review of Bally’s book, as far as I am aware.

172

Chapter 5

needs, in addition, a grammatically defined form of clefting. These English and French facts, far from being arbitrary whims of speech, are part of the repective languages, to be accounted for in the grammars of the English and French languages, respectively. If we want to be harsh on Sechehaye, we can say that he is open to the same criticism he himself voiced with regard to Wundt, as he declines to cross the barrier posed by the ‘grammar problem’, which is as much the problem of transforming topic-comment structures into surface sentences as it is the problem of transforming ‘semantico-psychological’ subjectpredicate structures into surface sentences. Sechehaye’s inability to decide between these two incompatible points of view appears, for example, from the following passage (slp: 126–127): We thus clearly see, in the simple format of subject and predicate, how speech [‘la parole’] makes up for the insufficiencies of the grammar by adding things that are independent of it. That […] nuances of intonation can, by habit of use, themselves become conventionalised and thus become part of the grammar, at least to some extent, is a point that we will not further investigate here. Suffice it to say that it is a system of expression independent of that which uses word classes and their syntagmatic properties. A footnote is attached at the end of the penultimate sentence (slp: 126): Since each language incorporates into its arbitrary grammar system certain elements taken from the flow of speech, in the form of quantity and pitch or loudness accent, it follows that the free flow of speech is subject to conditions that vary from language to language. […] It is relevant to observe here that the analysis whereby tcm is taken to be a surface overlay on top of a sentence already formed, a matter of last-moment ‘information packaging’, is the perspective universally taken nowadays in pragmatics, formal grammar and formal semantics. There seems to be little awareness, in these circles, that grammar is basically involved in all this, something which Sechehaye did see and which at the same time prevented him from reaching a unified solution to the subject-predicate problem. An alternative, and more adequate, analysis implies that the discoursedriven topic-comment structure is the first subject-predicate structure arising in any given discourse and that this initial structure is, as a first step, transformed into the corresponding discourse-neutral and fact-oriented structure suitable for logical operations and for memory storage. The original

Sechehaye and the Great Subject-predicate Debate

173

topic-comment structure must, however, not get lost or else the listener will forfeit the advantage of the question-answer game that gives coherence to a discourse. For that reason, the original topic-comment structure remains traceable in the form of special markings such as intonation or word order. In this perspective, the discourse-driven topic-comment structure arises in the speaker as a first product of what Slobin called ‘thinking-for-speaking’ (Slobin 1987), or of Levelt’s ‘microplanning’ (Levelt 1989: 107–110, 145–157), and thus lies at the very origin of sentences uttered in a running discoure, reflecting the way they have come about in the question-answer game that characterises the flow of speech. More specifically, one may propose that, at least in most languages, the ­topic-comment structure reflecting the thought structure as it arises in the context of current discourse, with the topic as subject and the comment as predicate, is first reduced to a standard subject-predicate (or predicate-argument) format suitable for logical operations and for memory storage, after which this reduced structure is processed by the grammatical machinery that transforms it into a surface structure, but always preserving the intonational or word/ constituent-order features required for marking the original discourse-driven topic-comment distinction (the precise nature of these markings has not so far been brought to light). For languages like French or to some extent also English, which require, or allow for, cleft constructions to reflect the underlying topic-comment structure, one may assume that the process of reduction from discourse-driven to the corresponding discourse-neutral propositional structure must, or may, be skipped, resulting in so-called ‘cleft’ constructions, so that this reduction to ‘neutral’ or fact-driven propositional form is not mediated linguistically (other than by nuclear sentence accent) but left to the standard cognitive interpretative or parsing powers of the hearer. It is clear, in any case, that we are faced here with a massive problem that has existed in theoretical linguistics since the 1850s and is still unsolved, despite modern attempts to neutralise it by relegating it to the ‘pragmatics of discourse’, where it is tacitly assumed that formally precise solutions are not needed—that is, where a reversal to romanticist varieties of analysis is considered acceptable. It is equally clear that Sechehaye, in dealing with this problem, did not wish to go that way. What present-day linguists can learn from this episode is that the standard modern notion that tcm phenomena are a late overlay over already formed sentences providing additional information as to how the uttered sentence is to be fitted into running discourse is inadequate. It fails to account not only for the grammatical facts involved but also for semantic facts such as those mentioned in Section 5.5.3 below, where it is shown that a sentence like Carl

174

Chapter 5

was surprised that JOHN had sold the car is truth-conditionally independent from a sentence like Carl was surprised that John had sold THE CAR. If one holds that the assignment of such emphatic or contrastive accents is merely a matter of late information packaging, one must also accept that such late accent assignments may change the truth conditions of the uttered sentence. This conclusion, however, is not drawn by those present-day linguists or pragmaticists who study such questions, nor would they, one suspects, be willing to accept it when faced with it. 5.3

Why Discourse-Driven and Fact-Driven Propositions?

What all this means, in effect, is that, in actual real-life linguistic interaction, there normally are, for each individual sentence, not one but two p ­ ropositional subject-predicate structures, distinct though systematically related, the one more relevant to factors that are operative in running discourse, the other more suitable for logical and other cognitive operations, including memory storage. As a rule, both are reflected in one single corresponding surface form, in which some elements reflect the subject-predicate structure as it is when shaped according to the requirements of the ongoing discourse and others the subject-predicate structure as it may be taken to reflect the factual state or situation of entities having properties. A natural question arising in this context is: what is the functionality of the conversion from underlying cleft structures to ‘canonical’ surface structures under a tcm pattern? Why should this take place at all? The answer to this question is not immediately obvious. There may, in fact, be a variety of answers. One answer comes to mind immediately.: if it is true that discourse is often, perhaps even in principle, organised in terms of a question-answer pattern as discussed above, so that each new utterance merely fills the WH-position of the question (see (5.1)–(5.2) below), then all the listener has to do, when interpreting a new utterance, is interpret the new predicate, since the topic has already been interpreted (which is why it is not in need of a prominent accent). This viewpoint is enlightening to the extent that, on the whole, psycholinguistic experiments on the interpretation of utterances have so far been based on the assumption that listeners have to interpret the whole utterance as a block, whereas in actual practice the listener, most of the time, merely has to interpret the predicate, that is, the new element that has just been added. This makes for a very considerable simplification of the psycholinguistic problem of the mechanics of utterance comprehension or interpretation. But there may well be a further rationale behind the reduction of discoursedriven to fact-driven propositional forms in the process of speech. This rationale

Sechehaye and the Great Subject-predicate Debate

175

has been adumbrated more than once above, as it was adumbrated also in Sechehaye’s writings about this question. It amounts to the following hypothesis. If it is true that the ‘cleft’ structure reflects discourse-dynamic propositions as they arise in the question-answer game typical for running discourse, it does not follow that this is also the form in which the propositions expressed are used in logical arguments or stored in memory, including the short-term memory known as the Discourse Domain (see, for example, Seuren 1985, 2010). It is far more likely that, for obvious functional reasons, memory storage and logical processing disregard the conditions under which propositions come about in the stream of thought or speech and operate in terms of structures that reflect more directly the world situation described in the sentence or utterance, rather than what gave rise, in the stream of thought or speech, to the expression of the propositions concerned, which helps the listener to reconstruct the thought processes gone through by the speaker. Sentences seem to be subject to two distinct functionality criteria. On the one hand, when realised as utterances, they make the listener follow the cognitive path gone through by the speaker so that the former is in a better position to reconstruct the thought processes developed by the latter. Let us call this the processing function of sentences. On the other hand, uttered sentences describe, or add to the description of, a (real or imagined) world situation as conceived by the speaker—the representational function of sentences. Without the former, the latter cannot be achieved: the listener first has to cotton on to what the speaker is doing before the result can be fitted into the situation model under development between speaker and hearer. (A human engineer trying to devise a language would probably not think of that, but nature has.) Therefore, each specific language system has to ensure that both functions can be satisfied, and languages do so by striking a balance between them. English, for example, has a preference for expressing the processing function by means of intonational and accentual means, giving grammatical prominence to the representational function. French, by contrast, puts greater store by preserving the structural features induced by the processing function throughout the syntactic mapping system, thus enriching the syntax with certain dynamic aspects reflecting the processing history of the text produced. French syntax, in other words, can be said to be more ‘rhetorical’ than the syntax of English or other similar languages. 5.4

Intermezzo on the General Structure of Propositions

Quite independently of the semantico-grammatical distinction between underlying discourse-driven propositions on the one hand and the more

176

Chapter 5

fact-oriented predicate- argument surface structures on the other, we must be aware of a development in logic, not in linguistics, whereby the traditional dyadic subject-predicate structure assigned to propositions was replaced with the polyadic predicate-argument structure modern logicians work with. Traditionally, in the wake of Aristotle’s pioneering work, logicians operated with unary predicates taking just a subject term, thus yielding a dyadic structure, as shown in Figure 6-a below. This remained so until the advent of modern predicate logic around 1900, when an analysis in terms of n-ary predicates, allowed to take more than one argument term and thus yielding polyadic structures, became preferred for propositions, as shown in Figure 6-b. The main reason for this innovation was that structures with n-ary predicates allow for a simplification and/or greater variation of logical operations—a fact that has been multiply confirmed since. The structure shown in Figure 6-a assigns a dyadic main structure to the proposition underlying the sentence ‘the boy give(s) the girl the book’, consisting of the subject the boy and the complex predicate (predicate-1) give(s) the girl the book, which in turn contains the terminal or bare predicate give (predicate-2) followed by the indirect object the girl and the direct object the book. The polyadic structure of Figure 6-b assigns an n-ary structure to the proposition, consisting, in this case, of the bare predicate give, followed by the subject term the boy, the indirect object term the girl and the direct object term the book, representable as ‘Give(the boy, the girl, the book)’. In the polyadic structure, the bare predicate give is a ternary predicate, that is, a predicate with three terms. The order of the constituents in the two kinds of structure is merely notational: in Figure 6-a, one could also, for example, place the predicate-1 first and the subject term second, or in Figure 6-b one could place the predicate at the end, and the terms themselves may be placed in different orders also, as long as they are marked for their semantic role vis-à-vis the predicate. Such variations are purely notational and immaterial from a strictly logical point of view, as long as each variation is properly defined. a.

Proposition

Subject the boy Predicate-2 give Figure 6

Predicate-1

b.

Proposition

Predicate Subject give the boy Indirect object direct object the girl the book

Indir. object the girl

dir. object the book

Dyadic and polyadic constituent structure of the proposition underlying ‘the boy give(s) the girl the book’

Sechehaye and the Great Subject-predicate Debate

177

A well-known and mathematically well-founded alternative for constituent structures as shown in Figure 6 are the dependency stemmas, proposed by the mathematically minded French linguist Lucien Tesnière (1893–1954) in Tesnière (1959). These stemmas are based on the mathematical notion that a predicate, or, generally, a logico-semantic operator (including the definite article), is a (unary or n-ary) function delivering a value, just as in the theory of Categorial Grammar developed by the Polish logician Kazimierz Ajdukiewicz (1890–1963), except that Categorial Grammar does not use Tesnière-type stemmas but constituent tree structures. Dependency stemmas differ from constituency trees in that the function (predicate or operator) dominates its terms, rather than being a sister node to them, as shown in Figure 7. In formal logic, the dependency-stemma format is not favoured, as it does not lend itself at all well to formal logical calculus. In this respect, the wellknown (polyadic) bracketing structures of modern predicate logic—which are equivalent to constituent tree structures—are much more convenient. The dependency-stemma format is popular in computational linguistics, due to the clear display of functions, their input and their output, but much less so in linguistics proper, the reason being that the structural properties of Tesnièretype stemmas have, so far, not proved useful in formulating generalisations that would help explain linguistic phenomena. Polyadic constituent tree structures, by contrast, have proved of crucial importance in this respect, which is why linguits do not, on the whole, work with Tesnière-type stemma structures. Within the constituent-tree format, some modern linguists prefer the subject-­ predicate division shown in Figure 6-a. Others, including myself, prefer the polyadic structure exemplified in Figure 6-b. The preference for the polyadic structures is based exclusively on their methodological and empirical advantages over the traditional dyadic structures. Polyadic constituent trees give = the boy give(s) the girl the book

Figure 7

the = the boy

the = the girl

N boy

N girl

the = the book

N book

Tesnière-type stemma structure of the proposition underlying ‘the boy give(s) the girl the book’

178

Chapter 5

transform much more easily and more regularly into surface structures than subject-predicate structures (or indeed dependency stemmas). Such questions are, however, not dealt with here any further. If, occasionally, I have spoken of ‘subject-predicate structure’, I have simply paid tribute to the tradition the author or piece of writing discussed is a part of. In general, I prefer to speak of ‘predicate-argument structure’. 5.5

An Analytical Synthesis of the Whole Question

The reader who has had the stamina to follow the analysis to this point may well be flustered by the magnitude, the complexity and the ramifications of the issue at hand. Yet it is important, in the context of the present study, to see that the path taken by Sechehaye, together with others, as regards the complex problem area of the subject-predicate division in natural language sentences is indeed one that leads to substantive new insights and an essential extension of current linguistic theory. Let me, therefore, try to present the issue de novo, as it were, and not in historical but in purely systematic terms.6 5.5.1 Definition of the Notion ‘proposition’ Like the term ‘logic’, the term ‘proposition’ has received a bewildering variety of definitions, meanings and applications through the centuries (see Nuchelmans 1973, 1980, 1983). In an effort to stay as theory-independent as possible, while avoiding undue complications, I have defined the proposition above (Section 3.4.1), as the mental act of truly or falsely—attributing a property to one or more entities. Correspondingly, a proposition, when cast into a format suitable for linguistic expression, is taken to consist of a predicate which denotes the property in question and one or more argument terms referring to the entities in question. Traditionally, as was explained in the previous section, only one argument term, the subject term, was taken into account, but nowadays we work with multiple argument terms, such as subject, object, indirect object, as has just been explained. Two caveats were issued in Section 3.4.1 in connection with this definition, one pertaining to the fact that the precise truth conditions for propositions are often hard to specify verbally, due to the extreme flexibility and evasiveness of predicate meanings, and the other to the fact that the term entities covers not what metaphysical philosophers may take to be the ultimate ‘furniture of the world’ (to use Bertrand Russell’s phrase), but whatever it is that corresponds 6 For an earlier discussion of the same complex of questions, see Seuren (2010: 378–408).

Sechehaye and the Great Subject-predicate Debate

179

in the outside world to the reifications constructed by the human mind, including reifications of what we see as single objects. Here a third caveat must be added, to do with the fact that propositions as they occur in the stream of speech (or thought) often have a propositional structure that is not, as such, reflected in the grammatical surface structure of the corresponding sentences, though the two structures are, in an overall but not absolute sense, truthconditionally equivalent. This, obviously needs further comment. As defined above, a proposition is a token event, that is, a single unique event taking place at a given time, at a given place, in a given context and in relation to a given (real or imagined) situation. It can be repeated in the form of a token copy just in case the same programmatic mental or cognitive procedure is gone through, regardless of the fact that the neurons involved have physically metabolized in the meantime and the sounds produced are not fully identical either (just as Heraclitus’ river never contains exactly the same water molecules: panta rhei). In this sense, we can utter the same proposition twice, as, for example, when I say something and my interlocutor makes me repeat what I said because, say, his or her attention was diverted as I spoke. We take it that a propositional token event goes through a number of preparatory stages until it is moulded into a structure that is fit for being fed into the type-level grammatical system of a specific language or language variety (Slobin 1987; Levelt 1989: 107–110). That structure, which has meanwhile been abstracted into a type-level system, we call the semantic analysis (SA) of the sentenceto-be and the resulting sentence will again be a type-level unit, which can be realised as a token-level actual utterance. The intervening grammar is thus a kind of algorithmic processing machine receiving type-level semantic inputs and delivering type-level outputs in the form of sentences. 5.5.2 Anchoring and Keying We now know (Szabolcsi 1981; Seuren 2009: 98–101) that for a type-level sentence to express a proposition and make sense in an actual token-level speech situation, at least two conditions must be fulfilled: the sentence must be both anchored in the running discourse and keyed to a given situation in the real or any imagined world. By this the following is meant.7 (i) The anchoring condition is one of coherence, or of comprehensibility in a larger context of discourse. Most sentences in natural languages are fit for 7 My use of the term anchoring conforms to the use found in some of the modern literature on value-assigning sentences, such as Comorovski (2008). Szabolcsi (1981: 513) uses the terms furthering function and anchoring function for what are called here (as in Seuren 2009: 98– 101) anchoring condition and keying condition, respectively.

180

Chapter 5

use only in a restricted class of given discourses. If I start a story by uttering the sentence It was the morning of my fiftieth birthday, and I continue by saying Genesis is the first book of the Old Testament, then my interlocutor will be puzzled, as the two sentences do not form a coherent text. Or if I say first I never had any students and then Most of my students passed their exams, my interlocutor will again be puzzled, since the second sentence presupposes that I had students, which the first sentence denies. Or if I say I saw a boy running out of the house; his radiator was sneezing, puzzlement will again arise, because boys don’t have radiators and radiators do not sneeze. Anaphoric expressions must have an antecedent: a text like We were having a good time; suddenly it started to jump up and down, is, taken by itself, incoherent because the listener has no clue what the it refers to. In other words, as a speaker, I must not only make clear to my audience what sort of situation I am building up in my discourse but I must also keep my discourse logically and textually consistent and, unless otherwise specified, in agreement with available world and situational knowledge. Sentences can only be used coherently in discourse contexts that satisfy these anchoring conditions. Presupposition and anaphora are central elements in proper anchoring processes and must, therefore, play a central role in the theory of discourse anchoring. (ii) The keying condition is a condition for truth and falsity. If the keying condition is not fulfilled, we have no proposition, only a propositional schema. Most sentences contain one or more referring and/or indexical expressions. If I say The door was locked, I have produced a correct English sentence, but it is impossible to say whether what I have said is true or false without a prior specification of what specific door and what specific past moment or time I am talking about. As long as those specifications are missing, I have not expressed a proposition: all I have produced is a type-level sentence, open to semantic and grammatical analysis but without a truth value.8 For a sentence to express a proposition and thus acquire a truth value, all referential and indexical expressions must have received a token level reference value, regardless of the grammar—that is, the sentence must be keyed to an actual or imagined situation. In other words, the grammar only produces unkeyed type-level sentences, which, as such, have a truth value only if they contain no referential or indexical ­expressions, which happens very rarely (in fact only in sentences 8 In Seuren (2009) and other publications I use the term L-proposition for the propositional schema underlying type-level sentences.

Sechehaye and the Great Subject-predicate Debate

181

c­ ontaining no definite but only quantified argument terms under generic tense, such as All humans are mortal). It is not until all referential and indexical expressions in a sentence have been keyed to well-defined elements in a well-defined situation that the sentence in question will actually express a proposition. Context, situation and available world knowledge are indispensable elements in keying processes and must, therefore, play a central role in any theory of keying, that is, of reference. The notion of reference is relevant to the grammar of a language only to the extent that potentially referring elements or phrases may have specific grammatical properties. 5.5.3 The Question-Answer Game: Underlying Cleft Constructions Why is it important to call attention to anchoring and keying processes? The reason is that subsequent utterances in a discourse tend to be answers to (mostly unspoken, or implicit) questions that are taken to have arisen in the course of the discourse (Seuren 1985: 297–304; Van Kuppevelt 1991), a fact that has an immediate impact on the propositional structures involved. Sechehaye’s role in this is that he, together with earlier authors such as Wegener, Stout or Jespersen, was one of the very few who helped prepare the setting for a discourse-based, yet nonpragmatic, answer to the subject-predicate problem. In this light, it seems worthwhile to see that this direction of research has proved productive, even if it is not (yet) generally accepted. Sechehaye’s work in this respect is thus seen to be prophetic beyond the present day. Let us, therefore, review the arguments for a discourse-semantic, not a discoursepragmatic, account of topic-comment modulation, even if that makes us stray a little, in this and the following subsection, from the study of Sechehaye in a narrow sense. The underlying semantic form of a question-answer pair, such as Who opened the door? My son opened the door is taken to be as follows: (5.1) a. Question:  the x such that [x opened the door] is WHO ? b. Answer: the x such that [x opened the door] is MY SON (5.1a) and (5.1b) thus have the structure of what is known as ‘clefts’ in the linguistic literature. Structurally, the predicate of (5.1a) is WHO?, which is not really a predicate but an open predicate-place, a cry to provide a predicate that will make for a true proposition. In the structurally parallel answer (5.1b), the cry is answered by providing the predicate MY SON, which is thus both a predicate and a referring expression. Both WHO? and MY SON are printed in small caps to make it clear that they carry main sentence accent, normal

182

Chapter 5

for the underlying semantic main predicate position.9 The grammar of English then ‘lowers’ the semantic predicate into the position of x in the embedded clause, which then becomes the frame for the main sentence structure:10 (5.2) a. Question: WHO opened the door? b. Answer: MY SON opened the door. The fact that (5.2a) and (5.2b) originate as clefts of the form (5.1a) and (5.1b), respectively, is manifested perceptually by the prominent accent on WHO and MY SON, which points to their status as underlying semantic predicates. From a purely grammatical point of view, however, the constituents WHO and MY SON must be reckoned to be the subject to the predicate opened the door. Some languages, French in particular, put a brake on the grammatical ‘lowering’ process converting clefts into what may be called canonical sentence structures. Without going into all the complex details of French question formation, let me just give a simple example. What is expressed by English What did you buy? is rendered in ordinary current French as (5.3a), originating from (anglicised) (5.3b): (5.3) a. Qu’est-ce que tu as acheté? what is it that you have bought? b. the x such that [you have bought x] is WHAT ? Here, only WH-Fronting has taken place, but the cleft structure has remained intact. There are many more indications that this ‘cleft’ analysis is correct. Thus, as mentioned in note 8 of Chapter 3, a sentence like (5.4a), with heavy accent on the grammatical subject Í, is fully grammatical and differs semantically from (5.4b), likewise with heavy accent on the grammatical subject Í, where normal reflexivisation has taken place: (5.4) a. Í didn’t educate me! b. Í didn’t educate myself! (5.4b) naturally means that the speaker was not a self-educator, whereas (5.4a) naturally means that the person who educated the speaker was someone else than the speaker himself. The ‘cleft’ analysis of sentences with such contrastive accents provides an immediate solution, in that (5.4a) is considered derived from (5.5a), and (5.4b) from (5.5b): 9 10

See Szabolcsi (1981: 515–517) for a similar analysis of main sentence accent in Hungarian. In English and many other languages, WH-question words like WHO or WHAT are placed in front position, which is not necessary in this example, because WHO already occupies that position. The effect of WH-fronting is visible, for example, in a sentence like WHAT did my son open? which is preferred to My son opened WHAT?, but this issue is left undiscussed here.

Sechehaye and the Great Subject-predicate Debate

183

(5.5) a. not [the x such that [x educated me] be I] b. not [the x such that [x educated x] be I] The ‘lowering’ process described above, before the incorporation of the negation not but after normal reflexivisation due to the repetition of x in the suchthat clause of (5.5b), yields (5.6a) and (5.6b), respectively: (5.6) a. not [I educated me] b. not [I educated myself] No alternative solution to this problem is known, nor, it seems, can any be envisaged without recourse to the ‘cleft’ analysis described here. A frequencybased account of grammatical structure (see Section 3.2.3) is, of course, out of the question here. In general we may say that, if only the surface constituent or ‘tree’ structure of sentences is considered, without taking into account accentual or intonational (tcm) features—as has been customary in twentieth-century formal theories of grammar—most sentences suffer from multiple ambiguity according to what element in the sentence is singled out as the predicate in the underlying SA-structure, that is, as the answer to an explicit or implicit question in the discourse running. A further argument for this ‘cleft’ analysis is drawn from the fact that sentences like (5.7a) and (5.7b) are not truth-conditionally equivalent. In fact, they are truth-conditionally independent of each other, in that each can be true or false independently of the other: (5.7) a. Carl was surprised that JÓHN had sold the car. b. Carl was surprised that John had sold the CÁR. We have here two different surprisals—a fact, again, that cannot be explained by an appeal to frequency of usage or to any pragmatic principle. But it can be explained by recourse to the ‘cleft’ analysis sketched above, which gives us (5.8a) and (5.8b) as the forms underlying (5.7a) and (5.7b), respectively: (5.8) a. Carl was surprised that [the x such that [x had sold the car] was JOHN] b. Carl was surprised that [the x such that [John had sold x] was the CAR] In (5.8a), the predicate in the clause describing Carl’s surprise is JOHN, whereas in (5.8b) it is the CAR. Since higher operators such as be surprised, but also the negation not and many others, primarily concern the very assignment of a property to the entities involved and thus focus on the underlying semantic predicate of their argument clauses,11 it follows that the surprisals expressed 11

The explanation of this fact, which is easily assented to but hardly ever mentioned explicitly, probably lies in the very character of a proposition as the mental act of assigning

184

Chapter 5

in (5.7a) and (5.7b) must be distinct from each other. (5.7a) may be true while (5.7b) is false or vice versa, and it is possible for both to be true or for both to be false: (5.7a) and (5.7b) are logically and thus semantically independent from each other, due simply to the different contrastive accents. A further confirmation of the reality of underlying cleft structures is provided by the Turkish question suffix -mI, which is always attached to what is the predicate at the discourse-driven propositional level, that is, to the main verb in ‘neutral’ questions and to the comment predicate in questions with a topic-comment profile (the capital letter I in-mI is a conventional phonological variable for the vowels i, ü, u or ı, selected according to the laws of Turkish vowel harmony). We thus have, for example (Seuren 2018: 409): (5.9) a. Bu adam cocuğ-a süt ver-di-mi? this man child-DATIVE milk give-PAST-QUESTION Did this man give milk to the child? b.  Bu adam-mı cocuğ-a süt ver-di? this man-QUESTION child-DATIVE milk give-PAST Was it this man who gave milk to the child? c. Bu adam cocuğ-a—mı süt ver-di? this man child-DATIVE-QUESTION milk give-PAST Was it the child this man gave milk to? d. Bu adam cocuğ-a süt-mü ver-di? this man child-DATIVE milk-QUESTION give-PAST Was it milk that this man gave to the child? The generalisation is that, in canonical Turkish question structures, the question suffix -mI is attached to the main predicate of the underlying discoursedriven proposition. If, at this level, which is input to the grammar, we have a predicate of a cleft construction, the suffix -mI is attached to that predicate, as in (5.9b–d), but if there is no question-answer game and thus no cleft structure, the suffix -mI is attached to the surface predicate, as in (5.9a). It would seem odd to speak of ‘pragmatic’ processes here, whereas speaking of ‘discoursedriven’ processes seems entirely to the point. Nothing prevents semantics, and thus also grammar, from incorporating discourse-sensitive parameters. Another noteworthy fact to be mentioned in the present context is the distinction made in German between the conjunctions aber and sondern, both a property to one or more entities. Those operators that pertain to this act of assigning, such as the negation, modal or emotive operators and others, will be interpreted as bearing primarily on the predicate of the argument proposition, leaving the (presupposed) argument term(s) unaffected. This question deserves further investigation in the context of the psychological aspects of how propositions come about and how their logical consequences are processed.

