Charles H. Kahn (Auth.), Simo Knuuttila, Jaakko Hintikka (Eds.) the Logic of Being- Historical Studies

THE LOGIC OF BEING

A PALLAS PAPERBACK

~p~

\]Q] DaDerbaCkS

THE LOGIC OF BEING Historical Studies

Edited by SIMO KNUUTTILA Dept. of Systematic Theology, University of Helsinki

and JAAKKO HINTIKKA Dept. of Philosophy, Florida State University

D. REIDEL PUBLISHING COMPANY A MEMBER OFTHE KLUWER

ACADEMIC PUBLISHERS GROUP

DORDRECHT ,I BOSTON / LANCASTER / TOK YO

Library of Co ngress Cataloging in Publica tion Data Main entry under title: The Logic of Bting. (Synthese historitallibrary; v. 28) Includes indexes. Contents: Introduction - Retrospttt on the verb ' to be' and the concept of being I Charles H. Kahn - Identity and predication in Plato I Benson Mates Aristotle and existence I Russell M. Dancy - letc.] I. Ontology - Histo ry Addresses. essays, lectures. I. Knuuttila. Simo, 1946II . Hintikka, Jaakko. 1929Ill. Series. 111 85- 19622 BD3)I.L826 1985 ISBN _13: 978_90_211_237 1_0 001 : 10.1007/978-94-009-4780-1

Published by D. Reidel Pu blishing Company P.O . Box i7, 3300 AA Dordrecht . Holland Sold and dist ributed in the U .S.A . and Canada by Kluwer Academic Pu blishers, 101 Philip Drive. As~in ippi Park, Norwell. MA 02061. U.S.A. In all other cou ntries. sold and dist ributed by Kluwer Acade mic Pu blishers Group. P.O. Box 322. 3300A H Do rdrecht , Holland Also published in /986 in hordbound edilion by Reidel in Ihe Synthese Historicol Library, Volume 18

All Ri ghts Reserved @ 1986 by D. Reidel Publishing Company, Dordrecht, Holland

No pan of the material protected by this copyright notice may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying, reco rding or by any information storage and retrieval system, without written pcrmission from the copy right owner

TABLE OF CONTENTS

ACKNOWLEDGEMENTS

vii

INTRODUCTION

ix

CHARLES H. KAHN / Retrospect on the Verb 'To Be' and the Concept of Being BENSON MATES / Identity and Predication in Plato

29

RUSSELL M. DANCY / Aristotle and Existence

49

JAAKKO HINTIKKA / The Varieties of Being in Aristotle

81

STEN EBBESEN / The Chimera's Diary

115

KLAUS JACoBI/Peter Abelard's Investigations into the Meaning and Functions of the Speech Sign 'Est'

145

HERMANN WEIDEMANN / The Logic of Being in Thomas Aquinas

181

SIMO KNUUTTILA / Being qua Being in Thomas Aquinas and John Duns Scotus

201

LILLI ALANEN / On Descartes's Argument for Dualism and the Distinction Between Different Kinds of Beings

223

JAAKKO HINTIKKA / Kant on Existence, Predication, and the Ontological Argument

249

LEILA HAAPARANTA / On Frege's Concept of Being

269

INDEX OF NAMES

291

INDEX OF SUBJECTS

297

ACKNOWLEDGEMENTS

.Tllakko Hintikka's work on 'The Varieties of Being in Aristotle' was supported by NSF Grant No. BNS 8119033. Btnson Mates's paper 'Identity and Predication in Plato' first appeared in Phronesis 24 (1979), 211 - 229. Russell M. Dancy's paper, 'Aristotle and Existence', appeared in Synthese 54 (1983), 409 - 442. laakko Hintikka's paper 'Kant on Existence, Predication, and the Ontological Argument' first appeared in Dialectica 35 (1981), 127 -146. Leila Haaparanta's paper 'On Frege's Concept of Being' contains some material which has also appeared in her dissertation Frege's Doctrine of Being (Acta Philosophica Fennica, vol. 39), Societas Philosophica Fennica, Helsinki 1985. All the previously published material appears with the permission of the author and of the editor or publisher, which the editors thus gratefully acknowledge.

vii

INTRODUCTION

The last twenty years have seen remarkable developments in our understanding of how the ancient Greek thinkers handled the general concept of being and its several varieties. The most general examination of the meaning of the Greek verb 'esti'/'einai'/'on' both in common usage and in the philosophical literature has been presented by Charles H. Kahn, most extensively in his 1973 book The Verb 'Be' in Ancient Greek. These discussions are summarized in Kahn's contribution to this volume. By and large, they show that conceptual schemes by means of which philosophers have recently approached Greek thought have not been very well suited to the way the concept of being was actually used by the ancients. For one thing, being in the sense of existence played a very small role in Greek thinking according to Kahn. Even more importantly, Kahn has argued that Frege and Russell's thesis that verbs for being, such as 'esti', are multiply ambiguous is ill suited for the purpose of appreciating the actual conceptual assumptions of the Greek thinkers. Frege and Russell claimed that a verb like 'is' or'esti' is ambiguous between the 'is' of identity, the 'is' of existence, the copulative 'is', and the generic 'is' (the 'is' of class-inclusion). At least a couple of generations of scholars have relied on this thesis and frequently criticized sundry ancients for confusing these different senses of 'esti' with each other. Others have found the distinction between the different Fregean senses in this or that major Greek philosopher, or otherwise used the distinction as an integral part of their interpretative framework. Kahn's results show that all these lines of argument are highly suspect. Independently of Kahn, Michael Frede (in his Habilitationsschrijt published in 1967 under the title Priidikation und Existenzaussage) reached the conclusion that Plato did not - at least not in the Sophist - accept anything like the Frege - Russell distinc~ion, thus striking another blow against the received views. We hoped to include excerpts of Frede's little classic here. Unfortunately, for reasons beyond our help this turned out to be impossible. IX

S. Knuullila and 1. Hintikka (eds.), The Logic of Being, ix-xvi. Co, 1986 by D. Reidel Publishing Company.

x

INTRODUCTION

Most philosophers and most classicists were not immediately convinced by Kahn and Frede. It seemed to them that to say that Plato (or any other Greek philosopher) did not show the Frege - Russell distinction amounts to accusing him of a logical howler. Isn't that distinction a purely logical one, an indispensable part of our logical apparatus? This question seems to have haunted even those students of Plato and Aristotle who had found the Frege - Russell distinction less a research tool than a Procrustean bed. They seem to have been reluctant to claim in so many words that the greatest Greek philosophers did not really operate with the Frege - Russell distinction. In the case of Plato, it took a scholar like Benson Mates, who combines a high degree of logical expertise with historical scholarship, to come out of the closet and argue expressly that to acknowledge that Plato Ia,cked the distinction is not to accuse him of any mistake; for there are alternative logical treatments of 'is' (or 'esti') which do not presuppose any irreducible ambiguity in these verbs between the allegedly different Frege - Russell senses of 'is'. Mates's article 'Identity and Predication in Plato' appeared in 1979, and is reprinted in this book. He argues there that it is a radical mistake to try to project the distinction onto Plato. In the same spirit as Mates, but with an entirely different alternative logical framework in mind, laakko Hintikka argued in his paper' "Is", Semantical Games and Semantical Relativity' (1979) that the Frege - Russell distinction is not only dispensable but indeed a mistaken representation of the logic of natural language. In the first half of Hintikka's paper 'The Varieties of Being in Aristotle', appearing for the first time here, he argues that Aristotle did not operate with the Frege - Russell distinction any more than Plato did (according to Frede and Mates), and traces some of the consequences of this insight. In some ways, these developments are only the tip of an iceberg. Much of the best work on Greek philosophy in the last twenty years has been inspired or otherwise influenced by the late G. E. L. Owen. He made it entirely clear that Aristotle handled the concept of being in a way essentially different from what the Frege - Russell logic leads us to expect. One of the many stimulating suggestions Owen made was that the purely existential uses of 'esti' in sentences of the type 'Homer is', in the sense 'Homer exists', are construed by Aristotle as being in the last analysis elliptical for 'Homer is a man', more generally 'Homer is what he essentially is'. Owen never argued for his position, however, and he may never have adopted it. In his paper, 'Aristotle and Existence', reprinted here, Russell M. Dancy sets Owen's idea against the background of Aristotle's

INTRODUCTION

xi

treatment of the concept of being in general and defends it at length. Dancy's paper serves at the same time as a useful introduction to the problems posed by Aristotle's use of the concept of being. The alleged Frege - Russell ambiguity does not exhaust the varieties of being that are relevant here. Although Aristotle fails to postulate the Frege - Russell distinction, he does assume a distinction between the different senses of 'esti' used in the different categories. Aristotle's doctrine of categories has recently been the subject of a great deal of interesting discussion. Much of it was prompted by G. E. L. Owen, who in 1960 published a paper entitled 'Logic and Metaphysics in Some Early Works of Aristotle'. The paper has since become a modern classic. It contains a discussion of how Aristotle sought to overcome the distinction of different senses of 'esti' in the different categories by means of the idea of pros hen multiplicity of uses. The contrast between these uses is therefore something less than a full ambiguity (or homonymy, in Aristotle's terminology). It is nevertheless far from clear what Aristotle's distinction between different categories really amounts to. In the second part of his paper 'The Varieties of Being in Aristotle' (see pp. 96 - 112) laakko Hintikka presents a systematic analysis of logical quantification in natural languages which yields as a by-product a rational reconstruction of Aristotle's theory of categories. Among other things, Hintikka offers an explanation of how the different aspects of Aristotle's theory go together - in particular, how Aristotle can consider the distinction between different categories sometimes as a distinction between classes of simple predicates, sometimes as a distinction between the different largest classes of entities one can speak of together, sometimes as a distinction between different senses of 'esti', and sometimes (especially in choosing his names for the different categories) as going together with the different kinds of wh-questions in Greek. It turns out that the topical imperfections of this rational reconstruction naturally lead us to some of the same conceptualizations as were used by Aristotle in further developing his theory of categories, especially to the matter-form contrast. Here some extremely interesting possibilities of cooperation between systematic and historical work seem to be opening up. From this survey, it is apparent that the first half of our volume is calculated to present a reader with the basic materials documenting the new perspective on the ways in which the major Greek philosophers dealt with the concept of being. The second half of our volume discusses some important aspects of the subsequent history of what we have called 'the

xii

INTRODUCTION

logic of being'. As shown in the papers of Klaus Jacobi and Sten Ebbesen, early medieval inquiries into the logic and semantics of 'is' were a part of an investigation of the nature of predication. It was usually thought that the standard logical form of an affirmative proposition could be thought of as a three-part form consisting of a subject term, a predicate term, and an interposed copula. In the Aristotelian manner, the three-part form was conceived of as an explanatory reformulation of a two-part form, in which a noun in the nominative case is combined with an inflected verb. On this approach, the question of the properties of the copulative 'is' became the main problem of the theory of predication. One of the difficulties was to understand the relation between 'is' as tertium adiacens, that is, as a copula, and 'is' as secundum adiacens, that is, without additions. Abelard's attempts to solve the question and his reports of the theories of his contemporaries are discussed by Klaus Jacobi in his contribution to this volume. In Abelard's time there were two main alternative positions. According to one theory, the function of 'is' as a copula is to join the semantic content of the predicate term to that of the subject term. It can exercise this function only when it has no semantic content of its own. No connection was seen between the copula and the 'is' used as secundum adiacens. The opponents of this equivocation theory argued that 'is' used as secundum adiacens expresses that the thing under discussion exists, but that, when 'is' is used copulatively, the predicate serves to determinate the manner in which the subject exists. According to this view, the actual multiplicity of uses if 'is' is not accompanied by any genuine multiplicity of meaning. The non-copulative use was thought of as an existential one, and it was suggested that the copulative propositions with non-existent subject terms can always be translated into forms having existential import. In many places Abelard seems to hesitate between these alternatives. This hesitation was connected with the fact, Jacobi argues, that both alternatives were based on the three-part analysis of a proposition, whereas Abelard's main interest was to develop a theory of predication in which a two-part form is preferred. From this vantage point he tried to interpret the copula as an auxiliary verb, which in conjunction with a predicate noun does duty for verbs which often are not yet invented. Abelard's ideas did not win any adherents, and as stated by Sten Ebbesen, in the thirteenth century the equivocation theory met with a general disapproval. In his paper Ebbesen delineates the ancient and medieval discussion of the problems of non-existent things, as ex-

INTRODUCTION

xiii

emplified by statements about chimera. One of the much discussed medieval examples was the consequence: Chimera is opinable, therefore, chimera is. This is, according to Aristotle, a fallacy secundum quid and simpliciter, not a fallacy of equivocation (Soph. EI. 167al - 6). Similarly, most of the mid-thirteenth century writers thought that 'is' as secundum adiacens signifies actual being simpliciter and 'is' as tertium adiacens signifies a special sort of being. This doctrine, influenced by Aristotle's logical writings, was already known in Abelard's time. Some developments can be seen from new attempts to define those special sorts of being which do not entail actual existence. It is historically interesting that the question of the existential import of 'is', when not used as secundum adiacens, was usually discussed in terms of the examples like 'chimera is opinable' on the one hand, and 'chimera is chimera', on the other hand. Although later medieval philosophers of being were more interested in other kinds of distinctions, this particular problem continued to be treated, too, and thus the question of the relations between something like the 'is' of existence, the 'is' of predication, and the 'is' of identity remained one of the live topics in the logic and semantics of 'is'. In the fourteenth century, this triple distinction also attracted attention because, in the logical analysis of complex propositions into immediately verifiable ones, the three uses of 'is' are found in the basic propositions. The fully analyzed form of 'Socrates is a man', for example, was presented as follows: 'This is & this is_ Socrates & this is a man'. In the second half of the thirteenth century, the completed reception of Aristotle's works in the Latin West gave rise to a new systematic approach to questions of epistemology, ontology, and meaning. Traditional problems in the semantics of 'is' were discussed on the basis of a general philosophical theory of modes of being, intellection, and signification. 'Is' was called an analogous term whose different syntactical uses and various contextual senses were taken to be bound together by a theory largely inspired by Aristotle's remarks on the pros hen multiplicity of the different senses of 'is'. This approach was part of the theoretical framework of the philosophy of language of the modistae, as mentioned by Ebbesen. Its role in Thomas Aquinas' thought is discussed by Hermann Weidemann and Simo Knuuttila. Weidemann first treats Aquinas' basic distinction between, on the one hand, the use of the verb 'be' to express the being of something which falls under one of the categories, and, on the other hand, the use of the verb to express the truth of a proposition. This

xiv

INTRODUCTION

dichotomy, based on Aristotle's Metaphysics (1 017a22 - 35), amounts to distinguishing between two different existential uses of 'is', which are called the use of the verb in an actuality sense and in a there-is sense, respectively. By taking this distinction as a starting point, Weidemann shows how one could find the parts of the Frege trichotomy of the uses of 'is' in Aquinas. He nevertheless also argues that, on the basis of Aristotelian metaphysics and some ideas of his own, Thomas Aquinas thinks that all uses of 'is' are based on the primary existential use in an actuality sense. Simo Knuuttila discusses the ontological and epistemological ideas on which the analogical focal meaning theory of 'is' is built in Aquinas and many other medieval thinkers. Peter Aureoli wrote in 1316 that, according to the common view, when something is spoken of as a being, it is immediately conceived as a substance or quality or a quantity and not as falling under some one common definition. The famous Scotist doctrine of the univocal metaphysical concept of being is discussed by Knuuttila against the background of this disjunctive concept of being. According to Duns Scotus, even in the disjunctive approach it is in fact presupposed that, when we speak of something as being, it is eo ipso conceived as having some kind of identity, although we cannot always actually identify it. This virtual identifiability is, according to Duns Scotus, the most primary constituent of the positive nature of whatever can be and as such the content of the univocal metaphysical concept of being. Virtual identifiability does not presuppose actual existence; it can be said of anything the existence of which is logically possible. In this new metaphysics, all possible individuals are considered as metaphysical beings, divided into possible worlds, of which the actual world is one. In Scotus's theory, the same individual can be considered as actual in several alternative worlds at the same time, although it has existence in the actual world only. Scotus's views on individual identity and existence were connected with his modal theory which influenced the non-Aristotelian trends in fourteenth-century modal logic and introduced a modal semantics having similarities with the modern possible worlds semantics. Scotus's doctrine of the univocity of the metaphysical concept of being was partially adopted by the fourteenth-century nominalists, because it suggested a univocal notion of existence, which was an important part of the nominalistic omology. The nominalists contested essential parts of the metaphysical framework on which the general theory of being of their predecessors was built. It seems that in their search for a new semantics of 'is' some of them were led to think that the basic types of

INTRODUCTION

xv

the uses of 'is' are fully inderendent. Sten Ebbesen mentions John Buridan as a representative of this trend. However, fourteenth-century discussions of these questions (and the later significance of these discussions) are still only partially known. One of the medieval conceptual tools which has also been used in the modern period is the doctrine of the different kinds of ontological distinctions, e.g., real, mental, and formal distinctions. These are discussed by Lilli Alanen. She shows that Descartes's well-known argument for the mind-body distinction is based on a specific interpretation of this traditional doctrine, and that the early discussions of the argument were largely concentrated on the peculiarities of Descartes's way of drawing the ontological distinctions just mentioned. The role of these issues concerning the logic of being has not received its due attention in the earlier literature. It is sometimes said, or implied, that the Frege - Russell distinction between the allegedly different senses of 'is' goes back to Kant and to Kant's idea that 'existence is not a predicate'. In his paper, 'Kant on Existence, Predication, and the Ontological Argument', Jaakko Hintikka argues that Kant does not in the last analysis assume anything like the contemporary Frege - Russell distinction. All we find in Kant is a contrast between relative and absolute uses of the concept of being. As a part of Hintikka's argument, he offers a largely new diagnosis of the fallacy in the ontological argument, denying Kant's dictum that 'existence is not a predicate'. Instead, he finds the crucial flaw in the ontological argument elsewhere, viz. in its tacit dependence on our being able to identify God (the being of whose existence is to be proved) between the different possible worlds presupposed in the argument. This provides a new perspective on historical as well as contemporary discussions of Anselm's argument. All these different investigations naturally raise the question: What is the origin of the Frege - Russell distinction? What is its background? In her paper, 'On Frege's Concept of Being', Leila Haaparanta discusses Frege's treatment of being in its historical setting. One of the crucial ingredients in Frege's treatment of being is his idea that existence is a second-level concept (property of a concept). Haaparanta sees the foundation of this assumption in Frege's ideas about the identification (individuality) and existence of individuals (objects), incorporated in Frege's treatment of the senses by means of which we can grasp an individual object. These were according to her inspired by Kant's ideas, especially by Kant's distinction between the predicative and existential

xvi

INTRODUCTION

uses of 'is'. Even though Kant did not subscribe to or even anticipate the Frege - Russell Jistinction, he thus seems to have inspired it. SIMO KNUUTIILA JAAKKO HINTIKKA

CHARLES H. KAHN

RETROSPECT ON THE VERB 'TO BE' AND THE CONCEPT OF BEING·

When I began work on the Greek verb to be in 1963, in the project that took shape in the article 'The Greek Verb "to be" and the Concept of Being,l and eventually resulted in a book on the Greek verb 'to be' in 1973, 2 my aim was to provide a kind of grammatical prolegomena to the study of Greek ontology. I wanted to give a description of the linguistic facts concerning the ordinary use and meaning of the verb, apart from its special use by the philosophers, in order to clarify the pretheoretical point of departure for the doctrines of Being developed by Parmenides, Plato, and Aristotle. I thought at the time (and still think) that the ancient use of the verb estileinail on was poorly understood, and that much of the modern discussion is vitiated by false assumptions, in particular by an uncritical application of the notions of existence and copula to the interpretation of ancient texts. I take the present occasion to summarize the results of my work both for the theory of the verb and for the interpretation of some of the early philosophical texts, referring to earlier publications for more detailed exposition and defense of the views outlined here. I. THE DISTORTING INFLUENCE OF THE TRADITIONAL VIEW

As I see it, confusion reigns both in the traditional account of the verb given by linguists and philologists, and also in much of the philosophical exegesis of ancient theories of being. The two lines of confusion have infected one another, since the linguists have borrowed their notions of existence and the copula from philosophy (and from rather superficial philosophy at that), while philosophers have in turn made use of linguistic doctrine as a basis for their own account of Greek ontology, and in some cases as the weapon for a general attack on the Greek notion of Being. I begin by stating what I take to be the principal errors in the standard view, by which I mean the views prevailing twenty years ago and still to be found in many handbooks and commentaries. (1) It was generally assumed that the uses of einai could be classified either as (a) meaning 'exists', or (b) serving only as copula. 3 But this S. Knuullila and 1. Hintikka (eds.), The Logic oj Being, 1-28. © 1986 by D. Reidel Publishing Company.

2

CHARLES H. KAHN

dichotomy is theoretically unsound and descriptively inadequate. It is theoretically unsound because (a) is a semantic and (b) is a syntactic notion. A rational theory would contrast 'exists' with other meanings of the verb, and the copula syntax with other constructions. The dichotomy could be justified only if there were a one-to-one correlation between sense and syntax, so that copula uses were all meaningless and the verb in absolute (non-copulative) construction always meant 'exists'. But both assumptions are false. (2) The traditional account of einai and its Indo-European root*es- in comparative grammar (which goes back to Brugmann and Meillet, and is reflected in some accounts of the copula in English) takes the existencecopula distinction for granted and proceeds as follows. The verb be (einai, *es-) was originally a verb like other verbs, with concrete meaning. The original meaning was to exist, or perhaps something even more concrete like to be present or to be alive. Predicate nouns and adjectives were originally expressed without any verb, in the so-called "nominal sentence" familiar from Russian and many other languages: John is wise was simply John/wise; John is a man was John/(a) man, and so forth. But in the course of time it became useful to introduce a verb into the nominal sentence in order to express the tense, person, mood, and other modalities carried by the finite verb in Indo-European. Hence the verb be (meaning exist) was introduced into the nominal sentence, where it gradually lost its original meaning and degenerated into an "empty" verb or "mere copula," a syntactic device which serves to satisfy the requirement that every sentence must contain a verb. This historical-sounding theory is enshrined in the textbooks,4 but there is really no evidence to support it. What looks like evidence is a misleading parallel to other verbs that take a predicate construction (like turn pale, grow tall) and that clearly had an independent meaning, but which in the course of time came to be used as substitutes or suppletives for be, and even provide forms that are now integrated into the conjugation. (Thus am, are, and is in English are derived from *es-, but be comes from I.-E. *bheu- 'to grow', 'become'; was comes from *wes- 'to dwell, stay in a place'; eta is, he in French, stato in Italian, estar in Spanish all come from Latin sto/stare 'to stand'.) However, these parallels prove nothing to the point, since in every case the known historical development presupposes the existence of a basic copula verb in Indo-European. 5 And there is no doubt that the original copula was *es-, our verb 'to be'. The notion of a prehistorical state of Indo-European without a copula verb is a pure figment of the imagination. !n Greek at any rate

3

THE VERB 'TO BE' AND THE CONCEPT OF BFINC;

the copula uses of einai are overwhelmingly more frequent than any other use in the earliest texts. The idea that the existential uses are somehow more fundamental or more primitive seems to be a mere prejudice, a prejudice based in part upon a mistaken view of existence as an ordinary predicate, taken together with an empiricist theory of meaning which assumes that the original sense of any word must have been something concrete and vivid, something like what Hume calls an "impression" . Hence I propose a modest Copernican revolution: to reinstate the copula at the center of the system of uses of einai. I do not claim that the copula uses are older, since for that claim also there is no evidence. In purely synchronic terms I propose that the copula uses must be regarded as more fundamental in three respects: (1) they are statistically predominant, (2) they are syntactically elementary, whereas other uses (existential, veridical, potential) are grammatically "second-order", operating as functors on a more elementary sentence, and (3) they are conceptually prior and central to the whole system of uses of the verb, in a sense that remains to be clarified, but which bears some analogy to the unifying role of a central term in Aristotle's scheme of "focal meaning" or pros hen legomena. Thus if we take the copula uses as given, we can see why the same verb may serve in other ways, for existence, truth, possession and the rest. But if we take any of the other uses as primary, the way back to the copula becomes difficult, if not impossible [Q understand. 6 (3) It was correctly noted by a number of linguists that the existence of a verb to be in our sense, which is at once a verb of predication, location and existence (to name only three of its functions), is a peculiarity of Indo-European. 7 As we can see from the monograph series on "the verb 'be' and its synonyms" edited by .I, W. M, Verhaar, the topic of be can itself be defined only by reference to Indo-European verbs from the root *es-. But why should a historical peculiarity of this kind be of any general significance, and how can a concept based upon the parochial usage of an Indo-Eurorcan verb provide a genuine topic for philosophical theory? Thus A. C. ~:(2harn has claimed, a rropos of the very different situation in Chinese. thJ.t there is no concert of Being which languages ar~ wei! Of iii cquiprcd to tipn." or 'to ht" as ('opld3 depend lIPon a grarnmatical rule t'O!' thl...' sentence. and it would b::.- merely a ·:elDcid'.:r~'(' ~r one found an\'thi:,,(~

ir-tr!!!uagc without this rule. S

-

-

l'-C",'flL:

tnr rune· ,)\" the

i:-" ,~-:atinn

Icscnl~)l;-:,'

11

in (:

.

