Meaning without Analyticity: Essays on Logic, Language and Meaning 184718975X, 9781847189752

Meaning without Analyticity draws upon the author s essays and articles, over a period of 20 years, focused on language,

210 19 68MB

English Pages [245] Year 2008

Report DMCA / Copyright


Polecaj historie

Meaning without Analyticity: Essays on Logic, Language and Meaning
 184718975X, 9781847189752

Table of contents :
Introduction • Paul Gochet
1. Does Language Determine our Scientific Ideas?
2. Semantic Theory and Language
3. Meaning without Analyticity
4. Sense and Mode of Presentation
5. Quine and the Pursuit of Truth
6. Meaning Holism and Semantic Realism
7. Logic Acquisition, Usage, and Semantic Realism
8. Davidson, Idealism, and Pragmatism
9. Intentionality Naturalized
10. No Need to Speak the Same Language?
11. The Electronic Dewey
12. A Role for Peirce’s Categories?
13. Old Pragmatists for New

Citation preview

Meaning without Analyticity

Meaning without Analyticity: Essays on Logic, Language and Meaning


H.G. Callaway

Cambridge Scholars Publishing

Meaning without Analyticity: Essays on Logic, Language and Meaning, by H.G. Callaway This book first published 2008 Cambridge Scholars Publishing 12 Back Chapman Street, Newcastle upon Tyne, NE6 2XX, UK

British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library

Copyright© 2008 by H.G. Callaway All rights for this book reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the copyright owner. ISBN (10): 1-84718-975-X, ISBN (13): 9781847189752

"Only in the context of a sentence does a word have a meaning." -Gottlob Frege, 1884. "For a large class of cases of the use of the word 'meaning'--even if not for all cases -the word can be explained thus: The meaning of a word is its use in language." -Ludwig Wittgenstein, (PI, §43). "To imagine a language means to imagine a form of life." -Ludwig Wittgenstein (PI, § 19). "While all language or symbol-meanings are what they are as parts of a system, it does not follow that they have been determined on the basis of their fitness to be such members of a system; much less on the basis of their membership in a comprehensive system. The system may be simply the language in common use. Its meanings hang together not in virtue of their examined relationship to one another, but because they are current in the same set of group habits and expectations. They hang together because of group activities, group interests, customs and institutions." -John Dewey, Logic, 1938.




























12. A











This volume assembles a baker's dozen ofmy papers focused on language, logic and meaning, 1981-2000. Five were originally published in Dialectica, the Swiss journal of the philosophy of knowledge, the title essay in Logique et Analyse from Belgium, two in Erkenntnis, the international journal, and five elsewhere. Nine of the papers were published in continental Europe and are chiefly focused on meaning and formal semantics, though these themes are here conceived from a contextualist point of view involving considerable sympathy for the pragmatic tradition. The papers become more explicitly oriented to the evaluation of the pragmatic tradition as they progress in the sequence provided. The papers were selected and ordered less with a view to chronology than with a view to present interest, thematic development and potential usefulness to the contemporary American-European interplay of pragmatic and analytic themes. I have chiefly omitted papers where their substance has already gone into my prior books. This set of essays develops themes from my Context for Meaning and Analysis (Rodopi, 1993). The first paper strikes a general theme of the volume: the relation of meaning and language in logic and linguistics to meaning and language as a social art and phenomenon within varying cultural settings. There one will find, as I see it, a distinctive approach to the theme of semantics for natural language. Something of this focus is maintained throughout the volume. While the papers earlier in the book concentrate on logic and language from a more technical point of view, the opening social and cultural context for these themes is recalled and elaborated as the sequence advances. The last three papers chiefly center on figures of the classical pragmatist tradition. I aim for a mid-Atlantic and Atlanticist book of a sort, suited to help take readers through the relationships of American analytic and American pragmatist treatments of some central themes in contemporary philosophy. These papers combine, as I estimate, an insider and outsider perspective on American developments and are somewhat colored by the perspective from afar. Though language does indeed have its influence on science, scholarship and cognition generally, the degree of this depends more on the cultural background and on the particular practices involved in scientific and scholarly inquiry.



Basic to the outlook of this book is a pervasive and underlying Peircean fallibilism, partly expressed in my Quine-like rejection of the analytic-synthetic distinction; and this combines with an anti-skeptical, contextualist theory of meaning and inquiry and a related view of the justification of belief or theory acceptance. Thus, in a sense, I take my start from Quine's "tum toward pragmatism" in his "Two Dogmas of Empiricism," though I have attempted to employ the resources of the pragmatic tradition in opposition to excessively formalist and more skeptical elements in the writings of Quine and Davidson. Along the way, there is much said about some other chief thinkers of my interest including Frege, Putnam, Fodor, Peirce and Dewey. The inspiration and composition of this volume owe a considerable debt of gratitude to W. V. Quine. This is evident in the many references to his work to be found below. Quine provides a critical standard in the mediation between American analytic philosophy and the pragmatist tradition-which it eclipsed at the start of the Cold War. I am more specifically indebted to some years of intensive correspondence in the late l 980's and early 1990's; and beyond that, Van kindly provided permission to translate and publish his Stanford University Immanuel Kant Lectures-presently available in my German edition: Willard Van Orman Quine, Wissenschaft und Empfindung, Die Immanuel Kant Lectures (Frommann-Holzboog: Stuttgart, 2003). This provided a marvelous opportunity to think through Quine's philosophy again, both the points of agreement and my critical departures. It is unfortunate that Van did not live to see the project completed. He did provide some advice on the early drafts for the translation. After thinking through Quine again in two languages, I return with new confidence to the essays here collected. Although in a less personal way, the essays assembled here are also especially indebted to the writings of Hilary Putnam and Donald Davidson. Putnam because his work has been an immense help to me in navigating both Quine and the classical pragmatists and Davidson because of the continuing interest and influence of his writings on meaning and interpretation. I believe that those who have followed these three writers or who have found significance in the contemporary recovery of the pragmatist tradition will also find the present book of interest. H. G. Callaway July 2008


Reading H.G. Callaway's Meaning without Analyticity, in depth is most rewarding. This may be especially true for those engaged in the technical side of logic who will not have always had time to read much philosophy in recent years. Reading this book, the informed reader will recognize that the prevalent problems and methods have changed a good deal. Yet, however that may be, this book helps us see the paths we need to follow to regain a deeper understanding of living issues in the field. This book recalls the relevancy of technical issues in logic and the philosophy of language to the broader realms of philosophic thought. The reader will be struck by the careful rendering of the arguments of opponents. The author explains the strengths of the arguments and positions of those under criticism, including contemporary giants such as W.V. Quine, Noam Chomsky, Donald Davidson, and Hilary Putnam, and only then does he start spelling out and displaying related weaknesses. This is a thoughtful book which assembles essays and reviews selected from 20 years of writings and publications on logic and language. The chapters take us back to 1981, and an early consideration of Chomskian semantics, originally published in Philosophical Topics and the Proceedings of the Southwestern Philosophical Association-well before the author's 1993 monograph, Context for Meaning and Analysis; and we come as far forward and up to date as a careful consideration of Susan Haack's Manifesto of a Passionate Moderate-a review published in Erkenntnis in 2000. The title essay first appeared in Logique et Analyse, 109, published in Belgium in March of 1985. This book will appeal to philosophers of an analytic persuasion for its summary and evaluation of debates of the late l 980's and 1990's focused on Quine's skeptical, behavioristic approach to linguistic meaning; yet it reaches beyond analytic philosophy, while retaining analytic interests, to include the American pragmatist tradition-rediscovered and revitalized at the end of the Cold War. (Much of the impetus for this new attention is to be credited to Hilary Putnam.) The author of Meaning without Analyticity is convinced that much more than narrowly scientific questions of formal and empirical semantics are deeply involved in the evaluation of Quine on meaning and translation. One center of interest in the present collection is Callaway's "Meaning Holism and Semantic Realism," originally published in the Swiss journal