Sechehaye and the Great Subject-predicate Debate

185

rendered in English as but. The German conjunction sondern is reserved for contrast between comments, whereas aber is used for all other cases where the contrastive particle but is appropriate. Consider, for example, the contrast between, on the one hand, Ich wohne nicht in BERLÍN sondern (*aber) in HÁMBURG (‘I don’t live in BERLÍN but in HÁMBURG’) and, on the other Ich wohne nicht in Berlin, aber (*sondern) ich weiß genau, wo das Brandenburger Tor ist (‘I don’t live in Berlin but I know exactly where the Brandenburg Gate is’). The frequency of the pattern ‘nicht A sondern B’ in German spoken and written texts is a measure for the frequency of the tcm pattern in spoken and written texts in any language. A further difficulty for a pragmatic account of the phenomena in question is the fact that in some sentence types the placing of emphatic or contrastive accent is impossible, due to grammatical, not pragmatic, reasons. Consider, for example, (5.10a) and (5.10b), the former being normal and fully grammatical, the latter ungrammatical (the asterisk marks ungrammaticality on the intended reading): (5.10) a. John isn’t in the least interested. b. *JÓHN isn’t in the least interested (KÉVIN is). c. √JÓHN isn’t in the least interested (not KÉVIN). The reason for the ungrammaticality (in the intended reading) of (5.10b) is not far to seek: the expression in the least is a strong Negative Polarity Item (npi), which means that it has to stand in the immediate scope of the negation operator in the semantic analysis of the sentence concerned, that is, in the propositional structure directly embedded under the negation operator, without any intervening operator. For (5.10a), this condition is fulfilled, as (5.10a) is considered derived from (5.11a). But the tcm over (5.10b) requires a derivation from a structure like (5.11b), where the expression in the least occurs in a relative clause, which does not stand directly under the negation operator. In (5.11b), the occurrence of in the least is, therefore, illicit: (5.11) a. not [John is in the least interested] b. not [the x such that [x is *in the least interested] is JOHN c. the x such that [not [x is in the least interested]] is JOHN (5.11c) shows the legitimate reading in which, as required, the negation stands directly over the propositional structure containing the npi in question. In fact, if JÓHN is the answer given to the WH-question WHÓ isn’t in the least interested?, as in (5.10c), (5.11c) is the only possible reading. In the reading in which (5.10b) is meant to be understood, it cannot function in the discoursedriven question-answer game. Another instance where this game does not work is provided by a sentence type that has so far, undeservedly, failed to catch the attention of linguists and is exemplified in (5.12b,d):

186

Chapter 5

(5.12) a. The first Americans crossed the Bering Strait some 15,000 years ago. b. The first Americans landed on the moon in 1969. c. The last Chinese emperor died in 1967. d. The last Americans left Vietnam in 1975. All these sentences are ambiguous. On the most natural reading, (5.12a) entails: ‘before the first Americans crossed the Bering Strait, there were no Americans’, whereas (5.12b), in the intended reading, does not entail ‘before the first Americans landed on the moon, there were no Americans’, but rather ‘before 1969, no American landed on the moon’. Analogously, (5.12c) entails ‘after the last Chinese emperor died, there were no Chinese emperors, whereas (5.12d) does not entail ‘after the last Americans left Vietnam, there were no Americans’, but, rather, ‘after 1975, no Americans (have) left Vietnam’. One notes, incidentally, that sentences like (5.12a,b) do not, salva veritate, allow for conjunction reduction or for anaphoric uptake, as shown in (5.13a,b) where the intended meaning of (5.12b) is lost: (5.13) a. The first Americans crossed the Bering Strait some 15,000 years ago and landed on the moon in 1969. b. The first Americans crossed the Bering Strait some 15,000 years ago and they landed on the moon in 1969. The main point here, however, is that sentences like (5.12b) or (5.12d), in the sense intended, do not allow for contrastive tcm accent assignments on argument terms (though they do on the adverbials in 1969 and in 1975). (5.14a,b), in the sense intended, are not well-formed and not acceptable (other than in a so-called ‘echo’-reading where an error in a previously uttered sentence is corrected):12 (5.14) a. *The first AMÉRICANS landed on the moon in 1969. b. *The last Americans left VIETNÁM in 1975. Whatever the analysis of sentences like (5.12b,d)—I know of no literature on this issue—it looks as if the impossibility of contrastive tcm accent just observed is directly related to the fact that they have no corresponding WH-question forms (other than echo-WH-questions, where the WH-element is not fronted but remains in situ, as in The first WHÓ landed on the moon in 1969?). If the question WHÓ landed on the moon in 1969? is answered by sentence (5.14a), the only possible reading of the phrase the first AMÉRICANS is the referential one

12

Camiel Hamans pointed out to me that (5.14a,b) are also acceptable in a so-called ‘list’ reading, as in The first RÚSSIANS landed on the moon in 1965; the first AMÉRICANS did so in 1969, and the first CHINÉSE in 1975. In such a ‘list’ reading, however, (5.14a) cannot be the answer to a corresponding WH-question WHÓ landed on the moon in 1969?.

Sechehaye and the Great Subject-predicate Debate

187

of (5.12a). And similarly for a question like WHÁT did the last Americans leave in 1975? and sentence (5.14b). Obviously, such cases ask for further analysis. A further interesting point is the following. The Oxford philosopher Peter Strawson (1919–2006), who initiated the modern study of presuppositions, published an article in 1964 in which he argues that, although normally a sentence like: (5.15) The king of France is bald. carries the presupposition that there is an actually existing king of France, this is no longer so when the noun phrase the king of France receives an emphatic or contrastive tcm accent, as in: (5.16) THE KING OF FRÁNCE is bald. The difference is, according to Strawson, that if there exists no king of France— as is indeed the case in our world—sentence (5.15) is (for Strawson) neither true nor false, whereas, in that case, sentence (5.16) is simply false. Much can be said about and against Strawson’s analysis of presuppositions, but the fact is that no-one at all informed about these questions nowadays denies that (5.15) carries a so-called existential presupposition, whatever the consequences for its truth-value. As regards (5.16), however, opinions are divided. Doubt has been expressed about the correctness of Strawson’s observation, and his account of the difference as he perceived it in his (1964) is unclear and not persuasive. Many have tried to interpret or elucidate his argument, and some think it is simply not valid. Here, however, we suddenly see that Strawson’s observation may well have been correct and, if so, why it is correct. The reason is now simple and straightforward. In (5.15) there is no tcm and thus no distinction between a discourse-driven and a fact-driven propositional form: in both cases the predicate is (be) bald and the subject is the king of France. But in (5.16), with its tcm, there is a difference between its discourse-driven and its fact-driven propositional form; in the former the noun phrase THE KING OF FRÁNCE is not the subject but represents the predicate, the ‘deep’ subject being ‘the x such that x is bald’. Since presuppositions, existential or other, are induced by the ‘deep’ predicate and apply to its argument terms, it follows that (5.16) does not carry the presupposition of the king of France‘s actual existence. Instead, it carries the presupposition that there is someone who is bald. To end my list of arguments against a pragmatic account of tcm, let me quote an example provided to me years ago by Keith Stenning of Edinburgh University. We were talking about the fact that when a new element has been introduced into the running discourse by means of an existentially quantified expression, such as a cat in (5.17a), this element can be taken up later in the text by a definite description, which may be a referring anaphoric pronoun, as in (5.17b):

188

Chapter 5

(5.17) a. I found a cat at my kitchen door last night. b. The animal (it) was obviously starving. Keith agreed, but presented me with what looked like a counterexample: (5.18) a. Yesterday morning I let my cat out. b. Coming home in the evening, I found a starving animal at the kitchen door. I had to agree that this was puzzling. Now, however, I can answer that tcmanalysis solves the puzzle. The NP a starving animal in (5.18b) is, in fact, not an existentially quantified NP but the predicate of an underlying cleft: (5.19) the x such that [coming home in the evening I found x] was a starving animal This solves the puzzle in what seems to be a convincing way. As has been said, it has become customary, since the 1950s, to maintain or imply that the emphatic or contrastive accent patterns caught under the term Topic-Comment Modulation or tcm are of a pragmatic nature and are a manifestation of what is generally called information packaging, without any connection with either the meaning or the grammar of the sentences involved. It will now be clear that this is untenable. 5.5.4 Formal Aspects of tcm: The Need for ‘Parameter Theory’ in Grammar Given the character of the present study, I will refrain from an attempt at further formalisation of the notions and the operations concerned, whether in a truth-conditional or a grammatical sense. Yet some further comment seems desirable. Grammatically speaking, as was shown in (5.1a,b) above, a discourse-driven proposition takes as subject (the ‘topic’) a phrase of the type ‘the x such that […x…]’ and as predicate (the ‘comment’) either a dummy, ‘question’ predicate that calls for the filling in of a real predicate answering the question, or an actual predicate that answers the WH-question. The structures of (5.1a,b) can thus be rewritten as (5.20a,b), where they appear in a slightly more technical garb:13 (5.20) a. WHO? (the x [open(x,the door)]) ⇒ WHÓ opened the door? b. MY SON (the x [open(x,the door)]) ⇒ MY SÓN opened the door Leaving all details aside, we simply say that a process of tcm-conversion (called ‘lowering’ above) turns the structures on the left hand side of the arrow into those on the right hand side, whereby the latter have a tcm accent on the converted (or lowered) element representing the ‘deep’ predicate. The ‘deep’ 13

Square brackets surround propositional structures; round brackets surround the ensemble (n-tuple) of argument terms of a predicate.

Sechehaye and the Great Subject-predicate Debate

189

accent on the predicate of the underlying tcm structure is thus preserved during the entire grammatical derivation, surfacing as the primary accent of the sentence or clause produced. In the literature on the theory of grammar, the grammatical aspects of such value-assigning sentences have so far not received much attention. A notable exception is Higgins (1979), which is about the grammatical properties of socalled ‘pseudocleft’ constructions as in What John did was clear up the mess or What John bought was a vacuum cleaner, a sentence type that is closely related to ‘cleft’ constructions as in It was clear up the mess that John did or It was a vacuum cleaner that John bought. Recently some authors have taken the issue up again, more from an observational than from a theoretical point of view, speaking of ‘specification clauses’ (see, for example, Mikkelsen 2005, 2011; Comorovski 2008, and the literature referred to there). A frequently made observation in this literature is the fact that the anaphoric reference in the second clause of a sentence like (5.21a) has to be made by means of the neuter anaphoric pronoun it, the masculine or feminine pronouns he, she or they being ungrammatical in the sense intended, as opposed (5.21b), where the opposite is the case:14 (5.21) a. In 1560 the pope was Pius iv, but now it (*he) is Francis I. b. In 1560 the pope wore a beard but now he (*it) is clean-shaven. This indicates that the antecedent of the pronoun it in (5.21a) is a mental discourse representation of a thing, namely the value of the parameter ‘the pope’. In (5.21b), by contrast, the antecedent of he is not a thing but, in a sense that has so far not become entirely clear, a person, or, in a more general sense, an ‘entity type’ in an organised setup (in this case the Church of Rome) that may be instantiated by different real token entities—a phenomenon usually referred to by means of the unilluminating cover term ‘sloppy identity’ (see Mikkelsen 2011: 1810 for some discussion). This takes us to the semantic aspects of discourse-driven tcm propositions, which are much more difficult. A good starting point is Donnellan (1966). This remarkable article made quite a splash when it appeared, but interest seems to have fizzled out in the meantime. Simplifying Donnellan’s sophisticated argument, we may say that he distinguished between a referential and an attributive use of definite noun phrases. In referential use, a phrase like the murderer 14

Note that when, instead of (5.21a), we say In 1560 the pope was Pius iv, but now the pope is Francis I, this does not imply that Pius iv somehow changed identity and turned into the different person Francis I: reincarnation construal does not seem to be an active process in human cognition. This is a complication in anaphora theory that has, to my knowledge, remained undiscussed so far.

190

Chapter 5

of Smith refers to a given known person to whom a property can be assigned in a proposition, as in the most obvious reading of The murderer of Smith has been arrested. In attributive use, however, the phrase does not ‘refer’, according to Donnellan, but is understood as ‘who(ever) murdered Smith’, without the identity of Smith’s murderer necessarily being known. In this interpretation, the phrase the murderer of Smith harks back to a so far unidentified entity postulated in de current Discourse Domain and described there by means of the parameter expression ‘the x such that x murdered Smith’, an expression that is, like the dummy WH-question predicates, in search of an identifiable entity as its value. Such an ‘attributive’ reading presents itself naturally in sentences like The murderer of Smith must be insane or The murderer of Smith has not so far been identified. In my terminology, I say that, in a sentence like The murderer of Smith is Jones, the definite noun phrase denotes a parameter functioning as the subject or topic, whereas is Jones provides the value of the parameter and is the predicate (comment). A specific notation for this category of sentences, or clauses, would be in order, but I will refrain from proposing one. What is clear, however, is that, both in semantics and in grammar, there is a need for what may be called parameter theory, which studies the semantic and grammatical aspects of sentences (clauses) assigning a value to a parameter. Such a theory would open unexpected new perspectives and would provide answers to problems that have so far remained either unsolved or unnoticed. Meanwhile we can ask what kind of predicate is the accented constituent (‘MY SÓN’ in (5.20b)) and what is the status of the subject constituent (‘the x [open(x,the door)]’? We begin by positing that a constituent like ‘the x [open(x,the door)]’ (or: ‘the x such that x open(ed) the door’) denotes a parameter and a constituent like ‘MY SÓN’ denotes the value of the parameter. This, of course, requires some comment. Defining the notion of parameter as a linguistically relevant category is not a simple matter and it is not possible, given limitations of time and expertise, to solve this problem here.15 All we can do is give an impression of both the complexity and the importance of this complex of issues. So let us have a closer look. Linguistically, parameter predicates are typically realised as definite noun phrases followed by the preposition of: the name of, the date of birth of, the address of, the age of, the capital city of, the father of, the profession of, the 15

In mathematics, the term parameter is used in a number of different senses (see Borowski & Borwein 1989: 435, sv ‘parameter’). The sense in which the term is used here is easily recognised in practice but appears to have lacked, so far, a formal mathematical definition.

Sechehaye and the Great Subject-predicate Debate

191

­temperature of, the number of, the captain of, the volume of, etc., as in My name is ­Pieter or The number of planets is nine. This means that parameter predicates are, linguistically speaking, typically realised as nominalisations or reifications, and thus occur grammatically as definite argument terms. They may, however, also occur as surface verbs of sentences, as in This parcel weighs five pounds—with the surface parameter verb weigh—which is an alternative way of saying The weight of this parcel is five pounds.16 Definite decriptions designating office holders, such as the pope, the president, the prime minister, appear to form a special category in that they often lack an of-phrase. In fact, however, the of-phrase is implicit: the expression the pope stands for ‘the head of the Catholic Church’, and if expressions like the president or the prime minister are used it is contextually understood what this person is the president or prime minister of. Such parameter-value assignments can be placed under a higher operator, especially an operator of time or place, which means that different values may be assigned for different times or places, as in (5.21a) above or His name used to be Bruce but now it is Ambrose, or My phone number used to be 736852 but now it is 5736852, or In Greece, my name is Petros, but here it is Pieter. Parameters thus may have different (unique) values under different higher operators. The (unique) values are often taken from conventionalised scales or measuring systems, such as numbers, temperatures, lengths, weights, dates, etc., but may also be uniquely determined by context and/or situation, the way definite descriptions are. Thus, one may say, for example, The DIRECTOR’S SÓN signed the order, or ‘the x such that x signed the order is THE DIRECTOR’S SÓN’, even though the director in question may have several sons, as long as the definite description the director’s son has been contextually keyed to one specific member of the set of this director’s sons. There is thus no intrinsic reason why parameter values should be taken from established scales. In fact, they are often (uniquely) selected from any cognitively or contextually restricted range of entities. This shows again that what counts in studies concerning language and cognition is the cognitive ontology created by the human mind as it construes reality in all its actual and virtual forms and with all its reifications. Other than metaphysical ontology, which may sensibly be subjected to a criterion of ‘ontological parsimony’ (Ockham’s razor), the ‘ontology’ created by the human mind 16

Static and dynamic measure specifications have been understudied in linguistics. After Klooster (1972) and Partee’s famous observation The temperatur is ninety degree and rising (see below), the literature on this topic seems to have virtually dried up.

192

Chapter 5

seems to be characterised by a tendency to be as luxurious as possible, congruent with its penchant for ‘object-oriented programming’. The study of valueassigning sentences thus leads to the much wider problem area of the human construal of reality, which is now seen to be no longer a question of metaphysics but of empirical science. For grammar, this means that we are faced with a remarkable fact that has, so far, found too little recognition in the worlds of linguistics, semantics and pragmatics: In natural language, the assignment of a value to a parameter is expressed, at the underlying level of Semantic Analysis and also often in surface structure, by the verb beval, a specific variety of the lexical verb be, especially reserved for the assignment of a value to a parameter. Thus, when I say The number of planets is nine, the verb be occurring in this sentence is not the identity predicate, as in The morning star is the evening star, nor is it the copula verb be used for the assignment of properties, as in John is a teacher or John is clever, nor is it the be of existence, as in I think, therefore I am, but the verb beval, symbolisable as ‘→’ and exclusively used for the assignment of values to parameters. This might seem mere fussing, of little moment in the theory of language, but the opposite is the case: the question at hand touches on the very foundations of linguistic theory, with ramifications into philosophy. To show this, I may point at a famous question in present-day pragmatics to do with the semantics of cardinal numbers. All leading authors on pragmatics discuss the fact that a sentence like John has five children is, in certain contexts, considered false when John has six children but true in other contexts, as when one has to have at least five children in order to qualify for extra tax relief: in that sort of context it is true that John has five children, since he satisfies the criterion for the tax relief in question. Wishing to avoid an across-the-board ambiguity for cardinal number assignments, the majority of pragmaticists, following Paul Grice, speak of an ‘implicature’ placed on top of a truth condition and based on some form of presumed social cooperativeness. In this analysis, the sentence John has five children is reckoned to be true in all cases where John has at least five, but possibly more, children, but it falls short of satisfying a condition of cooperativeness in those contexts where a precise number is required. One of the problems with this analysis is that it fails to account for the fact that natural intuition assigns straightforward falsity, not some form of uncooperativeness, to sentences meant as answers to questions about precise numbers. When I

Sechehaye and the Great Subject-predicate Debate

193

am stopped at a border being asked by an official how much money I am carrying and I answer ‘Twenty pounds’, whereas in fact I am carrying twenty thousand pounds, I will be taken away for lying, not for being uncooperative, as already pointed out in Fogelin (1967: 21). If John has five children answers the question How many children does John have?, it is false to say that he has five children when the number of his children is not precisely five (Seuren 1993): in ­sentences answering a specific WH-question regarding quantity, a cardinal number assignment is true just in case the assignment is precise. Otherwise, the number assignment usually is (lower-bound) ambiguous and is deemed to be true or false depending on whether or not there is a presumption of speaker’s adequate knowledge disambiguating the ambiguous number assignment (Seuren & Jaspers 2014: 619). 5.5.5 The Collapse of Quine’s Argument of the Opacity of Modal Contexts An anecdotal illustration of the wide-ranging relevance of the parameter analysis of value-assigning sentences is provided by a highly influential argument developed by the famous American philosopher Willard Van Orman Quine (1908–2000) in his equally famous article ‘Reference and modality’ (Quine 1961: 139–159). In this article, Quine slipped up precisely on the difference between the be of identity and the value-assigning beval (or →). Let us have a quick look at the argument and at the reasons why this question is of such importance for an adequate view of natural language. In the article in question, Quine wished to establish that modal predicates like possible or necessary create intensional contexts, just as Gottlob Frege, mentioned earlier in Section 2.4, had established, in his classic paper of 1892, that the that-clauses under predicates of propositional attitude, like believe, realise, know, hope, form intensional contexts. This question is of great philosophical importance in the context of Aristotle’s theory of truth as a proposition’s correspondence with what is actually the case. For if that theory is correct, two terms that differ in form but refer to the same entity, such as the morning star and the evening star, which both refer to the planet known as Venus, should always be mutually interchangeable in any proposition without its truth value being affected (that is, salva veritate), since what counts for truth or falsity is correspondence or noncorrespondence with what is the case, not what terms are used to denote the entities or properties involved. Frege (re)discovered17 that this principle of Substitution Salva Veritate (of referring terms), 17

The same discovery had been made by the Greek philosopher Eubulides (ca 405–330 bce), but this discovery went unnoticed until Frege, independently, made it again (see Seuren 2005).

194

Chapter 5

or ssv, does not hold in intensional contexts, since if the sentence John believes that the morning star is inhabited is true, it does not automatically follow that the sentence John believes that the evening star is inhabited is also true, as John may be under the misapprehension that the two terms refer to different entities. Frege’s discovery of this apparent anomaly, in fact, ushered in modern formal semantics, which started as an attempt to solve this problem in strictly mathematical terms, with the help of the notion ‘(sets of) possible worlds’.18 Quine’s argument, however, comes to nothing. It runs as follows (Quine 1961: 143): the sentence Nine is necessarily greater than seven may count as true, and so does the sentence The number of planets is nine. Yet the sentence The number of planets is necessarily greater than seven must clearly be taken to be false. Ergo, the context created by (or the embedded clause under) the modal operator necessarily (or it is necessary that) is intensional in the Fregean sense, which explains why ssv fails to apply under necessarily, and analogously for the possibility and other modal operators. The (main) error in the argument is now obvious: Quine treated the is in The number of planets is nine as the identity predicate be, not realising that it is the value-assigning predicate beval. This (together with other flaws, such as his failure to distinguish between epistemic and deontic modals) invalidates his argument—which, however, does not mean that his conclusion is false, as one may reach a true conclusion despite a defective argument.19 This problem has, till the present day, not found a satisfactory solution. In the late 1960s, Barbara Partee made an observation to the American logician and formal semanticist Richard Montague regarding the fact that a sentence like The temperature is 90 degrees and rising could not be handled in terms of 18

19

As was shown in Section 5.5.3 above (examples (5.5) to (5.8)), a similar problem arises in connection with tcm accent assignments: just as substitution of co-referring terms salva veritate (ssv) is blocked in intensional contexts generally, the assignment of tcm accents salva veritate is, analogously, blocked in intensional contexts under emotionally charged predicates such as regret that, be angry that, be happy that, be relieved that, be surprised that. No semantic theory, whether in terms of possible worlds or otherwise, has so far been able to account for these facts. The key to a solution lies in the assumption of an underlying tcm structure. In Seuren (2013: 333) it is argued that the blocking of ssv in intensional contexts is a result of cognitive modelling. If two distinct nominal expressions refer to the same entity in the actual world, they need not do so in a cognitive model reflecting an imagined world. Epistemic modals, which involve cognitive modelling, indeed disallow unlimited ssv. But deontic modals, which apply to states of affairs in the actual world, do allow for unrestricted ssv. For example, if may is taken as a deontic modal operator of permission, then You may now kiss the bride is exactly the same permission as You may now kiss Kate, provided Kate and the bride are the same person. This solution would make Quine’s conclusion (but not his argument) half correct and half false.

Sechehaye and the Great Subject-predicate Debate

195

his formal ‘possible world’ system of grammar-cum-semantics—a complication that is identical to that which made Quine stumble. She suggested that verbs like rise, change, increase should perhaps be treated as intensional with regard to their subject position. At the very end of his article on quantification in English, Montague reacts to this as follows (Montague 1973: 239): The next few examples concern an interesting puzzle due to Barbara Hall Partee involving a kind of intensionality not previously observed by philosophers. From the premises the temperature is ninety and the temperature rises, the conclusion ninety rises would appear to follow by normal principles of logic; yet there are occasions on which both premises are true, but none on which the conclusion is. He then proceeds to elaborate Partee’s suggestion, arguing that such verbs allow for definite descriptions in subject position that do not refer to an entity but ‘denote’ a so- called ‘individual concept’, taken to be the intension of a definite term, its extension being the world object(s) referred to. Subsequently, he presents a formalism intended to account for this distinction in terms of the logical system adopted by him. What has been called ‘parameter’ here, is thus treated by Montague as an ‘individual concept’, that is, in the terms of Frege (1892), the intension of an extensional referent, or, one may say, the mental representation of a (world) entity. Although it is clear what inspired Montague in regard of questions of this nature, whatever little there is in the way of literature on this question makes it clear that a final solution has not yet been reached. In fact, the very dimensions of this problem area have not even been clearly delimited so far. One immediate question raised in connection with Montague’s answer concerns the identity manifested by the use of the anaphoric pronoun it in cases like The captain of this ship used to be Erik Broadbeard, but now it (*he) is Sven Whiskers: what the it refers to in it is Sven Whiskers is not the ‘individual concept’ envisaged by Montague, but the value of the parameter ‘the captain of this ship’ (see also ex. (5.21) above). In any case, this entire area requires further theory.

…

Our discussions have taken us a long way from our two protagonists, Saussure and Sechehaye, which may make it look as if we are now treading on ground that is less relevant in the context of the present book. But the opposite is the case: a wider perspective allows for a wider bundle of light. Our study of Saussure and Sechehaye is not meant to be anecdotal but to provide a balanced

196

Chapter 5

assessment of the role and importance of their work in the history of twentiethcentury linguistics. For that reason, it is imperative that their work should be discussed in the light of more general, overarching aspects of linguistic theory. Taking distance has several advantages. It helps to see the two figures in a larger context—a necessary requirement for adequate history writing. Moreover, we now see even more clearly the seriousness of Saussure’s failure to discuss matters related to the great subject-predicate debate. Also, our digression shows that the accusation voiced by Sechehaye’s critics to the effect that he had been unable to cut himself loose from psychology was unjustified. Rather than criticising Sechehaye for being too psychological, we should criticise the more formally oriented sections of twentieth-century linguistics for being too nonpsychological.

Chapter 6

Structuralism, Rationalism and Romanticism in Psychology and Linguistics The great subject-predicate debate, discussed in the previous chapter, is not the only wider perspective in which to place our two protagonists. A different, equally relevant and much wider, backdrop is provided by the more general cultural currents and developments which include structuralism, rationalism and romanticism. For an adequate assessment of Saussure and Sechehaye and of their work it is imperative that we take a closer look at their position in the complex force field created by, in particular, these massive currents in Western intellectual life. We will start with structuralism. 6.1

What is Structuralism?