Such arguments from linguistic relativism tend to rei:lforce the

4

CHARLES H. KAHN

philosophical complaints which Mill and others have directed against "the frivolous speculations concerning the nature of Being", which Mill thought had arisen from overlooking the distinction between the verb of existence and the diverse uses of the copula, and from supposing "that a meaning must be found for it (namely, to be) which shall suit all these cases.,,9 So Mill's godson Russell insisted that "the word is is terribly ambiguous" and proposed that instead of a unified notion of Being we need to distinguish various "senses" of to be, including existence, identity, and predication. Both logical and linguistic criticisms thus tend to converge in a general suspicion that doctrines of Being in traditional ontology reflect a projection onto the universe of the linguistic structure of Greek or of Indo-European. I do not intend to do battle here against a general thesis of linguistic relativism, and I shall certainly not deny that the union of predicative, locative, existential and veridical functions in a single verb is a striking peculiarity of Indo-European. Whether this diversity is properly regarded as a case of ambiguity of meaning is a question on which I do not propose to take a stand. \0 What I do deny is that this cumulation of functions in the verb to be was necessarily a philosophical disadvantage. On the contrary, I want to suggest that the absence of a separate verb "to exist" and the expression of existence and truth (plus reality) by a verb whose primary function is predicative will have provided an unusually favorable and fruitful starting-point for philosophical reflection on the concept of truth and the nature of reality as an object for knowledge. This was due in part to the puzzling con verge'.lce of so many fundamental notions and functions in a single linguistic form. It was Parmenides who first introduced "what is" (to on) as a central topic for philosophical discussion, and the paradoxical argument by which he developed his thesis turned out to be one of the most creative innovations in the history of Western thought. In the first place he elaborated the stative-durative aspect of the verb into a systematic claim that what is must be ungenerated and incorruptible. And this claim provoked all the element theories of the fifth century, including the theory of atoms, as an account of how the most fundamental realities could remain immune to change. (If the doctrine of indestructible atoms is no. longer with us, we may perhaps recognize its Parmenidean shadow in the conservation laws of modern physics.) And it was the same Parmenidean thesis of unchanging reality that provided Plato with the ontological resources for his own account of immutable Forms. On the other hand, because esti is not only a verb of truth and reality but also the sign of predication, Parmenides'

THE VERB 'TO BE' AND THE CONCEPT OF BEING

5

paradoxical insistence upon the monolithic unity of what is stimulated both Plato and Aristotle into working out theories of predication, and eventually led Aristotle to propose his scheme of categories as an account of how "what is is said in many ways". Finally, the Aristotelian doctrines of matter and form, potency and act were also formulated as a response to Parmenidean paradoxes about the concept of change (as we can see from the structure of his argument in Physics I). Thus Western physics, logic, and metaphysics have all been the beneficiaries of a fierce century and a half of philosophical dispute generated by Parmenides' bold attempt to fuse into a single entity the diverse features of the verb to be. I hope to shed a little light on some of the earlier phases of this momentous episode. But first we need a realistic description of the prephilosophical uses of the verb. II. THE LINGUISTIC THEORY OF TO BE

1. The Copula

In order to see this as the central focus of the system of uses of einai, we must have a more adequate account of the copula itself. Linguists often speak of the copula as a dummy verb, as a merely formal "bearer" of the verbal marks of tense, person, and mood. 11 Hence a recent author can describe the entire function of the copula as "simply to act as a verbalizer," to convert "an adjective like 'cunning' into a verb-phrase 'is cunning', which is of the same category as 'snores' " or of any finite verb. 12 Abelard, who was either the inventor or at least the codifier of the classical theory of the copula, rightly associated the copulative function (vis copulativa) with all finite verbs. He saw is as distinctive in that it provides only the predicative link and not also the predicate (copulat tan tum et non copulatur); other finite verbs do both.13 Abelard's theory has the merit of focussing attention on the verbal function as such, and not simply on is as the verb for nominal predicates. What he saw is that the copula separates out the specific function of the verb, which is obscured in the case of other verbs such as runs, sleeps, just because they combine the information content of the predicate (running, sleeping) with the verbal form. This general verbal function is what one might identify as the propositional tie or mark of assertion; what we, following Abelard, call the copula is simply the canonical expression of this function in a sentence of the form X is Y.

6

CHARLES H. KAHN

In sentences of this form, the copula expresses what is otherwise indicated by the verb-ending: in either case we have a "sign of predication". What is meant by predication here has both a syntactic and a semantic aspect. (I) Syntactically, the copula or verb-ending serves to make a grammatical sentence out of two terms that would otherwise only form a list: John and running (or to run). It thus indicates not only tense, mood, and person but something more fundamental: sentencehood. In many cases the finite verb does this alone, without a noun, both in impersonal verbs like UH (pluit, "it is raining") and where the subject is understood from the context: 7f!€x.H (currit, "he is running"). (2) Semantically, if we take the indicative-declarative form as basic, the verb or copula gives formal expression to the truth claim of the sentence. (We need not attend here to the ways in which this truth claim may be modified, by interrogative or conditional sentence structure or by the various moods; in some but not all of these cases, the modification will be reflected in the verb ending.) Limiting ourselves to indicative forms used in declarative sentences we can say: tbe semantic function of is as copula or sign of predication is to bear the mark of sentential truth Claim, to serve as focus for the claim of the whole sentence. (The truth claim of a sentence corresponds roughly to the fact that it can have a truth value because it does have truth conditions.)14 This basic assertive function of the copula - more precisely, the intimate connection between the copula and the assertive function of the sentence - shows up when we stress the verb in pronunciation: "Margaret is clever, I tell you!" "The cat is on the mat after all." This semantic role of the copula as sign of sentential truth claim permits us to understand one of the most important special uses of einai in Greek, the so-called veridical, where the verb by itself (both in the third person indicative and in the participle) expresses the notions of truth and reality. If we lose sight of this connection between the copula and the truth claim that is fundamental for all declarative discourse, the fact that "being" in (;reek (to on) may mean reality will become a mysterious anomaly, quite independent of the predicative function of the verb. Hence philolog.ists have often tended to overlook the veridical use or to conflate it with the existential, despite the fundamental differences in sentence structure. This is a principal support for my claim that the copula use is fundamental; neither veridical nor existential use can be explained on the basis of the other, but both can be understood on the basis of the copula. Before considering these special uses of einai, I call attention to two features of the copula verb which are often on:rlooked.

THE VERB 'TO BE' AND THE CONCEPT OF BEING

7

(A) The verb einai, whether used as copula or in other constructions, has only durative (present-imperfect) forms, and no forms in the aorist, that is, in the punctual or non-durative aspect. (The future forms lie outside this aspectual contrast.) This formal peculiarity of einai (and of I.-E. *es-) is reflected in its semantic value as stative copula, in contrast to the mutative copula becomes (gignesthai in classical Greek). 15 The antithesis of Being and Becoming, opposed to one another as stability to change, was deeply built into the system of copula verbs, long before it was exploited by Parmenides and Plato. (B) Among the copula uses of be in a broad sense are what we may call ldcative uses, where the complement or predicate expression is not a noun or adjective but a local adverb (here, there) or a prepositional phrase of place (at home, in the marketplace). Some linguists may prefer not to count these as copula uses, since in this construction is cannot be replaced in English by the mutative copula becomes (though in Greek gignesthai may be used in locative sentences); whereas locative is functions (a) in contrast to a verb of motion (goes there, arrives in the marketplace) and (b) in parallel to a wide variety of verbs (John works at home, Socrates talks in the marketplace). This use of is, which I will call the locative copula, shows a special affinity with a small group of verbs of posture, which may serve as static replacers for copula is with predicate nouns and adjectives as well as with locative phrases: sits, stands, lies. 16 Because of these connections, and because of the more vivid or concrete sense that seems to attach to einai in locative sentences, and which (as we will see) often suggests an existential nuance that gets rendered in English by there is, some scholars have suggested that the locative or locative-existential use of einai represents the basic sense and function of the verb. l ? But although the locative uses are certainly important for understanding the intuitive force of the verb to be in Greek as a verb of state or station, I do not believe that they are more fundamental than the copula use with predicate nouns and adjectives. For one thing, it is possible to add locativt modifiers to copula sentences, as to many others, and then derive the locative copula by zeroing the nominal predicate: Athens is a city -+ Athens is a city in Greece -+ Athens is in Greece; John is busy -+ John is busy at home -+ John is at home. But there does not seem to be any plausible derivation in the opposite direction. Whereas locative phrases are optional modifiers for a wide variety of sentences, the introduction of a predicate noun or adjective presupposes the basic construction with a copula verb, as I have argued elsewhere. 18 Even more important, in my opinion, is that it is only the

8

CHARLES H. KAHN

general copula function as "verbalizer" and mark of sentencehood, and not the locative use, that can explain the deep connections of einai with the notions of truth and fact. I turn now to the two principle classes of non-copulative uses of the verb: the veridical and the existential. For present purposes we can ignore other non-copula constructions, such as the possessive (esti moi, "I have") and the potential (esti + infinitive, "it is possible to"). 2. The Veridical Uses

The Lexicon recognizes, what every good Hellenist knows, that in many cases the verb esti and its participle on must be translated by "is true" , "is so", "is the case" or by some equivalent phrase: esti tauta "these things are so" (cf. French c'est cela "that's right"), legein to onto "tell the truth", "state the facts" .19 Comparative grammar shows that this is a pre-historic use of*es- in Indo-European 2o ; and Aristotle himself notes this as one of four basically different uses of einai (Met. 117). All I have done is give this use a name, "the veridical," and correlate it with a definite sentence form. The veridical construction proper is characterized by three syntactic features: (i) the understood grammatical subject of esti is not a noun form (like man or hunting) but a sentential structure, as represented in English by a that-clause (that the man was hunting); (ii) the construction of the verb is absolute, i.e. there is no nominal, locative or other adverbial complement, except the comparative "so" (houtos), which introduces (iii) a comparative clause of saying or thinking. which is expressed in the full veridical construction and implied in every case: These things are as you say, fCTn raimx ovrws 'WCT1I"EQ CTU Af'YHS. Without claiming that the veridical construction is derived from the copula use, either historically or transformationally, I nevertheless believe that it is easy to see how the two uses logically and naturally belong together, as long as we keep in view the semantic function of the copula as the mark of truth claim. A properly veridical use of to be (as in "Tell it like it is") simply makes general and explicit the truth claim that is particularized and implicit in every declarative use of the copula ("The cat is on the mat"), as in every declarative sentence. Because the great flexibility of the copula construction makes it possible to produce an Sis P sentence that is roughly equivalent to every noun-verb sentence, the copula tends to serve as verb par excellence, that is, as representative for the finite verb as such and for its predicative force, what Abelard called its vis copulativa. It is because of this very general function of the copula as sign of predication and sentencehood that the very same form

THE VERB 'TO BE' AND THE CONCEPT OF BEING

9

(esti, "is") can serve to express the veridical idea as such: to bring out the implicit truth claim ("This is how I say things are") and the corresponding notion of reality ("This is how things really are"). Despite my general reluctance to decide when a different use becomes a different sense, I am inclined to speak here of a veridical meaning or connotation of einai, in cases where the Greek verb cannot be adequately translated by the copula or by an idiomatic use of is alone. This is most conspicuous when the participle (to) on is used to mean "truth" or "the fact of the matter", and when it may be replaced in Greek by a word like aletheia or (to) alethes. 22 And there are clear cases of the veridical connotation attached to a copula construction, as in the example which Aristotle cites of einai meaning "is true": esti S6krates mousikos, "that Socrates is musical, i.e. that this is true". 23 In English as in Greek, this force of the verb is typically brought out by a contrast between Being and Seeming: "He wants not to seem (dokein) but to be (einOl) the noblest" (Aeschylus Septem 592). Here again a basic philosophical contrast - between appearance and reality - is fully prepared in the pre-philosophical usage of the verb. 3. Existential Uses I briefly describe three uses of einai that we intuitively recognize as "existential" and are inclined to render by there is or even (in the third case) by exists.24 (1) The existential copula: (a)

"There is a city (esti polis) Ephyre in a corner of horsenourishing Argos". (Iliad VI. 152)

(b)

"There is a certain Socrates (estin tis Sokrates), a wise man, student of things aloft ... who makes the weaker argument the stronger". (Apology 18B 6)

Perhaps the most common of all "existential" uses of the verb in Greek are sentences such as these, where esti seems to functions twice: to assert the existence of a subject ("There is a city ... ") and then to say something about it: "The cityEphyre is in the corner of Argos". In most instances the predicative use will be locative, as in (a); (b) is one of the rare examples where a purely nominal copula takes an existential force. It is clear that the underlying syntax of the verb in such sentences is that of the copula, but that this construction has been overlaid with a secondary function, which I would analyze as introducing a subject for further

10

CHARLES H. KAHN

predication, and which accounts for the existential nuance we render by "there is". In most cases, where the construal is locative, the verb serves to introduce (posit, assert existence for) its subject by locating it in a definite place or context. But as (b) shows, esti may perform this function even without the support of a locative construction. In either case the verb typically occurs in the emphatic initial position (like "there is", which always begins its clause); but initial position is neither necessary nor sufficient for the Greek verb to play this role. 25 What do we mean by classifying the verb in such sentences as "existential"? It is in fact false t.J the intuitive force of the sentence to say that it asserts the existence of Ephyre or Socrates. The verb simply introduces its subject into the narrative, or into the stream of discourse, as a subject for further predication or, more often, as a local point of reference for the episode that follows. We may be inclined to connect the verb with the existential quantifier of formal logic, since the sentence does imply that the set of objects specified by the following predicates is not empty: "There is something, not nothing, which is a city in Ephyre, etc.". But the trouble with this analysis of the "existential" copula in sentences of type (1) is that it applies equally well to straight copula versions of the same sentences: "Ephyre is a city in Argos", "Socrates is a wise man". What is logically implied is one thing, what is expressed is another. Somehow, in virtue of initial position, locative function or more general contextual features (introducing an item for future reference), the verb in (a) and (b) gathers to itself the existential claim that properly belongs to the sentence as whole. How can the copula verb assume this function? Perhaps because the verb itself provides no lexical content but agrees formally (in person and number) with a subject term that does have content, emphasis on the copula serves to focus attention on the subject, to present it in an emphatic way (usually by localizing it), and thus to focus on the subject as such, as subject of "is" and hence precisely as a subject for predication. Whatever the explanation, the secondary existential force of the copula verb in type (1) points the way to a purely existential use, with no copula construction, in type (2). (2) The existential sentence operator: "There is someone (no one) who does such-and-such" (ouk)

esti has tis + relative clause. 26

In typical examples, sentences of this type refer to persons, but variants occur which give this form the full generality of the logical scheme (3x)Fx

THE VERB 'TO BE' AND THE CONCEPT OF BEING

11

where Fx stands for the relative clause and an initial esti functions as existential quantifier: "there is someone/something such that ... ". The verb esti serves precisely to affirm or deny (the existence ot) a subject for the following clause, to assert that the set specified by the following formula is or is not empty. There is no trace of the copula construction, nor any way to derive this form logically or syntactically from the copula construction. But there is a logical overlap with type (1), which can be seen as a copula construction overlaid with the existential function of type (2). This properly existential use is relatively rare: I found only 4 examples out of 562 occurrences of einai in the first twelve books of the Iliad. I think it would be unreasonable to suppose that (2) somehow represents the original, prehistoric value of*es-. Can we offer a historical explanation of this use of esti? My suggestion is that it arose out of the copula use by way of sentences of type (1), where the copula acquires "existential" connotations in virtue of its locative association and its rhetorical function of introducing a subject for predication. Given these connotations, it is natural that the existential function of (2) becomes one of the values esti can have when used alone, without nominal br locative complements. (The veridical, possessive, and potential uses represent other values esti may possess when it appears outside of the copula construction.) It is on the basis of the existential force of the verb in (1), where this force is secondary, and in (2) where it is primary but serves directly as the basis for ensuing predication, that we can understand the appearance of a new sentence type, in which esti itself becomes the grammatical predicate. (3) The existential predicate: (a) (b)

"There are (no) gods" (auk) eisi theoi, "Zeus does not even exist" oud' esti Zeus.

Type (2) is rare in Homer, but type (3) does not occur at all. My earliest specimens are from Melissus, Protagoras, and Aristophanes in the middle and second half of the fifth century B.C., and they clearly show the influence of philosophical speculation. 27 Sentences of this kind are sometimes cited as exhibiting the oldest meaning of *es- in IndoEuropean. On the contrary I regard this as a fifth-century innovation, based upon the existential force of the verb in the older types (1) and (2), but focussing attention on existence as such (i.e. on the question whether or not there is such a thing), as a result of philosophical speculation,

12

CHARLES H. KAHN

theological sceptIcIsm, and the general disputes about "Being" that begin with Parmenides. Whatever its date and origin, the verb in this sentence type is best understood as an abridgement from type (2), "There is someone (no one) who ... ". Thus "Zeus is not" and "Ther~jlre gods" are to be construed as generalized versions of "There is no Zeus who ... " and "There are gods who ... ", where the effect of generality is achieved by dropping the relative clause with its particular content and thus presenting (or rejecting) a subject for any and all unspecified predication. But the idea that in such sentences the existential verb would itself constitute the predicate is an illusion to which the Greeks seem never to have fallen prey.28 The functions of einai as instrument of predication were so fundamental that the same verb could not easily be seen as forming a selfsufficient predicate. 29 In Greek linguistic intuition, "There is no Zeus" (ouk esti Zeus) means that Zeus is not a subject for any predication, that there is nothing true to be said about him. The Greeks are thus untroubled by the modern puzzle of negative existentials, whicli arises from the temptation to assume that "Zeus does not exist" says something which is true of Zeus. III. THE EARLY PHILOSOPHIC USE OF TO BE

The outcome of my linguistic survey has been to underline the fundamental role of einai as copula verb and at the same time as verb of state and station, characterized both by locative and by durative-stative values. Among the non-copula uses I have called attention to the veridical expression for truth and fact, and I have insisted upon the very limited range of early existential uses, bound to a specific locative or predicative context. Thus while existential sentences of type 1 and 2 are well attested in Homer, the stripped-down "absolute" use of type 3, in which einai appears alone as existential predicate, is not found before the fifth century, and then only in contexts where philosophic or sophistic influence is clear. 30 My suggestion is that for understanding the early philosophical usage, both in Parmenides and in Plato, the veridical notion (whether or not it is the case that p) turns out to be more important than the idea of existence (whether or not there is such a thing as X), although both notions are present. To illustrate the new, quasi-technical use of the verb as a pure existential (my type 3), we may cite what is probably the earliest unambiguous example, the welI-known statement of Protagoras that, concerning the

THE VERB 'TO BE' AND THE CONCEPT OF BEING

13

gods, he does not know "either that they are (has eisin) 0;- that they are not (hos ouk eisin), or what they are like in form" (fr. 4). The contrast furnished by the last clause guarantees that einai here refers to the question of the gods' existence; and the verb itself might properly be translated as "exists". 31 However, in another even more famous quotation from Protagoras the natural reading of the verb to be is veridical: "Man is the measure of all things, of the things that are (ton onton), that they are (has estin) , and of the things that are not, that they are-not (has ouk estin)" (fr. 1). Here we have prefigured, in a slight modification of the old idiom for truth ("tell it like it is"), the distinction between ~he intentional being-so of judgment and statement (hOs estin) and the objective being-so of the way things stand in the world (Tex 11m). This intuitive distinction between the ways things are and the way they are judged to be, which Protagoras recognizes only to deny its validity, is precisely what we find in the two terms of Aristotle's definition of truth, where the participle (ta onta) is used for the facts of the case, as in Protagoras fr. 1, while the finite verb in Protagoras' formula is replaced in Aristotle by the infinitive, for the asserted einai of thought and statement. 32 The parallel being so exact, it is no accident that Protagoras' book was called "Truth". 33 Another early example from the philosophical literature shows how veridical and existential values can intersect in a single occurrence, or how an author can oscillate between the two. Melissus is conditionally assuming what he wants to deny: the reality of phenomenal diversity.

°

If there really is (ei eSlI) earth and water and air and fire and iron and gold, and living and dead, and black and white and all the other things men say are true or real (hosa phasin ... einai alethe), if these things really are (ei tauta estl), and we see and hear correctly ... (Melissus fro 8.2).

Melissus is actually insisting upon a radical discrepancy between the way things are and the way they seem to us. He is stating in an extreme form the distinction which Protagoras is attacking, and Protagoras' formula for truth could reasonably be read as a direct response to Melissus. Now in Melissus' own statement the first esti is pretty clearly existential: it looks like a pure existential of type 3. But the veridical undertone makes itself heard in the summary clause "all the things men say are true (einai alethe),', and the last occurrence of the verb in ei tauta esti is neatly ambiguous between "if these things exist" and "if these claims are true". And in what follows, the thesis about reality or existence is repeated in the veridical mode with a copula cC'nstruction: "these things would not change if they were true" (or "real": ei alethe en).