Dialectica in 1992-then under the editorship of Henri Lauener. The author shared Lauener's deep interest and accomplishment as a Quine scholar, and Lauener had related doubts on Quine's distinctive semantic doctrines of translational indeterminacy and indeterminacy of meaning and inscrutability of reference. While the article starts out as a detailed, critical examination of Jerry Fodor on the "internal" vs. "external" (a precursor distinction can be found in the writings of Josiah Royce, colleague of William James and George Santayana at Harvard at the start of the twentieth century), interest eventually shifts to the relation of discussions of the internal and external to indeterminacy of meaning and reference. Once we departing from behaviorism regarding language and meaning, there is reason to view the Quinean claims for indeterminacy and the nonfactual character of semantics as better understood and approached by elaborations on the traditional theme of vagueness. A paper not reproduced in the present collection is the author's "Semantic Competence and Truth-conditional Semantics" (Erkenntnis, 28, No. I, January 1988, pp. 3-27). There is reason for this in the desire not to reproduce themes already directly taken up in Context for Meaning and Analysis (1993), but a deeper consideration of the themes of the present book, suggests attention to that paper; and the present book does presuppose a somewhat similar account of the empirical business of the lexicographer in distilling meanings (hypothesizing word meanings) in the attempt to account for linguistic usage. The title of the present book is fully appropriate. We find here an approach to linguistic Meaning without Analyticity, one which substitutes for the analytic a priori in semantics the empirical based formulation of word meanings based on the study of usage; and in more theoretical contexts, meanings of theoretical terms become inseparable from the attempt to formulate hypotheses as needed to meet theoretical and empirical demands on hypothesis. This is a more thorough-going rejection of analyticity than we find in Quine, since Quine chiefly foresaw analyticity in the concepts of linguistic meaning which he criticized or rejected. The chief difference from Quine in the present collection arises from the author's contextualism. This is a moderately anti-formalist approach to interpretation. It presupposes no ultimate conflict with formalist methods, since formalisms do find their uses. The point of departure can be usefully thought of as building on the rejection of anachronism-the attribution to an author of what was only later known. In historical interpretations we need to respect the limits of past knowledge; and in a similar way, various logical and mathematical formalisms tum out most useful when they are applied with just appreciation of the limits and details of fact and theory



we may wish to formalize. Meanings may themselves be viewed as first explicitly arising in the lexicographer's minimal attempt to formalize and regularize pre-existing patterns of usage. Attentive students of later Wittgenstein may well be pleased: we are asked to first look to the use. Let me now briefly examine in more details a few points which strike me as major contributions to the solution of open problems in the field of analytic philosophy. Quine's most well known essay is unquestionably "Two Dogmas of Empiricism" in which he challenges the cleavage between analytic and synthetic statements. The paper prompted an impressive number of rejoinders and defenses. Besides philosophers, professional linguists entered into the arena and produced fresh evidence for restoring the dichotomy under attack. Ruth Kempson argued that it behooves semantic theory to give an account of intuitions regarding contradiction, entailment, presupposition and synonymy. She holds that the entailment between "John killed the warden" and "John caused the warden to die" is analytic. It differs both from the relation of causality which underlines the conditional "If John drank a bottle of sulfuric acid, he will die soon" and from the logical entailment which can be obtained by substituting predicates for predicate letters in the semi-formal "If John F-ies and G-ies, then John F-ies." Even Putnam who supported Quine's rejection of the dichotomy by bringing in new evidence against it insists on retaining a class of pure analytic statements. The statement "All bachelors are unmarried" is one of them. As opposed to statements which involve law-cluster concepts and are open to revision, statements like "All bachelors are unmarried" are not open to revision. Both arguments, that of Ruth Kempson and that of Hilary Putnam, are attractive. Yet I share the author's opinion that we should resist them. If we stick to standard formal semantics, only logical constants (connectives, quantifiers, modal operators) belong to the language. The interpretation of names, predicates and function symbols does not belong to the language. It belongs to the theory. As the author puts is "the semantic rules which interpret the non-logical vocabulary cannot be thought of as belonging to the language. Rather they belong to the theory expressed in the language." The bearing of this remark on the analytic synthetic-distinction is clear. Statements such as "If John killed the warden, John caused the warden to die" or "all bachelors are unmarried" belong to the theory embedded in common sense, not to the language. They are in principle revisable. A revision of the former statement is not so unlikely. The criteria of death are subject to change and the legal definition of the act of killing as well.



Imagine that a new weapon is discovered which disrupts the inter-relations of the neurons of the victim. After a while the action of disrupting someone's neurons in that way may be labeled "killing" even if the victim does not die. Among the many original theses put forward by Quine throughout his work, one of them has aroused permanent discussion, i.e. the claim that radical translation is irrevocably indeterminate. To set the stage, let me quote a passage from Quine's essay "Facts of the Matter" in which that claim is formulated briefly and sharply: " ...two translators might develop independent manuals of translation, both of them compatible with all speech behavior and all dispositions to speech behavior, and yet one manual would offer translation that the other translator would reject. My position was that either manual could be useful, but as to which was right and which wrong there was no fact of the matter." 1 Quine's well-known example is that of a heathen expression to which natives can be prompted to assent in the presence of a rabbit (the expression "gavagai"). The field linguist could equate it with any of the disparate English terms "rabbit," "rabbit stage," "undetached rabbit part" and still by compensatorily juggling the translation of numerical identity and associated particles, preserve conformity to stimulus meaning of occasion sentences. Among the numerous commentators who have criticized Quine's thesis, the author is the first, in my opinion, to have done it conclusively. His criticisms occur in several chapters of the book. In what follows I shall bring them together. In the first essay of the book, entitled "Does language determine our scientific ideas?" the author brings game theory to bear on language and belief-systems. He stresses that "the applications of game theory to problems of interpretation strongly argues that access to adequate evidence for translation/interpretation requires the researcher to enter into the culture under study as an active participant. The attitude of distant and "objective" observer, in contrast, blocks needed evidence. This suggests that if the two translators mentioned in Quine's quotation fail to reach agreement, this is due to the overly narrow concept of observational evidence which Quine allows. In chapter V, entitled "Quine and the Pursuit of Truth," the author shows by concrete example how acting with the natives might enable the translator to overcome indeterminacy. "Thus suppose that we can tell

I. W.V. Quine 1977, "Facts of the Matter" in Essays on the Philosophy of W.V Quine. Shahan, R.S. and Charles Swoyer eds. Norman, University of Oklahoma Press, p. 167.



when the natives are cooking and even help out, while ourselves not yet beyond the level of stimulus meanings. If we are also able to use number words at this level of understanding, then it seems that we will be in a position to decide between various "analytical hypotheses" in our interpretation of "gavagai." It is clear that if the natives ask for one gavagai in order to cook it, then it will also be clear that they will not be satisfied with either a temporal rabbit stage (as when a child asks "May I hold your rabbit?") nor with an undetached rabbit part. (May I hold your rabbit's leg?) ...". The problem is revisited again in Chapter VII entitled "Logic Acquisition, Usage and Semantic Realism." There the author spells out what seems to me the most powerful argument against Quine's juggling with analytic hypotheses to choose among different translations which are equivalent as far as stimulus meaning is concerned. As the author writes: "[ ...] we are naturally somewhat uncomfortable with the picture of attributed formal structure as merely a matter of analytic hypotheses-where there is no fact to the matter as to which formal structure is present. For, recursive procedures are thought to play a role in our ability to generate or understand indefinite or potentially infinite number of sentences." The next problem to which the author offers a new and illuminating solution goes back to Frege. The latter's achievements in Sinn und Bedeutung are well known. In that paper, Frege has spelled out a theory designed to solve two problems: (a) the problem raised by the difference of informative value between the two identity sentences "a = a" and "a = b," (b) the problem raised by the failure of the substitutivity of identity in intensional contexts. His solution of the second problem is open to a serious objection. As the author observes, when intensional contexts are iterated as in the sentence "John believes that Smith believed that Venus is a planet," the senses of referring expressions (such as "Venus" in the example) start to proliferate. At each iteration we need to bring in a new layer of senses. 2 The objection had already been made by Carnap. What is new is the author's solution. We owe him an alternative account of the mode of presentation which explains in a new way how the two identity sentences mentioned above differ in informative content. This account spares us Frege's ontology of senses and remains within the confines of standard semantics. I refrain from taking the bloom off this original contribution.

2. On this point see Gochet and Gribomont 2006, "Epistemic Logic" in The Handbook of the History of Logic, Vol. 7, Twentieth Century Modalities. Gabbay, Dov and John Woods eds. Amsterdam: Elsevier, pp. 141-146.