Around 1900 the air in Europe was thick with a pungent but as yet undefined je-ne-sais-quoi, vaguely to do with modernity in a general cultural sense. Like the London smog of those days, it was inhaled by all but noticed only by few, and there was as yet no word for it. It had started coiling from the chimneys of the human sciences workshops around 1850, as a hardly noticed by-product of rationalism, but it was not identified until its presence was felt so strongly and so widely that naming became inevitable. And when it was named, it ­received a variety of names in different spheres of life. In psychology this happened around 1900, and the name chosen was ‘gestalt psychology’ in Europe, next to ‘behaviourism’—which represented a very different but equally modernistic strand of thought—in America. Then, in linguistics, the name ‘structuralism’ was chosen around 1925. That name has stuck, more or less, until now. But what exactly was it that thickened the European air around 1900 and had been around without a name for half a century? This question does not have an easy answer. In a very general sense, we may say that, around the turn of the century, this ‘pungency’ began to permeate public life, in close connection with a semi-mystical feeling of ‘modernity’, and characterised by an urge to see and create ‘structure’ in human affairs from an overarching point of view, whereby the overarching ‘whole’ was seen as a piece of machinery considered to be more important than the component parts. But what was the ‘whole’ and what were the ‘pieces’? This question was answered in different ways in different spheres of life and in different academic disciplines. © koninklijke brill nv, leiden, ���8 | doi:10.1163/9789004378155_007

198

Chapter 6

On the European political scene, this structuralist urge led to a rash of dictatorial regimes during the first half of the twentieth century: the individual was made subservient to the overarching state—with dire consequences, as we all know. In the arts, we find it, without such dire consequences, in the conscious imposition of explicit overarching principles, often characterised by rectilinearity or other geometrical or formal properties. Style movements arose, such as ‘Bauhaus’ in architecture, which pursued the ideal of a geometrically regular ‘Gesamtkunstwerk’, a total piece of art, constructed according to certain minimalist principles. In Russia, there was the literary style movement called ‘Formalism’, with Roman Jakobson as a prominent member, characterised as follows in Steiner (1995: 18) “Literary works, according to this model, resemble machines: they are the result of an intentional human activity in which a specific skill transforms raw material into a complex mechanism suitable for a particular purpose”. These are metaphors, of course, but their origin speaks volumes. Ideologies arose which maintained that mankind and the world would be better off if logic reigned supreme as an overarching principle. In general, there was a feeling that we could do better than nature if only we applied logical and other f0rmal principles consistently throughout. Society came more and more under the spell of the all too short-sighted idea that we were only two steps away from having the whole of nature, especially human nature, in our grasp and under our control. Social engineering and psychological conditioning would solve all our problems. Science in general was seen as control, rather than as explanation, of the data—an attitude known as instrumentalism, part of positivism. We are a little wiser now, and a great deal sadder, but in hindsight we recognise the same mentality in all these developments and phenomena, even if there was no unifying term for it. Academia was not immune to this ‘zeitgeist’, which affected the human sciences particularly strongly. In fact, this is where the term structuralism arose, though it was never caught under the umbrella of a precise definition or set of criteria. That difficult task is now left to us. Part of the difficulty is that, in twentieth-century linguistics, anthropology and literary studies, the name has been appropriated by a variety of groups in a variety of senses that deviated, sometimes drastically, from the original sense in which the term was used, thus going the way of so many other ‘isms’. What had been happening in the human sciences since the middle of the nineteenth century may be described as a continuation of a process that had started in the physical sciences in the sixteenth century. Since that time, the natural sciences had developed a highly successful view of the physical world, including the human body, as a system of interconnected physical mechanisms

Structuralism, Rationalism and Romanticism in Psychology

199

or ‘machines’, the well-known ‘mechanisation of our world picture’, famously described in Dijksterhuis (1961). As from the middle of the nineteenth century, a feeling, perhaps an instinct, arose that drove some to apply this view to the realm of the mental, and this was intuitively felt to be ‘scientific’. This was, it would seem, the essence of that je-ne-sais-quoi that began to trickle into the European cultural air around 185o, thickening to an almost palpable substance half a century later. It was this that gave rise to all those thoughts about the ‘whole’, whatever that might be, being a functioning ‘machine’ that was more important than its parts. Such notions were necessarily diffuse and certainly not open to explicit formulation. But there was, as I see it, a general, implicit, underlying urge or motivation in that direction, making individuals, alone or in groups, grope for an understanding of the workings of the human mind the way the physical sciences had so successfully plumbed the depths of the physical world. These gropings were uncoordinated, taking place wildly in all sorts of directions. Although the common denominator may have been the search for an understanding of the system or systems underlying the manifold manifestations of the human mind, what appeared in the workplace was disparate, chaotic and often helpless, which makes it hard to detect, in hindsight, the unity or common motivation in what began to be happening all over. Below the surface, there was an identifiable stream of thought and feeling; on the surface, the events show confusion and chaos. But let us leave the lofty level of generalities and pursue the history of the term structuralism, which will help to get a more detailed picture of what was happening. According to Blumenthal (1985: 25), the term made its first appearance in psychology, in Titchener (1898). Since no earlier occurrence of the term seems to be attested anywhere, one may assume that this is the first occurrence ever. As far as linguistics is concerned, the terms structuralism and structuralist are much younger. Both Percival (2011: 243) and Levelt (2013: 217) note that these terms do not occur in Saussure’s Cours. The same holds for Sechehaye’s Programme et méthodes de la linguistque théorique of 1908 and his Essai sur la structure logique de la phrase, of 1926, where these terms are never used. In Bloomfield’s Language of 1933, the foundational publication of American structuralism, the term structuralism is likewise not to be found, though the term structure does appear occasionally. The index to the book gives three occurrences, one (p. 135) where the term structure is used in the sense of ‘structure of the system’ and two (pp. 264, 268), where it means ‘structure of a linguistic form’ (for this distinction, see below). Percival (2011: 243) states that, in fact, the term structuralism “is first attested in writings issuing from the ‘Linguistic Circle of Prague’ (Cercle

200

Chapter 6

linguistique de Prague)”, which had been founded by Vilém Mathesius (1882– 1945), Count Nicolai Trubetzkoy (1890–1938) and Roman Jakobson (1898–1982) in 1926, that is, a decade after the appearance of Saussure’s Cours. More precisely, Percival writes (2011: 244): Also in 1929, Roman Jakobson gave the following thumbnail definition of the term, which may be the earliest use of the word ‘structuralism’ by a linguist : Were we to comprise the leading idea of present-day science in its most various manifestations, we could hardly find a more appropriate designation than structuralism. Any set of phenomena examined by contemporary science is treated not as a mechanical conglomeration but as a structural whole, and the basic task is to reveal the inner, whether static or developmental, laws of this system. What appears to be the focus of scientific preoccupations is no longer the outer stimulus, but the internal premises of the development; now the mechanical conception of processes yields to the question of their functions. It thus seems that the term structuralism as used in linguistics is due to Jakobson, who, as has been said, had been a prominent member of the Russian literary movement known as Formalism. Within the Prague Circle, Saussure was not, or not wholeheartedly, seen as a role model. Trubetzkoy found the notion that his ideas were Saussurean in origin ‘absolutely revolting’ (Sériot 1999: 9). Jakobson was more inclined to make it look as if he fell back on Saussure, though it very much appears as if his frequent mentioning of Saussure’s name was due to his wish to gain publicity rather than to any serious intellectual connection (Gadet 1995).1

1 According to Gadet (1995: 457), Jakobson “used Saussure without recognising Saussure’s originality in general terms or following Saussure’s texts to the letter”. Referring to an interview she had with Cornelis van Schooneveld (1921–2003) in 1992, Gadet quotes the latter as saying that his friend Jakobson said to him, shortly before his death in 1982, in reply to the question what he found so inspiring in Saussure: “We needed a flag to cover the ship” (Il fallait bien un pavillon pour couvrir le navire). (I owe this reference to Andrew Nevins of University College London.) His attitude towards Sechehaye was quite different. Reminiscing about the past, Jakobson wrote (1975: 50): “But when we started to work on phonology, that is, on the strictly linguistic study of the phonic matter of language, it was rather the exposé by Saussure’s student Albert Sechehaye in his book Programme et méthodes de la linguistique théorique (1908), that directed me towards the basic entities of this discipline [i.e. phonology; pams]”. (See also Jakobson 1971a: 312, written in 1939.) (I must thank Dany Jaspers of Leuven University for drawing my attention to these passages.).

Structuralism, Rationalism and Romanticism in Psychology

201

The short-lived, highly programmatic and aprioristic, Copenhagen school of ‘glossematics’, founded by Louis Hjelmslev (1899–1965) (see Hjelmslev 1953), did more or less the same. Even though glossematics differed basically, in its philosophical and methodological principles (it never got any further), from the theories of the Prague School, Hjelmslev considered himself a structuralist, adopting Saussure, lavishly referred to in his (1953), as his icon. Yet his use of the term structure is restricted to the then still nebulous notion ‘structure of a language’: “Linguistics must attempt to grasp language […] as a self-sufficient totality, a structure sui generis” (Hjelmslev 1953: 2). The notion of structure of a sentence, phrase or word is not even hinted at. Similar liberties with regard to ‘structuralism’ were taken by some students and practitioners of literature and art, with the result that, in the end, the term meant to everybody what they wanted it to mean, with only one thing in common, at least in Europe: their claim of Saussure as their ancestor.2 Saussure and structuralism thus became brand names for a variety of disparate products of greatly varying quality. Meanwhile, other (schools of) linguists, whose work was strictly structuralist in the literal sense of the term, did not call themselves structuralists or Saussurean, or even refused to do so. But these schools, or individuals, never conquered the market. This explains why Saussure is now well-nigh universally called the ‘father’ of European structuralism. It shows, moreover, that the Saussure trademark was considered to be good for marketing. Who and what Saussure really was remained hidden behind the clouds of notional opacity and false ancestry claims. The situation was made worse by what happened in America. Here, the label ‘structuralism’ was shed by Noam Chomsky (and his followers) around 1960, when he turned not only against his teacher Zellig Harris, whose ideas on generative and on transformational grammar he had adopted (without proper attribution), but also against Harris’s predecessor Leonard Bloomfield, the real father of American structuralism (see Seuren 1998: 243–247). Yet Chomsky’s subsequent theories of grammar were as quintessentially structuralist as those developed by Bloomfield and Harris. It was all just a question of brand names and a fight over historical legitimacy.3 2 Calling the success of the term structuralism a “terminological pandemic” (2011: 256), Percival writes (2011: 254): “The term ‘structuralism’ was in effect an ideological slogan that could be moved about and re-used in whatever way was needed”. I take it that, for Percival, ‘ideological’ is a euphemism for ‘propagandistic’. 3 Percival writes (Percival 2011: 248): were an important component in the scientific window-dressing that professional linguists constructed for themselves from the early nineteenth century onward. Perhaps for that very reason, it was not necessary that

202

Chapter 6

I cannot, of course, keep anyone from using the term structuralism the way they want to. Nor can I, or do I wish to, impose my way of using that term on anybody else. But I can point out different uses of the term and I do have the right to ask those who use the term their way to provide some sort of definition or at least some justification for their use. And I am free to use the term my way. But I do not use the term loosely. On the contrary, I try to give as precise a definition or description as I can manage against the background of how both the term structuralism and the corresponding concept came into being in the first place. In this vein, I propose that the question of what structuralism actually amounts to in conceptually accurate and coherent terms is best answered by saying that structuralism in the human sciences in general amounts to: a point of view or manner of analysis whereby all attention is focused on the structure of the object of investigation—social structures for sociology, perceptual interpretations for psychology, sentences for ­ linguistics—whereby ‘structure’ is understood as the way the object of ­investigation has been put together from pre-existing elements, in the service of a meaningful, well-functioning whole. In a specifically linguistic sense that would be valid for linguistics as practised on both sides of the Atlantic, linguistic structuralism can then be described as: the static (synchronic) study of any language L, seen as a formal, partly rule-driven and partly memory-based system in each individual’s brain (or mind), allowing speakers of L to compose, and hearers to analyse and interpret, sentences of L.4

such key terms have unequivocal meanings. In this instance, the vagueness and openendedness of the term ‘structure’ may have contributed to its cachet. The very fact that it was non-committal may have made it an especially ideal slogan for the promoters of new methods of linguistic research in the interwar years and later who had otherwise very little in common. 4 Merely specifying structures without taking their meaning and function into account amounts to a mutilated form of structuralism, found mainly in the American structuralist tradition developed by linguists such as Leonard Bloomfield, Zellig Harris or Noam Chomsky, who, in the wake of positivism and behaviourism, failed to give meaning and function an appropriate place in their theories. Few will disagree that the expression of meanings under social (speech- act) commitments by speakers/writers in such a way that they can be reconstructed by listeners/readers in any given situation is the chief function of human language. Leaving that function out of account is missing out on the essence of language—no matter

Structuralism, Rationalism and Romanticism in Psychology

203

Linguistic structuralism is then, as a coherent notion, correlated with the assumed existence of an underlying system regulating the structure and interpretation of utterances. As this is now a standard assumption in all serious forms of linguistic theory, one may predict that the label ‘structuralism’ will eventually disappear from the linguist’s vocabulary, having become too selfevident, just as it has always been self-evident, and thus superfluous, in the physical sciences. But that stage of maturity has not been reached yet. For the time being, linguistic theory, whether or not it calls itself ‘structuralist’, in whatever variety on either side of the Atlantic, is still very much divided over the question of what structural format is to be taken as basic to all linguistic products in all languages. As a result, attention is focused in large part on this question: what kind of structural format is to be taken as the most appropriate for the products of human language in general? As long as this is the case, it makes sense to use the predicate ‘structuralist’, as structure is what most of the debate is about. Once unanimity on this issue is achieved, linguistics will be mature enough to shed the qualification ‘structuralist’. What one does find in linguistics is an overall negative unanimity regarding the question of what the basic structural format of natural language sentences should not be. There is, in linguistics, a totally misguided yet all-pervasive but never openly admitted strong aversion to the notion of propositional structure as being essential to any structural analysis of natural language sentences. This aversion is, as I see it, largely due to what was (and is) experienced as a traumatic remnant from ancient times, when no clear distinction was made between (logical) propositions and (linguistic) sentences, as a result of which the study of language was standardly mixed up with the study of logic. Those who wanted to establish linguistics as an autonomous science would thus cleanse themselves of logic as far as possible, throwing away the baby called ‘proposition’ with the bathwater thought to be polluted by logic. To make up for the loss, various linguistic schools or individual practitioners then thought up more or less ad hoc structures in terms of which it was thought that sentences could be adequately analysed and described. In this respect, Sechehaye made himself an outcast, as he had committed the mortal sin of following Wundt’s idea of the proposition as the structural format for the representation of sentence meanings. As was noted in Section 4.2.5 above, this may well have been one of the reasons why Sechehaye was

how elusive the notions of meaning, function and social commitment may thus far have proved to be.

204

Chapter 6

given the cold shoulder by the structuralist linguists of the twentieth century.5 Yet in fact, Sechehaye’s work is as ‘structuralist’ as it could possibly be, but the structuralist establishment shunned him, accusing him of ‘psychologism’, or ‘logicism’, or both. The notion of structuralism as sketched above in both historical and conceptual terms is, I think, clearly recognisable in the way the term has become accepted in the more substantial sections of twentieth-century linguistics. But even within this frame, the term has been applied inconsistently, as one often finds authors explaining the notion of structuralism in terms of the structure of the language system, not of the products whose structure is regulated by the system. In fact, in what is probably the first use in linguistics of the term structuralism, Jakobson takes that term to denote a property of the system, not of the system’s products, as shown above. Likewise for John Lyons, who wrote, using a basketful of undefined terms (1977: 231–232): What, then, is the central thesis of structuralism? To put it first in its most general form, it is this: that every language is a unique relational structure, or system, and that the units which we identify, or postulate as theoretical constructs, in analysing the sentence of a particular language (sounds, words, meanings, etc.) derive both their essence and their existence from their relationships with other units in the same language-­ system. We cannot first identify the units and then, at a subsequent stage of the analysis, enquire what combinatorial or other relations hold ­between them: we simultaneoulsy identify both the units and their interrelations. Linguistic units are but points in a system, or network, of relations; they are the terminals of these relations, and they have no prior and independent existence.6 5 The same fate befell the American linguist James McCawley during the 1970s, after he had, independently, proposed to reinstate the proposition as the structural frame for sentential meaning representations, each such structure to be transformed, by means of a grammatical rule system, into the well-formed surface structures of any given language. 6 The last two sentences of this quote are a direct reference to Saussure’s lucubrations on the purely negative and oppositional nature of linguistic units (see Section 3.5 above). It must have escaped Lyons’s notice not only that phonemes were identified thirty centuries before any notion of a phonological system arose (the identification of phonemes by the end of the second millennium bce underlay the development, by the Phoenicians and the Greeks, of alphabetic writing systems), but also that words in a sentence are effortlessly, and in largely identical ways, identified as such by illiterate speakers, who have no inkling at all of a language system (see Sapir’s famous footnote on page 35 of his 1921 book Language). Other

Structuralism, Rationalism and Romanticism in Psychology

205

Peter Matthews (2001: 48) quotes the Dutch Slavic linguist and early structuralist phonologist Nicolaas van Wijk (1880–1941) as saying (Van Wijk 1939: xiii): ‘In the mind of each member the language exists as a system with a determinate structure, and the task of a linguist is to work out such structures’. Van Wijk seems to be thinking here of the structure of the system, although the second clause of the quote may give the impression—one cannot really make out what exactly he meant—that he was also thinking of the structure of the products of the system. Matthews himself, moreover, strays from the one sense of ‘structuralism’ to the other. First (Matthews 2001: 49–50) he expands on the Saussurean distinction between ‘syntagmatic’ and ‘associative’ (later called ‘paradigmatic’) relations between words, interpreted by him as making up the ‘structure’ of a grammatical system, but then, on p. 51, in the context of American structuralism, he alludes in vague terms to the Bloomfieldian method of immediate constituent analysis, which, obviously, is to do with the structure of the products of the system. An explanation is not provided. By contrast, the Dutch linguist Albert W. de Groot (1892–1963) is clearly aware of the difference (De Groot 1949: 10): The terms ‘structure’ and ‘system’ are used in very different ways in the linguistic literature. […] One speaks of the structure of a given word, such as housing or houses. And one speaks of the structure of a phonemic system, or a case system, or a system of word classes, or of a language as a whole. […] We can thus speak of the structure of a linguistic element ánd of the structure of a language system or part thereof, a collection of categories of linguistic elements. De Groot’s 1949 book itself is then devoted for the most part to structures of linguistic elements, though in an earlier publication (De Groot 1940: 128), the author states, intriguingly: ‘[…] the structure of a system of signs is based […] on the structure of all single signs of this system’. Inductively, this is correct, since the underlying system is constructed hypothetically on the basis of intuitions about the structure of language elements (words, phrases, sentences). Deductively, however, De Groot’s statement must be reversed: from a deductive point of view, the structure of each composite linguistic element is based on, or defined by, the underlying system.

than Lyons has it, it is one of the aims of linguistic theory to make explicit and explain such pretheoretical native intuitions.

206

Chapter 6

Since the notion ‘structure of a system’ was never made very clear in the linguistic literature of the mid-twentieth century—in fact, all that could be said about the ‘structure’ of a linguistic system was that it consisted of a phonology, a morphology and a syntax—while the notion ‘structure of a product of a system’ was elaborated with great ingenuity and precision, and led to important developments, I will limit my use of the term structure to the structure of the products, leaving the notion ‘structure of the system’ aside as being left too vague, in the period at issue, and thus less relevant for the history of structuralist linguistics during its formative period. This does not imply, however, that the notion ‘structure of the system’ in itself should be considered less relevant or less interesting. On the contrary, the notion ‘structure of a linguistic system’ came to the fore in the works of Zellig Harris, especially in Harris’s 1951 book Methods in Structural Linguistics, and in the subsequent literature on formal grammar. All that is implied here is that this latter notion belongs to a different, in fact higher, level of theorising than the notion ‘structure of a product of a system’. What is known as ‘European structuralist linguistics’ never managed to reach this higher level of theorising in any explicit or precise way, remaining fixated on the structural properties of sentences and their parts. From the beginning, the term structure has covered a wider range than the term structuralism. The term structure is also much older than the term structuralism, which is first attested in Titchener’s article of 1898, as has been said. By contrast, according to The Oxford English Dictionary, the term structure is attested as early as the sixteenth century and occurs frequently through subsequent centuries, often, especially in the nineteenth century, in the context of the natural sciences, mostly biology, and always in the sense given above of the way the object of investigation has been put together from pre-existing elements. Talk about structure thus implies talk about component elements, in any sense whatsoever. For example, when I speak about the structure of a bridge, I imply that the bridge consists of component elements and I refer to the way these have been put together to form the bridge as a whole—which is thus more than the sum of its parts: it is the sum of its parts plus the whole resulting from the way these parts have been put together. But engineers have never called themselves structuralists. In medical science and biology, the terms structuralism and structuralist are not used either. Doctors and biologists speak of ‘physiology’ or ‘anatomy’, two branches of science studying, each from a different point of view, the way component elements have been put together to form a larger whole. Likewise, geologists have never been known as ‘structuralists’, even though they study the structure of the earth’s crust. In

Structuralism, Rationalism and Romanticism in Psychology

207

actual fact, the term structuralism has always been restricted to the human sciences, the traditional ‘arts’ or ‘humanities’, whereas the term structure has always had a wider and more general application. Since all human sciences ultimately revolve around the human mind (located in the brain), structuralism is ultimately a way of studying mental structures and systems. The human brain can be studied from an anatomical and a physiological or neurological, but also from a human-sciences point of view. In that context, the term brain is usually avoided and replaced with the term mind. The structures at issue in the human sciences are thus not physical but mental structures, which are not directly observable but must be reconstructed by hypothesis on the basis of observed phenomena. Not being of a physical nature, their ontological status is widely considered to be unclear, less clear than the ontological status of physical matter is, rightly or wrongly, thought to be. It would seem that in the natural sciences the notion of structure is so obvious that no need is felt for a separate term ‘structuralism’, whereas in the human sciences that notion had to force its way through, against the popular notion that the mind is a fluid, amorphous and largely unstructured cloud contained in a skull and subject to internal breezes that may grow into violent storms—the way Saussure imagined the mind, as was illustrated in Section 3.4.1 above. This takes us to a discussion of what distinguishes the human from the natural sciences, and also of what is meant when we speak of rationalism versus romanticism in the human sciences. Let us look at the latter term pair first. 6.2

Rationalism versus Romanticism: Clarifying the Terms

The distinction between romanticism and rationalism pervades the entire present study, and for a good reason. It has proved useful, if not indispensable, in that it provides a constant explanatory thread in the tohubohu of the countless interconnections and the ubiquitous interwovenness of the intellectual and cultural movements of the last two centuries. In the twentieth century, the gap between the two currents widened and became more clearly delineated, more fundamental and thus less bridgeable.7 7 The distinction is frequently found, under different names, in the literature of the period. Chapter 15 of Adolphe Pictet’s Du beau, entitled “Le classique et le romantique”, opens thus (Pictet uses the term classical (‘classique’) for what is here called rationalist) (Pictet 1856: 279): When a debate or a complex problem gets lost in ill-defined terms, one can be sure that all kinds of aberrations will result: this is what happened to the issue of the classical as against the romantic. In the wake of a long and ardent polemic, this row [‘querelle’]—as

208

Chapter 6

Let us begin by clarifying the terms themselves. The term rationalism is, in principle, used here to refer to a collection of ways of thinking current during the past two centuries, not to the rationalism of the seventeenth and eighteenth centuries, which was concerned with the epistemological question of the nature and the reliability of knowledge. In that context, rationalism was characterised by the thesis that reliable knowledge is possible only on the assumption of an innate mental machinery of categories and processing mechanisms. As such, it was opposed to empiricism, which preferred to start from the minimalist hypothesis of the mind as a tabula rasa, or, in the original sense, a wax tablet unblemished by any writing (the British empiricist John Locke spoke of ‘a white sheet of paper’). The rationalism referred to here is the modern variety, no longer opposed to empiricism but to romanticism, and characterised by the general understanding that: All actual (not virtual) reality is governed by systematic laws, rules or principles that are open to empirical testing and can, ideally, be captured in systems describing and, to some extent, explaining observed facts and processes in terms of formal (algorithmic) derivation.8 This modern, ‘Mark 2’ rationalism thus implies the tenet that actual reality is subject to mathematical necessity caught in algorithmic derivational systems and it follows the methodological principle that the systems postulated are open to systematic empirical testing, that is, verification and falsification. It is, in fact, no different from the ideology pervading all of modern science. And while romanticism plays no role in the natural sciences, this is not so in the human sciences, also known as ‘arts’ or ‘Geisteswissenschaften’, where ­romanticism keeps playing up and where the contrast between the two ways

it was rightly called—slowly settled down, tired as one was of fighting, without, however, any rational, that is, acceptable, if not generally accepted, solution being reached. Pictet then goes on for pages on end to describe how passionate, how ferocious, how chaotic and how widespread this ‘row’ was during the first half of the nineteenth century. That the issue did not disappear after Pictet is proved by Antoine Meillet, who wrote in 1930, in an article commemorating the late nineteenth-century French semanticist and theorist of language Michel Bréal (Meillet 1938: 227): ‘Bréal’s rationalism will no doubt displease the innovators who give pride of place to facts of emotion’. 8 The Greek word for Latin ratio, which underlies the term rationalism, is lógos, a key term in Aristotle and above all in Stoic philosophy. Lógos means ‘word’, ‘sentence’, ‘reason’, ‘rational thought’, ‘coherent story’, and later, in a stoic context, came to form the stem of the derived word logikḗ, our logic. (In the great Greek-English Lexicon of Liddell and Scott, the lemma lógos occupies three very large pages of very dense print.).

Structuralism, Rationalism and Romanticism in Psychology

209

of ­thinking has led to deeply rooted, often unbridgeable, oppositions and controversies. Materialism, or the doctrine that all actual reality consists of physical matter, has always been a faithful companion to rationalism, resulting in the close connection between the natural sciences and mathematics, considered to be the metaphysically necessary book of ground rules for physical matter. In general, the modern rationalists hold that insight and adequate understanding, as well as efficient action, are not promoted by following passions, emotions or fantasies but by the unimpassioned enquiry of the closest possible approximation of what may be considered to be actually the case. In philosophy, rationalism is, in some of its forms, supported by the underlying implicit notion that cognitive content is organised in the form of propositions, a proposition being taken to be the mental act of assigning a property to one or more entities. As has been said (Sections 3.4.1 and 5.3), the ‘entities’ are often of an ‘abstract’ nature, which means that they need not be discrete entities in the material world but may be what philosophers call ‘reified entities’. Nouns like justice, virtue, happiness, price do not denote world entities but reified entities. Reification is a cognitive process, reflected all over natural human language, which has the advantage that, once some cognitive complex has been reified, it is cognitively treated as an entity and can hence be assigned an appropriate property, thus forming a new, more abstract, proposition. Computer scientists found out how functional reification is in programming languages: they call it ‘object-oriented programming’ (see note 39 in Section 4.4.2 above). In other words, cognition does the opposite of what Ockham’s principle tells science to do: it multiplies entities as much as possible, so as to be able to assign them properties and thus reach higher levels of abstraction. Instead of saying, with Ockham, Entia non sunt multiplicanda praeter necessitatem (‘entities must not be multiplied beyond necessity’), the cognitive mind flaunts the slogan: Entia multiplicanda sunt ad maximam utilitatem, or ‘entities must be multiplied to the greatest utility’. Both Ockham and the mind (and objectoriented programming with it) are right, of course.9 The proposition, thus defined, provides the basic format not only for any logical system defining logically valid deductive processes and other operations, but also for any inductive (or abductive) generalisation over observed facts. This is how the ensemble of logic as a method of deduction, and i­nduction/ abduction as a method of generalisation, both using the propositional format as a basic structure, becomes the chief motivation for the projection of 9 That Sechehaye was aware of the function and importance of reification is shown in note 39 of Chapter 4. Saussure, as usual, does not touch on the issue at all.

210

Chapter 6

causality upon the material world, and thereby for a maximally coherent world construal: causality is induction-cum-logic projected onto what is taken to be the external world.10 Such projection is the first bit of natural science acquired, or developed, by the human mind as a result of an innate drive.11 This analysis hangs by the overall double assumption that the human race is equipped with (i) an innate drive to accept that perceptual input corresponds to a really existing external world (naïve realism) and (ii) an innate faculty of induction-cum-logic, whereby the implicit, nonconscious criterion holds that the maximisation of coherence is the most reliable guide to adequate world knowledge, that is, to truth. This double assumption, known as the inductive leap, is not grounded in formal logical argument but is nevertheless rational in an inductive sense, as it is continuously confirmed by the inductive experience that it provides optimal predictive power and thus helps to keep us safe from disaster and to realise our plans.12 10

11 12

See Michotte (1946) for experimental evidence that causality is indeed projected onto the world. Linguistic evidence comes, inter alia, from the fact that we say ‘The short circuit explains the fire’, with the verb explain in the present tense, referring to a here-and-now mental process, but ‘The short circuit caused the fire’, with the verb caused in the past tense, referring to a causal link between past events projected by the thinking mind onto the world. Logical priority is thus projected onto temporal succession: causes necessarily precede their effects, just as in logic premisses necessarily ‘precede’ their logico-semantic consequences. Hence Karl Popper’s saying (1959: 22): “Science is common sense knowledge writ large”. The inductive leap, which includes the process known as abduction, that is, the formation of explanatory causal hypotheses, is the centrepiece of the coherence theory of truth proposed by the thoroughly rationalist German philosopher and polymath Gottfried Wilhelm Leibniz (1646–1716). Nicholas Rescher put this in an admirably concise form as follows (Rescher 1973: 258–259): Leibniz clearly makes the following points: (1) that the coherence of lawful orderliness is the proper test as between real and illusory phenomena, (2) that this test is not and cannot be certified as correct by any demonstrative proof (and certainly not by an appeal to a Cartesian deus ex machina), and (3) that accordingly support for a coherence approach in the final analysis comes from pragmatic considerations relating to success in prediction and the guidance of ‘the practice of life’. Leibniz thus answered the Kantian paradox of knowledge even before it was put into words. This paradox, formulated in its most stringent form by the German philosopher Immanuel Kant (1724–1804), posits the basic unknowability of the ‘thing in and for itself’ (‘das Ding an sich’), and thus of the entire complex we call ‘world’. All empirical (or factual, or ‘synthetic’) knowledge of what presents itself to us in the guise of an external world (and thus not perceptions of the ‘ego’ or the ‘self’) may, after all, be doubted (the ‘Cartesian doubt’) and may even be taken to be one big illusion (solipsism): our mind may, after all, deceive us into thinking that what we perceive is real whereas in fact it is all our own dream or fantasy. There is no way of actually proving that it is not. The paradox consists in the fact that—but for pathological conditions such as severe autism—we proceed

Structuralism, Rationalism and Romanticism in Psychology

211

In linguistic structuralism, however, the proposition has hardly ever played any role—much to its detriment, I would say. In this respect, linguistic structuralism is not, or less, typically rationalist. In fact, we may distinguish between two kinds of structuralism in linguistics. On the one hand, we have the weakly rationalist kind, which dominates the market and seeks refuge in varieties of positivism and/or behaviourism, in frequency counts and statistically organised ‘big data’. On the other hand, there is the less known and rather isolated strictly rationalist kind, exemplified in Sechehaye’s work and in the tradition of Generative Semantics (including my own work in grammar and semantics).13 By way of contrast to rationalism in the human sciences as sketched above, I use the term romanticism, in the present context, for the academic doctrine or opinion that: cognitive content is an associative complex of notions calling each other up in ways that are determined by a conglomerate of psychological, in particular emotional, processes and thus bringing about a mental state composed of convictions, beliefs and/or emotions. Of course, this is a severe narrowing of the concept from the ways in which the term romanticism is used in a general sense, but, as has been said, this

13

as if what we perceive reflects actual reality, whereas our reason tells us that there is no guarantee that we can trust our senses. Our world construal is thus based on an overarching metaphysical assumption which is innate in human nature, and, therefore, inescapable in any process of acquisition of knowledge, and which instills trust in the reliability of our world construal—in sum, the inductive leap. Chomskyan linguistics is a curious hybrid, or better, a curious side alley leading nowhere: it makes no use of the notion of proposition nor does it want to fall back on positivist foundations like behaviourism, frequency counts or ‘big data’. It is thus neither ‘strictly’ nor ‘weakly’ rationalist. Instead, it seeks refuge in improvised ‘foundations’, such as the alleged overarching explanatory power, in linguistic theory, of the mathematical notion of recursion (unnecessarily renamed ‘merge’), or in biologically, linguistically and otherwise untenable and uninformed, even at times mystical, accounts of the origin and structure of language (see Behme 2014). Despite continuous explicit and implicit claims to profundity and to a historical lineage—all a deliberate part of his own myth-making— Chomsky’s work on the foundations of language and linguistics, no matter to what extent it has been glorified, has always suffered from a lack of proper philosophical and historical expertise, quite apart from its persistent linguistic inadequacies (see Seuren 2004). Even on a central point such as the ‘realist’ or ‘instrumentalist’ interpretation of grammatical descriptions, “Chomsky’s position is […] confused: his model seems to be described as realistic but the famous note 3 [Chomsky 1995: 380; pams] puzzles the reader. What is really Chomsky’s position? There is no answer but just recognizing that his account suffers from uncertainty, obscurity, contradiction and nonsense […]”. (Celano 2008).