14

CHARLES H. KAHN

This instability of the type 3 existential use is characteristic not only of the early texts but also of the verb in Plato and Aristotle. 34 On the one hand there is no doubt that, for Melissus as for Protagoras and for all later writers of Greek, esti used alone can mean "exists" or "there is such a thing". 35 On the other hand, the verb performs so many other functions, and its copula role is so prominent, that there is rarely any system:ltic reliance upon the fixed sense of the verb as "exists", except in certain special contexts such as the existence of gods and mythological creatures. Bur when Plato wants an unambiguous expression for an assertion of existence, he has recourse to a copula construction: dva,{ TL "to be something" rather than P-T]OEV dVaL "to be nothing". 36 Furthermore, although Plato and Aristotle both use sentences of type 3 to assert and deny existence, neither philosopher mentions existence as one of the basic notions of einai. When Aristotle applies his scheme of categories to show how Being (it is) "is said in many ways", we may prefer to describe his various modes of being as so many different kinds of existence, or even as so many different senses of "exists". But Aristotle does not speak in such terms, and he regularly illustrates his categorial distinctions by copula uses of to be: "Socrates is a man", "is wise", "is 6 feet tall", "is in the marketplace". For Aristotle as for Plato, existence is always dva,{ TL, being something or other, being something definite. There is no concept of existence as such, for subjects of an indeterminate nature. 37 Thus the limited literary use and unstable semantic value of type 3 existentials is reflected in the explicit philosophical doctrine that being (and a fortiori existence) is not a genos, not a definite kind of thing. So much by way of caveats before we turn to Parmenides, where I shall urge that the veridical use gives us a better initial grip on the argument than the existential does, although both are needed together with the copula construction in order to give a complete exegesis. In a deliberate challenge to what seems to be the prevailing interpretation, I want to claim that for Parmenides, as for Plato and Aristotle and also in the prephilosophic usage of the verb, existence is a subordinate and not a primary component in the concept of being. The notion of existence (or the use of the verb meaning "to exist") must be included in our account of Parmenides' argument, since "what is" (to on) is contrasted with "nothing" on the one hand and with coming to be (genesis) and perishing (olethros) on ;he 0ther. To sustain these contrasts, to on must be (a) something rather than nothing, (b) something that is already there, and (c) something that continues to be there, something that persists. These

THE VERB 'TO BE' AND THE CONCEPT OF BEING

15

contrasts define the sense in which for Parmenides esti means "it exists", where the durative and locative values of the verb give some definite shape to the claim of existence. But there is no reason, neither in the prephilosophic usage nor in the context of the poem nor in the later echoes of the Parmenidean thesis in Protagoras (fr. 1, cited above) and Plato (Rep. V, 476E - 477 A, cited below), to suppose that in the initial presentation of his thesis in the one-word sentence esti in fr. 2, the sense of Parmenides' claim can be adequately captured by the translation "it exists" . 38 How are we to construe this claim? The contrasts just cited require that, if the argument is to be coherent, the content of what is claimed must be such that it is (a) something rather than nothing, (b) already present, and (c) guaranteed to endure. But that gives us no clue as to where the argument begins, or how we are to understand Parmenides' initial presentation of the thesis so as to provide him with a plausible startingpoint. For this we must look at the context in the poem and above all at the preceding context: the allegorical proem. 39 Parmenides' thesis (that it is and that it cannot not-be)·is introduced as the acceptable member of a pair of alternative "ways of inquiry" for rational cognition (noesal) to travel on (fr. 2.2). Where is this inquiry supposed to lead? Obviously, to knowledge and to truth, as is clear both from the proem and from the words immediately following the thesis (2.4: "it is the path of Persuasion, who follows on Truth"). In the allegorical proem the voyager on the right road is a "knowing mortal", transported by wise horses and clever escorts, the daughters of the Sun who are leading him "to the light" (fr. 1.10). When he arrives, a goddess promises to instruct him in everything, but first of all in "the unshaken heart of persuasive Truth" .40 This is what Parmenides gives us as a background for understanding his thesis that it is, and that it cannot not-be. To interpret the thesis we must be able to say: what is the subject of the claim esti? And what is the content of that claim? I think the subject can be specified with some confidence, on the basis of clues from the proem and the immediate context. With these clues Parmenides makes quite clear that what the goddess holds out is a promise of knowledge, and that the path of it is must lead to truth. Hence the understood "it" which the goddess is referring to in the thesis must be located in the region of knowledge and truth; it can also be identified as the goal of inquiry and the object of that quest that began in the first verse of proem, where the horses are said to carry the youth "as far as his desire can reach". So we C:in ciescribe the subject

16

CHARLES H. KAHN

referred to by the goddess as what our youth has come to find out, and what will be made hIOwn to him in the revelation of persuasive truth. Such an initial characterization of the subject as "the object of inquiry" or "the knowable" can be taken for granted before the thesis is articulated, though any fuller characterization remains to be spelled out in the course of the argument. 41 What does esti say about the object of inquiry that (a) can be taken for granted as a point of departure, and (b) can justify the immediate, categorical rejection of the negative way, that it is not? Recent interpreters, looking ahead to see what content is given to the thesis later in the argument, propose to read esti as "it exists". But in addition to the dubious procedure of reading a poem backwards, this view has the disadvantage of saddling Parmenides from the outset with an essentially anachronistic notion of to be. If my interpretation of the linguistic evidence sUPlmarized above is even approximately correct, then it is highly unlikely that either Parmenides or his readers would understand a bare unadorned esti as meaning primarily or predominantly "it exists". Of course the parallel to sentence types 1 and 2 guarantees that "there is such a thing" will be there as a background meaning for Parmenides to rely upon. But the primary idiomatic sense of an unqualified esti in the early fifth century can only be the veridical, in this case taken objectively for the reality as known: "it is so" or "this is how things stand". And the logic in support of the initial thesis then becomes unassailable: what is known or knowable must be the case and cannot not be so. "For you could not know what is not (so)" (fr. 2.7).42 Thus Plato, when he echoes this argument in Rep. Y, 476E-477A, has Socrates ask: "Does a knower know something or nothing? ... Something which is (on) or which is not (ouk on)?" To which the interlocutor replies: "Something which is (on); for how could anything which is not (me on tl) be known?" Plato adopts Parmenides' starting-point here precisely because he wants to make the premises of his argument as plausible as possible. For the ancients as for the moderns, knowledge entails truth: what is known must be really SO.43 It is the veridical use, then, which not only provides the idiomatic background for understanding Parmenides' stark initial esti, but also provides the conceptual grounds for granting it,; necessary truth (best understood as necessity of the consequence: if p is knowable, then, necessarily, p is the case). Once this starting point has been granted, on the basis of a veridical what is (so), Parmenides will go on to unfold the richer implications of an esti whose full content will depend on other uses

THE VERB 'TO BE' AND THE CONCEPT OF BEING

17

of the verb, including the locative associations which justify his assumption that to eon is spatially continuous, indivisible, and sphere-like if not spherical. Among these other properties of Parmenides' being there will surely be some that depend upon the copula use (Being is unchanging, for if it is F, for any F, then it can never be not-F without falling into Not-being), and some that reflect the existential use in the sense specified above: if Being is, it is not nothing; if it is ungenerated it is already there; if it is imperishable it will persist. Whether Parmenides' move to these richer senses of einai necessarily involves him in a fallacy of equivocation is not entirely clear. We might well suspect something of this sort, in view of the astonishing nature of his conclusions. Plato is at pains to show, against Parmenides, that something can be X and also not be Y without falling into nonentity; whereas Aristotle distinguishes being not only in terms of the categories but also in terms of potency-act and substrateprivation-form (in Physics I) in order to avoid the conclusions which Parmenides draws by taking to on as univocal. Here I suggest that Plato's diagnosis cuts deeper into the actual structure of Parmenides' argument. Some unclarity but no radical incoherence results if Parmenides takes to on (I) veridically, as the objective state of affairs required for truth and knowledge, then (2) existentially, as a real, enduring object which is the "subject" of this state, and also (3) copulatively as being F for various F's, as well as (4) locative, i.e. spatially extended. 44 Fallacy enters only with negation, and the assumption that what-is-not in any respect must be a Non-being pure and simple. The inference from (I), "there is something which is the case, which is determinately so" to (2) and (3) "there is'something which exists as an enduring subject, and which is F" requires for its validity only the reasonable (if not inevitable) assumption that for a state of affairs to be definitely so there must be a definite subject with definite properties. The undeniable category-shift from a propositional entity that is the implied subject of esti in (I) to a substantial or thing-like entity for (2) and (3) is precisely parallel to the shift between "if these things are true (einai alethe), ' and "if these things exist" (ei tauta estl) in the text of Melissus cited above, and parallel also to the cat ego rial ambiguity of einai alethe in the same text: "if they are true" and "if they are real". Since similar shifts and ambiguities between propositional and substantial entities occur in Plato and Aristotle too,45 it would be surprising indeed if the paradoxical esti of the earliest Greek ontology were quite unequivocal in this respect. 46 This interpretation of Parmenides' thesis also provides a natural historical explanation for the paradox of false statement and false belief,

18

CHARLES H. KAHN

which seems to have been popular with some sophists and which persistently recurs in Plato's Cratylus (4290), Theaetetus (189A 10-12) and Sophist (236E, 237E), with an early variant in the Euthydemus (283E - 284C). If speaking falsely is saying what is not (the case), and what is not is nothing at all, then speaking falsely is saying nothing and hence not speaking at all. There may be other dimensions to this paradox, but the crucial move is clearly the slide from what is/what is not as the object of true and false statement and belief to what is not as that which is nothing at all, the non-existent - a slide precisely parallel to the one we have identified in Parmenides' argument, and which has its counterpart in Melissus' oscillation between doubts that the multitude of phenomenal things really exist, on the one hand, and claims that we do not see or hear correctly, on the other hand, or that there are not as many things "as men say are true". Thus "true being" (to on alethinon) for Melissus (in fro 8.5) is both (a) what really exists and (b) what is true, as the content of true statement and belief. This ambiguity is relatively harmless in the affirmative case, where (a) and (b) coincide (given the failure to distinguish between facts and things). But the corresponding negation leads to fallacy and paradox, if the denial of (b) in a reference to the object of falsehood is also taken as a denial of (a). Since I have treated Plato's use of esti and to on at length in a recent article, I will here simply list my chief conclusions concerning Plato's ontological vocabulary in the preliminary and mature statements of his theory of Forms. (1) In the so-called Socratic dialogues, the first philosophically relevant use of einai is its occurrence in connection with the What-is-X? question of Socratic definition. Examples: Laches 190B7-C6: "we ought to possess knowledge of what virtue is" eidenai hoti pot' estin arete); "we say we know what it is" (eidenai auto hoti estin); "But if we know, then we can say what it is" (ti estin). In the Euthyphro we find the contrast between such a "whatness" and other attributes of a thing hardening into a terminological distinction between ousia, "essence", i.e. the content or correlate of a true answer to the what-is-it? question, and pathos, any other property or attribute (Euthyphro IlA7-8). Here ousia is simply a nominalization for the verb estin in the what-is-it? question (to hosion hoti pot' estin, IlA7). In such contexts the verb is syntactically the copula, but logically or epistemically strengthened by the context of use into what we might call the definitional copula or the is of whatness, which aims at locating the true, proper, deep or essential nature of the thing under investigation.

THE VERB 'TO BE' AND THE CONCEPT OF BEING

19

(2) The first trace of a more technical use, growing directly out of the definitional copula, appears in the Lysis, where the attempt to explain what makes something dear, friendly, or beloved (phi/on) leads to the notion of "that which is primarily dear (ekeino ho esti prOton phi/on), for the sake of which we say that all other things are dear" (219C 7), these other things being potentially deceptive "images" (eidola) of "that primary thing, which is truly dear" (ekeino to prOton, ho hos alethos esti philon (219D 4). What is new here is (a) the use of the definitional copula as a kind of proper name for the concept under discussion, or for its primary instance: what is (truly, primarily) X, prefiguring the canonical reference to the forms in later dialogues as to ho esti X, and (b) the veridical strengthening of the copula in "what is truly (alethos) dear" cf. tpC)..OIl Of T~ olin at 220 BI and B4), by contrast with the "images" which are only "verbally" dear (220B 1), i.e., said to be dear because of their relation to the primary case (219D 1). Just as the Euthyphro adds precision to the is of whatness by a version of the essence-accident distinction, so the Lysis reinforces the metaphysical import of a privileged use of this formula by introducing a contrast between Reality and Appearance, between what really is F and what is only an image or a putative instance of F. (3) In Plato's first explicit statement of his mature doctrine, in the speech of Diotima in the Symposium, the Beautiful itself is announced as the goal of a process culminating in a final study "which is (ho estin) the study of nothing but that Beautiful itself", where the student will end by knowing "that which itself is beautiful (auto . .. ho esti kalon, 211 C 8). This is the formula of the Lysis, with its veridical force (' 'he will know what is truly beautiful") again underscored by contrast with appearance and images (2IIA 5, 212A 3). But in this case the formula unmistakably refers to the Form. For here we have a new (or newly formulated) doctrine in which, for the first time, Plato provides his specimen Form the Beautiful with a definite ontological status, based upon the Eleatic opposition between eternal, unchanging Being (aei on) and inconstant, perishable Becoming (211A I - 5). In this context the participle on is used both existentially ("it is forever") and as copula ("it is not beautiful in one respect, ugly in another,,).47 It is precisely in such Parmenidean contexts, where Being is contrasted with Becoming, that it seems most natural to regard to on in Plato as existential, though the aspectual value is that of the stative copula. (4) In the Phaedo and Republic, where the doctrine for Forms is systematically developed, the philosophical uses of einai become too

20

CHARLES H. KAHN

diverse for cataloguing here. I would emphasize only that (a) the veridical overtones of to on, used roughly as a synonym for "truth", are predominant in the initial presentation of the Forms in both dialogues: in Phaedo 65B - 67B, to on and ta onta occur together with aletheia and to alethes for the "true reality" which the philosopher's soul desires and pursues (and which is identified as Forms at 650); in Republic V, 476A the Forms are introduced by the contrast of Reality and Appearance: each of them is one but appears (phainesthal) as many (A7); and their ontological status is again expressed by a use of to on for the object of knowledge; that which is wholly real (pantelos on) is wholly knowable; that which is in no way real (me on medame) is in every way unknowable" (477A 3). The veridical-epistemic contrast between Being and Seeming (phainesthal) serves to distinguish the Forms and "the many" throughout this passage (cf. 479A 7 - BIO). (b) The formula auto to ho esti (ison), "that itself which is (equal)", familiar from the Lysis and the Symposium, is gradually developed in the Phaedo from idiomatic phrases into a semi-technical designation for the Forms (notably at 7502 and 7804, recalled at 9308), with a parallel use of ousia for the distinctive being, essence, or reality of the Forms. 48 The same designation is used to reintroduce the Forms into the central epistemological passage of the Republic: the Beautiful itself and the Good itself and the other unique entities, "each of which we call what it is" (ho estin hekaston prosagoreuomen, 507B 7). In this designation the predicative form is Fis taken separately, independent of all subjects, and made itself the target of the question 'what is it?' Thus ho esti serves in Plato, like to ti en einai in Aristotle, for the objective essence or definitional content given in a correct answer to the question "what is Fl" for a given predicate F. The syntax of the verb is still that of the copula, but its predicative role is reinforced now not only by the definitional search for the true nature of a thing but also by the ontological dualism of Plato's neo-Parmenidean opposition between Being and Becoming, the One and the Many, the Intelligible and the Visible. The specifically Platonic use of einai in the doctrine of the middle dialogues thus consists in a convergence between (i) the definitional copula from the what-is-it? question, (ii) the veridical Being that contrasts with Seeming, and (iii) the stative-invariant Being that contrasts with Becoming and Perishing. An unqualified use of to on, einai, or ousia may bear any and all of these connotations. 47 The predicative syntax is always latent if not manifest. The existential value appears above all in (iii), but even here the copula use, on which the stative-mutative contrast of Being-Becoming is found-

THE VERB 'TO BE' AND THE CONCEPT OF BEING

21

ed, may reappear at any moment. so The Platonic concept of Being is constituted not by a fusion of copula and existence but by the union of timeless-invariant Being (in contrast with Becoming) and cognitivelyreliable, veridical Being (in contrast with Appearance), both of them expressed or expressible in copula predications, but most rigorously distilled in the frozen auto ho esti version of the definitional is of whatness in application to the Forms. (c) In the Eleatic introduction to the doctrine of Forms at the end of Republic V, Plato has moved beyond Par men ides in a number of interesting ways. First, by accepting an intermediate "mind" reality between Being and Not-Being as object for the cognitive state of opinion (doxa) between knowledge and ignorance, Plato has provided an ontological basis for change and becoming, which was simply the domain of error and falsehood for Parmenides: Plato thus accounts for the possibility of true opinion short of knowledge by giving it an object of its own. In the second place, by his development of the copula use for parallel designations of Forms (as "what is F") and particulars (as "what is and is not F"), Plato opens the way to a philosophical analysis of predication and the diverse uses of to be which he will pursue in the Sophist and elsewhere, and which will lead eventually to Aristotle's theory of categories and his distinction between essential and accidental predication. On the other hand, where Plato in the Republic has not moved substantially beyond Parmenides is in his conception of the negation of Being as what is not in any way (to medame on); for this is indescribable and unintelligible, as Parmenides had insisted and as Plato in the Sophist will finally agree. The paradox of false statement and false belief will haunt Plato until he works out a way to negate the "being" of truth without falling into this hopeless region of blank non-entity. This is far enough to pursue a project that began as linguistic prolegomena to Greek ontology and not as a history of the subject. In conclusion, I want to say a word against the charge of linguistic relativism, in so far as it claims that ancient ontology was vitiated or distorted by the accidental possession of a verb that combines the functions of existence and predication. It is certainly true that the verb einai serves a multitude of functions that are rarely combined in languages outside of Indo-European. And if Greek-ontology had begun with a radical confusion between existence and the copula, then its first task should have been to distinguish the two, a task that neither Plato nor Aristotle undertook. On the contrary, both of them systematically subordinate the notion of existence to predication; and both tend to express the former by means

22

CHARLES H. KAHN

of the latter. In their view to be is always to be a definite kind of thing: for a man to exist is to be human and alive, for a dog to exist is to be enjoying a canine life. Instead of existence, which is a tricky notion at best, it was another use of to be that gave Parmenides and Plato their philosophical starting point: the veridical use of esti and on for "the facts" that a true statement must convey. Thus the Greek concept of Being takes its rise from that naive, pre-philosophic notion of "reality" as whatever it is in the world that makes some statements true and others false, some opinions correct and others mistaken. But this notion of what is as whatever distinguishes truth from falsehood, reliable information from idle rumor, is surely not peculiar to Indo-European. Some such notion will be functioning in any language in which questions can arise concerning what is true and what is false, what is knowledge and what is error. This notion is so essential to the basic descriptive or informative use of language that it is bound to be in some sense a linguistic universal. What is peculiar to Greek (and to Indo-European) is that a locution for "reality" in this sense should be provided by a verb whose primary function is to express predication and sentencehood for statements of the form X is Y. So doctrines of Being first arose in Greece in connection with the question: what must reality be like for knowledge and informative discourse to be possible, and for statements and beliefs of the form X is Yto be true? In principle, the question concerning knowledge and informative discourse is one that might have been posed in any language; the question about the sentential form X is Y reflects a point of view more specifically Greek. If anyone believes that it was a disaster for Greek theories of knowledge and reality to be concerned, from the beginning, with problems of predication and with the propositional structure of language and thought, let him blame the verb to be. NOTES • I wish to dedicate this review of my own work on einai to the memory of G. E. L. Owen. Rereading since his death his major articles on Greek ontology I see more clearly than before how he was a powerful ally in my campaign against the uncritically "existential" interpretation of is in Plato and Aristotle. In many cases we came by different routes to similar conclusions; in some cases I have been echoing his formulation without realizing it. Like all workers in this vineyard, lowe him a great debt of inspiration and encouragement. I Foundations of Language 2 (\966), 245 - 65. 2 The Verb 'Be' in Ancient Greek (Reidel, 1973), The Verb 'Be ' and Its Synonyms, Part 6, ed. by J . W . M. Vtrhaar (= Foundations of Language, Supplementary Series Vol. \6). This will be cited below simple as •Be' .

THE VERB 'TO BE' AND THE CONCEPT OF BEING

23

The earliest clear statement of the dichotomy known to me is that of J. S. Mill in his System of Logic (IB43), I. iv. § I, who attributes it to his father, James Mill, in the Analysis of the Human Mind{IB29). But the philologists were already using this dichotomy as early as G. Hermann in IBOI. (See the quotation in 'Be', p. 420, Note I.) Hermann in turn appeals to "what logicians call the copula", and is apparently dependent on the logic of Christian Wolff, ('Be', p. 423 with Note 5). 4 References to Brugmann, Delbriick, Meillet, Kiihner-Gerth, and Schwyzer-Debrunner in 'Be', p. 199, Note 21. Compare John Lyons, Introduction to Theoretical Linguistics (196B), p. 322: "Even in the Indo-European languages the copulative function of 'the verb to be' appears to be of secondary development". 5 For the argument see 'Be', pp. 199 - 207. 6 Thus to explain the stative value of copula*es- we would have to posit an original sense "to stay, remain" which is unattested, and turns out to be only a projection of the 'be''become' contrast for the copula. The priority of the copula uses is partially clarified below; for fuller discussions see Chapter VIII of 'Be', especially pp. 395 - 402,407 -409. For methodological remarks on the claim of priority here, see 'On the Theory of the Verb To Be', in Logic and Ontology ed. by M. K. Munitz (New York, 1973), pp. 17 - 20. 7 See in particular E. Benveniste, 'Categories de pensee et categories de langues' and , "~tre" et "avoir" dans leurs fonctions linguistique', in Problemes de linguistique generale, pp. 63 -74 and IB7 -193. 8 , "Being" in Classical Chinese', The Verb 'be' and Its Synonyms, ed. by J. W. M. Verhaar, Part I (Reidel, 1967), p. 15. Compare Lyons, Introduction to Theoretical Linguistics, pp. 322f. 9 See the reference to Mill's Logic in Note 3, above. 10 There has recently been a noticeable trend away from the Mill- Russell view that "is" has different senses, which the Greek philosophers should have distinguished. See, e.g., Benson Mates' suggestion that Plato's different uses of "is" can all be understood on the basis of a single, univocal use of the copula: 'Identity and Predication in Plato', Phronesis 24 (1979), 211- 229. And compare Jaakko Hintikka's paper in this volume. In my opinion, the question whether "is" has different meanings or only different uses cannot be answered without confronting certain very deep problems in the theory of meaning, which is ultimately a part of the theory of knowledge. For example, are senses of a word distinguishable as a matter of logical form and conceptual truth, independently of any factual question as to the kinds and natures of the things to which the word is applied? Up to a point, linguistics can settle questions of syntax and sentence structure. But epistemology and metaphysics must be called in to decide how linguistic "meanings" are related to the nature of things or to our "conceptual scheme". II See, e.g., Lyons, Introduction to Theoretical Linguistics, pp. 322f. 12 C. J. F Williams, What is Existence? (Oxford, 19BI), echoing Quine, Word and Object, pp. 96f. This view of the copula ignores the distinction between an 'is' of identity and of predication, a logical distinction which is not reflected in the syntax of the verb and is not plausibly regarded as a difference in meaning for 'is'. My argument for this view (in 'Be' p. 372, Note I and p. 400, Note 33) is defended by C. J. F. Williams, op. cit., pp. 10-12. For criticism of this view on philosophical rather than linguistic grounds, see Ernst Tugendhat, 'Die Seinsfrage und ihre sprachliche Grundlage', in Philosophische Rundschau 24 (I977), 164. 13 Logica 'Ingredientibus', ed. by Geyer, p. 351, cited with other passages from Abelard in 'On the Terminology for Copula and Existence', in Islamic Philosophy and the Classical Tradition. Essays presented . .. to Richard Walzer (Cassirer, 1972), pp. 146 - 149. 3

24

CHARLES H. KAHN

14 For the notion of truth claim, see 'Be', pp. 186f; 'Theory of the Verb', pp. Ilf. Compare Quine's statement: "Predication joins a general term and a singular term to form a sentence that is true or false according as the general term is true or false of the object, if any, to which the singular term refers" (Word and Object, p. %). This makes clt:ar the sense in which predication is more than a syntactic notion. 15 For the stative-mutative contrast see' Be' , pp. 194 - 198, following Lyons, Introduction to Theoretical Linguistics, pp. 397ff. Compare Benveniste, Problemes de linguistique generale, p. 198: "etre' ... est ... un verbe d'etat, ... est meme par excellence Ie verbe d'etat" . The durative aspect emerges as a distinct "sense" of the verb in the Type I ("vital") use with persons, where einai means "continue (in life), survive": eti eisi "they are still alive", theoi aiei eontes "the gods who live forever". See 'Be', 241 ff. 16 For the verbs of posture as static be-replacer, see' Be', pp. 217 - 219. 17 See my own exposition of this view in 'Be', pp. 225f, 375 - 379, with the work of J. Klowski cited there (p. 375n.) from Archiv fur Geschichte der Philosophie 49 (1967), 138ff. 18 'Be', pp. 201ff. 19 Cf. Thucydides VII. 8.2: tpo{3ov,uvos 010 1'17 oi 1rtl'1rO,uVOL ... ou Ta DVTCl Cx1ra-y-yi>V.w(lIv "(Nicias) fearing that this messengers might not report the facts (sent a written letter)". For additional examples, see 'Be', pp. 335 - 355. The veridical "is" appears in Shakespeare, e.g., King Lear IV.vi.l41: "I would not take this from report: it is,/ And my heart breaks at it". The idiom is still alive and well in contemporary speech: "Tell it like

it is" .

'Be', p. 332, Note 2. For the distinction between the intentional it-is-so of judgment and statement and the objective being-so of things in the world, see my article in Phronesis 26 (1981) 126f. This corresponds to the distinction between the roles of infinitives and participles, respectively, in Aristotle's definition of truth: "to say of what is (to on, objectively) that it is (einai, intentionally), to say of what is not (to me on) that it is not ... " (Met. r, 7, 101lb26). 22 See the passages from Phaedo 65B - 66C cited in Phronesis (1981), 109. Cf. the example from Thucydides in Note 19 above, and passages where 0 fWV M-yos means "the true report" (in' Be', p. 354). 23 Met. t. 7, discussed in Phronesis (1981), 106f. Other examples of veridical copula in 'Be', pp. 356 - 360. 24 Thus I ignore here two types (the "vital" use in Type I and the verb of occurrence in Type V) counted as existential in 'Be', pp. 239ff, 282ff. For Type III, see the next note. 25 In my existential Type III, which represents the plural of (I) above, instead of the verb in initial position we often have a kind of quantifier-word like "many" or "others": 1ro>v'al -yae Cxva UTeaTov tiUL xi>-.tVIJOL "For there are many paths up and down the encampment" (Iliad X.66). Further examples in 'Be', 261ff. 26 For examples, see 'Be~, 277ff. 27 Examples ln 'Be', 300ff. 28 See 'Why Existence Does Not Emerge as a Distinct Concept in Greek Philosophy', Archiv fur Geschichte der Philosophie 58 (1976), 323 - 34. 29 The one case where einai provides .m independent predicate is my Type I "vital" use for persons, where ouketi esti means "he is no longer alive". See Note 15 above. 30 Further discussion in 'Be', 301, 303 - 6, 320 - 323, 326 - 330. 31 With this "pure existential" use contrast a typical non-technical existential in Herodotus, with a locative restriction as in type I above: "There is no stag or wild boar in all of Libya" (IV, 192.2, cited in 'Be', p. 327).