The author has uncovered a questionable assumption in the argument by which Frege established the correct conclusion that statement of the form "a= b" can be informative. In his argument, the author observes that Frege assumes both that a sentence of the form "a= b" may be informative and that the corresponding sentence about the names, perhaps "There is something x which 'a' and 'b' both name," must be trivial or stipulative. This assumption however is false, the author observes, in so far as it overlooks the empirical elements in semantic claims. Let us tum to chapter VI, entitled "Meaning, Holism and Semantic Realism." Putnam's paper "The Meaning of Meaning" is a landmark in the field of analytic philosophy. It starts with a famous thought experiment which is designed to show that, contrary to what the received view takes for granted, meaning does not determine extension. Suppose that somewhere in the galaxy there is a planet-we call it Twin Earth-in which instead of water, there is a liquid whose chemical formula is not H2O but XYZ. XYZ is indistinguishable from water at normal temperatures and pressures. It tastes like water and quenches thirst like water. The extension of "water" differs although the psychological state of the speakers using the word on the two planets is the same. The aim of this thought experiment is to persuade us that the meaning of "water" can remain the same though the reference varies. This is by no means a conclusion which we should accept thoughtlessly. We are here on a slippery slope. If we accept this claim, we shall next be invited to accept that "mass" for Newton differs in extension from "mass" for Einstein. We approach the claim that their theories speak about different things, and we shall end in saying that they are incommensurable. If we want to avoid this devastating conclusion we should critically examine Putnam's thought experiment. This is done with great care by the author in chapter VI . Putnam's thought experiment might lead some to semantic skepticism concerning whether the objects of our best theories are really what we take them to be rather than something else inscrutably different. It should not. Putnam's story relies upon an unacceptable presupposition, i.e. the presupposition that we can take up a God's view on reference. But we can't; we can only sensibly talk of the reference of our words, on the basis of, and in relation to, our own context of knowledge. If we comply with this requirement, the problem vanishes. To sum up, "meaning or content determines reference in relation to the context of knowledge within which the words in questions are embedded." The author's rejoinder to Putnam's argument is very much in the spirit of the following passage in which Quine discards the skeptical interpretation of his views on the inscrutability of reference: "Staying aboard our



own language and not rocking the boat, we are borne smoothly along on it and all is well, 'rabbit' denotes rabbits and there is no sense in asking 'Rabbits in what sense of "rabbit"?' Reference goes inscrutable if, rocking the boat, we contemplate a permutational mapping of our language on itself, or if we undertake translation. But it is a confusion to suppose that we can stand aloof and recognize all the alternative ontologies as true ... Truth is immanent...We must speak from within a theory." 3 There is however a difference. Where Quine says that ontology and truth are immanent, the author says that meaning and reference are immanent. The two authors however fight skepticism in the same way, namely by speaking from within our best theory of the moment. In that sense both can be described as fallibilist realists and as spiritual heirs of Charles Sanders Peirce. The last comment I will make in this preface is about the claim that observation sentences can be theory-laden. According to Quine, an occasional observation sentence such as "The sun is rising" can be interpreted behavioristically as an unstructured whole. Under that interpretation, the author observes, the sentence can be thought to support both the Ptolemaic and the Copernican theory. However, the same sentence can also be interpreted as a structured whole-interpreted within one or the other theory. In that case, the truth value changes drastically. As the author puts it, "[ ...] the Copernican must reject the literal interpretation put upon it by the Ptolemaic theory." In what follows I shall first show the crucial importance of this twofold interpretation of observational sentences in Quine's philosophy. Next I shall quote an objection made against this twofold reading by a philosopher whose philosophical positions are often very congenial to those of the author, the late Professor Henri Lauener. In Pursuit of Truth, Quine downgraded reference and upgraded truth. In a passage quoted by the author, Quine writes: "what particular objects there may be is indifferent to the truth of observation sentences, indifferent to the support they lend to the theoretical sentences, indifferent to the success of the theory in its predictions." 4 The author however wonders whether it is possible to reconcile this discriminatory treatment of these two key notions with the following passage (almost contemporary) in which Quine put reference and truth on the same level: "As an individuative general term 'rabbit' denotes each rabbit. Such is reference in the home language, relative to the usual or homophonic 'manual of translation.' These paradigms are on a par with Tarski's familiar paradigm

3. Quine 1981, Theories and Things (Cambridge: Belknap Press), pp. 21-22. 4. Quine 1990, Pursuit of Truth (Cambridge: Harvard University Press), p. 31.



for truth." This passage leads the author to conclude that the demotion of reference might very well bring about the demotion of truth: "All things considered; this point appears to place the factuality of truth in question along with that ofreference." There is a loophole open to Quine. He might reply that the checkpoint of the truth of theory ultimately lies in occasion observation sentences. When these are understood holophrastically, Quine might have replied, they allow a definition of truth in terms of assent and dissent which, as opposed to Tarski's definition, does not presuppose the notion ofreference or the notion of satisfaction. Lauener however argues that this way out will not do. let me quote him in extenso: "I consider Quine's holophrastic treatment of observation sentences as the Achilles' heel of this naturalistic system because it compels him to introduce a primitive notion of truth which rests on the behavioristic criterion of assent or dissent prompted in the presence of appropriate stimuli and has nothing to do with Tarski's definition. I contend that such sentences do not belong to the proper language of the theory and consequently that they are of no avail when it comes to testing them." 5 If we agree with Lauener, and I think we should, we are bound to also agree with the author when he recognizes that a considerable tension remains between Quine's referential indeterminacy thesis and his talk about truth. Recognizing this tension does not mean, of course, that we should abandon the whole body of Quine's philosophy. The author's book is a responsible, innovative and convincing defense of a brand of empiricism which strikes me as very congenial to both Quine and Peirce. The author should be recognized and congratulated for having so illuminatingly exhibited the complex network of hidden relationships which obtain between philosophical theses currently under discussion. This is a book well worth thinking through in depth. It assists us in seeing the broad lay of the contemporary philosophical landscape and in overcoming Quine's limiting, behavioristic approach to empiricism. Paul Gochet Liege July 2008

5. Lauener, 2000, "Epistemology from a Relativistic Point of View" in Concept of Knowledge, East and West. (Prabhananda, Swami ed. Calcutta: The Ramakrishna Institute of Culture), p. 157.


This paper argues that the influence of language on science, philosophy and other field is mediated by communicative practices. Where communications is more restrictive, established linguistic structures exercise a tighter control over innovations and scientifically motivated reforms of language. The viewpoint here centers on the thesis that argumentation is crucial in the understanding and evaluation of proposed reforms and that social practices which limit argumentation serve to erode scientific objectivity. Thus, a plea is made for a sociology of scientific belief designed to understand and insure social institutional conditions of the possibility of knowledge and its growth. A chief argument draws on work of Robert Axelrod concerning the evolution of cooperation. 1

1. Introduction Disputing partisan political engagement in literature, Swiss writer Max Frisch argued that while we cannot perhaps escape ideology, still it is already a kind of engagement to "test language in use for its reality content." This is, as he puts it, "an engagement with reality, and thus a critique of ideology." Ideology requires this control-a control exercised by literature precisely when it lacks direct political engagement, practicing independent judgment and critique of language. For Frisch, advocate of self-knowledge and a Swiss tolerance, engagement with reality expresses itself in social criticism and innovative usage. The point and practice are important for literature, but also more generally. They suggest influence of language upon cognition, viewed as a sociological phenomenon-but also our ability to escape a role as passive victim. In fact, something similar to

Presented at a Colloquium of the Swiss Society for Logic and Philosophy of Science, Bern, Switzerland, May 1992 and published in Dialectica, Vol. 46, No. 3-4, 1992, pp. 225-242. Reprinted by permission. I. Axelrod, Robert 1984, The Evolution of Cooperation.




Frisch's "engagement with reality," and "critique of language in use"2 is expressed in philosophy by orientation to science. For science is both crucial in human contact with reality and a forum of linguistic and methodological innovations for cognitive purposes. Scientific orientation provides grounds for a critique of ideologies. From Frisch's perspective, given his sympathy for German, it becomes clear that an engagement with communities and their languages can be a form of engagement with reality-and at the same time a critique of language and ideology. Critical judgment is the link between social engagement and our orientation to reality. This requires strenuous attention to evidence, but beyond this it is a matter of commitment to specific intellectual values-including careful attention to argumentation and its socialintellectual roles. Of particular interest here are forms of communications and language involving "instrumentalized" distortions of interpretationoften functioning to further collective conflicts. Considering this or other phenomena, we must aim for as much scientific standing as possible: a sociology of belief, or better, of the institutions and practices of science and philosophy. The aim is to understand and insure the social conditions of the possibility of knowledge and its growth.