212

Chapter 6

d­ escription is meant for the present context, which is restricted to the human sciences. The link with romanticism in the various, much wider, senses in which the term is used generally I will leave undiscussed here, beyond saying that romanticism, in the wider, more general, sense, allows emotions to play a major part in the motivation not only of human behaviour but also in accounts of how adequate knowledge is acquired. For the rest, the reader will not find it too difficult to make the connection. 6.3

Human versus Natural Sciences

We are thus inevitably led to the question of what distinguishes the human from the natural sciences. The difference is profound and all-pervasive, and, though discussed intensively as well as extensively, still largely opaque. All I can do here is try to summarise my own thoughts, as they have shaped themselves over the years, often in interaction with others who had superior knowledge and insight. It seems to me that the distinction is grounded in the ontological and methodological differences between the two kinds of science. Ontologically, the difference seems obvious: the natural sciences deal with physical matter, while the human sciences deal with mind and its products. But what does this mean? I do not venture a complete answer but I can observe that, to the extent that the distinction is phenomenologically or experientially valid, matter has an extension in space and time, can be subjected to varying material conditions in what are called experiments, and is describable in terms of mathematical equations known as laws. Matter is considered constant in the sense that ­experiments are repeatable, meaning that they should yield the same results, ceteris paribus, in all cases. In the case of the human sciences all this is much less clear. Phenomenologically or experientally, we certainly see the mind as basically different from matter, but what we conceive as being ‘mind’ is much less clearly describable. It is obvious that the mind, somehow, depends on matter, in particular matter as instantiated in the brain. Putting it bluntly, if we receive a blow on the head, the mind may cease to function, temporarily or forever. So there is a relation of dependency between the two, but in what does it consist? One may think that ‘the mind’ is spatially and temporally enclosed in and by the human brain, but if so, how about ‘the social mind’, considered a reality by many human scientists and philosophers? For most mental systems, the prospect of capturing them in terms of mathematical equations seems remote. Speaking of ‘laws’ in the human sciences is controversial: one usually prefers ‘tendencies’. We can

Structuralism, Rationalism and Romanticism in Psychology

213

subject the mind to varying material and mental conditions, but in many cases it is unclear to what extent the mind has the constancy that is found in matter. Wundt, and with him all experimental psychologists, require that psychological experiments should be repeatable so that the same results are obtained, ceteris paribus, each time the experiment is carried out. But, as Wundt was well aware, masses of manifestations of ‘the mind’ do not lend themselves to carefully controlled experiments. The data necessary for verification and falsification are, on the whole, not of the material kind but consist in forms of behaviour categorised in nonmaterial terms (such as buying behaviour in economics), or in reaction times to stimuli, or in reports on experience or social acceptability (as in grammaticality judgments in linguistics). And, most importantly, the systems that are discovered or postulated are not, or not fully, implementable in either matter or mind (whatever the latter would mean), only, if you like, on paper. Even so, many practitioners in the human sciences, the rationalists in particular, felt that, to achieve the status of a real science one had to emulate the natural sciences as much as possible. This meant that they strove for the development of predictive theories and for methods of verification and falsification, no matter the status of the ‘data’, as long as these could be established to a reasonable degree of certainty and reliability, and no matter the ontological status of the elements posited in the theories, as long as the latter made the right predictions. This form of rationalism in the human sciences stood in radical opposition to what we have identified as idealism, which, in principle, is a nonrationalist but romanticist way of setting off the human from the natural sciences. It consists in denying, or leaving out of account, the materialist basis and saying that the reality studied in the human sciences is independent of its neurological substrate and is not just autonomous with regard to merely physical reality but actually independent from it. In this view, the application of the strict methodological criteria of the physical sciences to their human counterparts is considered inappropriate and even objectionable, since the two categories of ‘science’ are seen as being fundamentally different. The underlying idealist philosophy is known as German Idealism, founded in large part by the philosopher Georg Wilhelm Friedrich Hegel (1770–1831) as a romanticist reaction to the rationalism of Immanuel Kant (1724–1804). The man who was most influential in applying this romanticist-idealist philosophy to the new human sciences was the German philosopher Wilhelm Dilthey (1833–1911), who claimed that the human sciences could be as ‘scientific’ as the natural sciences without the assumption of a material substrate, since, for him, an experiential basis of deep empathetic ‘understanding’—the

214

Chapter 6

German term is ‘Verstehen’—on the part of the investigator sufficed as an empirical foundation for the human sciences. This was denied by the majority of Anglo-Saxon philosophers, who deemed this methodology basically flawed and unscientific, with the result that, in the English-speaking world, the word science is traditionally restricted to the natural sciences, excluding their human counterparts, which are traditionally known as ‘arts’ or ‘humanities’. In the German-speaking world, by contrast, one speaks, owing largely to Dilthey, of Naturwissenschaften versus Geisteswissenschaften, respectively (whereby mathematics was ranged with the former and philosophy with the latter). Since the late nineteenth century, this philosophical and methodological opposition has deeply split the academic world, leading to a division into two irreconcilable populations, the ‘continentals’ and the ‘Anglo-Saxons’, who are separated by an unbridgeable incompatibilité d’humeur, and are all too often unable even to talk to each other. One thing, however, is clear, in this complex field of tensions and oppositions: followers of Dilthey cannot be called structuralists. Structuralism is part of the rationalist tradition as it re-established itself during the nineteenth century and is alien to the romanticist tradition, which, on its part, provides a safe haven for a ‘verstehen’ philosophy of the human sciences. This point is of interest, since some, mostly French, representatives of the romanticist ‘verstehen’ philosophy have, since the end of the second World War, advertised themselves as structuralists, putting up Saussure as their icon—a clear example of the tendency, mentioned at the outset of this chapter, to appropriate the name ‘structuralism’ without it being applicable. 6.4 Reductionism The rationalist-materialist faction in the human sciences is inevitably faced with the fundamental question of reductionism: in what way are the systems specified in the human sciences implementable, or at any rate interpretable, in material terms? For the idealists, this question does not arise. But for the rationalist-materialists it does. The view that this is a real question that needs an answer is known as reductionism. Attempts at answering the question have, however, hit a wall that does not look as if it is going to be breached any time soon. Admitting this is the price to be paid by the human sciences for access to academia. It is generally held, nowadays, that reductionism must ultimately be the right course, mainly because the mind cannot function unless the material substrate is in good working order—a dependency relation that is not

Structuralism, Rationalism and Romanticism in Psychology

215

a­ ddressed in idealism. We are thus faced with the complex of issues related to the programme of reducing all human sciences to the material or physical basis or substrate of their objects of enquiry. Reductionism is thus a manifestation of the general position that characterises physicalism or materialism, both in metaphysics and in science. A distinction must be made between strong and weak reductionism, the latter being known as token physicalism. Strong reductionism holds (usually implicitly) that the systems found or postulated in the human sciences correspond one-to-one to physical systems in the brain. Token physicalism explicitly denies this, saying that, while each token manifestation of a mental system necessarily corresponds to some token event in the physical brain, the underlying mental systems are, to the best of present knowledge, not reducible as systems to the ones that are operative in the neurology of the brain. It is these autonomous mental systems that form the specific remit of the human sciences, which are thus taken to distinguish themselves crucially by being subject to supervenient principles that somehow emerge from or arise out of what is physically happening in the brain, forming systems of their own. But this is not enough, since the phenomenon of supervenient systems arising out of more basic ones is well-known within the complex of natural sciences themselves: biology arises out of chemistry, which arises out of physics. So where does the crucial caesura fall, the point where the natural and the human sciences come face to face? To say that it falls where the supervenient systems found or postulated are no longer implementable, or interpretable, in material terms means that the reduction problem has not been solved and thus takes us back to square one. Computer simulations of cognitive processes are a step in the right direction, provided they do not, as happens too often, take a shortcut making use of frequency statistics and degrees of success in terms of practical applications, rather than trying to model the actual system and thus be crucially correct for all cases. But they fail to provide a full answer as long as it is not shown how computer simulations are implementable, or recognisable, in terms of brain physiology. The problem will not be solved until mental systems can be reduced to, or identified in, the neurological structures and processes of the brain. When that is achieved, the distinction between natural and human sciences will either disappear or have to be reformulated. But we are still far removed from such a result, despite massive attempts at deciphering the code that will make us read psychological systems from physical token events. The insight that strong reductionism is untenable and that it is necessary to fall back on the position of token physicalism began to break through during the end of the nineteenth century. Hippolyte Taine, in his De l’intelligence of 1870, showed a clear awareness of this problem as he began to vindicate the

216

Chapter 6

rights of an autonomous psychology, not simply reducible to neurology, implying that the latter will not be able to provide the explanatory theories needed in the former (see Seuren 2016). But the gloves were off a few years later, when the twenty-nine-year-old Parisian Victor Egger (1848–1909),14 in a rash of youthful enthusiasm, published a spirited attack on a successful book (it had many reprints in the years following its publication in 1876) by the widely renowned Paris neurologist Jules Bernard Luys (1828–1897). In this book, Luys—still looked up to nowadays as a pioneer in neurology, yet, apparently, a poor theorist—reduced the whole of psychology to neurological brain functions. For Luys, it is the brain that is happy, suffers, sees, hears, feels, understands and remembers: that is all there is to it—an early form of behaviourism, if you like, but more crass. In his review article, Egger makes mincemeat of the great professor’s theory. He starts out as follows (Egger 1877: 193): ‘Scientific psychology really is brain physiology’: that proposition is considered a commonplace platitude nowadays in the prefaces to treatises on the nervous system published by the most renowned doctors, who repeat in all kinds of phrases that the old method in psychology, internal observation, is worn out, idle, impotent. One is given to understand that one gets to know the soul not by reflection but by the autopsy of damaged brain tissue: the drill and the scalpel have dethroned consciousness; schematic figures replace the lists of faculties so dear to the older school. The soul is nothing more than ‘the sum of the functions of the brain’ and psychology no more than a chapter in the physiology book. To raise it to the level of a science worthy of the name, one has to proceed the way one does for the heart or the stomach. The problem is of the same order, the method is identical. The most eminent and most independent minds of the Paris School of Medecine have made themselves the champions of this idea. Why is this ‘idea’ wrong? For one thing, Egger says, organs are material, they have an ‘extension’, whereas desires, thoughts, passions etc. are immaterial, having no ‘extension’. The psychological functions of the brain are ‘sui generis’. Anatomy and physiology on the one hand and psychology on the other

14

Victor Egger’s father, Émile Egger (1813–1885) was professor of Greek literature at the University of Paris. He was also one of the very first to write on child language (Egger 1879; see Levelt 2012: 100, 105).

Structuralism, Rationalism and Romanticism in Psychology

217

“are i­ncapable of offering any help to each other; they have no choice but to ­pr0ceed in perfect mutual independence; they do not presuppose each other; an unbridgeable gap separates their objects” (p. 197). Luys’s psychology is nothing more than “the psychology of a peasant”: he is totally uninformed about the results of modern, nineteenth-century psychology. To say that “the brain feels, remembers and reacts” is gobbledygook because it is me who feels, ­remembers and reacts, not my brain. Luys says “When light impinges on the retina, the nervous element is touched in its intimate sensitivity, it erects itself, stops and pays attention”. But what do these phrases mean? What is the “intimate sensitivity” of a “nervous element”? How can it “pay attention”? (p. 201) Nothing but empty words. It is impossible to “explain psychology by means of physiology” (p. 204). Science cannot be made by means of metaphors. Both the physiologist and the psychologist must speak in the well-defined terms of their respective trades: mixing up the two vocabularies results in a Babylonian confusion of tongues (p. 208). Suppose, ideally, that the physiologist has exhaustive knowledge of the physiology of the brain, which has no secrets for him any more, will he then understand what it is to doubt, to believe, to know, to desire, to fear? Of course not. To say, with Herbert Spencer, that the two realities are but two faces of the same reality, solves nothing. That is a metaphysics built on words, not on truth, and, as Taine says, only one of the two “faces” is true, the one visible to the psychologist, since that is the first immediately given datum. The situation is like that of a piano, which produces sounds when played upon. Clearly, the hammer striking a string and the tone heard are facts of different orders. If you denote both by using the same word, your language is absurd and your thought confused (p. 210). A new science is thus needed, to be called “psycho­physiology”, which will specify how the mechanisms of the brain give rise to those of the mind (p. 211). Egger’s 1877 review article had quite an impact, as it circulated as a pamphlet among French intellectuals. What Egger describes here is the problem of how matter generates mind, or, conversely, how mental phenomena are to be reduced to physiological phenomena. Egger advises the physiologist and the psychologist each to proceed, at least for the time being, according to the methods of their own métier, leaving the other to his own devices. Meanwhile, he recognises that there is a huge problem to solve: the problem that the systems found in neurology are not at all like the systems found in psychology, as was thought by early materialist psychologists like La Mettrie and Gall and by naïve neurologists like Luys. Yet the mental systems can only work if there is an underlying neurological system: the mind needs matter to run on. In other words, strong or strict reductionism is untenable: the mind, though written with the ink produced by the brain, has its own text. This

218

Chapter 6

makes it n ­ ecessary to consider the question of how mental systems arise out of, or emerge from, or are supervenient on, physical token events. The problem has proved persistent. Bloomfield (1936: 89; quoted from Koerner 1989: 441) proves a staunch, yet naïve, supporter of strict reductionism, or ‘type physicalism’: Linguistics as actually practised employs only such terms as are translatable into the language of physical and biological science; in this linguistics differs from nearly all other discussions of human affairs. Within the next generations mankind will learn that only such terms are usable in any science. It was still very much alive when Jerry Fodor wrote his 1976 book The Language of Thought. There he formulated the problem as follows (Fodor 1976: 2): [i]nsofar as psychological explanations are allowed a theoretical vocabulary, it is the vocabulary of some different science (neurology or physiology). Insofar as there are laws about the ways in which behavior is contingent upon internal processes, it is the neurologist or the physiologist who will, in the long run, get to state them. (italics original) This, he argues, is unacceptable. To show that it is, he draws a parallel with systems of monetary exchange (Fodor 1976: 15–16). Physically speaking, money cannot be defined in a way that makes sense in terms of any physical science. Money consists in coins, paper bills, bank accounts as they are kept in banks (nowadays as digitalised files), etc. Taken together, these forms of monetary existence, or rather subsistence, do not constitute a category that makes sense in the terms of any physical science. Though token events of monetary exchange will always correspond to some physical token event or series of events, no physical science will ever be able to provide systematic descriptions and analyses of the nature or function of such monetary exchange events, or, more generally, of the laws or tendencies that are operative in economic systems. For that, we need an autonomous science, called economics, which, therefore, is a human, not a natural, science. In the same way, we need an autonomous psychology to analyse and explain psychological phenomena, and, we may add, an autonomous linguistics to do the same for linguistic phenomena. Yet no matter how strongly we may agree with this, and no matter how successful we have been in the meantime in setting up those autonomous disciplines, the problem of how brain processes actually produce the autonomous structures

Structuralism, Rationalism and Romanticism in Psychology

219

s­ tudied in psychology, linguistics and economics is as far removed from a solution as it was a century and a half ago. We still are at the point where the famous Paris professor Luys was in his book of 1876, witness the bestseller published by the Dutch neurologist Dick Swaab in 2014, under the telling title ‘We Are Our Brains’. There we read (Swaab 2014: 4): The product of the interaction of all these billions of neurons is ‘mind’. Just as kidneys produce urine, the brain produces mind, as Jacob Moleschott (1822–1893) so inimitably put it. But now we know what this process actually entails: electrical activity, the release of chemical messengers, changes in cell contacts, and alterations in the activity of nerve cells […]. Brain scans are used not only to trace diseases of the brain but also to show which areas light up during different activities, so that we know which parts we use to read, think, calculate, listen to music, have religious experiences, fall in love, or become sexually excited. All true, one hopes, but, pace Moleschott, urine is physical matter while the mind is not, as Victor Egger already pointed out with youthful fervour in 1877. So, how can the production of urine by the kidneys be the same kind of process as the production of the mind by the brain? Cohorts of researchers and legions of lay readers may still hold on to the belief that the production of urine by the kidneys and the production of the mind by the brain are of the same order, but, alas, they are befuddled by empty words. All we know, or think we know, is that cerebral functions somehow cause or generate mental phenomena, which is also all Swaab is actually saying, though his claims reach far beyond, implying that the mind has now given up all its secrets. But it has not, because, unlike in the case of our kidneys, gall bladder or thyroid, we haven’t got a clue as to the details of the causal processes involved in making brain functions produce plans or theories, or experiences of emotions, loyalty, love, beauty, understanding and all the rest, despite 150 years of research (the fact that Swaab quotes Moleschott, an almost exact contemporary of Luys, speaks volumes). These causal processes are still one big black box with physical inputs and correlated mental outputs—correlations we may use to influence mental states—but what happens in between is, to us, still sheer magic. The mind is at least as complex as Humpty Dumpty: all the brain’s ‘messengers’ and all the brain’s scans have been unable to put together anything remotely like the mind. The mind has proved to be the bogeyman of all science. Token physicalism is thus provisionally accepted, and only in principle, as no actual solution has been offered to the problem it stands for. We must, in

220

Chapter 6

other words, learn to live with it. The excuse for this theoretical leap is that the hypothesised mental systems, no matter what their neurological substrate may turn out to be, are empirically testable on account of socially validated subjects’ reports of assent or dissent, of acceptance or rejection, or of courses of action, which are taken to play the role that empirical data play in the natural sciences. Moreover, it is felt that the results justify the theoretical leap, as the predictive and explanatory power of the theories that have been constructed on this basis has proved to be so substantial that it cannot be ignored. 6.5

The Coming about of the Human Sciences

The background to the birth of the human sciences lies for the most part in the socio-economic, and the concomitant socio-psychological, changes that took place in Europe, and to some extent also in America, during the eighteenth century. These changes had a twofold effect. Among the well-off, who had become wealthy as a result of colonial trade and further economic developments, they fostered romanticism, a craving for recognition of private emotional experiences and for satisfaction of emotional needs. But among those who had not benefited from the new capitalism but had become its victims, they led to social unrest. Around 1750, the bourgeoisie in western Europe had grown wealthy to very wealthy. Naturally, the jeunesse dorée of the late eighteenth century then began to take a serious personal interest in both the hitherto only partially known geographical expanse of the earth and the hidden treasures of the past—an interest that was driven by the urge to have direct, physical contact with both the remotest parts of the world with its exotic denizens, and the material remnants of the often romanticised past.15 The rich bourgeoisie craved for an ­expansion, both in space and in time, of their experiential world. It was only 15

This romanticist craving for direct, physical contact with the most remote and impenetrable parts of the earth has meanwhile found a continuation in the present-day urge to explore space. It is true, of course, that the exploration of the space immediately surrounding planet earth and the concomitant massive launching of satellites have brought about the enormous technological revolution of the internet and of cheap instantaneous worldwide communication that has radically changed our lives within the time span of a mere couple of decades. Even so, however, the initial drive to engage in space programmes arose to a large extent from the non-profit-oriented, in fact still romanticist, urge to expand our experience and our knowledge of solar space. The same urge is felt to be sufficient reason for present-day governments to spend very considerable sums of public money in an effort to explore the more remote parts of outer space beyond our solar system, without a realistic prospect of economic returns. We just want to fathom

Structuralism, Rationalism and Romanticism in Psychology

221

a matter of time before this romanticist groundswell of living out experiences should translate itself into a demand for official and public recognition by governments, legislative bodies, universities and other public institutions.16 It was thus that wealthy private individuals started off on world trips. One thinks, as just one example, of the Prussian aristocrat Alexander von Humboldt (1769–1859), renowned younger brother of the equally renowned diplomat/philosopher/linguist Wilhelm von Humboldt. Alexander von Humboldt travelled widely in Latin America from 1799 till 1804 and his travelogues influenced a number of sciences in important ways (Wulf 2015). At a different level, E ­ uropean monarchs, governments and learned societies organised large  ­exploratory expeditions, such as those of Louis-Antoine de Bougainville (1729–1811), Captain James Cook (1728–1779), and Captain Robert Fitzroy (1805–1865), who led the famous round-the-world expedition on hms Beagle which lasted from 1831 to 1836 and inspired Charles Darwin (1809–1882) to develop his theory of the evolutionary origin of species. All such activities were directly inspired by the romanticist urge for the satisfaction and public recognition of emotional needs. In the same vein, academics and serious amateurs began to look at exotic languages that came their way in the context of colonialism or expeditions, and also at the material remnants of the great cultures of the remote past, seeking as much as possible direct, even physical, contact. It was thus that, around 1800, the parallel disciplines of historical linguistics and archeology came into being, more or less at the same time, as an outgrowth of the luxury afforded by the gains of colonial trade and the general expansion of the economy. They were followed by history, anthropology and psychology, which are thus all seen to have arisen from, or been given a new boost by, the romanticist urge to ­confer academic legitimacy upon (the study of) experiential ­phenomena in individuals and groups of individuals. As the century unfolded, it became more and more evident that if these new human sciences were also to gain academic respectability, in addition to legitimacy, the old romanticist ways of looking at things had to be replaced with more rationalist methods and ways of thinking. Despite the pressure from philosophers such as Dilthey, mentioned above, many practitioners of the ­human sciences thus decided to go over to the rationalist camp, though a minority remained faithful to romanticism. We thus witness a partial shift

16

both the universe and its ultimate origin—an entirely legitimate urge for the inquisitive animals we humans are. A conspicuous example of this still ongoing process is the gradual recognition of lgbt rights in the countries of the world.

222

Chapter 6

from romanticism to rationalism in those nineteenth-century human sciences that originated in romanticism. It was to a large extent that shift which gave rise to structuralism in the human sciences.17 The result was that, as from roughly 1850, the two currents of romanticism and rationalism existed side by side as opposing, mutually hostile, parties in those human sciences that had arisen in the luxury of the wealthy bourgeoisie, whereby rationalism was gradually taking over from romanticism. That ­process is still going on in our day, as one can see from the fact that certain practitioners of disciplines or subdisciplines within the meanwhile enormously ­expanded conglomerate of activities falling under the rubric ‘human sciences’ still foster antirationalist feelings and have still not given up romanticist—that is, antirationalist and thus antiformalist—sympathies and patterns of thinking. To the extent that these disciplines or subdisciplines are still affected by such antirationalist tendencies, they cannot be called ‘structuralist’, as they live by experiences, impressions and statistics, rather than by formally explicit, explanatory and preferably predictive models. As such, they are, in general, more useful for practical applications than for theoretical insight. There was, however, another side to the coin of luxury, to do with those whose task it was to sustain the inequitable system. The new social, technical and economic developments had given rise to the emergence or increase of, first, a class of peasants and servants enabling the privileged to maintain their status, and then, as mass industry began to prevail, of a huge proletariat in the industrial centres. These people did not travel to exotic places to shake hands with tribal chiefs nor did they feel the urge to dig up the remains of ancient civilisations. On the contrary, they scarcely managed to make ends meet and their conditions of life were, in many cases, intolerable. There will be no need to remind the reader that this chasm between the rich and the poor led, from the 1840s on, to great social unrest and to violent, often ugly, clashes in the big industrial centres and capitals of Europe, in some cases leading to revolutions. It was only natural that some academics felt the urge to chart and understand these processes from an academic point of view. This radically changed the course of economics, and was the origin of sociology in the nineteenth century. These two disciplines aimed at satisfying the much more practical urge for knowledge and causal understanding of the radical social changes that were shaking up the Western world at the time. They were not fuelled 17

As is explained in Section 6.7.1 below, this transition from romanticism to rationalism was the main underlying feature characterising the Young Grammarians’ revolution of the late 1870s in linguistics. Seen from this angle, the Young Grammarians may be regarded as ‘prestructuralists’.