20 21

THE VERB 'TO BE' AND THE CONCEPT OF BEING

25

See Note 21 above. For this interpretation of Protagoras' homo mensura formula, see Phronesis (1981), 117 -119. 34 See the star example of this instability in Post. An. I!. I - 2, where the question ei esti is initially existential ("is there or is there not a centaur or a god"), but then gets divided into "particular" (epi merous) and "unqualified" (haplos) cases, where the unqualified cases are still more or less existential (fi fonv ~ /l~ OfA~VT/ ij vu~) "whether or not there is a moon or whether it is (?) night "), but the particular cases are not: "is the moon being eclipsed? or is it waxing?" (90" 1 - 5). This problem has been much discussed. (See Ross' commentary, pp. 610 - 612; A. Gomez-Lobo in Review oj Metaphysics 34 (1980), 71 - 89.) Barnes (Aristotle's 'Posterior Analytics, p. 194) takes the particular or "partial" (epi merous) question ei esti in Chapter 2 to be a reformulation of the hoti question in Chapter 1: " 'X is Y' says that X is 'partially' because 'X is' is a part of 'X is Y' ". Even if this is right, it would show that Aristotle did not consistently read ei esti (or hoti est!) as existential. The copula construction X is Y (fi rae ion rt ij /l~ fan n at 90" 4) is treated as a special case of the unqualified ("existential") esti. For what seems to me the most plausible explanation of this rather baffling fact, see the notion of predicative complex borrowed from Mohan Matthen in Note 46 below. According to this suggestion, Aristotle thinks of X is Yas equivalent to the YX exists. 3S Aside from the problematic example from Parmenides to be discussed in the text, the only type 3 existentials older than Melissus and Protagoras seem to be in the fragments of Zeno, where the syntax i~ uncertain. In frs. I and 3, fi 1rOMa fonv can be read either as "if (the) many exist" or as "if (beings) are many". In fro 3 aft rae fUeCX /lfrcx~V rwv ovrwv iort "there are always other things between the beings" the locative qualification recalls my Homeric Type III, the plural of type I above. (Compare the locativeexistential from Herodotus in Note 31 above.) But fiT/ and d Of fonv at the beginning of fro I seem to be straight-forward cases of type 3 existentials. 16 See passages cited in Phronesis 1981, p. 130, Note 17. 37 Here I am agreeing with, and in part echoing, a series of studies by G. E. L. Owen. Compare: "There is [for Aristotle) no general sense to the claim that something exists over and above one of the particular senses" ('Logic and Metaphysics in Some Earlier Works of Aristotle', in Articles on Aristotle Vol. 3, ed. by Barnes, Schofield, Sorabji = p. 165 in Aristotle and Plato in Mid-Fourth Century, ed. by DUring and Owen); "To be, then, is always to be something or other" (,Aristotle on the Snares of Ontology', in New Essays on Plato and Aristotle ed. by R. Bambrough, pp. 76ff); "The concept of being that he [viz. Plato) takes himself to he elucidating here [in the Sophist) is not that of existence" ('Plato on Not-Being' in Plato. A Collection oj Critical Essays I. Metaphysics and Epistemology, ed. by G. Vlastos, pp. 240f). 38 The view that I am opposing is defended by D. Gallop in ' "Is" or "Is not"?', The Monist 62 (1979),61 ff and J. Barnes, The Presocratic Philosophers, Vol. I, pp. 161 ff. BOlh Gallop and Barnes are following G. E. L. Owen, 'Eleatic Questions', Classical Quarterly 10 (1960), 84 - 102. It is ironicallhat these scholars should not have recognized the extent to which Owen's own work on einai in Plato and Aristotle (cited in the preceding note) has succeeded in undermining their assumption that the modern notion of existence is an appropriate instrument for capturing the sense of 'to be' in Greek philosophy. Owen himself later expressed some qualms about his use of the "conventional choice" between copula and existence to decide in favor of the traditional, uncriticized notion of existence for intepreting esti in Parmenides' thesis. (See 'Plato on Not-Being', p. 225.) 39 My interpretation here follows the main lines of the view developed in greater detail in 32

33

26

CHARLES H. KAHN

'The Thesis of Parmenides' and 'More on Parmenides', Review of Metaphysics 22 (1969), 700- 24 and 23 (1969),333 -40. But my earlier interpretation of esti and ouk esti in the thesis now seems to me too schematic. 40 Despite its recent reprinting by Tanin and other editors, I think that' A>4IEL'15 fU}(V}(Af05 in fro 1.29 is indefensible, both in terms of MSS. evidence and the rules for nounformation. (On this see Gnomon 40 (1968), 124.) The reading firndlfos is better attested and is guaranteed by the context: "persuasive truth" answers, with chiastic reversal, to pistis alethes in the next verse . The thought ("trust only in the truth") is fundamental, and recurs in 2.4 (the Way of Persuasion, who follows Truth) and 8.50 (the end of the pistos logos concerning truth). 41 G. E. L. Owen identified the subject as "what can be talked or thought about" ('Eleatic Questions', p. 95) without reference to the proem or the context. Gallop first suggests the vaguer subject "a thing", but ends by following Owen (Monist (1979),68 and 71). Barnes (The Presocratics I, 163) approximates to my identification by taking the subject of esti to be "the implicit object of dizesios", i.e., the object of inquiry . 42 "Nor can you point it out" (oute phrasais, 2.8), i.e., you can give no reliable information about what is not the case. Of course you can say what is not so, and hence some stronger notion than mere statement seems to be implied by Parmenides' claim "it is not sayable (ouphaton) . .. that it is not'" in 8.9. (The same problem arises for the rival interpretation of esti as "exists", for of course we can talk about what does not exist.) It was perhaps to strengthen this side of Parmenides' thesis that the paradox of false statement was first formulated. 43 Thus J . Hintikka rightly suggested that a rule like" 'm knows that p' entails 'p' " might be called "Parmenides' law", on the basis of fro 2.7. See his Knowledge and Belief, p. 22, Note 7. 44 Except for the locative-spatial implication, which does not hold for the Forms, Plato's adaptation of Parmenides' argument in Rep. V, 476 Eff follows the same steps, from the veridical on as object of knowledge (to on gnonai hos echei at 478 A 6) to the existential whose negation is nothing (meden at 478 B 12) and the copula at 479 B 9: "is there any one of these many things which is any more than it is not what one says it to be?" In view of this close affinity between copula and veridical uses (since the simpler examples of facts or states of affairs can always be framed in an X is Y construction), there is no real incompatibility between my reading of the Parmenidean esti and Mourelatos' proposal to regard the thesis as a predicative sentence frame: .. - - - is - - -" (with .. - - - is not - - -" for the negation). The basic copula function of esti in Greek will assure that, on any reading of the thesis, this sentence form is immediately felt to be implied. And I would not exclude Mourelatos' notion of "speculative predication" (as he says, this is a narrower concept which "falls within the range" of the veridical). But I do not think there is enough. early evidence for "quiddity" uses of einai to justify such a restriction on the primary reading of esti. See A. P. D . Mourelatos, The Route of Parmenides, pp. 55 - 59. 45 For Plato, see the preceding note; for Aristotle see Note 34 above. These remarks represent my answer to the second of Gallop's three objections to my interpretation, namely that a state of affairs is "of the wrong logical type to serve as the bearer of such attributes as 'ungenerable' ... and 'immovable' ", which belong rather to a thing-like entity (Monist (1979),66). I agree, and this might stand as an objection to Parmenides' argument. But it counts as an objection to my interpretation only if one assumes that Parmenides is in-. capable of overlooking such a distinction of logical type (as Plato and Aristotle certainly were not!). Much the same holds for Gallop's first objection, that a premise which presup-

THE VERB 'TO BE' AND THE CONCEPT OF BEING

27

poses that knowledge of the truth is possible is too weak to support Parmenides' argument, since "a sceptic might well respond, that no one knows anything, precisely because there is no truth to be known" (ibid). This objection seems to me to mistake Parmenides for Descartes. Why should Parmenides be thought of as arguing against a sceptic? His argument is about how to get to the truth and what one will find there, not about whether there is any such thing. Gallop's third objection is more substantial: if we start by rejecting "what is not the case" as an object for knowledge, how do we get to the rejection of non-existence that is required to disprove generation and perishing? (ibid., pp. 67 and 72). I agree that we must find in to me eon a sense of not-being which is equivalent to "nothing at ail", and if this is what is meant by an existential eon, then Parmenides' Being must be existential. But, as suggested above, if to eon as a determinate state of affairs is understood to contain or imply a real ("existing") subject and definite attributes, and if the negation (to me eon) is understood as denying everything contained in or implied by to eon, then "what is not" must be construed not as a well-defined, unrealized state of affairs but rather as a blank non-entity: no subject ("what does not exist") with no attributes ("is not P' for every Fl. It is obvious that Not-being so understood must turn out to be not only unknowable but indescribable. Plato will defuse Parmenides' argument precisely by distinguishing this hopelessly unqualified Not-being from the more precisely defined not-being-F for various values of F. 46 The most enlightening explanation known to me for the easy shift from propositional to existential and copulative construals of einai in Greek 'philosophy is the notion of a "predicative complex" proposed by Mohan Matthen in an unpublished paper. Matthen defines a predicative complex as "an entity formed from a universal and a particular when that particular instantiates the universar'. Thus artistic Coriscus is such an entity, which "exists when and only when Coriscus is artistic". In grammatical terms, a predicative complex (or rather, its linguistic expression) is the attributive transform of an ordinary copula sentence: corresponding to X is Y we may assume the existence of a logically equivalent predicative complex, the YX exists. Thus for (I) Socrates is healthy we have the corresponding (2) The healthy Socrates exists, where the truth conditions for (I) and (2) are assumed to be identical. Furthermore, truth conditions will also be the same for the veridical transform of (I), namely (lA): It is the case that Socrates is healthy. Aristotle in Met. 1!t..7 slides effortlessly between (1) and (lA). (See 1017" 33 - 35, as interpreted in Phronesis (l9SI), 106f) Now if (I) is transformed as (2), we see how the copula-veridical-existential slide can seem so natural in Greek, since all three formulations are logically equivalent. I believe this construal (following Matlhen) captures something quite deep, and quite strange to us, in the use of einai by the Greek philosophers. And it shows why our conventional dichotomy between existence and copula imposes a choice upon the interprett:r which corresponds to nothing in the Greek data. Also, our difference in "logical form" between propositional (fact-like) and substantial (thing-like) entities as subject of einai will reappear in this conception simply as a difference in formulation between (I A) and (2). 47 For the double construal of on at Symposium 21 IA I, see Phronesis (l9SI), lOR. 48 First in an idiomatic variant at 65D 13 - E I, then progressively from 74B 2 to 75D and 7SD. See Phronesis (1981), 109- Ill. 49 For a convergence of veridical (Being versus Seeming) and "existential" values (Being versus Becoming), see Rep. VI, 50SD 5 - 9, where to on is first paIred with atetheia, as radiating the light of rational cognition, and then contrasted with to gignomenon te kai apollumenon, the source of darkness and inconstant opinion. This is of course compatible with a slightly different contrast in the following section (50SE - 509B), where the role of

28

CHARLES H. KAHN

the Good as cause of truth and knowledge is distinguished from its role as source of Being (to einai te kai ten ousian) for the objects known. There Being for intelligible things is presented as parallel to generation (genesis) and growth for visible things and must refer to the stable existence of the Forms as appropriate objects for knowledge. (Shorey renders einai ... ousia as "existence and essence", thus recognizing that both ideas are in play; but it would be a mistake to look for any fundamental difference in sense between the verb and the noun. At 479C 7 he renders ousia alone as "existence or essence", again rightly.) so See, for example, the double syntax of aei on cited in Note 47. The stative, and hence potentially predicative rather than strictly existential force of expressions like (TO) ov ad is clearly indicated by the alternative formula with hELv: e.g. TO aft Xo/Ta TO/UTa WUo/VTWS fXOV for Forms as object of knowledge at Rep. VI, 484B 4 (cf. Phaedo 79A 9, 80B 2, etc.), which is immediately picked up by TO OV with veridical overtones (T~ ovn) at 484C 6: "those who are veritably deprived of the knowledge of the veritable being of things" (Shorey). Thus at 485B;2 ExfLVT/ ;, ouuLO/I;, ad oJu.i-YfTW }(CII')' aUTO. ob -yixe lcrTL TO crOL dva, TO JLovcr,)(~ dvw. -yae }(aTa cravTov d JLovcr,}(os. 0 aea }(aTa cravTov.)

ou

If we take this to give us different senses of 'to be' for Socrates on the one hand (for whom 'to be' means 'to be a man') and Bucephalus on the other (for whom 'to be' means 'to be a horse'), we are committing another blunder of the type mentioned above and to be dealt with below. We need not. We might just take it that filling in the predicate tells us under what conditions 'Socrates is' (i.e., 'Socrates exists') is true, and leave open the a)ternative possibility (which is, in fact, the right one) that what determines which predicate to fill in is, not 'is', but 'Socrates'. For the remainder of this section, I shall take it that this alternative possibility has not been ruled out, and leave further discussion for the next. Consider once more the two inferences

(E)

S is P

-+

S is

and (NCE) S is not P

-+

S is not.

It is not that the essentialist view we have just handed Aristotle requires him to reject the unrestricted use of (E) (again, leaving intensional and alienating predicates out of the picture). In the first instance, what he must reject is its converse, by restricting its application to cases in which 'P' is a predicate essentially true of S: cases in which S is by virtue of itself p.26 And so he must reject (NeE): only where S is P by virtue of itself, essentially, does it follow from'S is not P' that S is not. But all this has, given the background, an indirect impact on (E) as well. For the justification for the inference (E) was supposed to be simp)e simplification, the sort of move that justifies (P)

Socrates is a pale man

-+

Socrates is pale

58

RUSSELL M. DANCY

but breaks down for (L)

Socrates is a lousy provider

-+

Socrates is lousy.

For (P), that meant that 'pale' had to have the same role in antecedent and for (G)

Socrates is a good cobbler

-+

Socrates is good.

(L) and (G) were ruled out on the ground that the simplified residue had changed character in the course of the simplification. (P) was to be allowed because its residue was constant. And so, if (E) is to survive, it must show a constant residue. Consider, then, the instance of (E) that reads (el)

Socrates is pale

-+

Socrates is.

This is to be an inference with a detachable consequent. So we write its consequent on a slip of paper and put it to sea in a bottle. But when it washes up on a distant shore, its readers, who know that "what it is for each thing to be is what it is said t.o be by virtue of itself", and know that what Socrates is by virtue of himself is a man, will supply 'a man' as the completing noun. So (el) is a failure, and (E) is not in general true. But no such failure is involved with (e2)

Socrates is a man

-+

Socrates is.

(E), restricted to essential or by-virtue-of itself predicates, survives. In fact, confining (E) to essential predications is overly restrictive. Consider (e3)

That pale thing is a man

-+

that pale thing is.

We should not count 'that pale thing is a man' as a straightforward essentialpredication, and, in fact, Aristotle would not so count it. To the extent that he is willing to count it as a predication at all, he thinks of it as accidental: he thinks of it as grounded in, or having for its truth conditions, 'that man is pale' (see Section 4 below). And, as we have noted, in the Posterior Analytics he would prefer to legislate it away: "either it doesn't predicate at all, or it predicates not simply, but accidentally" (~TOL p:"ocxp.ws xcxrrnoeeiv, 7} xcxrq'Yoeeiv p.Ev 1'1, a1rAWS, XCXTCx CTvp.{3e{31/xos oE xcxrq'Yoeeiv, 83 a 15 - 17). For simplicity of outline, let us allow him the legislation he tries to pass in the Posterior Analytics. We shall expand our horizon slightly in Section 4 below.

ARISTOTLE AND EXISTENCE

59

We are now in a position to explain how Aristotle can suppose that to be is to be something or other without giving up the distinction between 'to be something or other' and just 'to be'. Take the latter, the distinction, first. According to this, dven n, to be something, or 'to be' + Pred., is not the same as dven cnrAws, just 'to be#'. The latter phrase is often translated 'to be without qualification', or, worse, 'to be absolutely'. These translations tempt one to make too much of the distinction. In fact, all Aristotle's phrase 'Elven a1rAWS' does is point to the 'is' in sentences such as (I)

Socrates is.

Here we have just 'is', not 'is' + Pred. Such occurrences mark existential sentences: if there were an existential sense of 'is' in Aristotle, this would be it. But there is plainly no special existential sense here; there is a merely syntactic difference between 'to be' followed by a predicate and 'to be #'. The latter results flom truncating a predication. So even 'to be #' is, implicitly, to be something. But, as we have seen, not just any predication may be so truncated: the predicate must be restorable. Under the simplifying assumption that such locations as 'that pale thing is a man' can be ruled out, the type of predication that allows truncation is an essential predication. So while to be# is to be something, it is not to be any old thing. In fact, (1) is elliptical for (3)

Socrates is a man.

That is the theory. Let us translate it into English . 2. BEING AND EXISTING IN ENGLISH

The word 'is' does not behave the same way in normal English: sentence (I) seems to be used in English primarily to translate philosophical Greek. There is a more or less archaic use of the verb 'be' that survives in recitations of Hamlet's soliloquy and Owen's example 'Arrowby is no more',26 and our ability to understand the locution may give a fingernail-hold for the theory in English, but it would be nice to have more. The first thing to notice is the oddity already mentioned about (4)

Socrates exists.

I have in mind not the oddity that can be corrected by changing tenses

60

RUSSELL M. DANCY

and adding adverbs, but the requirement placed on understanding the sentence of some background knowledge associated with the name 'Socrates': if you spring it on someone not in the know, he may not have the faintest idea what is going on. So also with the sentence-form (5)

There is (such a - as) Plato.

Without the parenthetical material, properly filled in, this does not even scan.!7 And properly filling in the blank requires some fairly substantial covering noun or noun phrase: just 'thing', for example, will only work in quite special cases ("My good man, there is such a thing as decency, you know", and so on). 'Plato' is the name of a computer language as well as that of a philosopher, and, for that matter, I could use it to name my pet cat, my pet theory, my favorite number, the national debt, the coin Washington skimmed across the Delaware, or the value of that coin in real terms on Christman Day 1776 given in terms of the Laspeyres index. In the absence of a covering phrase that has some bite to it, we do not know what is being talked about. It is not altogether clear, in that case, what pigeon-hole to place the ignorance in. Some will, no doubt, feel the urge to argue that it is not linguistically relevant. The pattern for such an argument might be this: invoke the sense-reference distinction and couple it (although this would not have made Frege very happy), with Mill's doctrine of proper names. According to that doctrine (translated into Fregese), proper names do not have sense. So when someone says 'I'm studying Plato' (or 'I'M STUDYING PLATO'), and it is not clear to you whether she is trying to grasp the theory of recollection or the theory of programmed learning, that does not mean you have failed to get the sense of what she is saying: you have got all the sense there is to get if you know the rest of her words and, perhaps, the minimal fact that 'Plato' (or 'PLATO') is a proper name. This seems to me one of the many places where the sense-reference distinction and its alliance with Mill's theory of proper names lead us astray. It seems to me clear that there is something about the sense of what she said that we are not getting, and somethin£ associated with the proper name in it. 28 But I am not able to layout a proper account of the matter, at least not here. So I shall simply assert that (4) and (5) require for their understanding, for grasping their sense, some fairly substantial covering noun or covering noun-phrase, and admit that I do not know whether the 'explication' (I shudder at the word) of making sense belongs to the science of semantics, that of pragmatics, or necromancy.

ARISTOTLE AND EXISTENCE

61

If this is correct, Aristotle is at least partly right: the demand for an informative covering noun or phrase is his demand for an answer to the question 'what is Socrates?' to complete the sense of 'Socrates is'. It is no accident that this is close to some of the things that have recently been said 29 in connection with identity by way of reviving Aristotle's essentialism. But as things are, an argument may suggest itself that would undermine any further steps in beating Aristotle's views into Anglo-Saxon shape. For it may seem that, whatever the facts about Greek (which is, after all, a dead language and perhaps should be allowed to rest in peace), we do in English have a special sense for 'is' that is plainly existential. Consider the 'is' in

(6)

There is such a person as Plato,

(again, embedded in a conversational context that gives it naturalness), or, better, the 'was' in 'there really was such a man as Prester John'. These verbs can be replaced by 'exists' and 'existed', respectively. But the 'is' in 'Plato is a person (who lives in Athens ... )' and the 'was' ·in 'Prester John was a man (who ... )' cannot be so replaced. So the 'is' in (6) is existential. I take it no one really falls for such replacement-arguments without further support. But, in the absence of any alternative account of the 'is' in (6), this one may cause consternation. So I shall say something about how an alternative account might go, without pretending to completeness or even correctness. I think we can see enough to tell that these are the right lines. 30 Consider the following sentences: (7) (8) (9) (10) (11) (12) (13)

There arose a furious clamor. There ensued a riot. There followed the entire retinue. There fought by his side the bravest of the Romans. There resulted the worst disaster he remembered. Outside her window there sang a popular singer. There sat next to me the saddest man in the world.

We do not need a special existential sense of 'arise', 'ensue', 'follow', 'fight', 'result', 'sing', or 'sit' to explain how we can have such sentences alongside their counterparts with the verb in the 'predicative' position. The latter are sentences such as

62

RUSSELL M. DANCY

( 7') (10') (13' )

A furious clamor arose. The bravest of the Romans fought by this side. The saddest man in the world sat next to me.

And what we require is (to put it in a way that was fashionable in linguistics last week or the week before) a transformation that will get us from such sentences (or the structures that underly them) as (7') - (13') to (7) - (13). The transformation would be labeled 'THERE-insertion' and would look a little like this: (T)

X - NP! - V (NP2) y X - there - V (such NP 2 as) Y NP!.

At least, this would generate (7) - (13) from (7') - 13'). And, just among others, it would generate 'There is such a man as Plato' from 'Plato is a man'. I said I would not pretend to correctness. And (T) is, accordingly, not correct. It would generate some things we - or at least I - do not want. E.g., starting from (14')

These books weigh a lot

We could use (T) to derive *(14)

There weigh a lot these books.

And it is not clear where the stuff represented by 'X' in (T) should go: for the move from (13') to (13), (T) seems all right as it stands ('next to me' is carried along with the verb), but with (15')

Mice are in the bathtub

(T) would generate *(15)

There are in the bathtub mice

where what we (at any rate, I) really want is (15)

There are mice in the bathtub.