2. Cultural Influences on Cognition Though fuzzy in details, some distinction between 'necessary' and 'possible' seems to be required to mark off things we can change from those which we cannot change. For example, we cannot change the basic laws of physics. They are beyond our control or influence. On the other hand, however difficult this may be at times, it is possible to change our forms of society and our patterns of social relationships; and, of course, it is possible to change our languages. An individual can give up one language in favor of another, and even entire societies can change their languagesleastwise slowly over time. Moreover, science shows that specialized forms of inquiry develop their own special languages. We change languages and change our languages intentionally, so we are also able to control, indirectly at least, the influence a language may have upon us. My claim is not that this is always easy, or always advantageous, but that it is possible. Hesitation on this claim points, in degree, to social constraints on language and its evolution: thus social constraints on cognition.

2. Frisch,Max 1985,Stichworte, pp. 118-119.



These points concerning language are as clearly true as things get. Still, there are considerable traditions of thought holding that conceptual systems embedded in languages, natural languages and the languages of philosophical systems in particular, constitute unavoidable, or virtually unavoidable, governing perspectives on the world. What other significance can we attach to the existence of absolutist metaphysics or ideology as a social-cognitive fact, and to the need to avoid them? We must assume that such systems have consequences in practice-often untoward consequences. Contrary to what a Peircean theory of meaning might lead us to expect, metaphysics and ideology have real effects-social effects. They organize or deform communities, and at the same time they can effectively organize (or distort) expectations, cognition and inquiry quite independent of scientific validity. 3 Languages, viewed semantically, contain their own particular Weltanschauungen, then: they even inform what we will find in perception. After all, we cannot think of observation reports, as "unvarnished news." We must have some particular conceptual system at a given time, and it follows that concepts or system will influence, both what we perceive and what we think, believe or say. The idea that systems of concepts may constitute metaphysical impediments to science is fundamental in the positivist critique of metaphysics. There is no point in attacking what could make no difference. Thus, the positivist and empiricist attack on metaphysics indirectly recognizes their social roles. Given these roles, we have every right to expect social influences on language and cognition. But, cognition and science also have their influence upon society, language and communicative practices. There are no general grounds for comfortable acquiescence here. Especially in light of the influence of B. L. Worf, Thomas Kuhn and Paul Feyerabend, related conclusions and suggestions have reverberated through contemporary philosophy, linguistics, and philosophy of science. Not without some reason, I think; though the points may be over-stressed. Thus, it has been argued that the use of a paradigm in research is " agreement, not because of compelling justifications. And the research in which it is used assumes it, and does not attempt its justification." 4 The reverberations are not empty, in spite of over-dramatization, partly because

3. Cf. Albert, Hans 1991, Traktat uber kritische Vernunft, Chapter IV; and Peter Janich et al. 1974, Wissenschaftstheorie als Wissenschaftskritique. 4. Barnes, Barry 1985, "Thomas Kuhn," in Skinner, Quentin (ed) 1985, The Return of Grand Theory in the Human Sciences, p. 88.



these themes make contact with more traditional scientific determinisms, in application to science itself-suggesting that cognition is under threat of extrinsic control just where we consider ourselves most sovereign. Even those most skeptical of Kuhn's work will agree, I think, that we stand in need of alternatives to such accounts of sociological influences within science. One crucial point here is that theories may be falsified even within the "theory-laden" terms which they contribute to observation. 5 Still, emphasis is needed upon sociological factors which contribute to, or detract from, our willingness to submit treasured conceptions to tests; and viewing absolutist metaphysics and ideologies as themselves sociological factors, embodying values, and at work as social forces, attention turns to examinations of social values embodied in the institutionalization of science and belief. There seem to be important philosophical insights at work in Kuhn's interpretations of science, for he provides a sophisticated perspective from which to question traditional assumptions of empiricism-and of freedom of inquiry in the sciences. If what we observe and perceive stands under the formative influence of prior, culturally induced conceptions and practices, then, contrary to traditional empiricism, it seems implausible to hold that the recognized validity of knowledge claims can be traced merely to origins in sense-experience. Further, one expects just the kind of claim which Kuhn has occasionally emphasized: that education in science, even on a "cursory inspection," is, as he put it, designed to induce "professional rigidity," 6 that commitment to paradigms and scientific consensus is induced by potent mechanisms of socialization. The challenge is that the rationality of science seems to be explained away, in sociological terms. The challenge calls for analysis of social and linguistic factors which make for independent judgment and those which discourage it. For, in these terms we can seek to distinguish reasonable commitment to the yet-to-beproved hypothesis, or theoretical approach, from institutionalized forms of self-immunization. If sociology ofbeliefis to have scientific standing, then this will depend upon a scientific standing for the study of language and communications: a theme to which I now tum.

5. Cf. Quine, W.V. 1990, Pursuit of Truth, pp. 7-8. 6. Kuhn, Thomas 1963, "The Function of Dogma in Scientific Research," in C.A. Crombie (ed) 1963, Scientific Research, p. 350.



3. Language, Community and the Sociology of Belief From a semantic point of view, it is best to consider a language as associated with a community: though there is a need for fine-tuned individuation of linguistic communities. To interpret language in use, we require recourse to evidence arising from (linguistic and non-linguistic) activities of the community in question-though we do not know at first who counts as a genuine member-we encounter the familiar phenomenon of recourse to theory to decide what to count as evidence. Quine's thought experiments in radical translation illustrate fundamentals of the connections between semantics, community and linguistic usage-the use of language in observation. However, this approach to radical translation depends upon evidence available within a limited kind of observation game. Other imaginable language games provide broader access to evidence for semantic hypotheses. 7 Although access to argumentation within a society presupposes some mastery of observational language and logical system, it seems clear as well that argumentation and the kind of participation which lead to it are crucial for an overall understanding of language-systems. This is a point I have argued at length in a recent article. 8 Conclusions there indicate that tolerance of argumentation is a crucial factor in our escape from socially induced rigidities or oversystemization. Language expresses and facilitates the typical activities and preoccupations of a community and constitutes the most articulate expression of the associated culture. Thus, though I resist the temptation to identify meaning and usage, it seems clear that broader access to the typical activities of a given community, and the associated linguistic usage-argumentation included-, will provide a broader range of evidence useful for interpretation. Since members of a given community have, in the end, no access to it not in principle available to a guest, the degree of self-understanding of a community is largely a function of the degree of communications and openness to participation which it provides-just as we expect the work of radical translation to be facilitated by an openness on the part of the community we wish to understand. Thus, the degree in which a society will be able to overcome the limitations of its own culture and language

7. Callaway, H.G. 1991,"W.V. Quine, Pursuit of Truth," below pp. 73-83. 8. Callaway, H.G. 1992, "Logic Acquisition, Usage, and Semantic Realism," reprinted below pp. 105-131.



depends upon sociological structures of interaction and flows of information. Meaning is better viewed as a matter of systems of hypotheses designed to explain abstract elements of usage: those which respond to truth-conditions of sentences in particular. Or, in other words, "usage" is itself theoryladen, and is, therefore, not something which we can access directly, without formulation of complex theory: usage is partly a function ofbeliefsystem and partly a function of non-linguistic intentions. Deeper access depends on getting involved. Mere imitation is one element in language acquisition as we are accustomed to think, but there is more. Imitation becomes less important as we move away from phonetic elements in the direction of syntax, semantics and pragmatics. Thus, optimal, as opposed to preliminary, situations for language acquisition are socially complex, eventuating in shared projects -where ends and the means to be employed are open to argumentation. This approach to semantics and interpretation will also facilitate discussions of themes in the sociology of belief. It is possible to imagine a community in which the social determination of belief approaches maximum. An example here might be the orthodox Marxist picture of capitalism as ruled by relations of the forces of production and ending in all-controlling class ideologies. While it is not my view that such a society ever existed, it is possible to imagine it, and it would be possible, at least in degree, to remake society so as to substantially instantiate, something like Althusser's cognitive-sociological nightmare-where all beliefs come under ideological suspicions. 9 Paradoxically, Stalinism perhaps best approximates this Orwellian image of thought-control. More generally, it is within the power of a given society (especially given hostile external conditions) to maximize the social determination of belief, and institutionally viable hypothesis, by means of control over the flow of information, and control over patterns of participation and interactions in the community. This requires, in addition, similar control over interaction between members of the community and various external influences. We should not expect overly conservative tendencies of cultural reproduction to fare extremely well in our age of electronic communications and global flows of information. Still, it seems we are at pains to escape such influences-and here rests the plausibility of sociological determinism. A chief antidote is emphasis upon the value of communications.