Structuralism, Rationalism and Romanticism in Psychology

223

by romanticism, nor were they, at least in the beginning, much bothered by German Idealism. On the contrary, they were just matter-of-fact and thus fitted into the rationalist pattern right from the beginning. Sociology thus has basically different roots from anthropology, which was prototypically romanticist from the start and has meanwhile adopted more rationalist ways of thought only to a limited extent. That the socio-economic and the romanticist-inspired new disciplines came to meet each other in the court of the human sciences is in itself a curious coincidence, which should be seen in the light of the fact that they all circled around the human mind. Moreover, contrary to the natural sciences which were far too technical for the general public, they appealed directly to large circles of culturally interested readers. In addition, they did not need experimental set-ups and thus no laboratories (psychology labs are a more recent phenomenon which did not become widespread until the early twentieth century). By 1850, when universities began to be organised in faculties and departments, it would thus have been natural for the university administrators to separate the natural sciences, with their experimental laboratories, from the ‘others’, which then became known as the ‘arts’, ‘humanities’ or ‘Geisteswissenschaften’, according to local preferences. This is also why philosophy, not needing any laboratories, became united with the humanities, while mathematics, indispensable to the study of physical matter, was assigned to the natural sciences. And there is something else that united the romantically inspired and the socio-economic branches of the human sciences. In the course of the nineteenth century, they both more and more strove to adopt the methods and viewpoints current in the natural sciences. For those that sprouted from ­romanticism this meant a shift away from their romanticist origins in the direction of a more rationalist outlook. For the socio-economic branch it meant a continuation and reinforcement of an already existing orientation. Thus running along parallel lines, both the socio-economic and the ­romanticist-turned-rationalist disciplines adopted the ‘machine view’ of the human mind, whether literally or metaphorically, as I have tried to show in Section 6.1 above. This happened first with regard to human societies, in the work of Auguste Comte, mentioned in Section 3.1.2, who proposed that s­ ocieties, like the physical universe, should be seen as complex systems of i­ nteracting social mechanisms. Then, in Comte’s wake, the ‘machine’ view was applied to h ­ istory and to the human mind in the work of Hippolyte Taine, and subsequently to language by linguists such as Meillet, Sechehaye or the P ­ olish linguist Jan Baudouin de Courtenay (1845–1929), mentioned in Section 2.2 above. (Saussure does not belong in this list as he never developed any serious thoughts on a formal rule system for language, other than in historical terms.) The mind,

224

Chapter 6

b­ eing the ultimate repository of all the systems studied in the human sciences, thus came to be seen as a complex ‘machine’, to be revealed by hypothesis and experimental testing. It is easy to understand how this ‘machine’ view of the mind acted as a catalyst for what was to become structuralism in the human sciences. 6.6

Early Structuralism in Psychology: The Theory of ‘Gestalts’

In psychology, the first steps towards using ‘structure’ as a tool in explanatory theories were taken by Wilhelm Wundt—both in his Logik of 1880 and in his Völkerpsychologie of 1900—but no more than the first steps: one cannot say that Wundt was a structuralist in psychology. Perhaps surprisingly, Wundt was more of a structuralist in linguistics than in psychology, as he stressed the primacy of the sentence over the word (Levelt 2013: 217) and proposed actual immediate-constituent diagrams for propositions as semantic representations underlying surface sentences, as was shown in Chapter 4 (Wundt 1880: 57; 1900: 329). In doing so, he inspired both the American linguist Leonard Bloomfield (see Percival 1976) and Albert Sechehaye—though they assimilated Wundt’s teaching in very different ways. Properly speaking, structuralism in psychology was developed by others during the early twentieth century, partly in opposition to Wundt’s voluntarist approach, which claimed a role for active, voluntary selection in the association of concepts. In Europe, the first real representatives of structuralism in psychology were found in the school known as gestalt psychology, even though the gestalt psychologists did not call themselves structuralists, the term gestalt being their logo. This school was presaged in a famous article by the Austrian nobleman, philosopher, writer and highly controversial (racist) world reformer Christian von Ehrenfels (1859–1932), who, in Ehrenfels (1890), introduced the term Gestaltqualität (‘gestalt quality’), but it was actually founded by Max Wertheimer (1880–1943; see Wertheimer’s powerful article of 1923). The school was further developed by Wertheimer’s students and test subjects Kurt Koffka (1886–1941; see Koffka 1935) and Wolfgang Köhler (1887–1967; see Köhler 1929). A further influential gestalt psychologist was Rupprecht Matthaei (1895–1976; see Matthaei 1929), who later turned to the question of the recognition of colours. Gestalt psychology (the German word Gestalt means ‘overall form or shape’) is characterised by the insight that perception is intrinsically holistic in that it is innately driven to construct an interpretable whole out of isolated single elements impinging on the senses: the mind projects interpretations, called

Structuralism, Rationalism and Romanticism in Psychology

225

gestalts, onto impressions and sensations.18 Or, as Matthaei put it, using a terminology we now consider outdated (Matthaei 1929: 65): The parts of a perceptual field are put together in virtue of an inner autonomy (‘Gesetzlichkeit’). Its structure (‘Ordnung’) is determined by the essence of the whole, its solidity (‘Festigkeit’) by the coherence of the parts. The function of the separate parts (‘Teilgegebenheiten’) is directed towards the whole. Gestalts are meaningful. This point of view is now widely accepted (though stripped of the metaphysical and phenomenological clutter), but it is not yet widely appreciated that perceptual gestalt formation is just one manifestation of an inherent property of cognition in general: human cognition is, in the most general sense, innately programmed to interpret, integrate and understand in terms of maximally ­coherent models (see note 12 above on Leibniz’s coherence theory of truth). These need not all necessarily correspond to physical reality, in that they may be false or somehow thought up, but, in their totality, they add up to what we may call an all-encompassing model of the actual world and of all those intensional or virtual ‘worlds’ that we, thinking humans, create as we dream, imagine, negate, predict or plan things to do or to make. The mind actively moulds incoming new experiences (including thoughts that arise in the mind) into structural models of all kinds. The notion that the human mind is nothing but a passive recipient of impressions is fatally false. It is a centrally important general property of the human mind that it is innately driven, at both the conscious and the unconscious level, to work actively to integrate ad hoc transient new experiences first into an episodic situational understanding of the new experience, then into larger, more comprehensive funds of knowledge until, in the end, it reaches the most general knowledge fund which may be called the overall world model, deposited in long-term memory (in ways that we don’t as yet understand very well). There is thus a continuing circular feedback process from episodic interpretation to permanent world knowledge and again back from permanent world knowledge to episodic interpretation: interpretation consists to a large

18

This insight had been formulated earlier by the in many ways visionary British ­ sychologist-philosopher George Stout (1860–1944), referred to extensively in the previp ous chapter, who developed the notion of ‘gestalt’, but without the term, in Stout (1899). See, for example, Stout (1899: 490): “The isolated facts of sense-perception are made continuous with each other by interposing between them ideally represented links”.

226

Chapter 6

extent in a circular bootstrapping process, whereby the fuel is supplied by incoming new experiences or sensations, whether they come from the outside world or from the inner world of thoughts and emotions. This process is deeply engrained in human cognition, which uses all the rules in the book—even, if need be, outside the book—to make sense of sensations, signals or experiences that come in from the outside or the inside world. The fact that gestalt psychology, on the basis of serious scientific results, made the world aware of this integrative function of the mind in the field of perception is what makes this school of psychology stand out as an early, crucial episode in the general cultural process of gaining knowledge of, and insight into, ourselves as cognising animals. Structuralism in the human sciences is thus much more than a passing episode in the intellectual history of ­mankind: it is a crucial step in the process of understanding the workings of the mind. But back to the nitty-gritty of our story. Evidence for the gestalt thesis in psychology came from experiments, mostly carried out between 1910 and 1940, and many involving either incomplete or ambiguous drawings, which were interpreted by the subjects as meaningful wholes, such as faces, cubes or other known classes of objects, as illustrated in the figures 8 and 9 below.19 But acoustic perception shows the same. We recognise a melody regardless of the key in which it is set or the instrument on which it is played (Matthaei 1929: 4, citing Ehrenfels 1890), which would be impossible if all that is perceived is physical sound qualities. The same is true, obviously, for the perception-andrecognition of phoneme sequences, which is independent of voice quality and fully dependent on the hearer’s construal of the incoming speech sounds.

Figure 8 The Kanizsa triangles, after the Italian psychologist and artist Gaetano Kanizsa (1913–1993). One triangle is only partially drawn, the other is ‘not there’ at all, yet both are clearly perceived.

19

An interesting sequel to the classic gestalt experiments are the more recent experiments in optical illusions. See Robinson (1998), Kitaoka (2007).

Structuralism, Rationalism and Romanticism in Psychology

227

Figure 9 Two faces or a goblet? Developed around 1915 by the ­Danish psychologist Edgar Rubin (1886–1951).

The school of gestalt psychology flourished in German and Austrian psychology between the two World Wars. It disappeared during the Second World War, as many of its prominent practitioners were either Jewish or closely connected to the German-Austrian Jewish community. Those who made it to America met with a lukewarm response there due to the dominant position of behaviourism, though its practical applications, such as the Rorschach (or ‘inkblot’) test, were widely adopted in psychiatry. A different brand of (short-lived) psychological structuralism was promoted in America by the British-American experimental psychologist and Wundt’s student Edward Bradford Titchener (1867–1927), psychology professor at Cornell University (Levelt 2013: 217). He was not a gestalt psychologist but promoted structuralism as a psychological theory aimed at reducing consciousness to constituent elements of experimentally registered ‘primary sensations’, whereby the experiments were based on data obtained from test subjects according to the method of ‘trained introspection’. American behaviourism was partly a reaction to Titchener’s rather overstretched reliance on his preferred form of ‘trained introspection’ (see Blumenthal 1985: 73–74). 6.7

Early Structuralism in Linguistics

6.7.1 The Young Grammarians Although the Young Grammarians were not themselves structuralists, an adequate account of the rise of structuralism in linguistics, and of Saussure’s position in this process, requires an understanding of their doctrine and of their place in history. One needs to understand how and why this remarkable ­nineteenth-century school of linguistics, and Saussure with it, failed to link up with twentieth-century theoretical linguistics. For that reason, I devote some

228

Chapter 6

space to this school, whose echoes are still resonating (see also Seuren 1998: 89–104). The Young Grammarians’ revolution, and their subsequent school, have already been mentioned a number of times. We have seen that Saussure, having studied at Leipzig, the epicentre of the Young Grammarian’s revolution, remained faithful to their doctrine till the end of his life, despite serious objections voiced by many, in particular by Hugo Schuchardt (1885), but which Saussure did not take into account and to which he did not react. What we have not seen is why the Young Grammarian movement was perceived as a revolution, not only by the initiators and their followers but also by the wider public. This is a point to be looked at, since a detailed study of the actual issues involved fails to show up the radical break with the past that the initiators claimed to have brought about. In actual fact, one finds many of the insights advertised as ‘revolutionary’ by the Young Grammarians already breaking through in the works of those they attacked and even earlier. If we look at the actual issues discussed, what we see is a gradual development rather than a radical break (Jankowsky 1972: 12–14). The ‘revolutionary’, and hence controversial, aura around the Young Grammarians was more a result of their noisy and unruly juvenile behaviour and their rhetoric than of a radical academic innovation, a real ‘paradigm change’. In the end, however, these angry young linguists turned out to be extremely serious and highly competent professionals. In this respect they differed from many other ‘angry young’ movements. The Young Grammarians’ ‘revolution’ was, in actual fact, nothing but a phase in the gradual process of emancipation from a largely romantically inspired historical linguistics to a more rationalist science, subject to stricter rules of method, argumentation and explanation. The romanticist spirit in which the ‘old guard’ did their work—August Schleicher, for example, the main butt of the Young Grammarians’ attacks, was a staunch Hegelian (Jankowsky 1972: 100–101) and thus, like Adolphe Pictet, belonged to the romanticist camp— gave way to a rationalist spirit: the Young Grammarians were the linguistic embodiment of the shift from romanticism to rationalism in those new ­human sciences that had their origin in romanticism, as explained in Section 6.5 above. The enormous upheaval caused by this ‘revolution’ thus seems to have been more a question of mentality than of substance. If so, it shows how high emotions could run when romanticism and rationalism clashed. But let us now look at what the Young Grammarians actually achieved and on what counts they have been, or can be, criticised. They cannot be labelled as structuralists, as they were mainly interested in linguistic change, less in linguistic systems in the sense that term is used today. Then, within the area

Structuralism, Rationalism and Romanticism in Psychology

229

of linguistic change, their main interest was in phonological change, though some, in particular Franz Bopp (1791–1867) and, later, Brugmann and Delbrück (but not Saussure), also carried out extensive studies of morphology and syntax, which they took to be necessary for the determination of (degrees of) cognateness.20 What sparked their movement was, in large part, the discovery that many apparent irregularities in the sound laws that had been proposed until then were, in fact, not irregularities but could be seen to be fully regular if more factors were taken into account. A well-known example is the correction to Grimm’s law by the Danish linguist Karl Verner (1846–1896). Jacob Grimm (1785–1863), elder brother of Wilhelm Grimm—both world-famous for their classic collection of German fairy tales—thought he had discovered, following earlier discoveries by the Dane Rasmus Rask, a regular pattern in the alternation between Ancient Greek voiceless plosives (p, t, k), the corresponding voiceless fricatives (f, θ, χ) in Old Germanic and the corresponding voiced plosives (b, d, g) in modern High German. He saw a parallel correspondence in Ancient Greek (b, d, g), Old ­Germanic (p, t, k) and modern German (f, z, χ), and again in what he took to be Greek voiceless fricatives (f, θ, χ), Old Germanic (b, d, g) and modern German (p, t,  k). This became known as Grimm’s law, or the Second Germanic Sound Shift. Grimm saw in this a wonderful circular movement (‘Kreislauf’) in the three languages concerned. He was, however, soon disabused of this illusion, as other linguists were quick to point out (i) that what he had taken to be voiceless fricatives in Ancient Greek were not fricatives at all but aspirated voiceless plosives (ph, th, kh), and (ii) that even with this fault corrected, there were large numbers of exceptions. 20

The German philologist Franz Bopp already showed a vivid awareness of grammatical phenomena in his extensive comparative work on the grammars of Indo-European languages, with Sanskrit as the central point of reference. Brigitte Bauer drew my attention to the fact that Berthold Delbrück took an interest in areas he saw as pertaining to syntax, witness his three volumes on the comparative syntax of the Indo-European languages (Delbrück 1893–1900). These three volumes are, respectively, the third, fourth and fifth volumes of the overarching work Brugmann & Delbrück (1886–1900), a total of, if I am right, roughly 6500 pages, together called ‘a concise account’ (‘kurzgefasste Darstellung’). In Part I of his Vergleichende Syntax der indogermanischen Sprachen, published in 1893, on pp. 3–88, Delbrück gives a very learned and informative but largely uncritical and superficial account of his theoretical views on, and the history of, the study of grammar. If it had been only for their work on syntax, the Young Grammarians would not have left a mark on the history of linguistics. Their positive contribution lay first and foremost in historical phonology and, to a lesser extent, in historical morphology, not in syntax or grammar as a whole.

230

Chapter 6

In his article of 1875, Karl Verner eliminated one class of apparent irregularities consisting in the fact that Greek intervocalic t sometimes corresponds to modern German d, as in German Ader (‘vein’) versus its Greek cognate ễtor (‘heart’), which is in accordance with Grimm’s law, but sometimes to German t, as in German Vater versus Greek patếr (‘father’), which violates Grimm’s law He showed that this is not irregular but regular, since the modern German intervocalic t occurs regularly when the following syllable carries the original word accent, as in patếr, whereas d occurs when the original word accent falls on the preceding syllable, as in ễtor. This has become known as Verner’s law.21 This in itself was hardly revolutionary and may be seen as a normal step forward in a beginning discipline. Yet it gave a group of young linguists at Leipzig and elsewhere the idea that regularity in sound change was the main thing to strive for in as thoroughly systematic a way as possible. This was not new, as it had been the main, though mostly implicit, guiding principle of all IndoEuropeanists since the early nineteenth century. What wás new was that the Young Grammarians made this principle explicit and placed it in the foreground, thereby latching on to the rationalist ideas and methods in the physical sciences, where indeed the search for regularity was, and is, a paramount methodological principle. Somewhat overconfident, to put it mildly, they thus posited the general principle of the exceptionlessness of sound change (‘die Ausnahmslosigkeit der Lautgesetze’), meaning that sound changes within one language or from one language to another are identical for any individual sound (phoneme) in well circumscribed (phonological or phonetic) environments during a given period and for a given language or dialect. I cite three classic passages from Osthoff and Brugmann’s ‘Vorwort’,22 known as the Young Grammarian Manifesto, to the first volume of the journal Morphologische Untersuchungen which they had founded (Osthoff & Brugmann 1878): It cannot be denied that the older linguistics approached their object of investigation, the Indo-European languages, without first forming a clear idea of how human speech really lives and develops, which factors are at work in speech, and how these factors, in concert, bring about the 21 Saussure’s Mémoire of 1879 was a further step forward along the same path, in that he eliminated a number of apparent irregularities in the Indo-European vowel system by positing the occurrence, in certain positions, of ‘laryngeal’ consonants that have since disappeared. As was said in Section 1.1, this hypothesis was dramatically confirmed when the Polish linguist Jerzy Kuryłowicz showed, in his 1927 article, that laryngeals did indeed occur in the predicted positions in Hittite. 22 Though, according to Brugmann (1900: 131–132), he was, in fact, the only author; Osthoff only made a few stylistic improvements and put his name to it.

Structuralism, Rationalism and Romanticism in Psychology

231

­ roduction and change of the material of speech. It is true that languages p were studied with great zeal, but the person who speaks was taken into account far too little. (p. iii) […] Only those comparative linguists who manage to leave the hypothesisladen atmosphere of the workshop where the Indo-European root forms are wrought and to enter the clear air of tangible actual reality so as to gain insight into things that grey theory will never show, only those who once and for all say goodbye to the once widely spread but nowadays still existing method that makes one consider language only on paper,23 that makes everything submerge in terminology, formulae and grammatical schemata and makes one believe that one has understood the essence of the phenomena as soon as a name for the thing has been invented,— only those can achieve an adequate picture of how linguistic forms live and change, and master those methodological principles without which no plausible results can be obtained in historical linguistic research and without which, in particular, any progress into the eras that lie behind the historically attested linguistic changes will be comparable to a sea voyage without a compass.24 (pp. ix–x) […] First, every sound change, in so far as it is mechanical, proceeds according to exceptionless sound laws. That is, the direction of the sound movement is identical for all members of a speech community, except when dialect split occurs, and all words in which the sound affected by the change occurs in similar circumstances are comprised in the change without exception. ( We are speaking, of course, only of mechanical sound change, not of certain forms of dissimilation or metathesis, which are founded in the idiosyncrasy of a single word and are always the physical reflection of some psychological movement. These have no bearing on the concept of sound law.) (pp. xiii–xiv) Thus even for them, sound change was not absolutely without exceptions, as they immediately added a few escape clauses, such as borrowing from one 23 24

I understand this to be a jab at Jacob Grimm, whose ‘Grimm’s law’, as shown above, was based on written letter signs, not on the corresponding sounds. The promise made here by Brugmann and Osthoff that they, as opposed to their predecessors, would study language as it is used in ‘real life’ was probably meant to appease their still largely romanticism-inspired audience (the same claim or promise is used again today by antirationalist, cognitivist linguists). In actual fact, however, that promise was never made good on: the works subsequently published by the Young Grammarians hardly show any concern with language as it is actually used in daily life. They themselves were at least as formalist as they accused their opponents of being.

232

Chapter 6

l­ anguage into another, or ‘dissimilation’ between two identical liquids in rapid succession, as in Latin caeruleus (‘skye-blue’) from caeluleus, derived from caelum (‘sky’). The most important escape clause, however, consisted in so-called analogy, a widely used concept in the linguistic theory of the day, standing for the suppletion of a missing term in an unresolved morphological equation on the analogy of a corresponding fully resolved equation. For example, Latin accented a in an open accented syllable regularly becomes French è (sometimes written ai): Latin mare (‘sea’) becomes French mer, Latin frater (‘brother’) becomes French frère, Latin manus (‘hand’) becomes main, etc. Yet Latin lavo (‘I wash’) does not become French ( je) *laive, but ( je) lave. Why this exception? The answer lies in analogy, which is, in principle, a sacrifice of the diachronic to the synchronic. The three singular forms and the third person plural of the French verb laver (‘wash’) follow the first and second person plural forms (nous) lavons (‘we wash’) and (vous) lavez (‘youpl wash’), where the vowel a is regular, since the corresponding Latin forms lavamus (‘we wash’) and lavatis (‘youpl wash’) have the accent on the second syllable.25 Since it is regularly so that the stem vowels of French present-tense verbs are the same for all persons, whether singular or plural, as in ( je) chante (‘I sing’) and (nous) chantons (‘we sing’), the incorrect form ( je) laive would be both regular and irregular: regular from a diachronic phonological, but irregular from a synchronic morphological viewpoint, whereas the correct ( je) lave is regular from a synchronic but irregular from a diachronic point of view. French speakers thus decided to sacrifice the diachronic to the synchronic and opted for ( je) lave. In the spirit of the day, this analogical process was called psychological, around 1880, not grammatical, whereas in fact it is both, and, in addition, also sociological (or else we could not speak of ‘correct’ and ‘incorrect’). We thus have the following two equations, the resolution of the second ­being determined by the first:26 (5.1) a. cantamus : chantons = canto : chante b. lavamus : lavons = lavo : x

25 26

Why the second, accented, a in Latin lavamus (‘we wash’) does not become è (or ai) is a different story, to do with analogical developments around the ending of first person plurals in French verb morpholoy (see, for example, Groult 1947: 197–198). Both Saussure (Cours: 222) and Sechehaye (pmlt: 188) speak of ‘proportional fourth term’ (‘la quatrième proportionnelle’).

Structuralism, Rationalism and Romanticism in Psychology

233

Clearly, according to this pattern, the x in (5.1b) is the correct lave, not *laive. But, this being so, one would expect the analogical adaptation always to ­proceed from the first (and second) plural forms to the others. But this is not what is found. Take the French verb aimer (love) from Latin amare. In this case, the diachronic sound law predicts ( j’) aime, (tu) aimes, (il) aime and (ils) aiment, versus (nous) *amons and (vous) *amez, whereas in actual fact French has (nous) aimons and (vous) aimez. Here the analogical process of adaptation has gone the other way: (5.2) a. canto : chante = cantamus : chantons c. amo : aime = amamus : x According to this pattern, the x in (5.2b) is indeed the correct aimons, not *amons. But the patterns in (5.1) and (5.2) are different: (5.1) goes from the first (and second) persons plural to the other persons, whereas (5.2) goes the ­opposite way. This complication was not taken into account by the Young Grammarians (nor, as far as I am aware, by their critics). Immediately upon the publication by Hermann Osthoff and Karl Brugmann, in 1878, of their so-called ‘Young Grammarian manifesto’ mentioned above, the principle of the unexceptionability of sound change came under heavy fire. The most prominent critic was Hugo Schuchardt, who pointed out—with a certain passion—that the restriction to given periods and given languages or dialects was sufficient to discredit the Young Grammarians’ principle of absolute regularity, since the notions of ‘period’, ‘language’ and ‘dialect’ are fluid: neither periods nor languages nor dialects have precise boundaries (Schuchardt: 1885). Schmidt (1872) had in fact shown, with Schuchardt’s help, that sound changes are like stains, each spreading primarily over a certain area in the basic vernacular or dialect, and only secondarily in the official standard language, if any is available. The boundaries of such areas are commonly called isoglosses,27 which may differ considerably from one sound change to another. As there usually is a lively trade between standard languages and dialects, sound changes may reach the standard language incompletely, leaving parts of the lexicon unaffected. 27

The term isogloss stems from the Latvian-German dialectologist and folklorist August Bielenstein (1826–1907). It was introduced as a calque of isotherm in Bielenstein (1892: 397) and rapidly gained acceptance in dialectology. Saussure (Cours: 277, 281–289) mentions the term but does not like it, still preferring ondes linguistiques or ondes d’innovation, both translating Schmidt’s Wellen (‘waves’).

234

Chapter 6

No modern linguist still upholds the principle of ‘Ausnahmslosigkeit’ in the strong sense in which it was proposed by the Young Grammarians. Even the Young Grammarians themselves, as they grew older, eventually admitted that the principle was flawed. It is useful as a working hypothesis, not as a statement of fact, since, as a working hypothesis, it puts an assumed regularity first and then makes one look for counterexamples, which is good scientific method. Nowadays it is held that there are many possible sources of irregular sound change. The following list, though no doubt incomplete, will give an idea (see also Bloomfield 1933: 386–391; 404–424): – interference of social class, dialect or register (formal versus colloquial, etc.), diglossia, the wish to be peceived as different or even the wish not to be understood (see Chapter 3, note 1); – language or dialect mixing, gradual transitions, substrate or superstrate influence, borrowings; – geographical shifts of a political or cultural power centre affecting the standard language; – onomatopoeia or sound symbolism (see Section 3.4.4); – morphological paradigm formation (the Young Grammarians’ ‘analogy’); – phonetic dissimilation (e.g. Latin caeruleus, ‘sky blue’, from caeluleus, from caelum, ‘sky’); – frequency (e.g. French très, ‘very’, irregular from Latin trans, ‘across’; English very for verily); – normative influence from schools; – folk etymology; e.g. English outrage, not from a supposed out+rage, as is widely thought, but from Vulgar Latin ultraticum, French outrage; the regular pronunciation would probably have been [′ʌtrıdž], like English courage ['kʌrıdž] from French courage (for the change from French ou to English [ʌ], cp. Engl. turret < Fr. tourette, Engl. crust < Fr. crouste, Engl. supper < Fr. souper, Engl. double < Fr. double, Engl. dozen < Fr. douzaine, Engl. mutton < Fr. mouton, etc.); – spelling pronunciation (e.g. [′oftǝn] for [′ofǝn], English often; [′mʌzlım] for [′muzlım], Muslim); – isolated idiosyncrasy (e.g. Portuguese erro, ‘error’; see Section 3.6 above). Although the Young Grammarians knew that a language is not a physical object, subject to physical laws, but a social institution regulated by a community according to its internal echelons and other subdivisions, and thus inevitably with manifold internal variations, this was not something that captured their interest (see note 24 above). They professed adherence to the principle that a language resides in a community, but they obviously did not draw the consequences from that principle, mainly because they were mesmerised by

Structuralism, Rationalism and Romanticism in Psychology

235

the notion of law as current in the physical sciences, which they were bent on emulating.28 Their critics were right, therefore, in rejecting the principle of the unexceptionability of sound changes. But does this mean that there are no ‘sound laws’ at all, that all is madness without any system in it? Clearly not. After all, there is a system in the madness. We do find clear regularities in sound change, which were described and analysed with great acumen by the linguists of the time. One cannot maintain that the entire machinery of Indo-European sound laws is mere fiction. So where do these regularities—I purposely avoid the term laws—come from? The answer was given in Section 3.6 above: regular sound changes reflect changes in the underlying system, while exceptions are due to a variety of factors, including the internal variability in each speaker’s internalised language system, their linguistic competence. The Young Grammarians may still not have seen explicitly the link between language change or variation on the one hand and language system on the other, their use of the notion of analogy as illustrated above shows that the notion of a synchronic system was beginning to germinate in that school. Yet it cannot be said that the Young Grammarians were structuralists, not even in spirit, since there is nothing remotely like a notion of putting elements together into larger wholes according to the instructions provided by a system of rules. They mostly failed to place regular linguistic changes in a wider systematic context. And when they did see a system in diachronic changes, it was never a system producing structures but rather a collection of (mostly phonetic) regularities, such as an overall tendency to palatalisation, or an overall loss of fricatives. Saussure went so far as to deny that there ís a system to diachronic changes (Cours: 134): ‘Diachronic facts are […] isolated and do not form a system among each other’. 6.7.2 Who Were, and Are, the Real Structuralists in Linguistics? The Young Grammarians were thus not, or not yet, structuralists, whether in fact or in name. But who were the real structuralists? We know who wished to be called structuralists: they were the ones who brandished Saussure’s name. But is this group coextensive with the group who actually did structuralist work during the mid-twentieth century? I have already intimated that this is not so. 28

The same holds for Saussure, who frequently mentions the social nature of language in the Cours but fails to draw the obvious consequence, namely that the system of each language must be taken to be internally variable, according (at least) to the position taken by the speaker on the parameters of geographical area, social scale and type of interaction (formal, informal, etc.)—the latter two usually being subsumed under the term ‘register’.

236

Chapter 6

But let me stress first that the label ‘structuralism’ is applicable in different ways in different branches of the conglomeration of disciplines falling under the general term ‘linguistics’.29 Speaking of ‘structuralism’ in phonology makes sense to the extent that there is a system behind phonological interpretations of phonetic sounds in each specific language, but the fact that such phonological systems produce, or generate, phonological structures is less pronounced. Phonologists do study accent assignment, syllable structure or vowel harmony in words, phrases or sentences, but—despite the fact that structuralism in linguistics started with phonology, during the 1930s—the structuralist element is much less prominent in phonology than it is in morphology or syntax, which revolve around the structural properties of the products of the grammatical system, that is, sentences, phrases and words. There is also a vast body of literature on structural(ist) semantics. What is known as ‘structural(ist) semantics’ studies mainly relations between word meanings in semantic fields, also called ‘word fields’ (‘Wortfelder’), distinguishing ‘semantic features’, whose configurations or combinations are represented as analytical representations of word meanings and their relationships in a given semantic, cognitive or factual ‘sphere’. One often speaks of ‘componential analysis’ in this context. As conspicuous examples of this approach we may mention Trier (1931), Lyons (1963), Fillmore (1982) with his ‘frame semantics’, but many other names and publications could be cited. This literature, no matter how relevant and how important, is only ‘structuralist’ in the sense that it detects structure within and between lexical meanings, but the question of what sort of ‘system’ could possibly be envisaged that would generate such ‘semantic structures’ remains in the background or is not taken into consideration at all. Only some of the authors in this area of studies claim a Saussurean lineage, Lyons being a prominent example. Other, equally structuralist, authors consider Saussure irrelevant. These do not, on the whole, call themselves structuralists. Formal semanticists, mainly in the tradition of Richard Montague’s model-theoretic or possible-word semantics (Montague 1973) or of Kazimierz Ajdukiewicz’s Categorial Grammar (Ajdukiewicz 1935), are bent on developing formal, strictly compositional, grammatico-semantic systems that take as input surface sentences of natural languages and produce as output logicosemantic propositional structures that are testable for truth or falsity in given models (their ‘interpretation’). These attempts are ‘structuralist’ in the full 29

In pragmatics the notion of structuralism is hardly alive: there is no branch of studies known as ‘structuralist pragmatics’, perhaps because of the romanticist origin of that field of studies.