So the rule requires a lot of tinkering with. The point is that, blemished as it is, this is the right sort of thing: we do not want to rewrite the dictionary to take account of all these newly discovered existential senses of all these verbs. But if this is the right 30rt of thing, 'is' becomes just one verb among others subject to the transformation, and bears no special sense when it follows 'there' as in (5) or (6):

ARISTOTLE AND EXISTENCE

63

it is simply the same verb as in 'Plato is a man' with the surrounding sentence reprocessed. 31 So we are left with 'simple sentences': subject-predicate sentences. These frequently employ 'is'. But this word is, after all, quite meaningless: it is merely a syntactic device for connecting subjects and predicates where the predicates are not already verb phrases, just as the word 'thing' is a device for turning adjectival phrases and others (e.g. 'of the mind') into noun phrases (e.g., 'things of the mind'), and just as 'to' is merely a device for turning verbs into infinitives (e.g., 'to walk', 'to be or not to be,).32 So Heidegger is right: But being remains unfindable, almost like nothing, or in the end entirely like that: The word 'being' is finally, then, only an empty word. It means nothing actual, tangible, real. Its meaning is an unactual vapor. 33

But for Heidegger, this represents a lamentable loss: apparently, for the Greeks before Plato, Erven was chockful to bursting with meariing. And about that, Heidegger is wrong. But fifty percent is not bad. So far all we have done is to show that the part of Aristotle's theory that seems at first provincial to Greek may, in fact, have something corresponding to it in English. But, of course, that is not all there is to the theory. For one thing, there is also its essentialism. This is not, on the fact of it, a linguistic matter: the doctrine of essentialism is as plausible or implausible in English, I take it, as it is in Greek. But it scmetimes sounds as if the doctrine might not be plausible in Chinese,34 or Nootka,35 or Rortyspeak .36 Here I am going to leave these languages, along with others of which I am innocent, on one side. But there is one thing about Aristotle's employment of his essentialism that must be mentioned, that is closely connected with the presence or absence of different senses of 'is' in English or in Greek. To get at this, first consider another distinction that some have alleged to be pertinent to English and Greek being: that between identity and predication. 37 There is certainly a distinction here: it is between claims like (16)

Dr. Jekyll is Mr. Hyde.

on the one hand, and claims like (17)

Dr. Jekyll is schizoid

64

RUSSELL M. DANCY

or (18)

Dr. Jekyll is an addict

on the other.(16) states an identity; (17) and (18), I think,38 do not, and I shall characterize both as predications. Philosophers are, notoriously, myopic, and when they apply their magnifying glasses to sentences that do different things, they usually can only manage to bring into focus a single word. 39 Here the word is 'is'. They profess to spot the 'is' of identity in (16) and the 'is' of predication in (17), and then fall to arguing over the character of the 'is' in (18). I shall call this habit of supposing that every difference in character from one sentence to the next must be locatable in single ambiguous words the 'fallacy of the magnifying glass'. We do not need to pull the word 'is' out of those sentences and go into Angst over the meaning of being. The situation is completely described by saying that 'is' is followed in (16) by a singular definite noun phrase, in (17) by an adjective, and in (18) by an indefinite noun phrase. Sentences that show the structure of (16) state identities; the others are predications. But this has nothing to do with the occurence of 'is' in different colors: it is what comes next that counts. 40 Aristotle does not commit the fallacy of the magnifying glass in connection with 'dvOIL', identity, and predication. 41 But he is prone to the fallacy; conceivably he is its inventor; probably42 he is its most influential perpetrator. And if he falls into it anywhere, he falls into it with 'is'. He thinks that the fact that 'man' and 'pale' relate differently to Socrates makes for a different 'is' in (3)

Socrates is a man

from the one in (2)

Socrates is pale,

the former is an 'is' xed)' CXUTO, a by-virtue-of itself 'is'; the latter is an 'is' XCXTO: avp.(3f(31]XOS, an accidential 'is' (see, e.g., Met. d 7. 1017a 7 - 8, and below, Section 4). Unless there is more to it, this is just as much a case of the fallacy of the magnifying glass as the idea that 'is' varies in sense from identities to predications. Perhaps it is a natural mistake. That Socrates is a man is essential to him; .that he is pale is accidental to him. These are different relationships. They are both covered by the word 'is'. So it is natural to talk as if there

ARISTOTLE AND EXISTENCE

65

were two varieties of 'is'. In the sequel, I shall occasionally go along with Aristotle's talking that way. But I do it under protest. If this way of talking is not to be explained away merely as a way of talking, but is to be taken as marking a genuine distinction in senses of 'is', or concepts of being, or whatever, it is a case of the fallacy of the magnifying glass. It would be a great relief to get Aristotle off this particular hook. I do not at present know how to do it. Rephrased in these terms, and still leaving 'that pale thing is a man' out of account, the theory of Section 1 becomes this. The 'is' of (3) is a by-virtue-of itself 'is', and permits cancellation of the predicate. The 'is' of the residue, 'Socrates is', is then a by-virtue-of itself 'is', and also an 'is' &'1I'~Ws, 'is#', since it no longer has an attached predicate. But these are two different characterizations: roughly, the latter is syntactic and the former semantic. To go one step farther along the road to perdition, we might say: the existential 'is' is, paradigmatically (see Section 4), a by-virtue-of itself 'is', but the by-virtue-of itself 'is' is, in the first instance, a predicative 'is', and only becomes existential by cancellation of the predicate. 3. DE INTERPRETA TIONE II. 21" 25-28.

In the final lines (21 a 18 - 34) of De into 11, Aristotle is investigating the rule opposite to the one just dealt with: before he was investigating when things predicated on their own could also be predicated together. but in these lines he is investigating when things predicated together can also be predicated separately.

So Ammonius (in De into 210. 17 - 20)43 rightly says. Aristotle has just been pointing out, for example, that the argumentative rule we might call 'Addition': (A)

S is Adj. & S is NP

-+

S is Adj. NP

does not always work: from 'Socrates is good', where we have "simply 'good' " or 'good #' (b1l'~ws ayat'Jos, 21 a15) and 'Socrates is a cobbler', you cannot use (A) to get 'Socrates is a good cobbler' (20 b 35 - 36).44 Now he notes that the rule we have been calling 'Simplification' also needs restricting. He is very brief about this: he mentions all three of the exceptions we listed in Section 1 above, namely things that are not (see 21 a 32 - 33), dead men (see 21 a 23), and inferences to existence based on the wrong predicate, and it is not easy to see what he is trying to say about them. I have excused myself from dealing with the first two here. What he has to say about the third is this. 21 a 18 -21 tell us:

66

RUSSELL M. DANCY

It is true to speak of the particular [more literally, 'of the something'] simply as well, for example, [to say that] the particular man [is a] man, or the particular pale man [is] pale, but not always ... (ah'1t?~~ 0' fUTiv fl1f'fiv xCiTa TOU TtVO~ xCii (X1f'hW~, olov TOV Ttva avt?eW1f'OV avt?eW1f'OV ij TOV Ttva hfUXOV avt?eW1f'OV hfUXCIV' oux afi Of. ... ).

It is not clear what the first of these examples is telling us we are permitted to do,45 but the second is plainly allowing us the inference labeled '(P)' in Section 1 above. 46 Aristotle then points out that simplification will not work with 'dead man': in cases of this kind, where there is a contradiction between the two terms antecedent to the simplification, the rule never works. (There is a puzzle here: apparently Aristotle·thinks 'this is a dead man' can be true, although 'dead man' involves a sort of contradiction.) But in other cases, sometimes it works and sometimes it does not; for example, Homer is something, e.g., a poet; well, is he, therefore, or not? For 'is' is predicated accidentally of Homer; for because he is a poet, not by-virtue-of itself, 'is' is predicated of Homer. (21 a 25 - 28: wune ·OJL'1eO~ fUT! Tt, oiov 1f'OI'1rT,~' &e' oOV XCi! EUTlV, ij ou; xCiTa UUJL(jf(j'1XO~ ,),ae XCiT7J')'OefiTCiI TO EUTIV TOU 'OJL~eOU' OTt ,),ae 1f'OI'1rT,~ iUTlV, aM' OU XCit?' CiUTO, XClT7J,),oefiTClI xCiTa TOU 'OJL~eOU TO EUTlV.)

The traditional way of taking these lines is as saying that it does not follow from (19)

Homer is a poet

that (20

Homer is.

And along with this, often the denial that 'is' by-virtue-of itself applies to Homer is understood to be the denial that the existential 'is' applies to Homer (and then Aristotle must have had advanced views about the 'Homeric question'). In my view, neither of these things is correct. That is, to be as explicit as possible: Aristotle is not denying that (20) follows from (19), and "'is' by-virtue-of itself" is not the 'is' of existence. 47 I take up these points in reverse order. Both bring in the theory of Section 1. The latter is a simple application of it. The theory of Section 1 tells us that the 'is' of 'Homer is a man' is a }(cxf)' CXUTO 'is', a by-virtue-of itself 'is', yielding 'Homer is #', and that this is the existential claim, 'Homer exists'. And the theory tells us that the 'is' of (19) is a }(CXTa (JUP.{3f{3.,.,}(O~ 'is', an accidental 'is', and so simplification may not be performed on (19). That is what Aristotle is here telling us. Since he is giving as a reason for prohibiting simplifica-

ARISTOTLE AND EXISTENCE

67

tion the claim that 'is' is not here used xat'}' aUTO, he cannot mean by this claim that 'is' is not here used existentially. That, after all, is the question. He is simply saying: since (19) is not an essential predication, and its 'is' is therefore not xat'}' aUTO, the predicate may not be canceled by simplification to yield 'is' a7rhWS, the "existential 'is'''. This way of putting it yields considerable ground to the fallacy of the magnifying glass, but it certainly shortens the work. So much for the second point: simplification cannot be applied to (19) because its 'is' does not apply to Homer xat'}' aUTO. But then the first point is clear: Aristotle is not saying that (20) does not follow from (19), but that it does not follow by simplification. He puts forth the question: "Homer is ... a poet; well, is he, therefore, or not?" In its context it is not hard to read this, not as asking "does it follow, by some devious means or other, that Homer is?", but "does the move we are talking about apply here?" This may, at first sight, seem unnecessary. But there are two good reasons why we should not take Aristotle to be saying that it does not follow from 'Homer is a poet' that 'Homer is'. The first is that we know from elsewhere that, when it comes to the question whether (20) follows from (19), he should say that it does. This is a point he considers in Categories 10, and quite unambiguously decides: when there is no Socrates, 'Socrates is healthy', 'Socrates is sick', 'Socrates is blind', etc., are false, and their negations true. The apparent conflict of that passage with De into 11 is a notorious cruX. 48 In my view, there is no conflict. The second reason for avoiding making Aristotle deny the entailment is that there is a very simple argument in favor of it, based in part on the De into passage itself. Homer is a poet. But poets are, after all, human; so they are men. So Homer is a man. But there the 'is' is a xat'}' aUTO 'is', and we can simplify. So Homer is. This argument is, admittedly, phrased in terms of my interpretation of the passage. But there is very little about it that is speci fie to that interpretation. On virtually any understanding of the passage, the inference has to fail because the 'is' of (19) is not xm'}' aUTO applied to Homer, where the 'is' in 'Homer is a man' is applied to him xat'}' aUTO. So all we need is the concession that poets are men, and we are away. It is extremely difficult to see how tha! could be denied. (It may be worth noting that it would not help to say: well, for all (19) has to say, Homer might have been a god. Then we would have another alternative, all right, but it would get us to the unwanted conclusion just as easily.)

68

RUSSELL M. DANCY

4. METAPHYSICS 67. 1017" 7 - 30

Here, especially, more needs to be said than I shall say. This chapter gives us four main headings 49 under which to rank 'that which is', 'to be', or 'is' (TO DV 1017 a6, bl,TO flvw a24, 31, 35, TO i!anv a31). About the latter two, the 'is' that signifies that something is true e31 - 35) and 'to be' signifying actually or potentially being (a35 - b9), our theory has nothing special to say. But the first two are accidental being and by-virtue-of itself being, and that is what our theory is about. But even here it will not explain everything. In particular, it will not explain, unaided, the most controversial of Aristotle's claims under these headings: the claim that 'to be' in the second of these ways of conceiving it shows variation from one category to the next. So, although I shall say a little to indicate how that claim might be dealt with, I shall not provide a full defence. It may seem that the sights have been lowered so far that we are in danger of shooting our toes off, but there are still difficulties to be met. Aristotle begins by stating that 'is' comes in an accidental variety and a by-virtue-of itself variety (a7 - 8), and promptly gives examples of the former 10):

e8 -

e.g., we say the just is cultivated, the man [is] cultivated, and the cultivated [is a] man. We may write: (21) (22) (23)

The just [one] is [a] cultivated [one]. The man is [a] cultivated [one]. The cultivated [one] is [a] man.

The English is pretty awful. The bracketed material has nothing to correspond to it in the Greek: it serves to remind us that, where in English the adjective 'cultivated' occurring as predicate is easily identified as a predicate because it is an unsupplemented adjective, in Greek the predicate adjective '/lovau(os' needs no supplementation in order to be treated as a noun phrase. These examples are to be construed as making reference to particular people in each case, the fellow holding up that lamppost over there, say: Aristotle is not talking about a maxim he and his friends like to utter to the effect that the just man is a cultivated man.

69

ARISTOTLE AND EXISTENCE

He adds another example which, he says, is similar (24)

nO - 11):

The cultivated [one] builds houses.

(Here 'builds houses' is a single verb, 'oixooop.f:iv'.) Notice that it is relevantly dissimilar to the preceding cases: it contains no 'is', and what we are talking about is accidental 'is'. Aristotle knows this, and promptly explains that (24) is so because it-is-accidental to the housebuilder to be cultivated, or to the cultivated [one to be al housebuilder (all - 12: OT! (Jup./3i/3rl'lt T~ oixooop..yJ P.OU(JLX~ dvOIL ij T~ P.OU(JLX~ oixooop.",,).

This imports the needed 'to be'. Aristotle is plainly not saying that (24) shows accidental 'to be', but its parapnrase, which he seems to think is more fundamental, does. So even (24) rests for its truth on accidental being. Aristotle's comment on (24) leads him to state the truth conditions for his examples: the general form for aJl, including the paraphrase for (24) (but not (24) itself) is "that this is this signifies that this is-accidental to this" (al3 - 14: TO -rae TOOf: dvcn TOOf: aT/p.CtLVH TO aup.{3f:{3T/XEvcn Tct>O(; TOOI':). More particularly, in cases like (21), both terms, 'just' and 'cultivated', are-accidental to the same thing e16) - namely, it appears, to the existent thing designated by the subject term; and in cases like (23), the subject term is-accidental to the predicate term (al7 - 18). These are paraphrases of sentences employing 'is' in predicative position. They do not themselves employ 'is' in that role, but in one of them, the paraphrase for (22), 'is' does occur. When Aristotle :.ummarizes in a19 - 22, it becomes clear that this occurrence is important: Well then, things said to be accidentally are so said either because both belong to the same [thing that) is, or because that [i.e., the predicate-term)so belongs to [a thing that) is, or because that, to which that of which it is predicated belongs, is.

These are occurrences of 'is # ': the claims are existential claims, and they are taken to justify the occurrence of 'is' in the predications (21) - (23). He is assuming stating (21) commits one to the existence of something both just and cultivated. And presumably he thinks that this requires the existence of, say a man who is both just and cultivated. 51 In a 18 - 19 he had made an incidental comment that pointed that same way: just as (23) is aJlowed because the cultivated belongs to the man, so also the not pale is said to be, because that to which it is-accidental is (OUTW Of 'Af-yf701L xed TO p.~ 'Awxov dvOlL, OT!

c;,

(JuP./3i/31/XfV, iXE;VO E(JT!V).

70

RUSSELL M. DANCY

So (25)

The not pale [thing] is #

is, marginally or parenthetically, acceptable. There are two possible derivations for its 'is # '. One may be suggested by the comment just quoted, viz.: (3) (I)

Socrates is a man. Socrates is # .

(I) is to follow from (3) in the prescribed way. Assume that it is Thursday night, after Socrates' appointment at the tanning parlor, and (26)

Socrates is [a] not pale [thing].

The using this and something like Leibniz' Law, we might get (25) from (I). If we did, the 'is # ' of (25) would be that of (I), and so, ultimately, that of (3), and so, again, a )Cod)' cxilTo 'is'. But the immediately preceding context suggests an alternative. The parenthetical comment of al8 - 19 is attached to a statement of the truthconditions for (23), 'The cultivated one is a man': this is so because the cultivated is accidental to the man, and so, Aristotle adds, even the not pale is said to be. That sounds as if Aristotle had the following derivation in mind. Start, as before, with (3), and assume (26); this yields (27)

The not pale thing is a man.

Then cancel the predicate, yielding (25). If we do it this way, the 'is #' of (25) is that of (27), and we have a case of 'is #' which Aristotle would describe as also an accidental 'is'. That was ruled out only on the theory under the restriction provided by An. post. A 22. 83 a l - 23, that made 'that pale thing is a man', and presumably (27) along with it, not a case of predication. But there was nothing intrinsic to the theory that brought this restriction on. We could not allow simplification to operate on (2)

Socrates is [a] pale [thing],

because, once context-free, its predicate would not be restorable: the answer to the question 'what is Socrates?' would lead to the completion 'a man'. But if we now count (27), et al., as accidental predications, we have cases in which the restoration would work as well as it ever does: the question 'what is the pale thing?' would be answered by 'a man'. (Of

71

ARISTOTLE AND EXISTENCE

course, to answer the question, you have to know what is being referred to, but that is just as much true of 'what is Socrates?' as of 'what is the pale thing?'.) The effect of this understanding of the passage, to which I am inclined, is to pry apart the 'existential "is" , and the xm'}' aUTO 'is' even farther: the 'is' of (25) is the former but not the latter. Still, it would remain so that the 'is' of (25) is dependent for its presence on a xat'}' aUTO 'is': that of (3). So the general message so far is that standing behind every accidental 'is' there is a by-virtue-of itself 'is'. What, then, about this latter 'is'? Aristotle has this to say (1017 a 22 - 27): As many things are said to be by-virtue-of themselves as the figures of predication signify, for in as many ways as [they) are said, in so many ways 'to be' signifies. So, since of things predicated, some signify what [itl is, some what-[it)-is-like, some how-big, some relative-to what, some to do or to undergo, some where, some when, 'to be' signifies the same things as each of these. (xali' aUTa Of dVaL hf-YfTat ouang Uf/p.aLVft Ta C1X~p.aTa rij5 xarrnogLa5' OUCa'ieav, a'A'ACx TOUT' ~V TO Ti ~V dvO't fxaTl'e~.] In the sense spelled out by Aristotle in Met. H 6, 1045 a 14 - 34, he thus ends up subscribing to the identity of a man and what it is for a man to be, alias the essence of man. Hence in the last Aristotelian analysis, essential predications are in a sense identities. Be this as it may, what has been seen suffices to show that there are

96

JAAKKO HINTIKKA

extremely important connections between the different uses (not senses!) of esti in Aristotle and his central metaphysical doctrines. Such uses mean that on certain occasions the force of to esti is purely existental or purely identificatory, even though these are not separate senses of the verb. For this reason it is misleading to describe the meaning of esti (absolutely or on a certain occasion) in ancient Greek philosophy as one in which the existential and copulative senses are "fused". (Cf. here Kahn, 1966, and Furth.) The converse image is more apt. Even though in the basic meaning of esti we cannot tell the allegedly different Fregean senses apart, contextual factors may on occasion have the effect of almost separating from each other the different Frege - Russell forces and nearly eliminating all but one of them. 4. INSTANTIATION IN NATURAL LANGUAGES: A SYSTEMATIC VIEW

Suffice this as an indication of one line of thought opened by our observations. To return to the main theme of this paper, notice that from the absence of the Frege - Russell ambiguity it does not follow that there might not be other ambiguities about esti, over and above the nonambiguous differences in use which Frege and Russell mistakenly raised to the level of ambiguities. Further light can be thrown on this question, too, by means of recent topical insights. There is in fact another major way in which recent logical and seman tical work on the concept of being puts Aristotle in an interesting new perspective. In order to see what it is, we have to stray temporarily away from Aristotle and discuss certain topical problems in the logic of natural languages. I shall discuss them in the case of English, even though similar things can be said of other languages, including ancient Greel(. These problems are as close to t.1e heart of all Sprachlogik as we can hope to get. Any logician knows that the lifeblood of virtually all interesting logical techniques in that basic part of logic which is variously known as first-order logic, quantification theory, or lower predicate calculus are the rules of instantiation (i.e., rules for substituting names or name-like terms for quantified variables). Now suppose w~ want to deal with the logic of natural languages directly, without first attempting the dubious and by this time largely discredited translation to formal languages. Then our first task is to formulate. likewise directly for natural languages, rules of instantiation far the quantifier phrases which take over the role of quantified variables in natural languages. How can we do that? How are we to deal with, say, a quantifier phrase like "every

THE VARIETIES OF BEING IN ARISTOTLE

97

white horse which Alexander rode" or "some small town where Socrates lived", occurring in a context X - W? (We take here the general form of these quanti fer phrases to be every } some

Y + wh-word + Z

where Z contains a "trace" to indicate where the wh-word was "moved away from".) Now the obvious way of formulating instantiation rules for such phrases is to legitimize a move from the sentence in which they occur to sentences like (I)

X - b - W if b is a white horse and Alexander rode b

or, respectively, (2)

X - d - Wand d is a small town and

Socrates lived in d, where" b" and" d" are the respective instantiating terms. In general, the output of an instantiation step is of the form if (3)

X - b - W

} b is a Y and Z' and

where b is the instantiating term and Z' is like Z except that the trace has been replaced by "b" with the appropriate preposition. (We have been assuming that Y and Z are here singular.) The details need not detain us here. What is of interest to us here is an important difference between the situation in formal first-order languages and natural languages. In the former, a single domain of values for the substituting terms (e .g. my "b" and "d") is given. In the latter, the entities referred to by the substitution-values have to be chosen from different subdomains in different cases. For instance, in (I) b has to be a living creature, whereas in (2) d has to be a location in space. It lies close at hand for a logician to say that the only novelty here is that natural languages employ many-sorted quantification theory (more generally, many-sorted logic). And this need not by itself introduce any complications (contrary to what is, e.g., implied in Moravcsik, 1976). In-

98

JAAKKO HINTIKKA

deed, many-sorted logics do not involve any serious new difficulties over and above one-sorted ones. Yet there is a new question present here. In many-sorted formal logics, the sortal differences are indicated by notational conventions. How are these differences marked in natural languages? How can one tell what subdomain b or d must belong to? Some clues are obvious, and the most obvious is the relative pronoun which disappears in the process of instantiation. (These relative pronouns can be taken to be question words in a new role, except that "what" is replaced by "that".) If the operative word is "who", the relevant subdomain consists of persons, if "where", of locations in space, if "when", of moments (and/or periods) of time, etc. Further subdomains are introduced by prepositional phrases containing similar words, for instance "like which" introducing a realm of qualities ("some color like which you have never seen"). Clearly there is not a sharp one-to-one correspondence between the ranges of natural-language quantifiers (my "subdomains") and different relative pronouns (or other wh-words, with or without prepositions or similar qualifiers), but a rough-andready correspondence certainly obtains. The main discrepancy here is the fact that "what" covers several different subdomains. By asking, "what is X?", we can mean at least three different things, to wit: (i) (ii) (iii)

Which partic4lar entity is X? What kind of entity is X? What does X consist of? (What is it made of?)

as illustrated by the following sample questions: (i) (ii) (iii)

What is Sirius? What is a gnu? What is cordite?

Similar things can be said of Aristotle's Greek word esti: This ambiguity is made especially important in Aristotle's case by the absence of the Frege- Russell distinction. For in terms of this distinction one could distinguish the different what-questions (i) - (iii) from each other by the different sense of "is" involved in them. But this distinction just is not available to Aristotle. In any case, the relative pronoun (or the corresponding question word) cannot be the only clue to the choice of the subdomain. For one thing,

THE VARIETIES OF BEING IN ARISTOTLE

99

the whole relative clause can be missing from the quantifier phrase in question, and hence be unavailable to supply any leads. Hence it is the meaning of Yin (3) which must supply the main in formation as to which subdomain (sort) we are dealing with. Presumably we must assume some kind of semantical categorization of the terms (phrases) that can serve as the Y in (3). In the case of simple terms these must be part of their lexical meaning. Since the Y's in (3) are basically predicate terms, we end up in this way postulating a classification of all simple predicates of English into certain equivalence classes. These classes wi\l be correlated one-to-one with those subdomains of quantification, which we are dealing with, when using quantifiers in English, i.e., the largest classes of entities we can quantify over, and also correlated in a loose way with certain wh-words and phrases. The need of relying on Y for our choice of the subdomain is vividly seen from the fact that if we try to eliminate Y (in the way in which we could dispense with the relative clause), we would end up with an ungrammatical expression. In order to preserve grammatically, we must amplify the quantifier word itself so as to make it capable of conveying the crucial information. For instance, some becomes someone, something, somewhere, sometime, somehow, etc., where the added handle serves to betray the relevant sort (subdomain). Furthermore, since each instantiation step (witness (3» introduces an occurrence of "is", these correlated classifications are likewise correlated with a distinction between different uses of "is", viz., those that could have originated from an application of the instantiation rules, plus of course those that are logically on par with them. Thus we are led to recognize four correlated multiple distinctions. They distinguish from each other (i) (ii) (iii) (iv)

Certain wh-words (and phrases with wh-words). Different kinds of simple predicates. The largest classes of entities we have to recognize in the logic of our language as domains of quantification. Certain semantically different uses of "is". (In them we of course cannot distinguish from each other the ises of identity, existence, and predication.) 5. WHAT DO ARISTOTELIAN CATEGORIES CATEGORIZE?