9. Cf. James, Susan 1985, "Louis Althusser," in Skinner 1985, pp. 141-58.



"Of all affairs," as Dewey says in Experience and Nature, "communication is the most wonderful." For by means of communications, things are able "to pass from the plane of external pushing and pulling," and "they come to that of revealing themselves to man ..." For, "when communication occurs, all natural events are subject to reconsideration and revision ..," w and since man and human cultures are themselves natural phenomena-in Dewey's "cultural naturalism"-it followings that communications has the power to open man and human culture to our view. We are able to pass from the plane of external pushing and pulling within a society to one where community and culture are open to our view and, moreover, subject to reconsideration and revision-in ways that physical laws are not. But where reconstructive possibilities are open to us, so are the opposite possibilities. We can, if we wish, remake society, in the opposite way, so that community comes to depend not upon our power to change it, or our power to communicate and form communities of discourse, but rather upon a refusal to make use of such powers-upon their proscription. As Popper has aptly put it, "the future is open." While we cannot always do what we want, we sometimes can. Moreover, such powers expand with our knowledge, and since knowledge cannot be predicted, neither can human potentialities. Knowledge of society and language has a particular importance here. It is even possible to change what we want, and what we may come to know about the viability of our projects and values has a crucial role to play in this. But our ability to change or preserve aspects of human society is ultimately dependent upon social conditions of the possibility of knowledge. Thus, the sociology of belief is important to science, and to the kinds of societies it makes possible. It can provide an understanding of the social-cultural presuppositions of science and of human control over these presuppositions. The sociology of belief could focus attention upon the values implicit in successful science. The enlightenment changed Western civilization in fundamental ways -most importantly to facilitate the growth of knowledge. But the social forces and individual preferences which brought Western civilization to its prior feudalistic conditions were not thereby abolished. It remains crucial to understand these forces and to see them at work in contemporary forms. Further, since communication regarding language is itself so crucial to our

10. Cf. Dewey, John 1927, Experience and Nature, 2nd ed., p. 166; Callaway 1993, "Democracy, Value Inquiry, and Dewey's Metaphysics," pp. 13-27.



capability to change and reform language, belief, and culture, discourse on language has a special role to play.

4. Socio-linguistic Regularity and Generalizations. The human sciences are "special sciences" in the sense that they concern a limited range of phenomena. 11 Often, delimitation of their range of application cannot easily be distinguished from the failure of their generalizations. Moreover, since they concern human-created phenomena, their generalizations-in contrast to those of the "hard sciences," range over changing and variable phenomena. It makes little sense to apply the generalizations of economic theory to a community without trade, money, or the division of labor. Generalizations concerning supply and demand apply to developed market economies; where there is no market, e.g., in a society based upon subsistence agriculture, one expects at best some precursors of the influence of supply and demand upon prices. Moreover, we expect to identify the presence or absence of markets and market-sustaining conditions by reference to the empirical failure or success of market mechanisms. Although high prices indicate scarcity, persistently high prices may indicate the artificial scarcity produced by restrictions on the entry of suppliers-i. e., connivance and/or market failure. Markets exist in various degrees, and hence the laws of the market apply only approximately as the idealizations are approached. Moreover, market-oriented societies may change, distorting themselves, or taking advantage of opportunities which become evident through economic theory itself. But none of this renders the theory of market economies nonempirical. Similarly, a linguistic community is partly identified by reference to linguistic rules or generalizations obtaining within it. Understanding this, we come to expect that distinct linguistic communities correspond in no simple way to easily distinguished groups of speakers. Rather, distinct linguistic communities overlap in complex patterns. For example, we might consider as a community all speakers of basic English-including both those highly accomplished in specialized areas of English and those just out of pidgin English. The domain of basic English might be made relatively distinct-for purposes of a test after a beginners course, or left vague where we wish to include those who might come to have a usable 11. Cf. Fodor, Jerry 1974, "Special Sciences,"and Callaway 1990,"J. A. Fodor, Psychosemantics."



English. Speakers of basic English will also belong to many other linguistic communities. But none of this demonstrates that there are no genuinely law-like generalizations within the domain. Instead, we find a bewildering complexity of generalizations interrelated so as to drive contemporary linguistics to recursive function theory; and we find as well an accentuation of the problems posed by the need to select evidence in light of unsettled theory. For present purposes, it is much to the point to consider semantic generalizations. I want to briefly consider semantic rules and regularities connected with the difference between classical and intuitionistic logic. Here the empirical differences between corresponding communities are relatively clear-in spite of the fact that they have not always been noticed. "Wanton translation," Quine had said, "can make natives sound as queer as one pleases. Better translation imposes our logic upon them ..." 12 But, if a community employs classical disjunction, then we expect, allowing for failures of understanding and vagaries in the individuation of communities, that they will assent to every disjunction of a sentence and its own negation. However, if the community employs an intuitionistic logic, then they will sometimes assent to such disjunctions and sometimes not. 13 Clearly this difference can be detected by empirical means. Following Quine and Nozick on verdict tables, 14 observation sentences or standing sentences may be interpreted by reference to stimulus conditions, and assent and dissent to compounds will then allow us to identify not only the native equivalents of 'or' but also differences in the underlying logic of sentential connectives. The point has considerable significance for debates concerning the objectivity of semantic structures and conceptual systems. For, while I borrow my emphasis upon the role of linguistic communities in the individuation of language-systems partly from Davidson, it is inviting to think ofDavidson's famous (or infamous) rejection of the notion of alternative conceptual systems 15 as rooted in a generalization

12. Cf. Quine, W.V. 1960, Word and Object, p. 58. 13. Nozick, Robert 1986, "Experience, Theory and Language," in Hahn L. and P. Schlipp (eds.) 1986, The Philosophy ofW V Quine, pp. 339-63, p. 361. 14. See also Quine, W.V. 1973, Roots of Reference, pp. 75ff. 15. Davidson, Donald 1974, "On the Very Idea of a Conceptual Scheme," reprinted in Davidson 1984, Inquiries into Truth and Interpretation, pp. I 8398.



of Quine's attributive approach to translating sentential connectives in Word and Object. 16 Thus in "Radical Interpretation," (1973) 17 Davidson claims that to devise "a theory of truth for an unknown language," we are to "first look for the best way to fit our logic ...on to the new language." He notes that "this may mean reading the logical structure of first-order quantification theory (plus identity) into the language ..." In a footnote, he mentions that his method, in contrast to that of Word and Object, "forces quantificational structure on the language to be interpreted." Quine emphasizes the principle of charity only with respect to "pure sentential connectives," (a view he substantially modified in Roots of Reference, 1973), while Davidson applies the principle of charity "across the board." 18 Paradoxically, such ill-conceived "charity" seems to extinguish all conceptual differences as a condition of the possibility of translation! Being less "charitable," we may actually detect what is distinctive among our linguistic neighbors. Though we may imagine various difficulties and complications in the application of empirical tests, there is little room to doubt that Nozick's test is serviceable. Thus, there is little reason to doubt that empirical evidence will allow us to distinguish versions of sentential logic and corresponding communities. Still, the discussion of this test is, quite properly, a scientific systemization of processes in normal language acquisition. If someone acquires classical logic by the normal routes of socialization, questions systematically revealing of logical system are unlikely to be involved. Other methods, focusing upon social cooperation, rather than linguistic investigations, suggest a more realistic approach to the details of language acquisition and the relation of this to social forces.