Structuralism, Rationalism and Romanticism in Psychology

237

sense of the term, but the practitioners of this branch of formal linguistics would balk at being called structuralists, just as they would reject any association with Saussure. It is in grammar that it makes most sense to speak of structuralism in the factual, not the flag-carrying, sense. Here it must be observed that there was a great deal of structuralism already implicit or covertly present in linguistics as it had developed over the centuries since the first Alexandrian grammarians during the third century bce. So-called ‘traditional grammar’ has always been, to a large extent, ‘structuralist’ even though that fact had never been noticed explicitly and no specific term had been coined for it. This structuralist streak in traditional grammar appears mainly from the simple and obvious fact that traditional grammar has always lived by what used to be called ‘parsing’,30 the technique of assigning constituent structure to sentences and of assigning linguistic elements (words) to classes according to the functional positions they can fill in any given sentence or sentence constituent. What we now perceive as structuralist grammar is, and should be seen as, a specifically structuralist, continuation of traditional grammar. It is impossible to present a representative list of grammarians that were, or are, structuralist in this sense. All I can do is mention a few outstanding authors who wrote detailed and systematic grammars of various languages. For English, one thinks in the first place of Otto Jespersen and his monumental Modern English Grammar on Historical Principles (1909–1949). But there is also A Grammar of Contemporary English (Quirk, Greenbaum, Leech and Svartvik 1972), and, more recently, The Cambridge Grammar of the English Language (Huddleston and Pullum 2002). For German, one thinks first of all of Gunnar Bech’s Studien über das deutsche Verbum Infinitum (Bech 1955, 1957), but there are also, for example, Heidolph, Flämig and Motsch (1981), Eisenberg et al. (1995), Zifonun et al. (1997), all monumental in their own specific ways. For French we have—besides the glorious eighteenth-century tradition that started with Claude Favre de Vaugelas in the preceding century—the classic Grammaire du français contemporain (Chevalier et al. 1964) and, of course, the incomparable Le bon usage (Grevisse & Goosse 1986) and a large number of other publications. The Dutch equivalent is the Algemene Nederlandse Spraakkunst (Geerts et al. 1984), next to Klooster (2001), but the singular Piet 30

The English verb parse [pa:z] is a late medieval derivation from the Latin noun pars (‘part’, pronounced [pa:s]), originating with English students in the ‘trivium’, who had to identify the partes orationis (‘parts of speech’) in given texts. In deriving the verb from the noun they simply followed the then productive rule of voicing final voiceless fricatives, as in house [haws] → house [hawz], calf [ca:f] → calve [ca:v], bath [ba:θ] → bathe [béjð], breath [brɛθ] → breathe [bri:ð].

238

Chapter 6

Paardekooper, with his concise syntax of Standard Dutch of 1963, besides many other studies, must also be mentioned. Similar works can be cited for all other European languages, next to, of course, the hundreds of studies produced over the past few decades on faraway languages in distant continents. Most of these works have, in different degrees, a distinct traditional flavour, which does not make them less structuralist in the notional, not the brand-name, sense of the term. Nowhere in any of these studies is any indebtedness claimed to Saussure, whose influence never extended into the actual practice of grammar writing, even if this practice is, in actual fact though not in name, of an outspoken structuralist nature. 6.7.3 Romanticist or Nonstructuralist Grammar? One could, conceivably, think of linguistic structures, and hence of linguistic structuralism, also in a romanticist, nonrationalist, context. This, however, has not so far materialized. Numerous attempts have been made, especially during the late nineteenth and the first half of the twentieth century, to provide explanations, mostly of a psychological nature, for linear word (or constituent) order and for intonational phenomena, but no clear notion of structure has so far arisen from such analyses. Attempts have been made to explain the collocation of words and intonation patterns in sentences in terms of ‘movements of the mind’ (‘Geistesbewegungen’) experienced as ‘what presents itself first to the mind’ and other phrases of that nature. The problem with such theories, apart from their obvious circularity, is that they fail to specify actual structures. Depending on the specific variety of psychology used—or simply in reliance on intuitive folk psychology—‘explanations’ for (usually unsophisticated) syntactic observations are proposed that are not only nonstructural but also imprecise, untestable, nonpredictive, unverifiable and unfalsifiable. They moreover fail to take into account plain facts of syntax that cannot possibly have anything to do with ‘movements of the mind’, such as those exemplified in the sentences (3.1) to (3.4) of Section 3.3. A truly romanticist structuralism thus never came off the ground. But how about other possible or thinkable nonpropositional and nonstructuralist analyses and decriptions of sentences? Here, too, attempts, or rather claims, have been made, sometimes complete with a purported philosophical foundation. I will mention three representative examples, Carl Svedelius (1861–1951), Theodor Kalepky (1862–1932) and Manfred Sandmann (1906– 1980), all three R ­ omance philologists by trade (one wonders why this should be so) and all three, each in their own way, denying the linguistic reality of the subject-­predicate division which, according to them, belongs in logic, not

Structuralism, Rationalism and Romanticism in Psychology

239

in ­linguistics.31 Taking advantage of the then prevailing hopeless confusion regarding the notions of subject and predicate and of the lack of any proper notion of syntax, which was considered to belong to the study of speech, not of the language system, these authors explicitly decided to do away with these notions altogether, but what they put in their place does not amount to much. Svedelius (1897) seeks refuge partly in terminological escapes, speaking of a terminus a quo, which amounts to what is commonly known as subject, terminus ad quem for object, and process (‘procédé’) for predicate. Partly also, he seeks relief in the concept of ‘communication’, but remains stuck in generalities and fails to propose a clear alternative. As regards word order, he says (Svedelius 1897: 49): ‘The order of linguistic units follows the order of thought’, a hypothesis that is not only circular and untestable but also highly unlikely when one considers, for example, cases like (3.1) to (3.4) in Section 3.3 above. Kalepky (1928), a work that arose out of despair over the fruitless subjectpredicate debate, claimed to present a new, truly scientific, theory of ­syntax without the subject-predicate division. But what this theory consists in remains opaque. With a great deal of benevolence, one may perhaps recognise a proposal that approaches a notion of grammar in terms of what are now called ‘thematic roles’: ‘bearer’, ‘undergoer’, ‘actor’, ‘goal’, ‘origin’, ‘tool’, all taking part in an overall ‘process’, but as there is not even an attempt at further elaboration, one remains in the dark about the details. Leaving aside the question of whether such notions have the potential to make for a serious theory of syntax, the proposal as such fails as a theory, owing to a total lack of specification, criteria, empirical testability and prediction, and an abundance of rhetoric and cheek.32 In his booklet of 105 small pages, Kalepky rejects traditional grammar in toto. All work in grammar of the past two millennia is worthless: there are no word classes, nor is there a distinction between main and subordinate clauses or between subject and predicate. Speaking is based on (Kalepky 1928: v): ‘the analysis of a content-of- conscience [‘Bewußtseinsinhalt’] pre-existing in the speaker’s mind, with concomitant conceptual subsumption of the parts found therein’—whatever that might mean (the notion of ‘Bewußtseinsinhalt’ ­remains totally undefined). A sentence is (Kalepky 1928: 8): ‘the smallest piece of communication offered by the speaker as rounded off in itself—or, shorter: a sentence is the smallest unit of communication’. We know, of course, that

31 32

See Elffers (1991: 294–312) for an excellent and more complete survey. In Kalepky (1928: 55), he likens his own effort to show Aristotle wrong as regards grammar to Galileo’s effort to show Aristotle wrong as regards physics.

240

Chapter 6

the notion of sentence is hard to define (see Section 3.3.2), but the ‘definition’ offered by Kalepky is, to put it mildly, not among the best. Sandmann (1954) is a very erudite work, full of history, philosophy, literary quotes and allusions, but, again, it fails to provide even the beginning of a theory of grammar, let alone actual structure assignments that could be of help in setting up a theory of syntax. These works thus do not provide any alternative way of assigning structures to natural language sentences and may thus be ignored in the present context. Other publications by other authors may be quoted, but the story simply repeats itself: in none of them does one find a viable nonpropositional but still structuralist alternative to the mainstream subject-predicate structure assignment in whatever form. Worse, all of them reject subject-predicate structure assignments without defining what could be meant by the terms subject and predicate, then to propose new concepts which remain equally undefined, thus creating a muddy arena where anything goes. In sum, this brand of linguistics is nothing but populism: the established order is attacked on purely rhetorical grounds and what is proposed to take its place may sound attractive to a naïve audience but is, in actual fact, ill-defined, ill-founded, ill-researched, ill-informed and ill-directed. Accordingly, authors such as those mentioned above have had little or no response in linguistic theory, which is as it should be. 6.8 Summary In this chapter, we have made the rounds of the various intellectual and cultural trends, movements and problem areas that were objects of debate during Saussure’s and Sechehaye’s lifetimes, thereby providing an insight into their world, the choices they faced and the positions they took with regard to them, if any. Writing such a chapter is not an easy matter, given the complexity of the issues themselves and of their interrelations and, often also, the vagueness of their boundaries. Overall cultural portraits of a period are among the most difficult to paint. Yet the exercise is necessary for an adequate rendition of the personae involved. We have seen how structuralism evolved during the early twentieth century and how it largely failed to affect Saussure, while Sechehaye was actively catching on to it, though not the way other linguists were, as he took the mental and psychological embedding of language more seriously than most linguists did and had the courage to stand by the notion of proposition. We have looked at the question of the scientificity of the human sciences in general and linguistics in particular and, consequently, at the problem of if and how matter can

Structuralism, Rationalism and Romanticism in Psychology

241

be seen to engender mind—a live issue in those days as it still very much is today. We have sketched the coming about of structuralism in psychology, in the shape of gestalt psychology, and in linguistics. And we have attempted to give an idea of how the school of the Young Grammarians, to which Saussure belonged, can be seen as a transition from a, mostly implicit, romanticist way of thinking about language to more systematic rationalist methods and approaches—all this in the hope that it will not only give more depth and relief to our protagonists but also place them in a wider perspective.

Chapter 7

Conclusions Now that we have come to the end of this study, what are our conclusions? They are of two kinds. First, there are the conclusions that pertain directly to the history of twentieth-century European linguistics. Secondly, and in direct relation to these, there are conclusions of a general theoretical nature, as the careful inspection of the questions at issue has revealed aspects and perspectives that have either been allowed to rest undiscussed, or have never been taken notice of—a fact that gave occasion to the subtitle of the present study. Looking at the historical conclusions, we note first of all that the close scrutiny carried out in the Chapters 2 and 3 will have made it clear that the title of ‘father of European structuralism,’ universally bestowed upon Ferdinand de Saussure over the past century, is not justified by the historical facts. On the contrary, once divested of the glamour clinging to him after a century of mythification, Saussure appears to us as a man whose real forte lay in IndoEuropean historical linguistics but who wanted it to lie in the general theory of language—an ambition that ended in failure. By contrast, Albert Sechehaye, his junior by thirteen years and his subordinate in the University of Geneva, now stands out as a powerful and independent thinker, who, without getting carried away by fashionable but superficial trends, developed ideas that have proved substantial and even prophetic. Other than Saussure, Sechehaye was denied recognition, owing, it seems, in part to his subordinate position in the ranks of his university, but probably more than anything else to the fact that his contemporaries failed to see, yet suspected, the force and the wide bearing of his thoughts, which were far more solid and substantial than theirs, with the result that they will have felt uncomfortable in his presence. I think I have shown the extent to which Saussure was out of touch with what was happening in linguistics in his day and the narrowness of his i­ntellectual horizon, while making it clear at the same time how Sechehaye differed from Saussure in these respects. Whereas Saussure never ventured beyond the philosophical and science-theoretical ideas instilled into him by his family, teachers and personal acquaintances, and never carried out any original grammatical analysis (though, of course, he gathered fame on the grounds of his youthful masterpiece on the Indo-European vowel system), Sechehaye analysed a number of specific syntactic constructions, both in child and in adult language, against the background of his general theoretical insights which he developed for himself, deriving much inspiration from Wundt and from the

© koninklijke brill nv, leiden, ���8 | doi:10.1163/9789004378155_008

Conclusions

243

notion of proposition placed at the centre of the cognitive mind in general and of language in particular. It has been shown, in the preceding pages, that of the ideas that Saussure has become famous for, all but one had been well known and widely discussed before him, while the one that was not—his notion that all in language is negative and oppositional—was obscure and vague, and has proved fruitless. By contrast, Sechehaye foresaw the notion of a grammar as an algorithmic procedure converting, or transforming, semantic representations of sentences into well-formed surface structures of any given language, a notion that has come to fruition in formal theories of grammar developed later on during the twentieth century. That Sechehaye was forced, all through his active life, to profess loyalty to Saussure’s meagre and partly unsound principles just shows the iniquity of academia, which, in this respect, does not differ too much from life in general. It has also become clear that Saussure lagged behind Sechehaye in that he had no idea of syntax or of syntactic constructions, which he saw as the products of a speaker’s individual creative activity and not of the rule system defining the grammar of any specific language. For Saussure, sentence structure was exclusively linear, so much so that he was unable to account for prosodic phenomena. He had no idea of sentence structure as a hierarchical complex of ­immediate, remote and ultimate constituents (to use Bloomfield’s terminology). Sechehaye, by contrast, was well on his way towards the discovery of ­syntax as a system of rules transforming underlying propositional structures into well-formed surface structures, both types of structure being analysable as, indeed, hierarchical complexes of immediate, remote and ultimate constituents. Moreover, Saussure’s notion of ‘sign,’ in relation to his analysis of the ‘speech circuit,’ has been shown to be inadequate in so far as (a) it remains restricted to words and fails to cover the sign-quality of sentences (or utterances), and (b) it is blind to the fact that word signs can be used, in actual speech, to refer to entities. The core function of language as a system enabling speakers to refer to entities in the real or any imagined world thereby remains unexplained: Saussure’s semantics fails to account for reference. Sechehaye rejects Saussure’s analysis of the ‘sign’ but makes no attempt at developing a theory of meaning and reference, his remit being the problem of grammar, his omnipresent “problème grammatical.” Then, Sechehaye further developed the distinction between a discoursedriven semantico-psychological sentence representation on the one hand and, on the other, the surface-grammatical fact-driven form in which sentences are normally uttered in discourse, where the ‘deep,’ discourse-driven predicate is

244

Chapter 7

marked by prominent sentence accent (Saussure never even mentioned this core element in linguistic theory). In other words, together with authors like Philipp Wegener, George Stout or Otto Jespersen, Sechehaye provided the germ of a solution to the problem of the great subject-predicate debate that raged from 1850 till 1930 and was central to linguistic theory during that period. With this, he contributed to the insight that the so-called topic-comment modulation of utterances is not a last-minute overlay over given sentences but, on the contrary, corresponds to the deepest underlying structure of sentences, displaying their form and structure as they take shape in running discourse. Chapter 5 has been devoted to this difficult but essential theme. I have, moreover, attempted to position both Saussure and Sechehaye in the wider societal and academic context of their day, sketching and analysing the great social and intellectual currents of the nineteenth century and the beginning of the twentieth. Against this background, both Saussure and Sechehaye stand out as rationalists, as opposed to romanticists: Sechehaye as one who did see the importance of the proposition as a central notion in the human sciences, Saussure as one who had no affinity to that notion yet wished to see a system behind the manifestations of language in speech. I have sketched the coming about, in the early 1800s, of the human sciences, or at least of a new conception of human sciences, whereby efforts were made to fit them into the mould of the (very successful) natural sciences. And I have placed these new developments in the context of the sociological, economic and cultural developments that dominated the nineteenth century. Pride of place has been given, in this respect, to the collection of movements now caught under the name of structuralism, a name that corresponds to an ill-defined concept covering a wide and heterogeneous variety of approaches in the human sciences. This made it necessary to search for unity, as far as possible, and thus to find a justification for the use of this term in relation to linguistics. The result of this search has been that under none of the possible senses of the term ’structuralism’ can Saussure be reckoned to be the ‘father’ of this movement. From a language-theoretical point of view, the preceding pages raise questions and perspectives that have been neglected, ignored or remained unnoticed in present-day general theories of language, such as the far-reaching importance of the notion of proposition, the nonpragmatic but fully semantic role of topic-comment modulation, or the cognitive nature of meaning, strangely studied nowadays as a mathematical phenomenon. If taken seriously, the results of this study will not only mean a radical reappraisal of the recent history of European linguistics, but will also induce the more theoretically minded linguists worldwide to rethink the foundations of their discipline.

Bibliography Aarsleff, Hans (1981). Bréal, la sémantique et Saussure. Histoire, Épistémologie, Langage 3.3: 115–133. Aarsleff, Hans (2004). Duality the key: Saussure’s debt to Taine in conceiving “the double essence of language”. Times Literary Supplement, Commentary, 20 August 2004. Ajdukiewicz, Kazimierz (1935). Die syntaktische Konnexität. Studia Philosophica 1: 1–27. (English translation by H. Weber in: Storrs McCall (ed.), Polish Logic 1920–1939. Oxford: Oxford University Press, 1967: 207–231). Allan, Keith (2010). The Western Classical Tradition in Linguistics. Second (expanded) edition. London: Equinox. Amacker, René (2000). Le développement des idées saussuriennes chez Charles Bally et Albert Sechehaye. Historiographia Linguistica 27.2/3: 205–264. Andresen, Julie T. (1990). Linguistics in America 1769–1924. London-New York: Routledge. Arnauld, Antoine and Claude Lancelot (1660/1966). Grammaire générale et raisonnée, contenant les fondemens de l’art de parler; expliquez d’une manière claire et naturelle; Les raisons de ce qui est commun à toutes les langues, et des principales différences qui s’y rencontrent; Et plusieurs remarques nouvelles sur la Langue Françoise. Paris: Pierre le Petit. (Facsimile reprint of 3rd ed. of 1676 edited by Herbert E. Brekle. Stuttgart– Bad Cannstatt: Friedrich Frommann Verlag). Ashworth, E. Jennifer (1973). The doctrine of exponibilia in the fifteenth and sixteenth centuries. Vivarium 11.2: 137–167. Auer, Anita (2009). The Subjunctive in the Age of Prescriptivism: English and German Developments during the Eighteenth Century. Basingstoke, Hamps.: Palgrave Macmillan. Balbuena, Francisco (2014). The pioneering work of Marguerite Sechehaye into the psychotherapy of psychosis: a critical review. Swiss Archives of Neurology and Psychiatry 165.5: 167–174. Bally, Charles (1905). Précis de stylistique: esquisse d’une méthode fondée sur l’étude du français moderne. Geneva: A. Eggiman. Bally, Charles (1926). La structure logique de la phrase. Review of Sechehaye (1926). Journal de Genève, July 5th 1926, p. 1. Bally, Charles (1944). Linguistique générale et linguistique française. (Revised 2nd edition; 1st edition 1932 published by E. Leroux, Paris.) Bern: Francke. Beauzée, Nicolas (1767). Grammaire générale, ou exposition raisonnée des éléments nécessaires du langage, pour servir de fondement à l’étude de toutes les langues. (2 vols.) Paris: J. Barbou (Modern edition: B.E. Bartlett (ed.), Frommann-Holzboog, Stuttgart-Bad Cannstatt, 1974, 2 vols.) Bech, Gunnar (1955, 1957). Studien über das Deutsche Verbum Infinitum. Det Kongelige Danske Videnskabernes Selskab. Historisk-filologiske Meddelelser, 35:2 (1955) and

246

Bibliography

36:6 (1957). (2 vols.) Copenhagen: Munksgaard. (2nd unrevised edition 1983, Tübingen: Niemeyer.) Behme, Christina (2014). A “Galilean” Science of Language. Journal of Linguistics 50.3: 671–704. Bielenstein, August Johann Gottfried (1892). Die Grenzen des lettischen Volksstammes und der lettischen Sprache in der Gegenwart und im 13. Jahrhundert. St. Petersburg: Eggers. Bloomfield, Leonard (1924). Review of Cours de linguistique générale by Ferdinand de Saussure. Modern Language Journal 8: 317–319. Bloomfield, Leonard (1926). A set of postulates for the science of language. Language 2.3: 153–164. Bloomfield, Leonard (1933). Language. New York: Henry Holt & Cy. Bloomfield, Leonard (1936). Language or ideas? Language 12.1: 89–95. Blumenthal, Arthur L. (1985). Wilhelm Wundt: psychology as the propaedeutic science. In: Claude Buxton (ed.), Points of View in the Modern History of Psychology. ­Orlando– San Diego–New York–London–Toronto–Montreal–Sydney–Tokyo: ­Academic Press: 19–50. Bolinger, Dwight (1949). The sign is not arbitrary. Boletin del Instituto Cam y Cuervo (= Thesaurus) 5: 52–62. Boring, Edwin G. (1950). A History of Experimental Psychology. Second edition. New York: Appleton-Century-Crofts. Borowski, Ephraim J. and Jonathan M. Borwein (1989). Dictionary of Mathematics. ­London/Glasgow: Collins. Bouquet, Simon (1989). Le Cours de linguistique générale de Saussure et la philosophie. Histoire, Épistémologie, Langage 11-II: 103–119. Bovet, Pierre (1917). Le Cours de linguistique générale de F. de Saussure. Wissen und Leben 1 Oct. 1916–15 March 1917: 306–307. Bréal, Michel (1883). Les lois intellectuelles du langage. Fragment de sémantique. ­Annuaires de l’association des études grecques en France 17: 132–142. Bréal, Michel (1891). Le langage et les nationalités. Revue des deux mondes 108 (1er décembre): 615–639. Bréal, Michel (1897). Essai de sémantique: science des significations. Paris: Hachette. Breva-Claramonte, Manuel (1983). Sanctius’ Theory of Language. A Contribution to the History of Renaissance Linguistics. Amsterdam/Philadelphia: Benjamins. Brouwer, Luitzen E.J. (1907). Over de grondslagen der wiskunde. (‘About the Foundations of Mathematics’). Amsterdam/Leipzig: Maas & van Suchtelen—Groningen: Noordhoff. Brouwer, Luitzen E.J. (1967). Intuitionistic reflections on formalism. In: Jean van Heijenoort (ed.), From Frege to Gödel: A Source Book in Mathematical Logic, 1879–1931, Cambridge, MA: Harvard University Press: 490–492. Brugmann, Karl (1900). Zu dem ‘Vorwort’ zu Band 1 der Morphologischen Untersuchungen von Osthoff und Brugmann. Indogermanische Forschungen 11. Anzeiger für

Bibliography

247

indogermanische Sprach- und Altertumskunde. Beiblatt zu den indogermanischen Forschungen 11: 131–132. Brugmann, Karl and Berthold Delbrück (1886–1900). Grundriss der vergleichenden Grammatik der indogermanischen Sprachen. Kurzgefasste Darstellung der ­Geschichte des Altindischen, Altiranischen (Avestischen u. Altpersischen), Altarmenischen, ­Altgriechischen, Albanesischen, Lateinischen, Oskisch-Umbrischen, Altirischen, Gotischen, Althochdeutschen, Litauischen und Altkirchenslavischen. (5 Bände; 8 volumes) Berlin- Leipzig: Walter de Gruyter / Strassburg: Karl Trübner. Bruner, Jerome S. (1990). Acts of Meaning. Four Lectures on Mind and Culture. Cambridge, MA: Harvard University Press. Bühler, Karl (1934). Sprachtheorie: die Darstellungsfunktion der Sprache. Jena: Fischer. Celano, Giuseppe G.A. (2008). Linguist List 19.2839. Charlton, Donald G. (1959). Positivist Thought in France during the Second Empire ­1852–1870. Oxford: Clarendon Press. Chevallier, Jean-Claude, Claire Blanche-Benveniste, Michel Arrivé and Jean Peytard (1964). Grammaire du français contemporain. Paris: Larousse. Chomsky, Noam (1995). The Minimalist Program. Cambridge, MA: The MIT Press. Clarapède, Édouard (1917). Ferd. de Saussure, Cours de linguistique générale. Archives de psychologie 16: 93–95. Comorovski, Ileana (2008). Constituent questions and the copula of specification. In: Ileana Comorovski and Klaus von Heusinger (eds), Existence: Syntax and Semantics, Springer: Berlin: 49–77. Comte, Auguste (1830–1842). Cours de philosophie positive. 6 volumes. Paris: Bachelier. Couturat, Louis (1901). La logique de Leibniz d’après des documents inédits. Paris: Alcan. De Groot, Albert W. (1940). Neutralisation d’oppositions. Neophilologus 25.1: 127–146. De Groot, Albert W. (1949). Structurele Syntaxis. The Hague: Servire. Delbrück, Berthold (1893–1900).Vergleichende Syntax der indogermanischen Sprachen. (Part I: 1893; Part II: 1897; Part III: 1900; = vols 3, 4, 5, respectively, of Brugmann & Delbrück 1886–1900). Strassburg: Karl Trübner. Delbrück, Berthold (1901). Grundfragen der Sprachforschung. Strasbourg: Trübner. Derrida, Jacques (1967). De la grammatologie. Paris: Éditions de Minuit. Derrida, Jacques (1982). Margins of Philosophy. Translated with Additional Notes by Alan Bass. Chicago: The University of Chicago Press. Dijksterhuis, Eduard J. (1961). The Mechanisation of the World Picture. Oxford: Oxford University Press. (translation from the Dutch De mechanisering van het wereldbeeld, Amsterdam: Meulenhoff, 1950). Dingemanse, Mark (2011). The Meaning and Use of Ideophones in Siwu. PhD-thesis, Radboud University. Nijmegen: Max Planck Institute for Psycholinguistics. Dingemanse, Mark, Damián E. Blasi, Gary Lupyan, Morten H. Christiansen and Padraic Monaghan (2015). Arbitrariness, iconicity and systematicity in language. Trends in Cognitive Sciences 19.10: 603–615.

248

Bibliography

Donnellan, Keith (1966). Reference and definite descriptions. Philosophical Review 75: 281–304. Dresher, B. Elan (2011). The phoneme. In: Marc van Oostendorp, Colin J. Ewen, Elizabeth Hume and Keren Rice eds., The Blackwell Companion to Phonology, Malden, MA/Oxford: Wiley-Blackwell: 241–266. Eisenberg, Peter, Herman Gelhaus, Helmut Henne, Horst Sitta and Hans Wellmann (1995). Duden Grammatik der deutschen Gegenwartsprache. (5th edition). Mannheim: Dudenverlag. Egger, Émile (1879). Observations et réflexions sur le développement de l’intelligence et du langage chez les enfants. Paris: Alphonse Picard. Egger, Victor (1877). La physiologie cérébrale et la psychologie. Revue des deux mondes, vol. 24: 193–211. Egger, Victor (1881). La parole intérieure: essai de psychologie descriptive. Thèse de doctorat, présentée à la Faculté des Lettres de Paris. Paris: Baillière. Ehrenfels, Christian von (1890). Über Gestaltqualitäten. Vierteljahrsschrift für wissenschaftliche Philosophie 14: 249–292. Elffers-van Ketel, Els H.C. (1991). The Historiography of Grammatical Concepts. 19th and 20th Century Changes in the Subject-Predicate Conception and the Problem of their Historical Reconstruction. Amsterdam: Rodopi. Elffers-van Ketel, Els H.C. (2016). Hard times for traditional grammar: the case of Martinus Langeveld. Beiträge zur Geschichte der Sprachwissenschaft 26: 171–186. Engler, Rudolf (1983). Das sprachliche Zeichen bei Saussure, Bally, Sechehaye. Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung 36.5: 533–542. Engler, Rudolf (2004). The making of the Cours de linguistique générale. In: Sanders (2004: 47–58). Everett, Daniel (2005). Cultural constraints on grammar and cognition in Pirahã: another look at the design features of human language. Current Anthropology 46.4: 621–646. Fillmore, Charles J. (1982). Frame Semantics. In: Linguistics in the Morning Calm. Selected Papers from SICOL-1981. Edited by The Linguistic Society of Korea. Seoul: Hanshin Publishing Company: 111–137. Finck, Franz Nikolaus (1905). Aufgabe und Gliederung der Sprachwissenschaft. Halle: Hauft. Fodor, Jerry A. (1976). The Language of Thought. Hassocks, Sussex: The Harvester Press. Fogelin, Robert J. (1967). Evidence and Meaning. Studies in Analytic Philosophy. London: Routledge & Kegan Paul / New York: Humanities Press. Fónagy, Ivan (1963). Die Metaphern in der Phonetik. Ein Beitrag zur Entwicklungsgeschichte des wissenschaftlichen Denkens. (= Janua Linguarum, series minor 25) The Hague: Mouton. Frege, Gottlob (1892). Über Sinn und Bedeutung. Zeitschrift für Philosophie und philosophische Kritik 100: 25–50.