At this point you are supposed to have a deja VII experience. For what I have arrived at by means of purely systematic (logical and semantical)

100

JAAKKO HINTIKKA

considerations is to aJl practical purposes tantamount to Aristotle's theory of categories. One of the most fundamental and most perplexing questions concerning Aristotle's distinction between different categories is: What is being distinguished from each other? What is Aristotle classifying in separating the different categories from each other? He uses different Greek question words or question phrases (7£ fait, 7roaov, 7rOLOV, 7r(!OS it, 7rOV, 7rOT€) as names for six of the categories, and the other labels likewise go naturaJly with certain types of questions in Greek. This is pretty much how he presents his categories in Top. 1,9. He envisages different kinds of entities "put before one" and classifies the different things that can be said (and by implication asked) of it. But when Aristotle introduces his categories in Cat. 4, they appear as classes of simple predicates or "things that can.be said" of an entity. Which are they? The plot is thickened further by Aristotle's deeply ingrained habit of considering categories as the widest genera of entities that can be logically considered together. This is seen for instance in Met. r 1, 1003 b 19ft" or Post. An. A 22,83 b 10-17. Furthermore, Aristotle repeatedly indicates that the distinction between the different categories goes together with a correlated distinction between different uses of esti. What is more, occasionaJly he seems to run the two distinctions together. For instance in Met. Z 1, 1028 a lOff "that which is" is said to signify the different categories. See also Met. Ll 7, 1017 a 23 - 30. Scholars have debated intensively which of these different things Aristotle "reaJly" meant. For example, one persuasion maintains that the categories represent the different kinds of questions one can (according to Aristotle) ask of a given entity. This view is in different variants held by among others Ockham, Charles Kahn, Benveniste, and Ackrill. Other scholars hold that Aristotelian categories are what he says they are, predicables. Others, led by the formidable Hermann Bonitz, have have held that categories were for Aristotle first and foremost the widest genera of entities. "Sie bezeichnen die oberst en Geschlechter, deren einem jedes Seiende sich muss unterordnen lassen", he proclaims (p. 623 of the original). Still others have held that Aristotle's category distinction is primarily a differentiation between several senses of esti, a reminder of the "systematic ambiguity" of words for being in Aristotle. This view is found, e.g., in Phys. A 2, 185 b 25 - 32. Among commentators, it has been represented by Heinrich Maier, and in a sense it can be maintained

THE VARIETIES OF BEING IN ARISTOTLE

101

that G. E. L. Owen is another case in point. He has certainly been followed by a host of younger scholars. If we had not seen that Aristotle is completely free from the Frege - Russell ambiguity assumption, we might also be puzzled by the fact that the distinction between the different uses of esti in the different categories sometimes appears prima facie as a distinction between different kinds of existential is (cf., e.g., the Topics discussion whether 'TO Oil is a genus), sometimes as a distinction between different kinds of predicative is (cf., e.g., Pro An. A 37, 49a 6 - 9, read in conjunction with the preceding chapter), and sometimes as one between different kinds of identity. These different emphases in Aristotle have found their fans. For instance, as Ross reports, "ApeIt regards the categories as primarily a classification of the meaning of the copula 'is' " whereas Bonitz stresses the existential and identity senses. Some of the shrewder scholars have responded to this problem situation by suggesting that Aristotle was led to his distinction between the different cate~ories by several convergent routes. For instance, AckriII suggests that in Aristotle's classification there are two elements, first the idea that different kinds of questions will have "categorically" different answers, and second the idea of categories as the highest genera. This is undoubtedly a step in the right direction. However, philosophers taking such a line will nevertheless face the almost equally perplexing problem as to why the different distinctions Aristotle had in mind should coincide - or at least why Aristotle should have thought that they coincide. By and large, they have not solved this problem. Ackrill says merely that, "It is not surprising that these two ways of grouping things together should produce the same results". This opinion simply will not stand up to scrutiny when viewed in the cold light of contemporary analyses of questions and answers. (Cf. Hintikka, 1976.) Contrary to what Ackrill suggests, it is not at all clear that answers to different questions fall into mutually exclusive classes which correspond to the widest classes of entities. For instance, it is perfectly legitimate to reply to the question "Who is the head of the Academy?" not only by saying "Plato" or "a man", but alternatively "a pale man", "the youngest brother of Potone", or even "he is lying there", alI of which have to be pigeonholed in different categories. Only by means of a further analysis can one perhaps hope to eliminate some of these replies as amounting only to partial answers (or as supplying collateral information to back up the conclusiveness of an answer). Worse still, AckrilI's account is intrinsically inconsistent. For if the appropriate answers to different questions belong

102

JAAKKO HINTIKKA

to different categories, it is impossible to construe Aristotle's categories as answers to one and the same question "What is it?", as Ackrill also suggests. Even if what he says can somehow be salvaged in the last analysis, it does not help us to understand what Aristotle's categories really were in the least. Prima facie, it is far from obvious that the four correlated distinctions we find in Aristotle should go together, and Aristotelian scholars certainly have not supplied valid reasons why they should do so. Aristotle seems to be aware of the objection and tries to handle it, not so much by explicit reference to questions as in terms of definition and sameness. In Post. An. A 22 he argues that to define an entity means to specify what substance it is and in Top. I 8 he argues that strict (numerical) sameness means identity of substance. These explanations presuppose Aristotelian distinction, however, and therefore cannot serve as an independent theoretical motivation for it. A systematic theory of questions and answers thus cannot in itself serve to explain the nature of Aristotle's theory of categories, in the way some scholars seem to expect. For the same reason, I shall not discuss the details of these two chapters here. 6. ARISTOTLE RECONSTRUCTED

Now the brief analysis of the conditions of instantiation which I carried out above, puts both Aristotle's theory and discussions thereof into a new perspective. Led by purely topical (logical and semantical) arguments, we have arrived at a remarkable reconstruction of Aristotle's theory of categories. (My arguments have an even stronger theoretical motivation than I have spelled out here, for they ensue from the basic ideas of the highly successful approach to language analysis which I have called game-theoretical semantics (see Hintikka, 1982, and Saarinen, 1979). We can now recognize all the apparently discrepant ingredients of Aristotle's doctrine in the systematic situation revealed by my analysis. Aristotle's use of question words and phrases as labels for categories matches my use of wh-words as a guide to the subdomain involved in an instantiation. His view of categories as the different kinds of simple things that can be said of an entity matches my classification of the meanings of simple predicates as guides to the logical "sort" intended. His use of categories as the largest classes of logically comparable entities amounts to the focal point of my quest of the different largest domains of quantification presupposed in a natural language,

THE VARIETIES OF BEING IN ARISTOTLE

103

and Aristotle's correlation of different uses of the word est; corresponds to the automatic alignment in my treatment of the other distinctions with certain differences in the use of the word "is". What is more important, the correlation of these several distinctions is seen not to be accidental or artificial. Its reasons lie deep in the logic of the situation. Charles Kahn has suggested that the different Aristotelian distinctions represent different strata in Aristotle's thinking. That may very well be so, but we don't understand Aristotle unless we also recognize the intrinsic logical connections between the different correlated classifications of his. No longer does it make any sense to ask which of the several distinctions Aristotle "really" means, for they are all inextricably intertwined. The extensive controversies that have been prompted by this question are simply otiose. (This does not mean that differences of emphasis are not called for here; cf. my comments below on those who stress the ties between categories and question types.) The interesting questions pertain instead (inter alia) to Aristotle's awareness of the connecting links between the different dinstinctions. Indeed, it is in spelling out the main interrelations between the distinctions which converge in Aristotle's theory of categories that my "transcendental deduction of the categories" goes essentially beyond those earlier scholars who have emphasized the multi-faceted character of Aristotelian categories. Even though the reconstruction of Aristotelian categories which we have just reached perhaps does not ipso facto solve any major interpretational problems, it yields valuable clues which help to understand Aristotle and in mar:ty cases even promise further insights. For instance, one problem we can now approach pertains to the relation of Aristotle's theory to the facts of the Greek language. Trendelenburg, Apelt, and Benveniste have claimed that the Aristotelian distinction between different categories reflects certain general features of the ancient Greek language. Ackrill's persuasive arguments to the effect that what is distinguished from each other in the category distinctions are not verbal expressions but entities may serve as an antidote to such excesses. However, Ackrill's thesis does not imply that Aristotle was not guided by logical structures which manifest themselves in the grammar of the Greek language. I cannot try to write either a transformational grammar or a game-theoretical semantics for the ancient Greek language. Suffice it merely to point out that the grammatical facts which are highlighted by my treatment are less eye-catching but subtler than those flaunted by Trendelenburg and Benveniste. They

104

JAAKKO HINTIKKA

pertain to such things as the identity (in form) of indefinite relatives with indirect interrogatives in Greek, and the close relationship of both with quantifier words. These features of the Greek grammar serve to link the different correlated distinctions explained above to each other especially closely, and thereby to motivate Aristotle's theory. If I had to find linguistic evidence for my interpretation of Aristotle, that is the direction in which I could (and would) go. Even on the present superficial level, it is not hard to see that my treatment of instantiation works mutatis mutandis even better with Greek thari with English. Likewise, we are now in a position to draw an interesting conclusion from our observations. The different classes of questions with which Aristotle correlated his other distinctions were primarily indirect questions. The correlation depends crucially on an analogy between relative pronouns and question words, and this analogy (or near identity) can obviously be best argued for by comparing with each other the logical behavior of relative clauses and indirect questions. (An especially useful Mittelglied here is the class of relative clauses without antecedents. Their logic is remarkably similar to that of indirect questions.) Aristotle's distinction between different categories is less a distinction between different question types as between question words, and it pertains to these words in so far as they are doing duty of their relative clause twins. This observation reflects somewhat unfavorably on those scholars ·who have made much of the classification of questions as the alleged cornerstone of Aristotelian categories. It seems to me that their thesis remains unproven. Admittedly, the importance of the dialectical questioning games practiced in the Academy for Aristotle can scarcely be exaggerated. However, there is little evidence in the Topics or elsewhere that the theory of categories was developed for (or from) such games. 7. CONCLUSIONS FROM THE RECONSTRUCTED ARISTOTELIAN THEORY. CATEGORIES VS LOGICAL TYPES

One respect in which my reconstruction matches Aristotle's ideas is what might first have seemed a blemish in it. It is the ambiguity of whatquestions (and of the corresponding mUltiplicity of that-clauses in quantifier phrases) registered above. In my reconstruction certain whatquestions correspond to the category of substance. Aristotle, too, relates the category of substance to suitable what-questions. As Aristotle puts in Met. Z I, 1028 b 2, the old question, "What is that w!1ich is?" really amounts to "What is substance?" Furthermore, and importantly,

THE VARIETIES OF BEING IN ARISTOTLE

105

Aristotle's lists candidates for the status of substance in Met. Z, both in Met. Z 3 and in Met. Z 13. These lists contain matter, form, and the individual consisting of both. These three correspond to the three senses of what-questions listed in Section 4 above (in the reverse order). Moreover, the fourth main candidate, the universal, can perhaps be considered (as Russell Dancy has suggested) a hangover from the earlier academic practice of answering what-questions by reference to a genus ("the universal") and its differentiae. (This view is certainly supported strongly by such passages as Met . .::l 28, 1024 b 4 - 6, Top. A 9, 103 b 36 - 37 and A 18, 108 b 22 - 23.) We also have to raise one of the most crucial questions concerning Aristotle's theory of categories. Is the theory correct as an analysis of the "logic" of the Greek language (or of the English language)? Are there differences between different languages vis-a.-vis Aristotelian categories? The "transcendental deduction" of Aristotelian categories presented above might seem to vindicate the main features of Aristotle's theory. In spite of its persuasiveness, it nevertheless gives us only an approximation to the true seman tical theory of natural language categories. It is based upon assumptions which are only partly true, and hence it cannot be taken as the final word on the subject. In the next section, I shall indicate one specific limitation of my argument and consequently of our reconstructed theory of Aristotelian categories. On a general theoretical level, another major shortcoming of the theory is obvious, connected with its relation to logical type distinctions. It is in a wider logical and philosophical perspective clear that even Aristotelian category distinctions must in the last analysis be based on type distinction. Th(' latter distinctions may not coincide with Russell's. Indeed, the types (categories) of Frege and Russell seem to me too few and too far apart to serve as a realistic basis of our Sprachlogik. But, whatever the requisite types are, they must serve as the foundation of any viable distinction between different categories. In other words, some bridge has to be constructed from Aristotelian categories to logical ones to vindicate them. How foreign modern type distinctions were for Aristotle is also illustrated by his deeply ingrained habit of bracketing together the obtaining of (what we would call) facts and the existence of individuals. (This habit was noted above in Section 2.) This assimilation offers us in fact an additional illustration of the absence in Aristotle of any real distinction between the is of predication, which expresses the obtaining of facts and the is of existence. This general problem is highlighted by the more specific observation

106

JAAKKO HINTIKKA

that Aristotelian categories turn out on my analysis to be quite different from logical categories in the sense of logical types . (This point is relevant here among other reasons because the contraty has been maintained by Gilbert Ryle; cf. Ryle 1937 - 1938.) Not only is it the case that entities of a different logical type (in what is roughly Russell's sense) belong to the same category, as Socrates, man, and animal all belong to the socalled category of substance. There is a sense in which all categories come close to containing entities of the same logical type. After all, they all contain items which can be said of a substance like Socrates. For instance, the members of the so-called category of relation are not relations for Aristotle, but relatives (relational predicates). This is amply shown by his discussion of this category in Cat. 7, 8 a 35ff, especially his comments on correlatives and their epistemic and bntological interdependence. (Cf. Cat. 6,6 a 35ff; De Soph. EI. 31, 181 b 26 - 28; Top. VI, 4, 142 a 28 - 31; and VI, 8, 146 b 3 - 4.) Likewise, quantities are not for Aristotle what we would think of as quantities (e.g. a certain length), but quantitative attributes (e.g. being of such-and-such length). These observations are perhaps not very surprising. There is a sense ill which the very "category" of relation (as distinguished from relational predicates) came to its own only much later in the history of philosophy. (Cf. Weinberg 1965.) However, the absence of relations proper from Aristotle's categorical scheme highlights the problems it leads into. For where else can he put relations? The only propositional form he seems to recognize is the subject-predicate one. If some of those predicates are relational, we need an account as to how some of them can be built of relations. Alternatively, we need a reduction of relational propositions to subject-predicate propositions. Neither task was attempted by Aristotle, although the latter one was undertaken by Leibniz, whose philosophy is in the last analysis much more Aristotelian than is usually recognized. (Cf. here Hintikka 1972b.) Another instance of the same unsatisfactory state of Aristotle's theory is the discrepancy between Aristotle's treatment of time in the Categories and in the Physics. Treating time as a category is simply inadequate for a satisfactory categorial analysis of time. Merely including temporal predicates as a class of predicables says next to nothing about the true "categorial" structure of time. No wonder, therefore, that Aristotle offers an entirely different (and deeper) analysis of time in the Physics. Similar remarks can be made about several other categories, especially about the categories of quantity, place, and action. These are illustrations of deeper and more widespread tensions in

THE VARIETIES OF BEING IN ARISTOTLE

107

Aristotle's thought. In treating (at least in its first stage) all categories on a par Aristotle (as well as my rational reconstruction of his theory) fails to give a deeper account of the rationale of category distinctions. It is for this reason especially important to realize the differences between Aristotelian categories and logical types. It is here that Aristotle's relative neglect of the Frege - Russell distinction (even as a difference in use and not just as a difference in meaning) becomes a handicap for him. Admittedly, it was claimed by Maier that Aristotle's ·theory of categories was calculated to accommodate certain distinctions between different senses of "is" (see Vol. 2, p. 291 ff). Maier's distinctions include most of the Frege - Russell ones. Indeed, Maier's first two distinctions are identificatory being vs accidentally predicatory being, p. 280, and existential vs copulative being, p. 282. No major insights are forthcoming from Maier, however, into the way Aristotle managed to combine the Frege - Russell distinction with his. doctrine of categories. For he firmly believes that. according to Aristotle, "immediate reflection on the concept of being [Maier's emphasis) ... forms the principle of division for the table of categories" (pp. 298 - 299). Maier's immediacy claim notwithstanding, Aristotle himself does not trust immediate intuition here, but discusses the relation of other categories to that of a substance. However, these arguments are either calculated to show the dependence of other categories on that of substance, or (which may come down to the same thing) to point out the role of focal meaning in relating the being of the other categories to that of substance. They do not rely on the kinds of distinctions which Maier mentions or which are likely to be made by a twentieth-century logician. Instead, they mark a slightly different point of partial contact between Aristotle's theory of categories and modern type distinctions. For the primacy of substances over members of categories came to mean for Aristotle something very much like the claim that only substances are individuals in a modern philosopher's sense. This is shown by Aristotle's frequent reference to a substance as a "this" (TC>O€) or as a "this something" (TOOf n). Another indication is given by Aristotle iq Met. Z, 3, 1029 a 27 - 29, where he says that "it is taken to be chiefly true of a substance that it is separable and a certain this". [}(~ )'ae TO xWeWTOP }(at TO TOOf n lJ7rCxexHP oO}(fi p,CxALaTa Til ova[~.) More generally, Aristotle's use of separability (TO xwew rop) as a characteristic of substance (cf. e.g., Met. Z 1, 1028 a 24 - 25) points in the same direction. In Met. Z I, 1028 a 17 - 20 he says of the members of the other categories that they "are called beings because they are, some of them, quantities

108

JAAKKO HINTIKKA

of that which in that way is [sc. of substances], some qualities, some affections, some something else" . Similar statements are found elsewhere, e.g., in Met. r 2, 1003 b 5 - 10. This passage is an especially clear indication of the fact that the Aristotelian primacy of substance is not due to any related recognition of the Frege - Russell distinction, for it is in the very same chapter that Aristotle denies (as we saw in Section 1 above) in effect most emphatically this distinction. However, Aristotle provides little by way of closer analysis of this mode of dependence of the other categories on substances. Nevertheless it seems fairly clear that Aristotle's celebrated manoeuvre of considering the differences between the uses of esti in different categories as not being homonymous but instances of focal meaning (7I'eos gv) is squarely based on the idea of treating substances as something very much like individuals in the sense of a Frege - Russell ontology. (Concerning Aristotle's attempted Aufhebung of category distinctions along these lines, see Owen (1960) and (1965).) 8. CATEGORIES, MATTER, AND FORM

We have found plenty of indication that Aristotle did not consider the doctrine of categories as being completely satisfactory in itself. On systematic grounds, too, it can be argued that the doctrine so far expounded is only an approximation to the real analysis of the relevant parts of our Sprachlogik. As a matter of historical record, in the Organon Aristotle seems to be satisfied with this approximation. However, in the Metaphysics, especially in rand Z - H, he realizes that he has to go beyond it. In order to understand how he does this, it is once again advisable to turn again to topical considerations. My discussion in Section 4 above was based on a simple analysis of quantifier phrases. If we look away from the relative clause, the structure we have presupposed is simply something like the following:

--

QUANT

I

some every each

NP ...............NP

109

THE VARIETIES OF BEING IN ARISTOTLE

This is too simple to be realistic, however. Indeed, Joan Bresnan has proposed the following more refined analysis: NP

/~

/'" QP

Q

Det

/~

some every

NUM

I

PP

P

/"'"NP

I

I

of

e:~h r;~ Here Q marks among other things quantity classifiers, such as the following:

Q

I

one (s) number part herd gallon ton

However, instead of this kind of Q we may have almost any correct noun. Examples of a quantifier phrases (sans the relative clause) to which Bresnan's analysis applies are the following: every two statues of bronze many tables of rosewood few cubes of ice. Roughly, we can thus think of the rightmost NP (under the PP) of Bresnan's analysis as specifying at least in a number of clear-cut cases,

110

JAAKKO HINTIKKA

the material of which the individuals over which the quantifier ranges as being formed by imposing the individuation principle that goes together with Bresnan's Q. The rightmost NP must in such cases involve a mass term. The whole phrase thus expresses a kind of combination of matter and a certain individuating principle. This principle is what Aristotle calls the form, whereas the last NP specifies what Aristotle would call (he matter. Thus the closer analysis of quantifier phrases just sketched is closely related to the important Aristotelian contrast between matter and form. It is seen that this refinement is our analysis of quantifier phrases goes beyond the conventional logical languages which can be traced back to Frege and Russell. For in these languages, one starts from a given class of basic individuals. The way they are constituted from more basic ingredients, such as matter and form, does not come up in them at all. Hence the logic of Bresnan's refined analysis cannot be captured by means of the usual logical languages, even when they are turned into many-sorted languages capable of incorporating the reconstruction of Aristotle's theory of categories outlined above. From what has been said-it also follows that the systematic reconstruction of Aristotelian categories presented above can only be an approximation to the truth of the matter. It is based on an oversimplified analysis of quanti fer phrases. This need not make my reconstruction of Aristotelian categories any less interesting historically, however. On the contrary, it seems that the pressures on the reconstruction due to its approximative character are very closely related to the reasons why the simple picture of categories so far adumbrated did not satisfy Aristotle, either. Thus we are beginning to see what light do these systematic observations throw on Aristotle's argumentation in the Metaphysics. First of all, we can understand the role of one of the main concepts which Aristotle did not use in the Categories but which he relies on heavily in Metaphysics. That is the concept of matter. It is one of the main novelties of the Metaphysics treatment of being and substance as compared with the Categories. Now the role of this concept is roughly the same in my systematic treatment as it is in Aristotle, for Aristotle, too, discusses how what in the earlier approximation were unanalyzable values of quantifiers must now be thought of as if they were combinations of matter and some individuality principle. This principle Aristotle labels form. Several further similarities between Aristotle and my analysis can be registered.

THE VARIETIES OF BEING IN ARISTOTLE

III

For one thing, Aristotle says that substances consist of matter and form (cf. e.g., 1029 a 27 - 30). Moreover, apparently it is only substances that do so, not members of other categories. For instance, in Met. H 4. 1044 b 8 - 9 Aristotle says that "things which exist by nature but are not substances have no matter; their substrate is their substance". [ouo' Qaa oh cj>vaH J1.fV, J1.h ouaLW Of, oux Ean TOVTOLS VA1], aAAO! TO V7rOXfLJ1.fVOV ~ oua[a.] On the systematic side, too, it is at least questionable whether the full Bresnan analysis can be found among entities other than individuals. Thus we can see how the introduction of the matter-form contrast seriously upsets the symmetry between different categories which they originally enjoyed in Aristotle's Categories. Another relevant observation is that in many instantiations of the Bresnan analysis we don't literally have to do with a clear matter-form combination. In a large number of cases, we may have as the lexical instantiation of the rightmost NP, not a mass term but the plural of a count noun. Then Q must be instantiated, not by a quantity word, but by a term which indicates a structure the entities referred to by the count noun can instantiate. Cases in point are the following: some discrete set of moments of time every large school of fish many ordered groups of numbers. We may describe these cases by saying that in them higher-order entities are thought of as being from lower-order one. In contrast, earlier we were dealing with the formation (construction) of individuals out of matter and form. It seems to me that Aristotle's notions of matter and form are calculated to cover both formation processes. It is very dubious that any unambiguous concept can bear such a burden. Hence we find here some reasons to be suspicious of Aristotle's concept of matter. We have to be very cautious, though. It might for instance seem that the more refined analysis of quantifier phrases indicated above embodies a mistaken admission to Aristotelian ways of thinking in that it in effect disregards the modern contrast between count nouns and mass terms. This distinction is linguistically much more dubious, however, than has been realized in recent literature. Perhaps this is another direction in which Aristotle is closer to the semantics of natural languages than Fregean logic. Be that as it may be, Aristotle's conception is far from unproblematic. Aristotle's problems are increased by the fact that he assimilated the form-matter contrast also to the traditional subject-predicate contrast

112

JAAKKO HINTIKKA

and to the actuality-potentiality contrast. It is doubtful that anyone concept can happily cover all these cases. Furthermore, the assimilation increases once again the distance between Aristotle's conceptual framework and that of modern (Frege - Russell) logic. However, it would require more space than I have here to follow Aristotle in these ventures. BIBLIOGRAPHY Ackrill, l. L.: 1963, Aristotle's 'Categories' and 'De Interpretatione', Clarendon Press, Oxford. Albritton, Rogers: 1957, 'Forms of Particular Substances in Aristotle's Metaphysics', Journal of Philosophy 54, 699 - 708. Apelt, 0.: 1891, Beitriige zur Geschichte der griechischen Philosophie, Leipzig. Balme, D. M.: 1980, 'Aristotle's Biology was not Essentialist', Archiv fur Geschichte der Philosophie 62, 1 - 12. Benardete, Seth: 1976 - 77, 'The Grammar of Being', Review of Metaphysics 30, 486-496. Benveniste, Emile: 1966, 'Categories de pen see et categories de langue', in Problemes de linguistique generale, Gallimard, Paris, pp. 63 - 74. Bluck, Richard S.: 1975, Plato's 'Sophist': A Commentary, Gordon C. Neal (ed.), Manchester University Press, Manchester. Bonitz, Hermann: 1853, 'Uber die Kategorien des Aristoteles', Sitzungsberichte der Kaiserlichen Akademie der Wissenschaften, Phil.-Hist. Klasse, Vol. 10, pp. 591 - 645; reprinted as a separate volume 1967 by Wissenschaftlicbe Buchgesellschaft, Darmstadt. Bresnan, loan: unpublished seminar notes, 'On the Syntax of English Quantifiers', Stanford University. Dancy, R. M.: 1975, Sense and Contradiction: A Study in Aristotle, D. Reidel, Dordrecht. Dancy, R. M.: 1975, 'On Some of Aristotle's First Thoughts about Substances', Philosophical Review 84, 338 - 373. Dancy, R. M.: 'Aristotle and Existence', this volume, pp. 49-80. Frede, Michael: 1967, Priidikation und Existenzaussage, Hypomnemata, Vol. 18, Vandenhoeck & Ruprecht, Gottingen. Furth, Montgomery: 1968, 'Elements of Eleatic Ontology', Journal of the History of Philosophy 6, 111 - 132. Gomez-Lobo, Alfonso: 1976 - 77, 'Aristotle's Hypotheses and Euclidean Postulates', Review of Metaphysics 30,430-439. Gomez-Lobo, Alfonso: 1980 - 81, 'The So-Called Question of Existence in Aristotle's An. Post. 2. 1 - 2', Review of Metaphysics 34, 71 - 89. Gomez-Lobo, Alfonso: 1981, 'Aristotle, Metaphysics H 2', Dialogos 38, 7 - 12. Graeser, Andreas: 1978, 'Aristoteles und das Problem von Substanzialitat und Sein', Freiburger Zeitschrift fur Philosophie und Theologie 25, 120 - 4 I. Hintikka, laakko: 1959, 'Aristotle and the Ambiguity of Ambiguity', Inquiry 2,137 - 151: reprinted with changes in Hintikka, Time and Necessity, pp. 1 - 26. Hintikka, laakko: 1972a, 'On the Ingredients of an Aristotelian Science', Nous 6, 55 - 69. Hintikka, laakko: 1972b, 'Leibniz on Plenitude, Relations, and the "Reign of Law" " in Harry G. Frankfurt (ed.), Leibniz: A Collection of Critical Essays, Doubleday, Garden