5. Cooperation and the Prisoner's Dilemma One conclusion here concerns the importance of information concerning action to the acquisition of language. As Davidson has urged in other connections, game theory and considerations of action have a role to play

16. Cf. Callaway, H.G. 1992, "Logic Acquisition, Usage, and Semantic Realism," below. 17. Davidson, Donald 1984, "Radical Interpretation," reprinted in Davidson 1984, Inquiries into Truth and Interpretation, pp. 125-140. 18. Ibid., p. 136.



in theories of the evidence available for interpretation. 19 Linguistic action is nested within systems of other sorts of action: thus both initial and more sophisticated interpretations of linguistic actions and linguistic expressions are able to draw upon a broad context of knowledge concerning who is doing what in particular situations. Acquisition of linguistic competence, and semantic competence in particular, is thus shown possible in ways dependent upon systems of interaction and the understanding of action in concrete situations. The point here depends upon a recognition that it is possible to (fallibly) note specific causal connections without, at the same time, acquiring a knowledge of covering laws.20 Though retaining the external perspective of one's own language and belief-systems, the application of game theory to problems of interpretation strongly argues that access to adequate evidence for translation/interpretation requires the researcher to enter into the culture under study as an active participant. The attitude of distant and "objective" observer, in contrast, blocks needed evidence. So long as linguistic meaning is not identified with actual usage, it is evident that meanings are only imperfectly reflected in usage. Thinking of sentence meanings as truth-conditions, for instance, usage is not merely a function of what is believed true (or any other single factor). A great variety of extra-linguistic intentions and purposes (including many linguistic intentions beyond that of stating what one takes to be true) also play a role in determining the usage of language in a given community and on a given occasion. One objective here will be to explore this point, in a fairly rigorous way. The point is worthy of illustration and deeper examination, because where we know independently what someone is trying to do, and especially what he is trying to do by means of his words, this information is of considerable import for understanding or interpreting what he has to say. Being in a position to distinguish a statement from an imperative, for example, is only the most obvious illustration of this type of phenomenon. We depend upon a wealth of common-sense knowledge of ordinary purposes and intentions to disambiguate and interpret as the need arises. Similarly, we can appreciate how the game-theoretic concept of cooperation has a role to play in theories of the evidence available for interpretation. Where we attempt cooperation, goals and intentions are 19. See Davidson, Donald 1974, "Belief and the Basis of Meaning," reprinted in Davidson 1984, Inquiries into Truth and Interpretation, pp. 141-54; Davidson I 980, "Toward a Unified Theory of Meaning and Action," pp. 1-12. 20. Davidson, Donald 1963, "Actions Reasons and Causes," reprinted in Davidson 1980, Essays on Action and Events, pp. 3-20.



communicated and become clear on either side, or relatively clear in any case; and to the extent that someone's extra-linguistic goals and intentions can be treated as known, this will constrain interpretation of utterances and ultimately constrain the interpretation of the relevant linguistic system. It will be useful, therefore, to consider some recent advances in game theory, highly relevant to the notion of cooperation, and capable of application to problems of interpretation and language acquisition. These are results related to the iterated prisoner's dilemma and strategies which have special relevance to social situations or processes which can be understood as instances of the iterated prisoner's dilemma. The development of cooperation depends in part upon a transfer of information between the parties involved; and conventions of language can be thought of as instances of the kind of information we need to communicate (or possess) in order to facilitate the special kind of cooperation involved in belonging to, or participating in, a particular linguistic community. All of this suggests modeling language acquisition, acquisition of semantic competence, and the transfer of semantic information in general, upon the development of cooperation which is possible under conditions of the iterated prisoner's dilemma. The prisoner's dilemma is defined in terms of a payoff matrix such as the following which is adapted from Axelrod's The Evolution of Cooperation (1984):

Column Player Cooperate Defect

******************************* Row * Cooperate R=3, R=3 Player * Defect T=5, S=O

S=O, T=5 P=l, P=l

Each player chooses simultaneously (or in ignorance of the other player's choice) one of the two options, either to cooperate or to defect. (What counts as cooperation or as defecting is usually assumed to be clear from the description of the specific situation, in other cases, this becomes clear by means of a series of interactions.) The two choices together determine the payoffs to the two players. Thus, for instance, if the column player chooses to defect while the row player chooses to cooperate, then the result is to be found in the upper right-hand comer where the row player gets S = 0, the "sucker's payoff' and the column player gets T = 5 which is "the temptation to defect." The lower right-hand comer shows the result when both players choose to defect-each gets one point. This is P = 1, the



"punishment for mutual defection." Where each player chooses to cooperate, they each get 3 points-as is shown in the upper left-hand comer. This is R = 3, "the reward for mutual cooperation." What is crucial to the prisoner's dilemma is not how these scores are interpreted-the particular rewards or punishments-, but rather the relative values of the outcomes (as defined for each player independently): T > R > P > S. So long as the temptation to defect is greater than the reward for mutual cooperation, and this in turn is greater than the punishment for mutual defection-which is greater in turn than the sucker's payoff-, then we have an instance of the prisoner's dilemma. This involves a dilemma in view of considerations which arise independently for each player. Thus suppose you are the row player and you are trying to decide what to do. You must consider two options of the column player: the column player will either cooperate or defect. Suppose that the column player is going to cooperate. lfyou cooperate too then you get 3 points, the reward for mutual cooperation. However, if you defect then you get 5 points, the temptation to defect. Thus, where the other player will cooperate, then it is best to defect. But suppose, on the contrary, that the other player will defect, you still have two choices. If you cooperate, then you get O points, the sucker's payoff; however, if you also defect, then you will at least get I point, the punishment for mutual defection. Thus, where the column player will defect, then it is better to defect yourself. The dilemma is, then, that it appears to be always better to defectregardless of what you think the other player will do-, and this in spite of the fact that both players will do better from mutual cooperation. "What makes it possible for cooperation to emerge," Axelrod remarks, "is the fact that the players might meet again." 21 For, if we assume only one interaction, or that the interactions will be limited to a finite and know number, then there will be no incentive to cooperate. Assume, for instance that there will be only one interaction. The reasoning above will thus govern this single interaction and both players will have an incentive to defect. But, if they will both defect in the case of a single interaction, then the same reasoning leads to mutual defection on the last of a series of interactions of known and finite length. This reasoning tends to move back along the chain to the present. 22 (The point is evident, e.g., where employ-

21. Axelrod 1984, p. 12. 22. Cf. Luce, R.D. and H. Raiffa 1957, Games and Decisions, pp. 94-102.



ees change jobs in a bureaucracy. As the date approaches cooperation declines and hostility becomes overt.) This quite simple scheme for a non-zero sum game expressed in terms of the iterated prisoner's dilemma has found application in analysis ranging from international relations to biological symbiosis, and it casts light upon a great variety of applicable situations. In a zero-sum game such as chess, in contrast, it makes no sense to attempt to cooperate with one's opponent, for there is no possibility in the nature of the game for a player to benefit by cooperation. Thus, the strategies of zero-sum games are typically quite complex and elaborate. It makes little sense in a zero-sum game to allow one's opponent to become aware of one's strategy or intentions. Strategies become Byzantine. However, Axelrod shows that in the iterated prisoner's dilemma, the best overall strategy to follow, in most situations, is a very simple one which combines a willingness to cooperate with the refusal to be exploited. Moreover, there is an advantage to having one's opponent aware of the strategy one intends to employ.

6. TIT FOR TAT and Linguistic Cooperation The most important strategy in Axelrod's work is TIT FOR TAT. While this is not the best strategy to follow in every possible circumstance, it turns out to be a quite crucial strategy for a large number of possible situations. For example, if one's opponent is using the strategy ALL D, defecting on every move, then one cannot do better than to use the same strategy in reply. However, Axelrod does show, by means of computer simulations of tournaments involving a great variety of possible strategies, that TIT FORT AT, is extremely robust and successful in a vast number of possible situations: both where it plays against similar strategies and where it plays against strategies which attempt exploitation. Yet, TIT FOR TAT is a relatively simple strategy to apply and follow. It calls for a player to cooperate on the first move of a sequence and thereafter to do whatever the opponent did on the last move. Thus TIT FOR TAT cooperates with those willing to cooperate and defects from those who show unwillingness to cooperate in their actions. Making use of TIT FOR TAT, one stands the best chance of finding those players who are in fact willing to cooperate. It will be useful here to think of language learning situations as involving a version of the iterated prisoner's dilemma where both sides make use of TIT FOR TAT (or other strategies). As we shall see, making use of these assumptions, it will be possible to think of a sequence of moves in a



game of iterated prisoner's dilemma as a means for either side to communicate, quite wordlessly, what is to count as cooperation as viewed by the other side. This is a kind of information which is not always available at the start of a sequence of interactions-even where there is mutual intention to cooperate. Thus, focusing on linguistic cooperation, and especially the observance of semantic conventions, the iterated prisoner's dilemma provides an approach to the acquisition of semantic competence. In this context, willingness to cooperate amounts to a willingness to communicate. The point suggests that complexity of strategy is a direct functional indication of reluctance to communicate, and this suggests in tum that complexity of linguistic forms is indicative of reluctance to cooperate: as one might well expect, clarity and simplicity of expression are very important to wider communications. Further, since evaluation of claims made depends upon their being communicated and understood, we fully expect that scientific communications will be bound by simplicity and clarity of expression; obscurantism is simply a refusal to cooperate in the scientific enterprise.