Bibliography

249

Frýba-Reber, Anne-Marguerite (1994). Albert Sechehaye et la syntaxe imaginative. Contribution à l’histoire de la linguistique saussurienne. Geneva: Droz. Frýba-Reber, Anne-Marguerite (1995/1996). Charles-Albert Sechehaye: un linguïste engagé. Cahiers Ferdinand de Saussure 49: 123–137. Frýba-Reber, Anne-Marguerite (2001a). La revanche de la stylistique: hommage d’Albert Sechehaye à son prédécesseur et ami Charles Bally. Cahiers Ferdinand de Saussure 54: 125–144. Frýba-Reber, Anne-Marguerite (2001b). De la cohésion et de la fluidité de la langue. Textes inédits (1908–1943) de Charles Bally et d’Albert Sechehaye édités par AnneMarguerite Frýba-Reber. Cahiers Ferdinand de Saussure 54:429–487. Frýba-Reber, Anne-Marguerite (2017). La réception du CLG en Suisse. Paper read at the Colloquium “The Course in General Linguistics, 1916–2016. The Emergence”, held in Geneva 9–14 January 2017. Gabelentz, H. Georg C. von der (1869). Ideen zu einer vergleichenden Syntax. Wort- und Satzstellung. Zeitschrift für Völkerpsychologie und Sprachwissenschaft 6: 376–384. Gabelentz, H. Georg C. von der (1901). Die Sprachwissenschaft. Ihre Aufgaben, Methoden und bisherigen Ergebnisse. Leipzig: Tauchnitz. (2nd posthumous edition revised and ­expanded by Albrecht Graf von der Schulenburg). Gadet, Françoise (1995). Jakobson sous le pavillon saussurien. Linx. Revue des linguistes de l’université Paris Ouest Nanterre La Défense 7: 449–459. Gambarara, Daniele (1972). La bibliothèque de Ferdinand de Saussure. Bulletin du ­Musee d’Art et d’Histoire de Genève 20: 319–368. Gardiner, Alan H. (1932). The Theory of Speech and Language. Oxford: Clarendon Press. Gasparov, Boris (2013). Beyond Pure reason. Ferdinand de Saussure’s Philosophy of ­Language and its Early Romantic Antecedents. New York: Columbia University Press. Gautier, Léopold (1916). La linguistique générale de Ferdinand de Saussure. Gazette de Lausanne 13 Aug. Geerts, Guido, Walter Haeserijn, Jaap de Rooij and Maarten C. van den Toorn (1984). Algemene Nederlandse Spraakkunst. Groningen: Wolters-Noordhoff / Leuven: Wolters. Godel, Robert (1957). Les sources manuscrites du Cours de linguistique générale de F. de Saussure. Geneva: Droz / Paris: Minard. Godel, Robert (1966). F. de Saussure’s theory of language. In: Th. A. Sebeok (ed.), Current Trends in Linguistics. Vol. III, Theoretical Foundations, The Hague-Paris: Mouton: 479–493. Graffi, Giorgio (1995). Old debates and current problems: ‘Völkerpsychologie’ and the question of the individual and the social in language. In: Lia Formigari and Daniele Gambarara (eds), Historial Roots of Linguistic Theories. Amsterdam/Philadelphia: Benjamins: 171–184. Graffi, Giorgio (2001). 200 Years of Syntax: A Critical Survey. Amsterdam/Philadelphia: Benjamins.

250

Bibliography

Grevisse, Maurice and André Goosse (1986). Le bon usage. Grammaire française. 12th edition. Paris-Gembloux: Duculot. Groult, Pierre (1947). La formation des langues romanes. Tournai-Paris: Casterman. Gundel, Jeanette and Thorstein Fretheim (2004). Topic and focus. In: Laurence R. Horn and Gregory Ward (eds.), The Handbook of Pragmatics, Malden, MA/Oxford: WileyBlackwell: 175–196. Harris, Zellig S. (1951). Methods in Structural Linguistics. Chicago: The University of Chicago Press. Havet, Louis (1908). Review of Mélanges de linguistique offerts à M. Ferdinand de ­Saussure. Bulletin de la Société Linguistique de Paris 16: xxi–liv. Heidolph, Karl Erich, Walter Flämig and Wolfgang Motsch (1981). Grundzüge einer deutschen Grammatik. Berlin: Akademie-Verlag. Helbig, Gerhard (1989). Geschichte der neueren Sprachwissenschaft. 8th edition. Wiesbaden: Springer Fachmedien. Herder, Johann Gottfried von (1772). Abhandlung über den Ursprung der Sprache, welche den von der Königl. Academie der Wissenschaften für das Jahr 1770 geseztenPreis erhalten hat. Berlin: Christian Friedrich Voß. Hewson, John (1990). Un système où tout se tient: Origin and evolution of the idea. In: Hans-Josef Niederehe and E.F.K. Koerner (eds.), History and Historiography of ­Linguistics: Proceedings of the Fourth International Conference on the History of the Language Sciences (ICHoLS IV), Trier, 24–27 August 1987. Amsterdam/Philadelphia: Benjamins: 787–794. Higgins, Francis R. (1979), The Pseudo-Cleft Construction in English. New York/London: Garland Publishing. Hjelmslev, Louis (1953). Prolegomena to a Theory of Language. Supplement to International Journal of American Linguistics 19.1. Indiana University Publications in ­Anthropology and Linguistics. Memoir 7. Baltimore MD: Waverly Press. Huddleston, Rodney and Geoffrey Pullum (2002). The Cambridge Grammar of the ­English Language. Cambridge: Cambridge University Press. Humboldt, Wilhelm von (1836). Über die Kawi–Sprache auf der Insel Java, nebst einer Einleitung über die Verschiedenheit des menschlichen Sprachbaues und ihren Einfluß auf die geistige Entwickelung des Menschengeschlechts. Königliche Akademie der Wissenschaften. Berlin: F. Dümmler. Hymes, Dell H. (1983). Essays in the History of Linguistic Anthropology. Studies in the History of the Language Sciences 25. Amsterdam/Philadelphia: Benjamins. Iemhoff, Rosalie (2013). Intuitionism in the philosophy of mathematics. In: Stanford Encyclopedia of Philosophy. http://plato.stanford.edu/entries/intuitionism/. Jaberg, Karl (1916). Ferdinand de Saussure’s Vorlesungen über allgemeine Sprachwissenschaft. Der Bund 17 & 24 Dec.

Bibliography

251

Jakobson, Roman (1971a). Un manuel de phonologie générale. In: Roman Jakobson, ­Selected Writings, Vol. I: Phonological Studies. The Hague: Mouton: 311–316. Jakobson, Roman (1971b). Retrospect. In: Roman Jakobson, Selected Writings, Vol. II: Words and Language. The Hague: Mouton: 711–724. Jakobson, Roman (1971c). La première lettre de Ferdinand de Saussure à Antoine Meillet sur les anagrammes. L’Homme 11.2: 15–24. Jakobson, Roman (1975). Structuralisme et téléologie. L’Arc, Revue trimestrielle, 60: 50–54. Jakobson, Roman and Morris Halle (1956). Fundamentals of Language. (= Janua ­Linguarum 1) The Hague: Mouton. James, William (1890). The Principles of Psychology. (2 vols.). London: Macmillan. Jankowsky, Kurt R. (1972). The Neogrammarians. A Re-evaluation of their Place in the Development of Linguistic Science (= Janua Linguarum 116) The Hague: Mouton. Jespersen, Otto (1909–1949). A Modern English Grammar on Historical Principles. 7 vols. London: Allen & Unwin / Copenhagen: Ejnar Munksgaard. Jespersen, Otto (1917). Negation in English and Other Languages. Det Kgl. Danske ­Videnskabernes Selskab, Historisk-filologiske Meddelelser, I.5. Copenhagen: Høst & Son. Jespersen, Otto (1924). The Philosophy of Grammar. London: Allen & Unwin. Jespersen, Otto (1933). Essentials of English Grammar. London: Allen & Unwin. Jespersen, Otto (1937). Analytic Syntax. Copenhagen: Levin & Munksgaard. Joseph, John E. (1996). ). ‘Undoubtedly a powerful influence’: Victor Henry’s Antinomies linguistiques (1896) with an annotated translation of the first chapter. Language & Communication 16.2: 117–144. Joseph, John E. (2003). Pictet’s Du beau (1856) and the crystallisation of Saussurean linguistics. Historiographia Linguistica 30.3: 365–388. Joseph, John E. (2004). Root and branch: Pictet’s role in the crystallization of Saussure’s thought. Times Literary Supplement 9 January 2004. Joseph, John E. (2012). Saussure. Oxford: Oxford University Press. Kalepky, Theodor (1928). Neuaufbau der Grammatik als Grundlegung zu einem wissenschaftlichen System der Sprachbeschreibung. Leipzig: Teubner. Kitaoka, Akiyoshi (2007). Tilt illusions after Oyama (1960): a review. Japanese Psychological Research 49.1: 7–19. Klippi, Carita (2010). La vie du langage: la linguistique dynamique en France de 1864 à 1916. Lyon: ENS Éditions. Klooster, Wim G. (1972). The Structure Underlying Measure Phrase Sentences. (= Foundations of Language/Supplementary Series/Volume 17.) Dordrecht: Reidel. Klooster, Wim G. (2001). Grammatica van het Nederlands. Een volledig overzicht. The Hague: Sdu Uitgevers.

252

Bibliography

Koerner, E.F. Konrad (1973). Ferdinand de Saussure: Origin and Development of his Linguistic Thought in Western Studies of Language. Braunschweig: Vieweg. Koerner, E.F. Konrad (1988). Meillet, Saussure et la linguistique générale. Histoire, Épistémologie, Langage 10.2: 57–73. Koerner, E.F. Konrad (1989). Practicing Linguistic Historiography: Selected Essays. ­Amsterdam/Philadelphia: Benjamins. Koerner, E.F. Konrad (1993). The natural science background to the development of historical-comparative linguistics. In: Henk Aertsen and Robert J. Jeffers (eds), ­Historical Linguistics 1989. Papers from the 19th International Conference on Historical Linguistics, Rutgers University 14–18 August 1989. Amsterdam/Philadelphia: ­Benjamins: 1–24. Koerner, E.F. Konrad (1997). Notes on the history of the concept of language as a s­ ystem ‘où tout se tient’. Linguistica Atlantica 19: 1–20. Koerner, E.F. Konrad (1999). Linguistic Historiography: Projects and Prospects. Amsterdam/Philadelphia: Benjamins. Koffka, Kurt (1935). Principles of Gestalt Psychology. London: Lund Humphries. Köhler, Wolfgang (1929). Gestalt Psychology. New York: Horace Liveright. Kuryłowicz, Jerzy (1927). ‘ǝ indo-européen et ḫ hittite’, Symbolae grammaticae in honorem J. Rozwadowski Vol. 1. Krakow: Gebethner & Wolff: 95–104. Kuteva, Tania (2001). Auxiliation: An Enquiry into the Nature of Grammaticalization. ­Oxford: Oxford University Press. Kutschera, Franz von (1989). Gottlob Frege. Eine Einführung in sein Werk. Berlin: de Gruyter. Lakoff, Robin (1969). Review of Herbert H. Brekle (ed.), Grammaire générale et raisonnée, ou La grammaire du Port Royal. Stuttgart-Bad Cannstatt: Fromann, 1966. ­Language 45.2: 343–364. Lancelot, Claude and Antoine Arnauld (1660). Grammaire générale et raisonnée contenant Les fondemens de l’art de parler; expliquez d’une manière claire & naturelle; Les raisons de ce qui est commun à toutes les langues, & des principales differen­ ces  qui s’y  rencontrent; Et plusieurs remarques nouvelles sur la Langue Françoise. ­Paris: Pierre le Petit. (Facsimile reprint by The Scolar Press Ltd., Menston, England, 1967.) Lazarus, Moritz (1851). Über den Begriff und die Möglichkeit einer Völkerpsychologie. Deutsches Museum. Zeitschrift für Literatur, Kunst und öffentliches Leben 1: 112–126. Lazarus, Moritz and Heymann Steinthal (1860). Einleitende Gedanken über Völkerpsychologie, als Einladung zu einer Zeitschrift für Völkerpsychologie und Sprachwissenschaft. Zeitschrift für Völkerpsychologie und Sprachwissenschaft 1: 1–73. Lepschy, Giulio C. (1962). Ancora su “l’arbitraire du signe”. Annali della Scuola Normale Superiore di Pisa, Classe di Lettere e Filosofia, 31: 65–102. Levelt, Willem J.M. (1989). Speaking. From Intention to Articulation. Cambridge MA: MIT Press.

Bibliography

253

Levelt, Willem J.M. (2013). A History of Psycholinguistics. The Pre-Chomskyan Era. ­Oxford: Oxford University Press. Luys, Jules Bernard (1876). Le cerveau et ses fonctions. Paris: G. Baillière. Lyons, John (1963). Structural Semantics. An Analysis of Part of the Vocabulary of Plato. Publications of the Philological Society XX. Oxford: Blackwell. Lyons, John (1968). Introduction to Theoretical Linguistics. Cambridge: Cambridge University Press. Lyons, John (1977). Semantics. (2 vols). Cambridge: Cambridge University Press. Lyons, John (ed.) (1970). New Horizons in Linguistics. A Pelican Original. Penguin Books. Martinet, André (1960). Éléments de linguistique générale. Paris: Colin. Martinet, André (1965). La linguistique synchronique. Études et recherches. Paris: Presses Universitaires de France. Marty, Anton (1908). Untersuchungen zur Grundlegung der allgemeinen Grammatik und Sprachphilosophie. Vol. I. Halle a. S.: Max Niemeyer. Marty, Anton (1950). Über Wert und Methode einer allgemeinen beschreibenden Bedeutungslehre. (Nachgelassene Schriften. Aus “Untersuchungen zur Grundlegung der allgemeinen Grammatik und Sprachphilosophie” III (Schluß). Edited by Otto Funke. Bern: Francke. Matthaei, Rupprecht (1929). Das Gestaltproblem. München: J.F. Bergmann. Matthews, Peter H. (2001). A Short History of Structural Linguistics. Cambridge: Cambridge University Press. McCawley, James D. (1967). Meaning and the description of languages. Kotoba no Uchū 2.9: 10–18; 2.10: 38–48; 2.11: 51–57. (reprinted in McCawley 1973: 99–120). McCawley, James D. (1970). Semantic representation. In: Paul M. Garvin (ed.), Cognition: A Multiple View, New York: Spartan Books: 227–247. (reprinted in McCawley 1973: 240–256) McCawley, James D. (1972). A program for logic. In: Donald Davidson and Gilbert Harman (eds), Semantics of Natural Language. Dordrecht: Reidel: 498–544. (reprinted in McCawley 1973: 285–319). McCawley, James D. (1973). Grammar and Meaning: Papers on Syntactic and Semantic Issues. Tokyo: Taishukan Publishing Cy. Meillet, Antoine (1908). Review of Sechehaye (1908). Bulletin de la Société linguistique de Paris 15.56: xxiii–xxv. Meillet, Antoine (1910). Le langage. L’année sociologique 11 (1906–1909): 789–798. Meillet, Antoine (1916). Review of Cours de linguistique générale. Bulletin de la Société de Linguistique de Paris 64.20: 32–36. Meillet, Antoine (1917). Review of Cours de linguistique générale. Revue Critique d’Histoire et de Littérature 84.1: 49–51. Meillet, Antoine (1921). Linguistique historique et linguistique générale. Collection linguistique publiée par la Société de Linguistique de Paris, VIII. Paris: Édouard Champion.

254

Bibliography

Meillet, Antoine (1938). Linguistique historique et linguistique générale. Vol. II. Collection linguistique publiée par la Société de Linguistique de Paris, XL. Paris: Klincksieck. Michotte, Albert (1946). La perception de la causalité. Louvain: Institut Supérieur de Philosophie. (The Perception of Causality, translated by T.R. Miles and E. Miles, New York: Basic Books, 1963.). Mikkelsen, Line (2005). Copular Clauses: Specification, Predication and Equation. Amsterdam/Philadelphia: Benjamins. Mikkelsen, Line (2011). Copular clauses. In: Klaus von Heusinger, Claudia Maienborn and Paul Portner (eds), Semantics: An International Handbook of Natural Language Meaning. Vol. 2.  Series: Handbücher zur Sprach- und Kommunikationswissenschaft / Handbooks of Linguistics and Communication Science (HSK) 33/2, Berlin: De Gruyter Mouton: 1805–1829. Montague, Richard (1973). The proper treatment of quantification in ordinary English. In: K.J.J. Hintikka, J.M.E. Moravcsik and P. Suppes (eds), Approaches to Natural Language. Proceedings of the 1970 Stanford Workshop on Grammar and Semantics, Dordrecht/Boston: Reidel: 221–242. Morpurgo-Davies, Anna (1975). Language classification in the nineteenth century. In: Th. Sebeok (ed.), Current Trends in Linguistics 13, Historiography of Linguistics. The Hague-Paris: Mouton: 607–716. Morpurgo-Davies, Anna (2004). Saussure and Indo-European linguistics. In: Sanders (2004: 8–29). Müller, Otto (1922). La stylistique de M. Charles Bally. The Modern Language Journal 7.1: 5–16. Naville, Adrien (1901). Nouvelle classification des sciences: étude philosophique. 2e édition entièrement refondue. Paris: Félix Alcan. Niedermann, Max (1916). De Saussures Vorlesungen über allgemeine Sprachwissenschaft. Neue Zürcher Zeitung August 3rd 1916. Nuchelmans, Gabriël (1973). Theories of the Proposition. Ancient and Medieval Conceptions of the Bearers of Truth and Falsity. Amsterdam: North-Holland. Nuchelmans, Gabriël (1980). Late-Scholastic and Humanist Theories of the Proposition. Amsterdam: North- Holland. Nuchelmans, Gabriël (1983). Judgment and Proposition. From Descartes to Kant. ­Amsterdam: North-Holland. Ogden, Charles K. and Ivor A. Richards (1923). The Meaning of Meaning. A Study of the Influence of Language upon Thought and of the Science of Symbolism. London: Kegan Paul. Ohala, John (1997). Sound symbolism. In: Proceedings of the Fourth Seoul International Conference on Linguistics (SICOL), Seoul: Linguistic Society of Korea: 98–103. Oltramare, André (1916). La résurrection d’un génie. Semaine littéraire 27 May.

Bibliography

255

Osthoff, Hermann and Karl Brugmann (1878). Vorwort. Morphologische ­Untersuchungen auf dem Gebiete der indogermanischen Sprachen. Vol. I. Leipzig: Hirzel: iii–xx. Paardekooper, Piet C. (1963). Beknopte ABN-syntaxis. (Concise Syntax of Standard Dutch). Den Bosch: Malmberg. Paul, Hermann (1920). Prinzipien der Sprachgeschichte. Halle: Max Niemeyer. (5th ­revised edition; first edition called Principien der Sprachgeschichte, 1880). Peirce, Charles Sanders (1906). Prolegomena to an apology for pragmaticism, The ­Monist 16: 492–546. Percival, W. Keith (1976). On the historical source of immediate constituent analysis. In: James D. McCawley (ed.), Notes from the Linguistic Underground (=Syntax and Semantics Vol. 7). New York–San Francisco–London: Academic Press:229–242. Percival, W. Keith (1977). Review article of: E.F.K. Koerner, Ferdinand de Saussure: Origin and Development of his Linguistic thought in Western Studies of Language. A Contribution to the History and Theory of Linguistics. Braunschweig: Vieweg, 1973. Language 53.2: 383– 405. Percival, W. Keith (2011). Roman Jakobson and the birth of linguistic structuralism. Σημειωτική—Sign Systems Studies 39.1: 236–262. Pictet, Adolphe (1856). Du beau dans la nature, l’art et la poésie. Études esthétiques. Paris: Sandoz & Fischbacher. Pictet, Adolphe (1859–1863). Les origines indo-européennes ou les Aryas primitifs. Essai de paléontologie linguistique. (2 vols.) Paris-Genève: Cherbuliez. (2nd ed. 1877, 3 vols., Paris: Sandoz & Fischbacher). Popper, Karl R. (1959). The Logic of Scientific Discovery. London: Hutchinson. (English translation of Karl Popper, 1935, Logik der Forschung). Quine, Willard Van Orman (1961). From a logical Point of View. Logico-Philosophical Essays. Cambridge, MA: Harvard University Press. Second revised edition (first edition 1953). Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik (1972). A Grammar of Contemporary English. London: Longman. Rescher, Nicholas (1973). The Coherence Theory of Truth. Oxford: Clarendon Press. Ries, John (1894). Was ist Syntax? Ein kritischer Versuch. Marburg: Elwertsche Verlagsbuchhandlung. Ries, John (1931). Was ist ein Satz? (= Beiträge zur Grundlegung der Syntax 3). Prague: Taussig & Taussig. Robins, Robert H. (1967). A Short History of Linguistics. London: Longmans. Robinson, James O. (1998). The Psychology of Visual Illusion. Mineola NY: Dover. Ronjat, Jules (1916). Le cours de linguistique de Ferdinand de Saussure. Journal de Genève 26 June. Russell, Bertrand (1900). A Critical Exposition of the Philosophy of Leibniz. Cambridge: The University Press.

256

Bibliography

Sanders, Carol (ed.) (2004). The Cambridge Companion to Saussure. Cambridge: Cambridge University Press. Sandmann, Manfred (1954). Subject and Predicate. A Contribution to the Theory of Syntax. Edinburgh: Edinburgh University Press. Sapir, Edward (1921). Language. An Introduction to the Study of Speech. New York: ­Harcourt, Brace and Company. Sapir, Edward (1929). A study in phonetic symbolism. Journal of Experimental Psychology 12: 225–239. Saussure, Ferdinand de (1879). Mémoire sur le système primitif des voyelles dans les langues indo-européennes. Leipzig: Teubner. Saussure, Ferdinand de (1881). De l’emploi du génitif absolu en sanskrit. PhD-thesis, Leipzig University. Geneva: Jules-Guillaume Fick. Saussure, Ferdinand de (1916). Cours de linguistique générale. Publié par Charles Bally et Albert Sechehaye, avec la collaboration de Albert Riedlinger. Paris: Payot. Saussure, Ferdinand de (1968). Cours de linguistique générale. Édition critique par Rudolf Engler. Vol. I. Wiesbaden: Harrassowitz. Saussure, Ferdinand de (1972). Cours de linguistique générale. Édition critique préparée par Tullio de Mauro. Paris: Payot. Saussure, Ferdinand de (1974). Cours de linguistique générale. Édition critique par Rudolf Engler. Vol. II, Appendice, notes de F. de Saussure sur la linguistique générale. Wiesbaden: Harrassowitz. Saussure, Ferdinand de (2002). Écrits de linguistique générale. Établis et édités par Simon Bouquet et Rudolf Engler avec la collaboration d’Antoinette Weil. Paris: Gallimard. Schmaus, Warren (1994). Durkheim’s Philosophy of Science and the Sociology of Knowledge: Creating an Intellectual Niche. Chicago: The University of Chicago Press. Schmidt, Johannes (1872). Die Verwandtschaftsverhältnisse der indogermanischen Sprachen. Weimar: Böhlau. Schuchardt, Hugo (1885). Ueber die Lautgesetze. Gegen die Junggrammatiker. Berlin: Robert Oppenheim. Sechehaye, Ch. Albert (1905). L’imparfait du subjonctif et ses concurrents dans les hypothétiques normales en français. Esquisse de syntaxe historique. Romanische Forschungen 19.2: 321–406. Sechehaye, Ch. Albert (1908a). Programme et méthodes de la linguistique théorique: psychologie du langage. Paris: Champion / Leipzig: Harrassowitz / Genève: Eggiman. (Facsimile reprint 2011, Nabu Press, Charleston, South Carolina) Sechehaye, Ch. Albert (1908b). La stylistique et la linguistique théorique. In: Mélanges de linguistique offerts à M. Ferdinand de Saussure. Collection linguistique publiée par La Société de Linguistique de Paris—II. Paris: Honoré Champion: 155–187. Sechehaye, Ch. Albert (1916). La méthode constructive en syntaxe. Revue des langues romanes 59.1/2: 44–76.

Bibliography

257

Sechehaye, Ch. Albert (1917). Les problèmes de la langue à la lumière d’une théorie nouvelle. Revue Philosophique de la France et de l’Étranger 84 (juillet à décembre 1917): 1–30. Sechehaye, Ch. Albert (1926). Essai sur la structure logique de la phrase. Paris: Champion. (Reprint 1950) Sechehaye, Ch. Albert (1930). Les mirages linguistiques. Journal de psychologie normale et pathologique 27: 337– 366. Sechehaye, Ch. Albert (1933). La pensée et la langue, ou comment concevoir le rapport organique de l’individuel et du social dans le langage? Journal de psychologie normale et pathologique 30: 57–81. Sechehaye, Ch. Albert (1941). Les classes de mots et l’imagination. Cahiers Ferdinand de Saussure 1: 77–88. Sechehaye, Ch. Albert (1944). La méthode constructive en grammaire et son application. Cahiers Ferdinand de Saussure 4: 22–25 (reprinted from: 59. Jahrbuch des Vereins schweizerischer Gymnasiallehrer, 106–108, 1930). Sériot, Patrick (1999). Structure et totalité. Les origines intellectuelles du structuralisme en Europe centrale et orientale. Paris: Presses Universitaires de France. Seuren, Pieter A.M. (1982). Internal variability in competence. Linguistische Berichte 77: 1–31. Seuren, Pieter A.M. (1985). Discourse Semantics. Oxford: Blackwell. Seuren, Pieter A.M. (1993). Why does 2 mean “2”? Grist to the anti-Grice mill. In: Eva Hajičová (ed.), Proceedings of the Conference on Functional Description of Language, Prague Nov. 24–27, 1992. Faculty of Mathematics and Physics, Charles University, Prague: 225–235. Seuren, Pieter A.M. (1998). Western Linguistics: An Historical Introduction. Oxford: Blackwell. Seuren, Pieter A.M. (1999a). Topic and comment. In: Carol F. Justus & Edgar C. Polomé (eds), Language Change and Typological Variation. Papers in Honor of Winfred P. Lehmann on the Occasion of His 83rd Birthday. Journal of Indo-European Studies Monograph Series 31: 348–373. Seuren, Pieter A.M. (1999b). The subject-predicate debate X-rayed. In: David Cram, Andrew Linn & Elke Nowak (eds), History of Linguistics 1996. Volume 2: From Classical to Contemporary Linguistics. Selected Papers from the Seventh International Conference on the History of the Language Sciences (ICHOLS VII), Oxford, 12–17 September 1996. Amsterdam/Philadelphia: Benjamins: 41–55. Seuren, Pieter A.M. (2001). A View of Language. Oxford: Oxford University Press. Seuren, Pieter A.M. (2003). Verb clusters and branching directionality in German and Dutch. In: Pieter A.M. Seuren and Gerard Kempen (eds), Verb Constructions in German and Dutch, Amsterdam/Philadelphia: Benjamins: 247–296. Seuren, Pieter A.M. (2004). Chomsky’s Minimalism. New York: Oxford University Press.