THE VARIETIES OF BEING IN ARISTOTLE

113

City, pp. 152 - 190; reprinted (with an important change) in Simo Knuuttila (ed.), 1981, ReJorging the Great Chain oj Being, D. Reidel, Dordrecht, pp. 259 - 286. Hintikka, laakko: 1973, Time and Necessity, Clarendon Press, Oxford. Hintikka, laakko: 1976, The Semantics oj Questions and the Questions oj Semantics, North-Holland, Amsterdam. Hintikka, laakko: 1979, '''Is'', Semaantical Games, and Semantical Relativity', Journal oj Philosophical Logic 8, 433-468. Hintikka, laakko: 1980, 'Aristotelian Axiomatics and Geometrical Axiomatics', in laakko Hintikka et al. (eds.), Theory Change, Ancient Axiomatics, and Galileo's Methodology, D. Reidel, Dordrecht, pp. 133 - 144. Hintikka, laakko: 1982, 'Game-Theoretical Semantics: Insights and Prospect', Notre Dame Journal oj Formal Logic 23, 219 - 241. Hintikka, laakko: 1983, 'Semantical Games, the Alleged Ambiguity of "Is", and Aristotelian Categories', Synthese 54, 443-468. Kahn, Charles: 1966, 'The Greek Verb "To Be" and the Concept of Being', Foundations oj Language 2, 245 - 265. Kahn, Charles: 1973, The Verb 'Be' in Ancient Greek, D. Reidel, Dordrecht. Kahn, Charles: 1973, 'On the Theory of the Verb "To Be" " in Milton K. Munitz (ed.), Logic and Ontology, New York University Press, New York, pp. 1-20. Kahn, Charles: 1976, 'Why Existence does not Emerge as a Distinct Concept in Greek Philosophy', Archiv Jur Geschichte der Philosophie 58, 323 - 334. Kirwan, Christopher: 1971, Aristotle's Metaphysics, Books r, ~, E, Clarendon Aristotle Series, Clarendon Press, Oxford. Loux, Michael 1.: 1974, translator and editor, Ockham's Theory oj Terms, Notre Dame University Press. Maier, Heinrich: 1896-1900, Die Syllogistik des Aristoteles, Vols. 1- 2, K. F. Koehler, Stuttgart; reprinted 1969 by Georg Olms, Hildesheim. Mates, Benson~ 1979, 'Identity and Predication in Plato', Phronesis 24, 211 - 229. Moravcsik, 1. M. E. (ed.): 1967, Aristotle: A Collection oj Critical Essays, Doubleday, Garden City. Moravcsik, 1. M. E.: 1967, 'Aristotle's Theory of Categories', in Moravcsik (ed.), Aristotle: A Collection oj Critical Essays, pp. 125 -145. Moravcsik, 1. M. E.: 1976, 'The Discernability of Identicals', Journal oj Philosophy 73, 587 - 598. Owen, G. E. L.: 1%0, 'Logic and Metaphysics in Some Earlier Works of Aristotle', in 1. During and G. E. L. Owen (eds.), Aristotle and Plato in Mid-Fourth Century (Studia Graeca et Latina Gothoburgensia, Vol. ii), Goteborg, pp. 163 - 190. Owen, G. E. L.: 1965, 'Aristotle in the Snares of Ontology', in R. Bambrough (ed.), New Essays on Plato and Aristotle, Routledge and Kegan Paul, London, pp; 69 - 95. Owen, G. E. L.: 1978-79, 'Particular and General', Proceedings oj the Aristotelian Society 79, I - 21. Ross, W. D.: 1924, Aristotle's Metaphysics, A Revised Text With Introduction and Commentary, Vols. 1-2, Clarendon Press, Oxford. Ryle, Gilbert: 1937 - 38, 'Categories', Proceedings oJ the Aristotelian Society, Vol. 38, pp. 189-206; reprinted in 1971, Collected Papers, Vol. 2, Hutchinson, London, pp. 170-184. Ryle, Gilbert: 1939, 'Plato's Parmenides', Mind 48, 129 - 151, 302 - 325. Saarinen, Esa (ed.): 1979, Game-Theoretical Semantics, D. Reidel, Dordrecht. Trendelenburg, Adolf: 1846, Geschichte der Kategorienlehre, Bethge, Berlin.

114

JAAKKO H!NT!KKA

Weidemann, Hermann: 1980, 'In Defense of Aristotle's Theory of Predication', Phronesis 25,76-87. Weinberg, Julius R.: 1965, 'The Concept of Relation: Some Observation on Its History', in Abstraction, Relation, and Induction: Three Essays in the History of Thought, The University of Wisconsin Press, Madison, pp. 61 - 119. Woods, M. J.: 1974-75, 'Substance and Essence in Aristotle', Proceedings 0/ the Aristotelian Society, Vol. 75, pp. 167 - 180.

Dept. of Philosophy, Florida State University, Tallahassee, FL 32306- /054, U.S. 4.

THE CHIMERA'S DIARY Edited by STEN EBBESEN

Institute of Greek and Latin Medieval Philology, University of Copenhagen

[Editor's note: This paper reproduces the manuscript left by the chimera, but I have added references to books and manuscripts, plus a few notes which appear in square brackets. The reader will notice that the chimera has wisely disregarded accidental changes of philosophers' choices of example when they need a composite animal. The chimera takes remarks about, e.g., the goat-stag as remarks aimed at itself. As a matter of fact, Aristotle and the Greek Aristotelian commentators prefer the goat-stag (TQOI-YMOIIPOS) and the centaur ({1r1rOXEJlTaIlQOS). In the Hellenistic period, the centaur, the scylla and the chimera are the standard examples. In Latin medieval texts the chimera (inherited from Manlius Boethius) is vastly more popular than any of the other composite animals.)

My feelings towards philosophers are mixed. For centuries they have used me as an experimental animal, keeping me on a minimum of being. In a way I may owe them my "life", but their experiments have weakened me so much that the end may be drawing near. If my weakness proves fatal, please inform the Centaur, Goat-Stag and Pegasus, who are my next of kin. If the philosophers kill me, I expect them to keep at least one of my relatives alive in order to continue the experiments. If we are all doomed, I would like to secure us a place in man's memory. This is why I have put together these eXtracts from my diary, recording the sufferings to which I and my tribe have been subjected. 1. LONG B.C.

"In front a lion, in the rear a serpent, in the middle a goat" - that's what I am. Zeus bless Homer for his excellent description [Iliad 6.181]. 2.400 B.C.

Philosophers exhibit an un savoury interest in the being and the nonbeing. Gorgias claims that I am not. He seems to say, though, that my claim to being is as good as anyone's because I may be thought about 115 S. Knuuttila and J. Hintikka (eds.), The Logic of Being, 115-143. © 1986 by D. Reidel Publishing Company.

116

STEN EBBESEN

('PQoP€iat?m). But he does not intend to save me that way. Only to show

that being thought about is no criterion of being, so that even if something else is and I am not, men cannot single out that something else. [Here the chimera may be wrong in detail, though hardly in substance. Sextus Empiricus, Adversus Mathematicos 7.80, mentions the chimera in a paraphrase of Gorgias' On the Non-Being (n€QL TOU p.r, OPTOS). He also mentions the Scylla. I doubt whether these examples occurred in Gorgias' text. They do not occur in our other source for that work, viz. Ps.-Aristotle's De Melisso, Xenophane, Gorgia (979a - 980b in Bekker's edition of Aristotle; new edition ill Cassin (1980) 610-643, at p. 64) = Bekker 980a).] 3.320 B.C.

This is an exciting century. Philosophers now seem to agree that I am not, but thanks to Plato [Sophist] they no longer throw all us non-beings into the waste-bin with the remark that "it shall never be established that things which are not, are" [Parmenides F. 7]. One brilliant representative of this new flexibility is Aristotle whose lectures I used to attend (invisibly, of course) until his recent decease. He once stated as a simple matter of fact that people do say "the non-being is non-being" [Metaph. 4.2 1003blO]. But he would not accept an unqualified 'the non-being is'. Thus he said that it does not follow that the non-being is because it is anobject-of-belief (oo~aaToP, I think I will henceforward render this 'opinable'), "for it is not the same to be something and to be simpliciter though the similarity of expression makes it seem so" [Sophistici Elenchi 5 167al - 6]. He also explained the failure of the inference with the words "for there is belief (or: opinion - oo~a) about it [i.e. the non-being] not because it is but because it is not" - or rather (because his use of OTt was ambiguous): "for there is belief concerning it, not [a belief] that it is, but [a belief] that it is not" [De interpretatione II 21a32 - 33. Cf. Analytica Priora 1.38 49a24 and the Greek commentators on the latter text, viz. Alexander, APr., CAG 2.1 :368 - 369; Philoponus APr. J, CAG 13.2:345; Leo Magentinus, APr. J, MS Vat. gr. 244: 229r, scholium P1'/'.] Whichever interpretation is correct, I think he wanted to say that the very point of saying 'the non-being is opinable' is to make it clear that one will not subscribe to the unqualified 'the non-being is'. In the same lecture he indirectly informed me that my friend Homer is no more, because he treated the inferences 'Homer is a poet, therefore Homer is' and 'the non-being is opinable, therefore the non-being is' as similar in structure [De interpretatione II 2Ia25sqq.].

THE CHIMERA'S DIARY

117

Other memorable remarks of Aristotle's include, "Not-man is not a name ... Let us call it an infinite name since it holds indifferently of anything, whether being or non-being". [De interpretatione 2 I6a30- 33; 'since' etc. omitted by the best mss. and in Boethius' Latin translation, but cf. Boethius, Int. £'C.'. 2a p. 62.7 Meiser: "et aequaliter dicitur vel in eo quod est vel in eo quod non est".] He made a similar remark about verbs with a prefixed 'not-' [De interpretatione 3 I6bI2-15]. Also, he said that "even the goat-stag signifies something, but not, as yet, anything true or false, unless 'is' or 'is not' is added" [De interpretatione 1 16al6 -18]; and "of that which is not, no one knows what it is, only what the account or the name signifies when I say 'goatstag', but it is impossible to know what a goat-stag is" [Analytica Posteriora 2.7 92b5 - 8; cf. Topica 4.1 121a21 - 25]. I must admit that I do not quite grasp all the implications of Aristotle's statements about the non-being and fabulous animals. I hope that future generations of men will help me understand him better. 4.50 B.C.

I have just read a beautiful poem by an Epicurean called Lucretius. He gives us monsters a physical existence, thinking that the atomic pictures thrown off by the animals of which we are composed may get mixed up so as to produce monstrous pictures which men may perceive. But he will not give us physical existence as sources of those pictures. The freedom of atoms to combine is not so great, he holds, that beings consisting of parts belonging to different natural species can arise [Lucretius, De rerum natura 2.700-717,4.722-748,5.878-924]. Others will only give us conceptual existence. It seems to be the common notion these days that there are two ways of forming concepts, (a) directly on the basis of things met with in nature; (b) through creative work on the materials gathered through direct confrontation with the things. Creative work which consists in joining things which have not been received together is called 'secondary combination' (€7rtavpt'hUL~) and is responsible for concepts of such monsters as me. [See Ebbesen (1981), I: 191.] 5. A.D. 650

I once wished for help to understand Aristotle. During the last 500 years or so there have been many Aristotelian commentators. By now they are

118

STEN EBBESEN

few. I therefore think it may be time to try to summarize what they have said about us non-beings. Some of the more recent writers have the rather crude notion that our names are non-significative in the same way that 'blityri' and 'skindapsos' are; or that 'goat-stag', 'blityri' and 'skindapsos' share the property of signifying non-existent things. [Stephanus, Int., CAG 18.3: 7.17 - 18; Elias, Intr., CAG 18.1: 3.7 - 8; Ps.-Elias, Intr., 25.8, p. 52; David, Intr., CAG 18.2: 1.16-18; Anonymus Moraux, Intr., 111.107 -109 p. 80; cf. Eustratius, APo., CAG 21.1: 95 - 96.] As if it made no difference that you can describe a goat-stag or me, but not a blityri or a skindapsos since no sort of meaning or notion is associated with those words - in fact, they are "words" coined by philosophers precisely to show that it is . possible to have an articulated string of sounds with n.9 meaning at all. The better commentators - and even the naive ones, on occasion agree that 'chimera' and 'goat-stag' are significative and not to be confused with such nonsense "words" as 'blityri'. (After all, I have a nominal definition [cf. Simplicius, Ph., CAG 9: 696.3sqq; Ps.Philoponus, APo. II, CAG 13.3: 359.26-360.9]). In their standard theory, the semantic relation has three terms. The word signifies a concept which signifies a thing [cf. Ebbesen (l981a), 1: 141ff]. In my case, they hold, the semantic relation is not satisfied as far as the third term is concerned since there is no independently subsisting thing (or: "nature") to be signified by the word via the concept; but at least there is a concept of sorts. With 'blityri' there can be no semantic relation at all, since both concept and thing are lacking. [Ammonius, Int., CAG 4.5: 29.8-9 (cf. 184-185); Elias, Cat., CAG 18.1: 129.15-17, Cf. Boethius, Int. ed 2a p. 50; Ps.-Philoponus, APo. II, CAG 13.3: 362:32 - 363.2.] I should like to know whether the commentators will allow me to be in some Aristotelian category. The general attitude seems to be negative. [See lamblichus apudPhiloponus, Cat., CAG 13.1: 9; Ammonius, Cat., CAG 4.4: 9 - 10; Dexippus, Cat., CAG 4.2: 7.20 - 24.] But perhaps Porphyry accepted me as a quasi-substance, in the sense that my name names a concept of something which is not, as if it were a substance [Porphyrius apud Simplicius, Cat., CAG 8: 11.6 - 12]. At any rate, if there is a concept corresponding to my name, it may be argued that I have a sort of conceptual existence as long as people think of me - though they may kill me by ceasing to do so - and that it is possible to know and understand (E7rLamaiJw) what I am. This knowledge and understanding (E7rtaT~/L7J, scfentia) will be as perishable

THE CHIMERA'S DIARY

119

as its object (E7rtaT7]TOII, scibile), being totally dependent on the human mind which possesses the understanding. My being known and my being become inextricably joined. Some have suggested that this may in fact be the case [Porphyrius, Cat., CAG 4.1: 121.4-15. Cf. Boethius, Cat., PL 64: 229; Simplicius, Cat., CAG 8: 191; Ammonius, Intr., CAG 4.3: 39.14-40.6]. However, this line of though makes me similar to such abstract universals as the species "man". Too similar, most commentators think. Hence thei introduce a distinction between "constructive thought" (E7riIlOteX) and "mere constructive thought" (1/tt)..~ E7riIlOta). Abstract, post-rem, universals are (concepts) derived from extra-mentally existing things, and so they "are in constructive thought", but not "merely"; whereas the concept of a chimera "is in mere constructive thought" for lack of an extramental correlate. [See David, Intr., CAG 18.2: 116-117. Cf. Elias, Intr., CAG 18.1: 49.19 - 20; Anonymus Moraux, Intr., 111.102 - 109 p. 80. Cf. also David, Intr., CAG 18.2: 46.35 - 47.1, where 1/tt)..~ E7rillOta is distinguished from rpaIlTaaia]. By this move the commentators actually assign me a baser sort of concept than real things have, and so they can deprive me of my "scibility", leaving only things associated with the finer sort of concept as possible objects of genuine knowledge and understanding (scientia). They can reaffirm the Aristotelian dictum "about that which is not, one cannot know what it is" and they can deny that I have a being which consists in being known (in the pregnant, Aristotelian sense of the word), though they concede that man's thought may give me a precarious existence which lasts as long as the thought about me. [See Ammonius, Intr., CAG 4.3: 39.14 sqq; David, Intr.,CAG 18.2: 108.25-109.5 & 114.1-6; Elias, Intr., CAG 18.1: 47.3-9; cf. Boethius, Cat., PL 64: 229.] Since the commentators allow me and my kin a being in the mind, they can also allow us some predicates, such as 'not-just', 'nameable', 'opinable'. Ammonius in one place [Int., CAG 4.5: 184.25sqq.] imagines somebody who [in imitation of Aristotle, Physics 4.1 208a30] asks, "Where is that goat-stag of which we truly predicate 'not-just'?". The answer given by Ammonius is, "In constructive thought". Predication presupposes that the subject has been conceptualized - constructed in thought - but not that it has existence before being constructed in thought. The predicates I truly have include infinite nouns and verbs, 'not-just', 'not-man', 'does-not-run', for instance [Ammonius, Int., CAG 4.5: 42, 52,184-185; Boethius, Int. ed. la, p. 60; ed. 2\ pp. 62, 69-70]. This

120

STEN EBBESEN

opens an interesting perspective. If these infinite terms can be truly predicated of me, does it not follow that some finite predicates of the same type can also be so? If I am not-just, not-man and do-not-run, doesn't this mean that I am some other substance than man, equipped with some other quality than justice, and exercising some other activity than running? I have little hope, though, that the commentators would admit as much. It looks as though the only finite predicates they will allow me are such as indicate being in thought and having a name ('opinable', 'nameable' etc.), plus, I suppose, my nominal definition ('animal composed of goat, lion and snake'). My manuscript of Boethius' second commentary on De interpretatione has a lacuna at the vital point [Boethius, Int. ed. 2 a p. 70], but I think he wants to say that in 'a/the centaur runs' running is predicated, but falsely. This indicates that he thinks that all categorical propositions with finite predicates and me for their subject are false when affirmative and true when negative. But he does not say this in so many words. Anyhow, the commentators agree that 'the chimera is not' is true. Ammonius [reported by Philoponus, APo. I, CAG 13.3: 323.12sqq.] once added that since this cannot be otherwise, to know it is to have genuine, "scientific" knowledge (€7rL(JT~p,.,.,). Some others similarly state that I neither am, nor will I possibly be [Boethius, Int. ed. 1a p. 60; Simplicius, Ph., CAG 9: 517.9 - 10; also Michael Ephesius (12th c.), Metaph., CAG 1: 573.33 - 34]. They do not say whether the impossibility of my being can be deduced from the principles of biology (essentially = Lucretius' position) or whether I am ex hypothesi a creature that cannot be. In either case, this bodes ill. If people become convinced that it is no contingent, but a necessary fact that I do not exist, my scarce ration of being may be curtailed even further. As I have mentioned, the commentators accept the truth of 'a/the chimera is opinable', but they do their best to explain away its apparent existential implication. One approach consists in saying that 'opinable' means 'being in thought, but not extra-mentally" [Boethius, Int. ed.2 a , p. 376; cf. ed. p. 166]. This interpretation of 'opinable' can be bolstered up with the following interpretation of Aristotle's remark in De interpretatione [11 21 a32 - 33]: "for there is opinion (and not scientific knowledge) about the non-being not because it is (because then there might be scientific knowledge about it) but precisely because it is not" [Boethius, II. cc.]. Secondly, they sometimes follow Plato in distinguishing between several kinds of being and not-being. The being which the proposition 'the chimera is opinable' truly asserts I have is a

r,

THE CHIMERA'S DIARY

121

vile one - viler even than being-different-from namely being-inopinion, from which it is not permissible to detach a simple "being" as it stands in an indissoluble relationship to opinion [Thus Ammonius, Int., CAG 4.5: 213; cf. Anonymus Tanin, Int. p. 98]. Thirdly, following a suggestion of Aristotle's (De interpretatione 11 21 a26 - 28], they say that in 'the chimera is opinable', 'is' is not primarily predicated of me but of the opinable, so that the sense is "there is an opinion, and that is an opinion of a chimera"; whereas 'is' is directly predicated of me in the false proposition 'the chimera is' [Boethius, Int. ed. 2 a p. 376; cf. Ammonius, Int., CAG 4.5: 184 & 211]. For good measure, rabid Platonists even deny me a concept (vorIlLCx). The Peripatetic commentator Alexander of Aphrodisias has used me in an explanation of one of Aristotle's arguments against the ideas [Alexander, Metaph., ad A 9 990b 14, CAG 1: 81 - 82; see also the text in Harlfinger (1975),26-27]. If, the argument runs, thinking (vofiv) of something requires a permanent, non-particular object - an idea -, there must also be an idea of me, since people think of me. Unfortunately, Alexander's intention is not to equip me with an idea. He uses me in much the same way as Gorgias did. Assuming that it would be silly to posit an idea of me, he concludes that one should not posit ideas at all. About a century ago the more Platonically oriented Asclepius delivered a counter-attack against Alexander, in order to save the ideas - but not mine. He pointed out that believing there to be an idea when something is intellectually grasped, with the grasp called vorWL~, does not entail believing in ideas of things which are not. For the sort of intellectual grasp people have of such things is not a VOTJal~ but a cpCivTCiaiCi [Asclepius Metaph., CAG 6.2: 74-75]. This seems to be the same sort of trick as the one used to save the abstract universals. A distinction between two sorts of intellectual grasp is introduced and the inferior one is associated with me. But I just wonder whether Asclepius' argument is not circular. How does a man know that a thought of me is no vOTJaL~ except by knowing that I have no idea? The Aristotelian may claim that he empirically knows that I do not exist in the particulars. and that therefore the concept of me has no real foundation. But can the Platoni~t have a similar knowledge of which ideas there are? Is the non-existence of an idea of me vouched for by the empirical fact that he sees no mirrorimage of such an idea in the world of coming-to-be and passing-away? Or did Asclepius have a reason for considering my non-existence a necessary fact? I just wonder.

122

STEN EBBESEN

6. A.O. 1260

After several dull centuries, the last one and a half have been exciting. I have often wondered whether I am an individual or a universal. [The same doubt has haunted the editors of CAG. Some print XLJLmQex and give the references in the index nominum; others use lower-case and put the word in their index verborum.] It is now clear that I am a universal. An endangered species, in fact. Thus one introduction to logic says, Est quoddam universale quod praedicatur de nullo actualiter, sed de pluribus secundum intellectum, ut chimaera [Anon., Logica Cum sit nostra, LM 11.2: 432. 11 - 12; cf. Anon., Logica Ut dicit, LM 11.2: 387). [There is one sort of universal which is predicated of nothing actually but of several things in thought; thus (the) chimera.)

One recent philosopher claims that the opinable is my genus, saying, cum dicitur 'chimaera est opinabilis', haec est enuntiatio, et est de universali non universaliter sumpto. Unde si dicatur 'chimaera est opinabilis', praedicatur hic genus de specie. Et sciendum quod istae differentiae 'universale', 'singulare' ut hic sumuntur non sunt differentiae entis sed rei, et non sumitur ibi res secundum quod dicitur res quae habet esse actuale, sed dicitur res omne illud quod potest apprehendi ab anima [Anon., Int., MS Oxford Bod!. Canon. misc. 403: 43rB; minor scribal errors have been tacitly corrected). [When we say 'a/the chimera is opinable', this is a statement and it is about a universal not taken universally. In fact, when we say 'a/the chimera is opinable' a genus is predicated ofa species. It should be understood, though, that the differences 'universal' and 'singular' in the sense employed here are not differences of (a/the) being but of (a/the) thing; and 'thing' is not used in the sense in which a thing is so called because it has actual being, but 'thing' is used to cover everything which may be grasped by the mind.)