7. Language, Cooperation and Social Structures Axelrod explores factors giving rise to social structures among interacting groups in terms of the iterated prisoner's dilemma. Of special importance to language and language acquisition is what is called labeling. "A label is a fixed characteristic of a player, such as sex or skin color, which can be observed by the other player. It can give rise to stable forms of stereotyping and status hierarchies." 23 Clearly, there are linguistic as well as nonlinguistic forms of labeling, and the effects connected with the labels of sex or skin color are also associated with observable linguistic features: dialect, social register and accents. In a highly intellectual atmosphere, the effects of labels can also be found associated with semantic differences, reflecting points of view and intellectual affiliations. Labels are important because people can begin their interactions with strangers on the basis of expectations that they will act and react in known ways. The advantages of observing labels and of being labeled are directly connected with the creation of expectations: One of the most interesting but disturbing consequences of labels is that they can lead to self-confirming stereotypes. To see how this can happen,

23. Axelrod 1984, pp. 145-46.


MEANING WITHOUT ANALYTICITY suppose that everyone has either a Blue label or a Green label. Further, suppose that both groups are nice to members of their own group and mean to members of the other group. For the sake of concreteness, suppose that members of both groups employ TIT FOR TAT with each other and always defect with members of the other group .... Then a single individual can do no better than to do what everyone else is doing and be nice to one's own type and mean to the other type. 24

In view of the incentives, "stereotypes can be stable, even when they are not based on any objective differences." 25 Moreover, the assumed conditions are not in the least unlikely. We are all aware, in degree, of effective stereotypes: those based on race, or skin color or national origin. Still, it is of special interest to appreciate how each individual has a strong incentive to participate in systems of stereotyping and discrimination. This is what Axelrod finds so disturbing in his results: he captures the social phenomenon of self-serving discrimination in game-theoretical terms. The account is compelling and chillingly simple. Strategies tend toward collectively stability, as a function of the importance of future interactions in comparison with the present. If we assume the payoff matrix given above, then TIT FOR TAT will be collectively stable provided that the next interaction is 2/3 as important as the present interaction. Under such conditions, if everyone else is using TIT FOR TAT, then an individual player can do no better than by employing the same strategy. A group of players using this strategy in their interactions with each other cannot be invaded by players employing an alternative strategy. By definition, an alternative strategy will not come to replace a collectively stable strategy. Thus, in a range of situations described, the stereotypes lead to incentives to conform to the system. Anyone who departs from it will see scores drop. For example, if a blue player attempts to break the stereotypes and use the strategy TIT FOR TAT in interactions with a green player, then since all green players use ALL D with the blue players, the blue player will get O points, the sucker's payoff, rather than I point, the punishment for mutual defection, whenever attempting cooperation with a green player. There are incentives, for each individual, to maintain the system and to act in accordance with the stereotypes. Now imagine the prospects of cooperative research between distinct traditions under similar conditions.

24. Axelrod 1984, p. 147. 25. Ibid.



Labels also play an important role in maintaining status hierarchies. Thus, suppose that everyone in a particular group has a particular characteristic in some definite degree. Examples are strength, or height, or skin tone, or an assigned position. Axelrod describes a status hierarchy in terms of strategies: "everyone is a bully toward those beneath them and meek toward those above them." 26 (Interestingly, German has a derogatory word for this: 'Radfahrer'.) We can imagine that everyone uses a certain strategy in interactions with those above in the hierarchy: "cooperate unless the other player defects twice in a row, in which case never cooperate again." This strategy is relatively meek. The player allows himself to be a sucker on alternative moves. Still, it also shows some provocability-it will not tolerate more than a certain amount of exploitation. In contrast, when meeting subordinates in the hierarchy, one uses a "bully" strategy: "alternate defection and cooperation unless the other player defects even once, in which case never cooperate again. " 27 Obviously, the players at the top of the hierarchy will do very well, and those at the bottom less well. However, it is important to see that the situation will be stable where "the discount parameter is high enough"-that is, where future interactions in the same group are relatively important. (This factor is effected, e.g., by physical propinquity and geographical and professional immobility.) For in the case of an isolated revolt from below, that player's scores will sink drastically. The reason is that "it would be better to take one's medicine every other move from the bully than to defect and face unending punishment. " 28 Thus, even those at the bottom of an insulated social hierarchy have considerable incentive to maintain itespecially given expectations of later advances in the hierarchy. Now, consider the effects on career advancement of a critical attitude expressed by a student within a hierarchical institutional environment. These analyses illustrate the facility of game theory to illuminate the inner structure of social phenomena. We see a quite pervasive human social structure which can be made more specific by examples of tribalism and nationalism-or intellectual affiliations. So imagine that each tribe or nation uses a label which maintains its distinctness from every other and that a hierarchy is instituted within each group on the basis of the degree to which a particular person exemplifies the positively evaluated tribal or national traits. Whatever the value of the system to those who participate in

26. Ibid., p. 149. 27. Ibid. 28. Ibid., p. 150.



it, disadvantages are also evident: quite pervasive human problems centered on collective conflict, conflicts we know to be avoidable in principle but which may appear inevitable. We know that such conflicts are avoidable in principle because we know that groups structured in terms of the above considerations could often benefit by breaking down stereotypes preventing wider cooperation. Still, conflicts among groups often appear unavoidable because we are familiar with the incentives which maintain the system.

8. Conclusions I can only briefly suggest some implications of Axelrod's work for the sociology of belief and related issues regarding language and communicative practice. The chief implication is that communications highly structured by labeling with regard to intellectual affiliations and dependent upon strongly hierarchical institutional structures tend toward insularity and dogmatism. 29 Under such conditions, one also expects a development of language-systems suited to facilitate insularity and dogmatism: key terms playing the roles of labels for intellectual affiliations tend toward greater obscurity, general claims with similar social-institutional functions tend toward multiple and complex ambiguities. Language becomes stilted by attempts to insulate key expressions from deeper examination or criticism. These changes function to guard insiders against outsiders, to carry on mock debates with opponents, and to maintain social hierarchy. Special dangers exist in socially defined and intensely hierarchical schools of thought-a point reflected in early modem disdain for "scholasticism." For, we expect that membership or advancement is only possible by strict adherence to stereotypical thinking, and disdain for critical attitudes-styles of education which over-emphasize reproduction of isolated and internally defined tradition. The points here help clarify the connection between science, democratic participation and the social character of language. More democratic social relations are required for broader, deeper argumentation. Otherwise, emerging problems and innovative viewpoints remain under-developed, unarticulated or ignored. Unfortunately, in situations of intensive competition (of whatever sort) recognition for outsiders (or those with low status-ranking) is often the first victim. Yet these people are often key social representatives of criticism.

29. Cf. Albert 1991, pp. 116-17.



Failing deep moral concern for democratic participation based on the cognitive value of contributions, existing viewpoints and schools may reproduce in hot-house monocultures, cut off from improvements and corrections. We need ladders within existing social and institutional structures of scientifically oriented thought and bridges between traditions-things which over-competitive conditions strongly tend to eliminate. The hypothetical-deductive method, in particular, may come to resemble a rationalization for ideology or absolutist metaphysics, without due stress on the search for potentially falsifying evidence and deep consideration of alternative hypotheses. The social-institutional correlate of this, in science and philosophy as elsewhere, is engagement to "test language in use for its reality content:" an emphasis upon clear, tolerant, and unpartisan participation, including searching and critical argumentation and research. Failing this, language tends to deteriorate to the level and type of conflicts it serves. The root plausibility of sociological determinism appears linked to strategies of social-institutional reproduction arising from excessive competition and overly collectivized career identities. Objectivity is fostered by reaching, with tolerance and a critical eye, across the boundaries of cultures, generations, schools, and institutions.