258

Bibliography

Seuren, Pieter A.M. (2005). Eubulides as a 20th-century semanticist. Language Sciences 27.1: 75–95. Seuren, Pieter A.M. (2008). Apollonius Dyscolus en de semantische syntaxis. In: Lo van Driel and Theo Janssen (eds), Ontheven aan de tijd. Linguïstisch-historische studies voor Jan Noordegraaf bij zijn zestigste verjaardag. Amsterdam: Stichting Neerlandistiek VU—Münster: Nodus Publikationen: 15–24. Seuren, Pieter A.M. (2009). Language in Cognition. (= Language from Within, Vol. I). Oxford: Oxford University Press. Seuren, Pieter A.M. (2010). The Logic of Language. (= Language from Within, Vol. II). Oxford: Oxford University Press. Seuren, Pieter A.M. (2013a). From Whorf to Montague. Explorations in the Theory of Language. Oxford: Oxford University Press. Seuren, Pieter A.M. (2013b). The logico-philosophical tradition. In: Keith Allan (ed.), The Oxford Handbook of the History of Linguistics, Oxford: Oxford University Press: 537–554. Seuren, Pieter A.M. (2014a). Universe restriction in the logic of language. In: Jack Hoeksema & Dicky Gilbers (eds), Black Book. A Festschrift in Honor of Frans Zwarts. Groningen: University of Groningen: 282–300. Seuren, Pieter A.M. (2014b). The cognitive ontogenesis of predicate logic. Notre Dame Journal of Formal Logic 55.4: 499–532. Seuren, Pieter A.M. (2016). Saussure and his intellectual environment. History of European Ideas 42. 6: 819–847. Seuren, Pieter A.M. (2018). Semantic Syntax. Second Revised Edition. Leiden/Boston: Brill. Seuren, Pieter A.M. and Camiel Hamans (2009). Semantic conditioning of syntactic rules: evidentiality and auxiliation in English and Dutch. Folia Linguistica 43.1: 135–169. Seuren, Pieter A.M. and Camiel Hamans (2010). Antifunctionality in language change. Folia Linguistica 44.1: 127–162. Seuren, Pieter A.M. and Dany Jaspers (2014). Logico-cognitive structure in the lexicon. Language 90.3: 607–643. Slobin, Dan (1987). Thinking for speaking. In: Jon Aske, Natasha Beery, Laura A. ­Michaelis and Hana Filip (eds), Berkeley Linguistics Society. Proceedings of the 13th Annual Meeting. Febr. 14–16, 1987. Berkeley CA: Berkeley Linguistics Society: 435–445. Sofia, Estanislao (2015). La “Collation Sechehaye” du ‘Cours de linguistique générale’ de Ferdinand de Saussure. Édition, introduction et notes. (= Orbis Supplementa vol. 43). Leuven: Peeters. Spencer, Herbert (1855). The Principles of Psychology. London: Longman, Brown, Green, and Longmans.

Bibliography

259

Starobinski, Jean (1971). Les mots sous les mots. Les anagrammes de Ferdinand de Saussure. Paris: Gallimard. (English translation by Olivia Emmet: Words upon Words: The Anagrams of Ferdinand de Saussure. New Haven/London: Yale University Press, 1979.). Stawarska, Beata (2015). Saussure’s Philosophy of Language as Phenomenology: Undoing the Doctrine of the Course in General Linguistics. New York: Oxford University Press. Steiner, Peter (1995). Russian Formalism. In: Raman Selden (ed.), The Cambridge History of Literary Criticism, Vol. 8: From Formalism to Poststructuralism, Cambridge: Cambridge University Press: 11–29. Steinthal, Heymann (1860). Charakteristik der hauptsächlichsten Typen des Sprachbaues. (Neubearbeitung von Dr. Franz Misteli). Berlin: Dümmler. Stout, George F. (19184; 18961). Analytic Psychology. 2 vols. London: Allen & Unwin / New York: Macmillan. Stout, George F. (1899). A Manual of Psychology. Vol. I. London: W.B. Clive / New York: Hinds & Noble. Strawson, Peter F. (1964). Identifying reference and truth-values. Theoria 30.2: 96–118. Stuart Mill, John (1843). A System of Logic, Ratiocinitive and Inductive. Presenting a Connected View of the Principles of Evidence and the Methods of Scientific Investigation. (2 vols.) London: John W. Parker. Svedelius, Carl (1897). L’analyse du langage appliquée à la langue française. Uppsala: Almqvist & Wiksell. Swaab, Dick (2014). We Are Our Brains: A Neurobiography of the Brain, from the Womb to Alzheimers’. New York: Penguin Random House. Szabolcsi, Anna (1981). The semantics of topic-focus articulation. In: Jeroen A.G. Groenendijk, Theo M.V. Janssen and Martin B.J. Stokhof (eds.), Formal Methods in the Study of Language (3 vols) (= Mathematical Centre Tracts 136), Amsterdam: Mathematisch Centrum: 513–540. Taine, Hippolyte (1870). De l’intelligence. (2 vols.) Paris: Hachette. (3rd edition 1878). Taine, Hippolyte (1872). History of English Literature. (2 vols.) Translated by H. van Laun. Edinburgh: Edmonston & Douglas (3rd edition; original French edition 1863–1864). Taylor, John R. (2012). The Mental Corpus. How Language is Represented in the Mind. Oxford: Oxford University Press. Tesnière, Lucien (1959). Éléments de syntaxe structurale. Paris: Klincksieck. Tieken-Boon van Ostade, Ingrid (ed.) (2008). Grammars. Grammarians and GrammarWriting in Eighteenth-Century England. Berlin: Mouton de Gruyter. Titchener, Edward B. (1898). The postulates of a structural psychology. Philosophical Review 7: 449–465. Trier, Jost (1931). Der deutsche Wortschatz im Sinnbezirk des Vertsandes. Die Geschichte eines sprachlichen Feldes. Heidelberg: Winter.

260

Bibliography

Van Ginneken, Jac. (1926). Notice on F. de Saussure, Cours de linguistique générale, 2nd. ed. 1922. Indogermanisches Jahrbuch 10.5: 37. Van Kemenade, Ans (1992). The history of English modals. A reanalysis. Folia Linguistica Historica 13.1/2: 143–166. Van Kuppevelt, Jan C.J. (1991). Topic en Comment. Expliciete en impliciete vraagstelling in discourse. PhD-thesis, Radboud University Nijmegen. Van Wijk, Nicolaas (1939). Phonologie: een hoofdstuk uit de structurele taalwetenschap. The Hague: Nijhoff. Verner, Karl A. (1875). Eine Ausnahme der ersten Lautverschiebung. Zeitschrift für vergleichende Sprachforschung 23.2: 97–130. Wackernagel, Jacob (1916). Ein schweizerisches Werk über Sprachwissenschaft. Basler Zeitung 15 & 22 Oct. Wegener, Philipp (1885). Untersuchungen über die Grundfragen des Sprachlebens. ­Halle: Niemeyer. Wells, Rulon S. (1947). De Saussure’s system of linguistics. Word 3.1/2: 1–31. Wertheimer, Max (1923). Untersuchungen zur Lehre von der Gestalt II. Psychologische Forschung. Zeitschrift für Psychologie und ihre Grenzwissenschaften 4: 301–350. Whitehead, Alfred N. and Bertrand A.W. Russell (1910–1913). Principia Mathematica. 3 vols. Cambridge: Cambridge University Press. Whitney, William D. (1867). Language and the Study of Language. Twelve Lectures on the Principles of Linguistic Science. London: Trübner & Co. Wulf, Andrea (2015). The Invention of Nature: Alexander von Humboldt’s New World. New York: Alfred A. Knopf. Wunderli, Peter (1976). Saussure als Schüler Sechehayes? In: Hans-Josef Niederehe & Harald Haarmann (eds), In Memoriam Friedrich Diez, Studies in the History of the Language Sciences 9, Amsterdam: Benjamins: 419–460. (reprinted in Wunderli 1981: 180–200). Wunderli, Peter (1981). Saussure-Studien: exegetische und wissenschaftsgeschichtliche Untersuchungen zum Werk von F. de Saussure. Tübingen: Narr. Wundt, Wilhelm (1880). Logik. Eine Untersuchung der Principien der Erkenntnis und der Methoden Wissenschaftlicher Forschung. Stuttgart: Enke. Wundt, Wilhelm (1900). Völkerpsychologie. Eine Untersuchung der Entwicklungsgesetze von Sprache, Mythus und Sitte. Volume 1: Die Sprache. (2 parts.) Leipzig: Kröner. Zifonun, Gisela, Ludger Hoffmann, Bruno Strecker et al. (1997). Grammatik der deutschen Sprache. (3 vols). Berlin: Walter de Gruyter.

Index Aarsleff, H. ix, 84, 95 abduction 210 accent 55, 87, 88, 136, 137, 163–165, 171–175, 181–184–189, 194, 236, 244 African languages 82 Ajdukiewicz, K. 177, 236 Alexandrian grammarians 63, 237 algebra 26, 112 Boolean 26 algorithm 68, 69, 131, 179, 208, 243 Allan, K. 1, 27, 34, 63, 101 allomorphy 85, 86 Amacker, R. 111, 117, 120, 122, 125 American-Indian languages 82, 109 analogy 57, 97, 232, 235 anaphora 180, 186, 189, 195 anarchism 46 anchoring 66, 179–181 Andresen, J.T. 110 Apollonius Dyscolus 63 arbitrariness of linguistic sign 18, 34, 80–88 archeology 221 Aristotle 25, 62, 83, 126, 127, 130, 161, 162, 165–167, 176, 193, 208, 239 Arnault, A. 26, 27, 63, 68, 130, 142 Ashworth, E.J. 157 Asian languages 82 associative relations 34, 60, 117–119, 205 Auer, A. 28 Augustine 73, 78, 79 auxiliation 94 Aymara 155 Balbuena, F. 103 Bally, Ch. 12–14, 35, 41, 42, 99, 104, 105, 111, 119–124, 149 Baudouin de Courtenay, J. 15, 223 Bauer, B. x, 229 Bauhaus 198 Bass, A. 40 Beauzée, N. 27, 28 Bech, G. 62, 237 behaviourism 50, 55, 127, 148, 197, 202, 211, 216, 227 Behme, C. 211

Bernoulli, J. 26, 28 Bielenstein, A. 233 bleaching (semantic) 94, 113, 156 Bloomfield, L. 32–34, 37, 39, 50, 62, 81, 87, 124, 131, 153, 199–202, 205, 218, 224, 234 Blumenthal, A.L. 199, 227 Bogrea, V. 32 Bolinger, D.L. 81 Boole, G. 26, 28 Bopp, F. 229 Boring, E.G. 49 Borowski, E. 190 Borwein, M. 190 Bougainville, L.-A. 221 Bourdeau, C. x, 160 Bourdon, B. 32 Bovet, P. 32 Bréal, M. 94, 95 Brentano, F. 23 Breva-Claramonte, M. 27, 63 Brouwer, L.E.J. 137 Brugmann, K. 10, 17, 18, 104, 229, 230, 233 Buffier, C. 27 Bühler, K. 49, 79 Calvin, J. 31 cardinal numbers 192, 193 categorial grammar 177, 236 causality 210 Celano, G.G.A. 211 characteristic function 75 Charlton, D.G. 47, 59 chess comparison 89–91 Chevalier, J.-Cl. 237 Chinese 109 Chomsky, N. 52, 54, 68, 153, 201, 202, 211 Clarapède, E. 32 clefting 172, 173, 181–184, 189 communism 46 Comorovski, I 179, 189 competence 54–56 complementation 152, 153 componential analysis (of lexical meaning) 236 compositionality 85, 86, 236

262 Comte, A. 24, 25, 46–48, 94, 223 constituent structure 39, 87, 183 Cook, J. 221 coordination 157, 158 Copenhagen School 201 Cratylus 83, 86 Curtius, G. 10 Cuvier, G. 38 Darwin, Ch. 221 Davidson, D. 142 De Mauro, T. 9, 14–19, 125 De Morgan, A. 141 deduction 131, 140, 143, 148 Delbrück, B. 10, 21, 35, 100, 229 dependency stemma 177, 178 Derrida, J. 39, 40, 92 Descartes, R. 59, 137, 210 description, definite 7 Desmarais R. 27 diachrony 23, 24, 34, 45, 51, 57, 89, 90, 94–101, 143, 144, 149 diglossia 234 Dijksterhuis, E.J. 199 Dilthey, W. 213, 214, 221 Dingemanse, M. 81 direme 121, 152 Discourse Domain 175 dissimilation 231, 232, 234 Donnellan, K. 189, 190 double articulation 15, 39, 93 Dresher, B.E. 15 Dumarsais, C.C. 27 Durkheim, É. 47, 48 Dutch 154, 155 dyadic propositional structure 176 Dyirbal 81 ‘Echo’-reading École Pratique des Hautes Études 11, 29, 30 Egger, É. 216 Egger, V. 69, 78–80, 84, 216, 217, 219 Ehrenfels, Chr. von 224, 226 Eisenberg, P. 237 Elffers-van Ketel, E. x, 21, 123, 160, 239 empiricism 208 Engler, R. 13, 26, 125 enlightenment 28, 46, 110, 162

Index Esperanto 10 Eubulides 193 Euler, L. 26, 28 European languages 86 Everett, D. 153 exponibilia 157 Feeder 169 Fichte, J.G. 48 Fillmore, Ch. 236 Finck, F.N. 54 Firth, J.R. 33 Fitzroy, R. 221 Flämig, W. 237 Fodor, J.A. 218 Fogelin, R.J. 193 folk etymology 234 Fónagy, I. 81 Foolen, A. x formalism, Russian 198, 200 Fourier, J.-B. 28 Franklin, B. 110 Frege, G. 26, 141, 193, 195 Frei, H. 105 Fretheim, T. 160 Freud, S. 104 Frýba-Reber, M. x, 12, 13, 32, 102–107, 110, 111, 113, 114, 122–124, 141, 142, 155 Gabelentz, G. von der 160, 163, 164 Gadet, F. 200 Galileo 239 Gall, F.J. 47, 217 Gambarara, D. 27 Gardiner, A.H. 54, 67, 77, 165, 166, 169 Gasparov, B. 41, 42 Gautier, L. 31, 32 Geerts, G. 237 General Semantics 73 generative grammar 63, 125, 130, 201 generative semantics 68, 130, 211 Geneva school of linguistics 5, 7, 105, 111 Geneva University 5, 12, 13, 19, 29, 31, 104, 133, 242 genitive absolute 11, 60 German 62, 98, 108, 132, 184–185, 229, 230 gestalt 197, 224–227, 241 gestural language 88, 97

263

Index Girard, G. 27 glossematics 201 Godel, R. 12, 13, 19, 51, 57, 59, 89, 90, 96, 107, 125, 144 Goldbach, C. 28 Goosse, A. 237 Gothic 11, 29 Graffi, G. 49, 60 grammaire générale 26, 28 grammar See linguistics grammatology 39, 40 Grammont, M. 32 Greek 16, 27, 60, 81, 86, 97, 99, 229, 230 Greenbaum, S. 237 Grevisse, M. 237 Grice, Paul H. 192 Grimm, J. 229–231 Grimm, W. 229 Groot, A.W. de 205 Groult, P. 232 Gundel, J. 160 Halle, M. 143 Hamans, C. x, 43, 98, 113, 186 Harris, Z.S. 37, 131, 201, 202, 206 Hausa 81 Havet, L. 107 Hebrew 27 Hegel, G.W.F. 41, 213, 228 Heidolph, K.E. 237 Helbig, G. 33 Heraclitus 179 Herder, J.G. 48 Hewson, J. 38 Higgins, F.R. 189 Hittite 11, 230 Hjelmslev, L. 201 Hrozný. B. 11 Huddleston, R. 237 Humboldt, A. von 221 Humboldt, W. von 28, 94, 109, 132, 221 Humpty Dumpty 219 Hymes, D. 1 Idealism 67, 143, 213, 223 ideophones 82 Iemhoff, R. 137

immediate constituent analysis 124, 153, 205, 224, 243 implicature 192 incrementation 168 induction 131, 143, 148, 210, 211 information structure 165, 166 Institut Jean-Jacques Rousseau 105 instrumentalism 198, 211 intentionality 58, 74, 76, 80 Interlingua 10 intonation 55, 87, 88, 136, 165, 169, 172, 173, 175, 183, 238 introspection 49 intuitionism 137 Irish 81 isogloss 233 Italian 98, 108 Jaberg, K. 32 Jankowsky, K.R. 94, 132, 228 Jakobson, R. 16, 18, 37, 38, 143, 165, 198, 200, 204 James, W. 75 Jaspers, D. x, 142, 193, 200 Jefferson, Th. 110 Jena University 10, 11 Jespersen, O. 32, 67, 94, 97, 149, 154, 164–167, 169, 170, 181, 237, 244 Jones, W. 1 Joseph, J. 9, 10, 12, 16–18, 24, 31, 32, 57, 69, 83, 86, 120 Junggrammatiker See Neogrammarians) 11 Kalepky, T. 165, 238–240 Kanizsa triangles 226 Kant, I. 28, 59, 137, 210, 213 Karcevski, S. 105 keying 66, 74, 77, 179–181, 191 Kilivila 81 Kitaoka, A. 226 Klippi, C. 28 Klooster, W.G. 191, 237 Koerner, E.F.K. 32, 33, 35, 38, 218 Koffka, K. 224 Köhler, W. 224 Korean 128 Kripke, S. 142

264 Kruszewski, M. 15 Külpe, O. 49 Kuryłowicz, J. 11, 230 Kuteva, T. 94 La Mettrie, J.O. de 217 Labov, W. 45 Lagrange, J.L. 28 Lakoff, R. 63 Lambert, C.H. 28 Lancelot, C. 26, 27, 63, 68, 130, 142 ‘langage’ 53, 57, 130, 136, 139 Laplace, P.-S. 28 laryngeals 93, 230 Latin 16, 27, 60, 97, 98, 99, 232, 233 Laun, H. van 59 Lazarus, M. 49 Leech, G. 237 Legendre, A.-M. 28 Leibniz, G.W. 28, 210, 225 Leipzig University 10, 11, 21, 104 Lepschy, G. 13 Levelt, W.J.M. x, 7, 21, 36, 44, 48–50, 81, 173, 179, 199, 216, 224, 227 Leskien, A. 10 Lewis, D. 142 lexical items as predicates 128, 153, 154–156 lgbt rights 221 linear order 132, 243 linearity of signifier 87, 88, 243 linguistics anthropological 110 comparative-historical 11, 16, 44, 51, 99, 100, 221, 242 computational 177 frequency 54, 55, 127, 183 Lipps, Th. 160, 165–167, 169 Locke, J. 208 logic 25, 26, 44, 66, 67, 100, 112, 126, 127, 140–143, 147, 148, 156, 162, 168, 169, 176, 198 Longfellow, H.W. 98 Luther. M. 98 Luys, B. 216, 217, 219 Lyons, J. 33, 34, 36, 100, 204, 205, 236 Malay 81 Malinowski, B. 33

Index Maltese 81 Martinet, A. 14, 15, 39, 93 Marty, A. 23, 95 Marx, K. 47 materialism 209 Mathesius, V. 165, 200 Matthaei, R. 224–226 Matthews, P.H. 205 Mattmüller, M. 28 Mauritian Creole 155 McCawley, J.D. 130, 153, 156, 204 Meillet, A. 11, 15, 30–32, 38, 43, 54, 96, 99, 106, 107, 120, 125, 129, 149, 223 Mettrie, J.O. de la 47 Meyer-Lübke, W. 161 Michotte, A. 210 microplanning 173 Mikkelsen, L. 189 Mill, J.S. 28 modal verbs 97 modernity 197 modularity 45, 46 Moleschott, J. 219 monoreme 151, 152, 167, 168 Montague, R. 73, 142, 194, 195, 236 morphology 60, 61, 63, 65, 87, 127, 132, 145 Morpurgo Davies, A. 11, 17, 54 Morris Ch.W. 79 Motsch, W. 237 Müller, F.M. 19, 20, 25 Müller, O. 122 Naville, A. 22, 25, 133 negation 184, 185 copying 94 negative polarity item (npi) 185 Niedermann, N. 32 Nevins, A. x, 16, 200 Newton, I. 28 Noordegraaf, J. x Noreen, A. 149 Nuchelmans, G. 178 Object-oriented programming 192 Ockham’s razor 191, 209 Ogden, C.K. 77, 79 Ohala, J. 81 Old-Germanic 16, 62, 229

265

Index Old High German 11, 29 Oltramare, A. 32 onomatopoeia 234 Osthoff, H. 10, 17, 18, 230, 233

Qualia problem 71 quantifiers 128 Quine, W.V.O. 193–195 Quirk, R. 237

Paardekooper, P.C. 237 paradigmatic relations 60 paradox of knowledge 59 parameter (theory) 7, 170, 188, 190–195 parole 52–61, 64, 71, 74, 76, 88, 121, 170–172 parsing 237 Partee, B. 191, 194 Paul, H. 21, 34, 64, 95–97 Peano, G. 26, 141 Peirce, Ch.S. 75, 78 Percival, K. 1, 23, 29, 36–39, 130, 199, 200, 201, 224 performance 54 phoneme 15, 108, 119, 204, 230 phonology 15, 87, 94, 97, 108, 132, 133, 145, 236 Pictet, A. 18, 83, 84, 86, 109, 207, 228 Pirahã 153 Plato 83, 86 polyadic propositionl structure 176 Popper, K. 140, 210 Port Royal grammar 26, 27, 28, 63, 68, 130, 141, 142 Portuguese 98, 99, 234 positivism 46, 47, 198, 202, 211 Pourtalès, Louise de 9 pragmatics 45, 136, 137, 148, 160, 165, 172–174, 181, 184, 185, 187, 192 Prague School 165, 170, 199, 200 predicate-argument structure 128, 134, 142, 143, 147, 176–178 pregrammatical 116, 138, 145 presupposition 168, 180, 187 Priscian 68 proposition 2, 7, 8, 25, 28, 50, 61–68, 71, 72, 76, 119, 126, 127, 129, 141–143, 147, 148, 151, 152, 160, 161, 178–180, 184, 185, 203, 204, 209, 211, 240, 243, 244 prosodic structure 38, 39 Provençal 98 pseudocleft 189 Pullum G.K. 237

Racism 109 Rask, R. 229 rationalism 7, 24, 28, 63, 81, 83, 93, 99, 141, 166, 197, 207–209, 211–214, 221–223, 228, 230, 241, 244 reductionism 47, 137, 214, 215 reference 66, 74, 80, 180–181, 189–190, 243 reflexives 182 regrammaticalisation 94 Rescher, N. 210 reification 71, 155, 156, 179, 191, 209 rhotacism 97 Richards, I.A. 77, 79 Riedlinger, A. 13, 96, 144 Ries, J. 63, 65–67 rituals 18 Robins, R.H. 27, 96 Robinson, J.O. 226 Rojas-Berscia, L.M. x, 155 Romance languages 98, 99 romanticism 7, 24, 28, 46, 62, 63, 67, 81, 83, 93, 143, 162, 166, 173, 197, 207, 208, 211–214, 220–223, 228, 231, 238, 241, 244 Ronjat, J. 32 Rorschach test 227 Rubin, E. 227 Russell, B. 26, 142, 178 Sanctius F. 26, 63, 141 Sandmann, M. 165, 238, 240 Sanskrit 11–13, 30, 86, 104, 111 Sapir, E. 33, 81, 149, 204 Saussure, Alphonse Jean François de 9 Saussure, Henri Louis Frédéric de 9 Saussure, Horace-Bénédict de 9 Saussure, Léopold 10 Saussure, Nicolas-Théodore de 9 Saussure, Raymond de 104, 114 Saussure, René de 10, 26 Schleicher, A. 10, 19, 32, 228 Schmaus, W. 48 Schmidt, J. 20, 233 Schooneveld, C. van 200

266 Schuchardt, H. 20, 21, 32, 149, 228, 233 Sechehaye-Burdet, J.M. 103, 104 Sechehaye-Duess, L.L. 103 semiology 22, 39, 40, 51, 73 Senft, G. x sentence 64–69 serial verb 157 Sériot, P. 200 Seuren, P.A.M. 11, 16, 18–20, 28, 39, 43, 55, 58, 59, 62, 63, 73, 79, 86, 87, 90, 94, 98, 109, 113, 126, 142, 166, 167, 175, 178–181, 184, 193, 194, 201, 211, 216, 228 Siebers, J. x Sievers, E. 104 sign – 69–74, 78, 97, 117, 119, 136 arbitrariness of 80–88 history of notion 78–80 signified/signifier 69, 87, 92, 243 visual 87 Slobin, D 173, 179 sloppy identity 189 social engineering 198 social reality 45, 47 socialism 46 Société de Linguistique de Paris 16 sociolinguistics 45, 46, 100 Sofia, E. 13, 42, 120 solipsism 210 sound symboliwsm 234 speech act operator 62, 66, 90, 127, 202 theory 158 speech circuit 58, 243 spelling pronunciation 234 Spencer, H. 47, 217 Square of Opposition 161 Sranan 156 Starobinski, J. 16 Stawarska, B. 41, 42 Steiner, P. 198 Steinthal, H. 49, 160, 162, 163 Stenning, K. 187, 188 Stoa 73, 78, 208 Stout, G.F. 82, 160, 165–169, 181, 225, 244 Strawson, P.F. 187 structuralism 6–8, 14, 29, 37, 50, 63, 102, 125, 132, 197–207, 211, 214, 222, 224–227, 235–242, 244 stylistics 104, 120–122

Index subject-predicate debate 21, 22, 97, 99, 130, 157, 166–169, 181, 196, 244 subject-predicate structure 25, 26, 119, 124, 130, 132, 134, 142, 143, 146–148, 151, 152, 157, 160–196, 162, 167–169, 172, 174, 176, 238–240 subordination 157–159 substitution salva veritate (ssv) 193, 194 substrate 234 superstrate 234 surface structure 127, 128, 130–134, 143, 148, 154 Svartvik, J. 237 Svedelius, C. 165, 238, 239 Swaab, D. 219 Swahili 81 synchrony 23, 24, 34, 45, 51, 57, 89, 94–101, 143, 144, 149 syntagm 61 syntagmatic relations 60, 61, 117–119, 205 syntax 36, 52, 54, 60–67, 87, 98, 99, 119, 123, 127, 132, 145, 146, 175, 243 Szabolcsi, A. 179, 182 Taine, H. 16, 47, 48, 59, 69, 75, 77–80, 137, 215, 217, 223 Tarski, A. 142 Taylor, J. 54 Terracini, B. 32 Tesnière, L. 15, 177 thematic roles 239 thinking-for-speaking 173 thought 70, 71, 132, 137, 150, 164, 168–170, 173, 175 Tieken-Boon van Ostade, I. 28 Titchener, E.B. 199, 206, 227 token physicalism 47, 138, 215, 219 topic-comment distinction 121 modulation (tcm) 165, 168–174, 184–189, 194, 244 transformation 63, 64, 127, 129, 130, 131, 146, 243 transformational grammar 132, 134, 140, 201 Trier, J. 236 Trubetzkoy, N. 200 Turkish 184 type-token distinction 74–80, 85, 179, 180, 218

267

Index Unexceptionability of sound laws 147 ‘Valeur’ 89, 91, 92, 94, 118 Van Ginneken, J. 33 Van Kemenade, A. 97 Van Kuppevelt, J. 169, 181 Van Wijk, N. 205 variable 91, 93, 113 Vaugelas, Cl. F. de 27, 237 Vedic 16 Vendryes, J. 149 Verner, K. 229, 230 virtual reality 58 völkerpychology 48–50, 140, 161, 224 Vufflens, Château de 9, 10, 36 Wackernagel, J. 32 wave model 20 Weber, M. 47 Wegener, Ph. 21, 160, 165–167, 169, 244 Wells, R.S. 6 Wertheimer, J. 12, 104

Wertheimer, M. 224 Whitehead, A. 26 Whitney, W.D. 18–20, 34, 109 Whorf, B.L. 72 Wikipedia 35, 36 Wilmotte, M. 122 Windisch, E. 104 Wittgenstein, L. 15 ‘wortfeld’ 236 Wulf, A. 221 Wunderli, P. 23 Wundt, W. 2, 21, 26, 34, 35, 39, 49, 50, 64, 87, 91, 99, 100, 112, 124, 126, 130–136, 140, 143, 145, 146, 149, 161, 164, 172, 203, 213, 224, 227, 242 Young Grammarians 11, 18, 19, 21, 24–26, 52, 83, 84, 90, 95, 97, 100, 104, 132, 147, 222, 227, 228, 230, 231, 233–235, 241 Zifonun, G. 237