Like most people these days he holds that any proposition of the form 'A is (a) B' may be expanded ("expounded", they say) so as to yield "A is (a) B being", in which 'B' modifies ("restricts") 'being' in the same way that 'rational mortal' restricts 'animal' in 'rational mortal animal' [Gp. cit. 48rA, quoted in lewry (1978),128. lewry mistakenly gives the reference as 47vB]. The implication is that 'a/the chimera is opinable' may be expounded as 'a/the chi mara is (an) opinable being'. This philosopher is in express opposition to another group which tried to get rid of the being that appears to be assigned me in that proposition by distinguishing between two senses of 'is'. According to them, 'is}' = "'is' secundum adiacens" is short for 'is 2 being'; whereas 'is 2 ' = "'is' tertium adiacens" is a mere copula, i.e., solely a sign of composition of subject and predicate. Their 'is 2 ' cannot be subjected to "exposition", i.e. it cannot be considered shor.t-hand for 'is being' or any other phrase. Consequently, the move from "the chimera is opinable' to "he chimera

THE CHIMERA'S DIARY

123

is' would be a move from 'is 2 ' (the "substantive 'is' ", as some say) to 'is l ' (which they call "adjective"); and so the move is illegitimate as it hinges on the wrong assumption that 'is' has one sense only. [An excellent collection of 13th-century texts bearing on this question (as well as the others dealt with by the chimera in the present entry) is found in Lewry (1978) and (1981a - b). The interpretation of est tertium adiacens as a mere copula is at least as old as Abelard (see, e.g., his Dialectica, pp. 135, 162, 164; cf. De Rijk (1981a- b)). It is earnestly defended by Robert Bacon (early 13th c.) in his Syncategoremata; text in Braakhuis (1979), I: 131 - 135). Robert's view is criticized in an anonymous scholium on Sophistici Elenchi 166b37, in M. Osterr. Nat.bib!., lat. 166: 184v: Item quaeri potest de paralogismis quos ponit Aristoteles. Sunt autem huiusmodi: 'quod est opinabile, est; sed quod non est, est opinabile; ergo quod non est, est'; eodem modo: 'chimaera est opinabilis, ergo chimaera est'; alius paralogism us talis: 'hoc non est homo, ergo non est'. Videtur quod in primo paralogismo sit aequivocatio, quoniam proceditur ab hoc verbo 'est' secundum quod est substantivum ad idem secundum quod adiectivum. Dicitur 'est' dupliciter sumi: quandoque scilicet est adiectivum, quando[que] simpliciter praedicatur et non ponitur in numero, ut 'Socrates est'; quandoque est substantivum, ut quando ponitur in numerum, ut 'Socrates est homo'. Fit ergo processus in praedicto paralogismo secundum diversas acceptiones huius verbi 'est', scilicet in una est secundum quod est substantivum, in altera secundum quod adiectivum; et ita erit ibi aequivocatio vel nulla est praedicta distinctio. [Further, questions may be raised concerning the paralogisms in Aristotle's text. They are as follows: 'what is opinable, is; but what is not, is opinable; therefore what is not, is'; in the same manner: 'althe chimera is opinable, therefore althe chimera is'; another paralogism goes like this: 'this is not a man, therefore this is not'. It is arguable that in the first paralogism we have to do with equivocation because there is a move from the verb 'is' in its substantival function to 'is' in its adjectival function. 'Is' is presumed to be capable of two uses; sometimes it is adjectival (namely when it is precticatcd absolutely and does not enter in a count, as in 'Socrates is'); sometimes it is substantival (namely when it does enter in a count, as in 'Socrates is a man '). So, in the above paralogism the move from premisses to conclusion is accompanied by a shift in the interprelation of the verb 'is', which in one proposition is laken in its substantival funclion, in another in the adjectival one; this, then, must be a case of equivocal ion - or else the said dislinclion is void.]

This question is solved as follows: Ad hoc quod obicilur de primo paralogismo dicendum quod non eSI ibi aequi\ ocalio, sed processus secundum quid el simpliciler, quoniam 'esl' secundum sC pracdical esse simpliciler el secundum aClum, haec aulcm determinatio 'opinabik' Irahil ipsum ad nominandum esse quodam modo, scilicet esse secundum opinionem; unde proccdilu ibi ab esse quodam modo, scilicel secundum opinionem, ad esse simpliciler el in aL·IU. Unde dici POlcsl quod nihil esl dicere [ctislingucre cod.] hoc verbum 'esl' e"c aeqlliv()cum, nee quod quandoquc CSI subslanlivum, quandoque adieClivum: semper cnim esl substanli\ulTl, sive ponilur in numerum sive praedicelur simpliciICr.]

124

STEN EBBESEN

[As regards the criticism raised concerning the first paralogism, the answer must be that there is no case of equivocation but a move from in-some-respect to absolutely, because 'is' in itself predicates being absolutely and actually, whereas the determination 'opinable' makes it name being in some way - being in opinion, that is. Hence a move is performed from being in some way - in opinion - to being absolutely and actually. Thus it may be said that there is no foundation for claiming that the verb 'is' is equivocal, nor for claiming that it is sometimes substantival, sometimes adjectival. For it is always substantival, whether it enters into a count or is predicated absolutely. 1

The distinction between the predicative and the existential 'is' has met with general disapproval, at least from the time of Robert Kilwardby (c. 1240). He flatly denies that 'is' is equivocal, and holds that the right way to expound both 'is l ' and 'is 2 ' is "is being" [see text in Lewry (1978), 128]. This exposition brings out that a predicate has both matter and form. The fo!'m is the means relating the matter to..the subject. When 'is' is secundum adiacens, as in 'a/the man is', the matter of the predicate is being simpliciter. If 'is' is tertium adiacens, the matter is a specified sort of being - substantial in '(a/the) man is (an) animal', accidental in '(a/the) man is just'. In his commentary on De interpretatione, Kilwardby says: Dubitatur postea si praedicetur tertium, propter hoc quod in omni enuntiatione est medium hoc verbum 'est' et subiectum et praedicatum extrema, cum praedicatur tertium adiacens; et nihil unum et idem potest esse medium et extremum; ex quo sequitur quod non praedicatur tertium. § Sed intellege quod uno modo extremum, alio modo medium: ratione compositionis medium, ratione substantiae sive rei verbi extremum. Est enim hoc ipsum 'est' praedicatum secundum materiam et formam; et dico formam praedicati medium per quod comparatur praedicatum [medium) subiecto, et praedicatum secundum materiam dir:o rem verbi; et ex hiis fieri unum praedicatum, ut cum dico 'homo est' id est 'homo est ens', et sic praedicatur hoc ipsum 'est' secundum adiacens. Et quam vis sit copula aliquo modo tertium (scilicet non ordine sed numero), quia iIIud cui adiacet (scilicet praedicatum secundum materiam) non ponit in numero cum eo, non dicitur hoc ipsum 'est' esse tertium adiacens, sed secundum, cum subiectum sit quod adiacet et praedicatum est quod adiacet; et quia in hac 'homo est iustus' et 'homo est animal' praedicatur esse specificatum (scilicet per substantiale et accidentale), quod quidem ponit in numerum cum esse simpliciter, ideo dicitur hoc verbum 'est' in talibus praedicari tertium adiacens, ita scilicet quod sit tertium numero, non ordine. Sic ergo aliquando praedicatur hoc verbum 'est' tertium adiacens. [MS Cambridge, Peterhouse, 206: 7?rB.) [Then doubt is raised whether it is predicated third. For in every statement the verb 'is' is the mean and the subject and the predicate the extremes, when it is predicated as a third supplement; and no one thing can be both mean and extreme. It follows that it is not predicated third. § But it should be realized that it is an extreme in one way, a mean in another way: a mean as far as the composition is concerned, an extreme as far as the substance or content of the verb is concerned. For this very 'is' is a predicate in respect of matter and form. By a predicate's form I understand the mean through which the predicate is related to the subject. By predicate in respect of matter I understand the verbal content. Now, my claim is that these two together constitute one predicate, as when I say '(a/the) man is' = '(a/the) man is being', and thus 'is' is predicated as a second supplem~nt. And

THE CHIMERA'S DIARY

125

although the copula is in a way a third - not in order, that is, but in number -, as that to which it is a supplement - viz. the predicate in respect of matter - does not add to the count together with it, 'is' is not said to be a third supplement but a second one, since the subject is one supplement and the predicate is another supplement. And since in the propositions '(a/the) man is just' and '(a/the) man is an animal' the being which is predicated is specified (as substantial or accidental), and this adds to the count together with being absolutely, for this reason the verb 'is' is said to be predicated as a third supplement in such propositions, i.e., in such a way that it is third in number, not in order. Thus, then, 'is' is sometimes predicated as a third supplement.)

There can be no doubt that for Kilwardby the matter of the predicate in 'a/the chimera is opinable' is opinable being. In fact, there has been quite a debate about the status of opinable being vis-a-vis simple being. Is the one somehow included in the other? If opinable being were a subjective part (species) of being, it would be permissible to argue, 'this is opinable, ergo this is'. The argument would be as good as 'Socrates is a man, ergo Socrates is an animal'. But since the substitution of 'the chimera' for 'this' would render the antecedent true and the consequent false, people deny that opinable being is a species of simple being. But then it might seem that opinable being is a wider term than being since both the (actually) being and the non-being is opinable. [So already Alexander Aphrodisiensis, Topica, CAG 2.2: 359.17 -18 ad Top. 4.6 127a34sqq.). This view would imply that 'this is, ergo this is opinable' is a sound inference. But then the converse inference, 'this is opinable, ergo this is' would commit the fallacy of consequent rather than the fallacy secundum quid et simpliciter which Aristotle says it commits. Kilwardby, and many others, solve this difficulty by distinguishing between a proper and an improper sense of 'opinable'. In the proper sense, they hold, it means "in opinion only, not really". When 'opinable' is taken in this sense, the actually being does not fall under the opinable, and so there is a fallacy secundum quid et simpliciter, but not a fallacy of the consequent in 'this is opinable, ergo this is'. However, in an improper and general sense 'opinable (being)' does comprise actual being as well as non-being or imaginary being. When 'opinable' is taken in this sense, there is a fallacy of the consequent in 'this is opinable, ergo this is' - but then, did not Aristotle himself say that one and the same argument may be fallacious for more than one reason? [For an early (12th century) treatment of the above que~tion, see Anonymus Cantabrigiensis, Commentarium in Sophisticos Elenchos, MS Cambridge St John's College D.12: 90rB - vA (on Arist. SE 166b37). The following texts are all doctrinally close to Kilwardby's: Anonymus Monacensis, Commenlarium in Sophislicos Elenchos, MSS Admont

126

STEN EBBESEN

241: 28rA - B & CLM 14246: 11 vA (presumably a little earlier than Kilwardby); Albertus Magnus, Expos. SE 1.3.7, ed. Borgnet 2: 570 (directly influenced by Kilwardby, as shown in Ebbesen (1981 b»; Anon., Scholium in Arist. SE 166b37, MS Cambridge Gonville and Caius 466/573: 137v; Anonymus e Musaeo, Commentarium in Sophisticos Elenchos, MS Oxford Bodleian E Musaeo 133: 3rA. Kilwardby's own text (Commentarium in Sophisticos Elenchos, MSS Cambridge Peterhouse 205:291rB-vA & Paris BN lat. 16619: 18rB-vA) runs like this (discrepancies between the mss. are few and not noted here): Adhuc dubitatur de primo paralogismo quem ponit in littera (SE 167al), in quo dicit esse fallaciam secundum quid et simpliciter. Videtur enim quod ibi sit magis fallacia consequentis quam secundum quid et simpliciter cum dicit 'est opinabile, ergo est', hac ratione: quandocumque ali quod argumentum sic se habet quod sequitur econverso et non sic, ibi est fallada consequentis; sic autem est in praedicto paralogismo; ergo in ipso est fallacia consequentis. Maior huius ration is patel. Minor etiam manifesta est, cum omne quod est potest cad ere in opinione et non econverso. Penitus eodem modo opponitur de secundo paralogismo. Ibi enim est fallacia consequentis a destructione antecedentis cum dicit 'non est homo, ergo non est'; econverso enim sequitur, sic: 'non est, ergo non est homo' .... Ad primum dicendum quod, sicut dicit Aristoteles in secundo huius, nihil impedit plures occa· siones fallendi esse in una eademque oratione. Unde dico quod in prima oratione, si sumatur 'opinabile' communiter, est fallacia consequentis. Uno modo opinabile enim communiter est omne illud quod cadit in opinione, et hoc modo non sequitur 'est opinabile, ergo est', immo tenet econverso. Sumpto ergo opinabili communiter est in hoc paralogismo tam consequens quam secundum quid et simpliciter. Ipso autem sumpto proprie non est ibi fallacia consequentis: tunc enim non sequiter econverso, sic: 'est, ergo est solum in opinione'. Hoc enim est proprie opinabile quod est solum in opinione. Un de dico quod Iicet in hoc paralogismo sit tam fallacia consequentis quam secundum quid et simpliciter, est tamen hic proprius secundum quid et simpliciter, et dico "proprius" quia secundum quid proprie sumitur opinabile. In alio autem paralogismo, cum dicitur 'asinus non est homo, ergo non est', est fallacia consequentis et similiter fallacia secundum quid et simpliciter. Inducit tamen ipsum propter peccatum secundum quid et simpliciter. Dicunt tam en quidam quod non est ibi fallacia consequentis. Ponunt enim quod ista duo possunt simul stare 'est homo' et 'non est', et dicunt quod Caesar est homo et cum non est. Et cum arguitur contra eos sic: 'est homo, ergo est', dicunt quod hoc argumentum non valet, et hac ratione: omne praedicatum essentiale et universale dicit esse habitudinis; quandocumque autem praedicatur hoc verbum 'est' secundum adiacens, dicit esse temporis sive esse ut nunc. Unde dicunt quod cum arguitur sic 'Caesar est homo, ergo Caesar est' fit fallacia aequivocationis, quia proceditur ab esse habitudinis ad esse temporis vel ad esse ut nunc. Utrum autem dicant verum vel falsum non magnam habet dubitationem. "j [Next, doubt is raised concerning the first paralogism which he gives in the text claiming that it commits the fallacy of "in-some-respect-and-absolutely". For it is arguable that the fallacy involved in that of consequent rather than that of in-so me-respect· and-absolutely, and for the following reason: Whenever an argument is thus constituted that the reverse inference holds, but not the one stated in the argument, there is a case of the fallacy of con· sequent. But that is the situation with the paralogism in question. Therefore there is a fallacy of consequent in it. The major of this reasoning is evident. The minor is also clear

THE CHIMERA'S DIARY

127

since everything that is can occur in opinion, but not vice versa. Exactly the same objection is raised concerning the second paralogism. For there the fallacy of consequent from destruction of the antecedent is committed when he says 'it is not a man, therefore it is not, as it follows vice versa, in this way: 'it is not, therefore it is not a man' .... As for the first point, the response must be that, as Aristotle says in Book II of this work, nothing prevents joint occurrence of several possible reasons of fallacy in the same utterance. Accordingly, my position regarding the first utterance is that if 'opinable' is taken in a general sense, there is the fallacy of consequent. For in one way the opinable is, in a general sense, all that which can occur in opinion; and in that way 'it is opinable, therefore it is' is a non sequitur, whereas it does hold vice versa. So when the opinable is taken in the general sense, there is in this paralogism as well a fallacy of consequent as in-some-respect-andabsolutely. But when the opinable is taken in its proper sense, no fallacy of consequent occurs. For then it does not follow vice versa; thus 'it is, therefore it is in opinion only', as that is properly speaking opinable which is in opinion only. Accordingly, I submit that although there is in this paralogism as well a fallacy of consequent as in-some-respect-andabsolutely, more properly speaking there is here a fallacy in-some-respect-and-absolutely - and I say "more properly" because the opinable is taken in the proper sense in somc respect. But in the other paralogism, when it is said 'a/the donkey is not a man, therefore it is not', there is a fallacy of consequent and likewise a fallacy in-some-respect-andabsolutely. But he introduces the paralogism for the sake of its vicious in-some-respectand-absolutely. Some, however, claim that no fallacy of consequent occurs there. For they submit that the claims 'it is a man' and 'it is not' are compatible, and they say that Caesar is a man even when he is not. And when they are confronted with the arg).lmcnt 'it is a man, therefore it is', they say that this argument is not valid, and for the following reason: Every essential and universal predicate expresses habitudinal being; but whenever the \crb 'is' is predicated as a second supplement it expresses temporal being (or "being as of now"). Accordingly, they say that when it is argued as follows, 'Caesar is a man, therefore Caesar is'. a fallacy of equivocation is committed because a move is performed from habitudinal being to temporal being (or: "being as of now"). Whether they are right or wrong on this score is scarcely a mattcr of doubt.]

Long ago, after reading Manlius Boethius, I asked whether my infinite predicates confer some sort of existence on me. This question, whether infinite terms "posit anything" is now being eagerly discussed. [The question "utrum nomen infinitum ponat aliquid" is a standard one in 13th-century commentaries on De interpretatione. For Kilwardby and his contemporaries' treatment of the matter, see Lewry (1978). 45fT For later treatments, see (e.g.) Petrus de Alvernia, Quaestiol1c\ sUf7cr lihrulll Peri hermeneias, MS Paris BN lat 16170: 102rB -103r13; Simon de Faverisham, Quaestiones super libra Perihermeneias pp. 159 -- 160.] Kilwardby holds that an infinite noun, such as 'not-man', docs signify substance with quality like any other noun. But the quality is lack-of-aquality (privatio qualitatis) and the substance, which the term does posit or at least "leave" (derelinquere) , is unspecified being, common to real physical being and being in notion (ratio), expression (dictio), or (mere) opinion. [Texts in Lewry (1978) 48ff. A characteristic passage from his

128

STEN EBBESEN

commentary on Anatytica Priora, MS Cambridge Peterhouse 205: 1l6vA: Si autem tunc quaeratur expositio, dico quod sic exponitur 'non homo': "ens quod non est homo", sed hoc quod dico "ens" in hac expositione non dicit aliquid ens secundum naturam, sed accipitur communiter ad ens secundum naturam et ad ens secundum rationem vel intellectum, et sic ponendo ens non ponit aliquid simpliciter.] [But if an exposition is demanded, I say that 'not-man' is to be expounded as "being which is not (a) man". But when I say "being" in this exposition, this does not mean something which is physically, but the exprrssion refers indifferently to what is physically and what is notionally or conceptually; and when positing being in that way, it does not posit anything in an absolute sense.]

It follows that, in Kilwardby's view, 'the chimera is (a) not-man' does

not claim that I have some substantial being. Nor does 'althe chimera does-not-run' contain a claim that I exercise any activity; all that is posited or left is an unspecified being - in fact just as is the case when the predicate is 'is opined' (opinatur) [see Lewry (1978), 62J or 'is opinable' in the broad, improper sense. I am but the most extreme example of that which is not actually. Less extreme examples are extinct or not yet actualized species, and individuals of the past. Philosophers often perform an experiment in thought, saying "suppose there were no men". Their standard example of an individual of the past is Caesar (who has taken over the role Homer used to have). Now, if people can admit that there is a sort of being which is predicable of both me and actually existing things, they should have little difficulty in admitting that there can be true affirmative 'is'propositions about non-actual natural species or a defunct individual. Some do indeed claim that there is a way of being which renders this possible. They call it "habitual being" (esse habituate) and explain that no stronger being is needed for purposes of verification, and that all essential predications may be interpreted as not positing any stronger being. Hence 'Caesar is a man' is true even when Caesar is not (because he is dead), and 'man is an animal' is true whether the species has actual representatives or not. [For the 13th-century debate about habitual being, see Ebbesen and Pin borg (1970), Braakhuis (1977, 1981), Lewry (1981a, b), De Libera (1981), Fredborg (1981).J There would seem to be several interpretations of esse habituate, viz. (I) = esse habitu (as dinstinct from esse actu) , i.e. an incomplete way of being, being as a tendency to be actualized in a certain way. Thus 'man is (habitually) an animal' means "the nature of man is such that man can be actualized and an actual man will be an animal"; (2) = esse habitudinis or esse consequentiae (distinguished from esse temporis or

THE CHIMERA'S DIARY

129

esse ut nunc), i.e. a way of being which consists in entering into a relationship. Thus 'man is an animal' may mean "there is a relationship

(habitudo), viz. that of species to genus, between man and animal, such that man entails animal and 'this is a man' entails 'this is an animal". In other words, 'man is an animal' contains the true claim that there is a relation of entailment (consequentia) between man and animal. [The reader should be aware that the chimera follows general 13th-century usage in talking about entailment as a relation between terms as well as propositions. However, Roger Bacon (Compendium Studii Theologiae p. 57) shows awareness that this usage is dangerous: Adhuc cavillant de esse habitudinis, sed hoc in propositione (pro nomine ms & ed.) habet locum, et ideo destruetur postea, cum de propositionibus fiet sermo.l [They further use the trick of talking about habitudinal being. But this belongs in a proposition, and so it will be demolished below when I shall talk about propositions.l

(3) = esse habitum, i.e. to be present in human minds as a habitus, a structure reflecting an intellectual grasp of something. In this sense a habitus de homine is knowledge or opinion about man. Though not invented to help me, habitual being is tailored to my measure. On int!!rpretations (I) and (2), all I need in order to become as respectable as the natural species is a quidditative definition, such as 'an opinable being with such and such differentiae" or a recognition that my nominal definition is as good as a quidditative one. On interpretation (3) I just need people not to stop recognizing opinion as a habitus alongside the true grasp, understanding - and then I defy them to find a waterproof criterion by which to distinguish opinion from understanding! As a matter of fact, Nicholas of Cornwall seems willing to declare 'the donkey is a not-man' and 'the chimera is a not-man' equally true when interpreted as propositions about habitual being [Commentarium in De interpretatione, MS Oxford Corpus Christi 119: 126vB: De nullo (hoc MS) non ente praedicatur vere actus entis, ut currere etc. Sed de termino infinito vere praedicatur actus entis. Ergo etc. Minor patet per hoc quod sequitur 'asinus currit, ergo non homo currit'. [About no not-being can an act of the being be truly predicated, such as running, etc. But of an infinite term an act of the being can be truly predicated. Therefore etc. The minor is evident from the fact that this follows: 'a donkey is running, therefore a not-man is running' .1

The answer to this argument is: Potest dici secundum sententiam Bocthii quod terminus infinitus nihil ponit, respondendo huic rationi sic, quod nomen infinitum commune est ad ens et ad non ens secundum Boethium: eo quod est in quolibet, quod est et non est, ideo neutrurn ponit; de co tamen vere potest dici utrumque et ipsurn de utroque, surnpto esse habituali.]

130

STEN EBBESEN

[Answering the argument one may follow the view expressed by Boethius, viz . that an infinite term does not posit anything, and say that an infinite noun extends indifferently to the being and the not-being according to Boethius. Inasmuch as it is in just anything that is or is not, it posits neither, but either may be truly said of it, and itself it may be truly said of either, when the being in question is taken to be habitual.l

But habitual being has its enemies. Even Kilwardby dislikes it. For it is clear that 'Caesar is a man' can have an esse actuate - interpretation, not only an esse habituate - interpretation. But then, why should it be different with 'Caesar is'? In short, 'is' becomes equivocal and 'Caesar is a man, therefore Caesar is' will be a valid inference if 'is' is taken in the same sense in the antecedent and the consequent, but invalid if it is taken in the habitual sense in the antecedent and in the actual sense in the consequent. Although he himself uses the habitus/actus distinction in another, but not unrelated, context [De artu scientiarum § § 433 - 434, p. 150], Kilwardby will have none of such "an equivocal 'is' [See the end of the extract from his Etenchi commentary, supra .] Yet, doesn't he make being equivocal when he says that simple real being has a finite meaning and may be infinitized so as to leave unspecified being, whereas unspecified being has no finite meaning and cannot be infinitized (or, if it can, the result will be "in-no-way-being")? [Kilwardby, Cammentarium in Anatytica Priara, MS Cambridge Peterhouse 205: 116vB Notandum autem quod non dicitur proprie infinitari ens nisi secundum (n.s. iecrio incerra) quod dicit ens simpliciter secundum naturam et veri tat em et tunc privatur simpliciter ens et derelinquit ens secundum opinionem sive secundum rem, illud tamen non ponitur per ipsum. Si autem accipiatur ens commune secundum animam sive opinionem et additur negatio, puto quod sola negatio erit, quia ilia privatio nihil derelinquit, cum privet tam ens secundum rationem quam ens secundum rem. Sic enim omnino privatur ens, et sonat idem quod nullo modo ens. Ens autem in tali communitate a