1. The Object of Semantic Inquiry In recent years linguists have devoted a good deal of attention to semantic problems. At the center of this work is the concept of the semantic component of a transformational grammar. The objective is an account of semantic competence in a language, integrated in some fashion with rules of syntax and phonology, and sharply distinguished from the influences of speakers' beliefs and knowledge of matters of fact. In spite of important traditions skeptical of such a conception of semantics, generative grammarians have generally not departed from it. According to Chomsky, the attempt to incorporate factors such a belief or attitude into grammar, " ...amounts to a rejection of the initial idealization to language, as an object of study." 1 While he is not willing to rule out such a development entirely, Chomsky's view is that there is insufficient empirical motivation. If such a development did prove correct, it would strongly affect the Chomskian conception of the object of study in linguistics. Speakers of the same language differ in beliefs and attitudes. Thus, if these differences are fully relevant to semantic theory, the very notion of semantic competence in a language fails of application. As Chomsky puts it, " .. .I would conclude that language is a chaos that is not worth studying ,,2

In spite of such commitment, Chomsky's conception of semantics must contend with both philosophical skepticism and contrary traditions in linguistics. In "Two Dogmas" Quine argued that " is non-sense, and the root of much non-sense, to speak of a linguistic component and a factual


Originally presented to the Southwestern Philosophical Society, Denton, Texas, November 1980 and published in the Proceedings of the Southwestern Philosophical Society/ Philosophical Topics, Summer 1981, pp. 61- 70. This paper is indebted to Henry Hiz, Monroe Beardsley and Stephen Stich for helpful comments on earlier versions. Reprinted by permission of the editor. I. Chomsky, Noam 1977, Language and Responsibility, pp. 152-53. 2. Ibid. p. 153.



component in the truth of any individual statement.'' 3 If so, it follows that language as the object of semantic investigation cannot be separated from collateral infonnation. F. R. Palmer pursues a similar contention in his recent survey of issues in semantic theory: " .. .it is impossible even in theory to draw a clear line between the meaning of a word or sentence and all possible relevant infonnation about it." 4 In spite of such skepticism, and through a variety of theories, devotion to lexical decomposition and truth 5 dependent on language has not abated. An interesting assault on recent semantic theory, based on Quinean grounds and directed against the semantic usage of the concept of logical fonn, has served to place some recent semantic work in sharper focus. For, according to Stephen Stich, such theories involve no account of logical fonn at all! Briefly, since the attribution of logical fonn to sentences of a natural language builds upon speaker's intuitions regarding relations between sentences, Stich argues that the theorist has not sufficient grounds to claim that his "logical truths" are true. "To substantiate the claim he needs an argument that data about intuitions and beliefs can yield conclusions about truth." 6 Speakers' intuitions cannot justify an account of validity, logical truth and logical fonn. Rather, the standard of the logician " .. .is the production of the best theory adequate to the needs of empirical science." 7 Thus, although identification of semantic structures in natural language may account for a speaker's intuitions regarding sentence relations such as consequence and contradiction, this is not at all the same as identifying logical fonn or logical relations. Like any other scientist, the logician is successful " ...when he can replace his current theory with another which, while equally adequate for the purposes of science, invokes a simpler logical grammar, making due with a sparser regimented language."8 In short, the logical truths are those fonnulated in the logician's language. "There is no cause to protest if the truths of logic cannot be

3. Quine, W.V. 1951, "Two Dogmas of Empiricism," in Quine 1953, p. 42 4. Palmer, Frank 1976, Semantics A New Outline, p. 44. 5. Cf for instance Chomsky 1977, Language and Responsibility; J.D. Fodor 1977, Semantics, Theories of Meaning in Generative Grammar; Bever, Katz, and Langendoen 1976, An Integrated Theory of Linguistic Ability; Ruth Kempson 1977, Semantic Theory. See also Chomsky 2000, New Horizons in the Study of Language and Mind, pp. 61-62. 6. Stich, Stephen 1975, "Logical Form and Natural Language," Philosophical Studies 28, p. 414. 7. Ibid., p. 406. 8. Ibid.



formulated in the language of the market place, for neither can the truths of physics." 9 While linguistics may set itself the goal of identifying semantic rules constitutive of the speaker-hearer's competence, identification of such rules does not serve to establish correlative truths based on language. Even supposing that human beings are endowed by evolution with a full-blown set of semantic universals, Chomskian innate ideas, corresponding to a universal conceptual system, truth is an entirely separate matter. An analogy may be helpful here. Suppose that the mammalian eye and brain are pre-wired so that the world is perceived in Euclidean terms. This might account for certain abilities and also for the early pre-eminence of geometry as a science, but even given such an hypothesis, if we perceived space to be Euclidean as a matter of biological necessity, it by no means follows that physical space is Euclidean. Emphasis upon "logical form" arose in the wake of the massive retreat from the Katz-Postal-Aspects standard theory. In Chomsky's recent work, he " ...use(s) the expression 'logical form' to designate a level of linguistic representation incorporating all semantic properties that are strictly determined by linguistic rules." 10 But the problem is whether any such rules could serve to establish truth based on language alone, as Chomsky supposes. Commenting upon Katz' work, Chomsky has pressed a similar point against the broader claims of the standard theory: There are, I believe, good program. It seems that other of beliefs concerning things judgments of meaning and semantic representation from

reasons for being skeptical about such a cognitive systems---in particular, our system in the world an essential part in our reference ... I doubt that one can separate beliefs and knowledge about the world. 11

But while rejecting a fuller theory of meaning in Katzean style, Chomsky has not given up the goal of accounting for semantic competence independent of matters of fact and belief. Rather, semantic theory is now to be restricted in order to better attain this goal: My own speculation is that only a bare framework of semantic properties, altogether insufficient for characterizing what is ordinarily called "the

9. Ibid. 10. Chomsky, Noam 1977, Language and Responsibility, p. 145. 11. Ibid., p. 142.



meaning of a linguistic expression," can be associated correctly with the idealization "language." 12

This minimal semantic component is still to make room for truth based on language . . . . it seems reasonable to suppose that semantic relations between words like 'persuade,' 'intend,' 'believe' can be expressed in purely linguistic terms (namely: if I persuade you to go then you intend to go; If I persuade you today is Tuesday, then you believe that today is Tuesday. These are facts of language and not of the external world). 13

The view is, then, that although " ...a full dictionary cannot be distinguished in a principled way from an encyclopedia, ... 14 and it" ... is in fact false that the interpretation of sentences by hearers is independent of questions of fact and belief, ... 15 still, " ... the contribution of grammar to semantic interpretation is very different in kind from the contribution of fact and belief." 16 Chomsky still envisages the semantic component of a generative grammar as " ...a separate idealized structure, ...with certain aspects of dictionary entries as parameters." 17 Thus he continues to consider it a good working hypothesis that " ...there are semantic properties that are general, universal, of the type proposed by Katz and others." 18 These will require some version of lexical analysis and provide for truths resting on language, independent of the facts of the external world. This view is not Chomsky's alone. For example, in her recent text, Semantic Theory, Ruth Kempson argues, " ...that the problem presented by Quine's attack [on meaning] has to be bypassed if we are to make the correct predictions about sentence relatedness ..." 19 Semantic theory is to give an account of intuitions regarding contradiction, entailment, presupposition and synonymy. This account must correctly predict, according to Kempson, the relation of entailment between 'John killed the warden' and 'John caused the warden to die.' 20 which goes beyond the logician's con-

12. Ibid., p. 143. 13. Ibid., p. 142. 14. Chomsky, Noam 1977, Essays on Form and Interpretation, p. 36. 15. Ibid., p. 38. 16. Ibid. 17. Ibid., p. 36. 18. Chomsky 1977, Language and Responsibility, p. 141. 19. Kempson, Ruth 1977, Semantic Theory, pp. 186-187. 20. Ibid., p. 187.



cept of logical implication, taking in a supposed analytic truth. Kempson's survey and many other works show a wide acceptance of similar views among linguists.

2. Language and Theory The syntactic component of a transformational grammar includes a lexicon, phrase structure rules, and transformation rules, and these correspond rather directly to the vocabulary and formation rules of an artificial language, which together serve to recursively specify the set of wellformed formulas. The semantics involved in the interpretation of an artificial language, however, does not have the same direct relevance to the language specified by the formation rules. When an interpretation is provided in standard model-theoretic semantics, the interpretation represents a choice between alternatives and is directly correlated with a system of theory L expressed in the language. A theory, in this sense, is a deductively closed set of sentences, including all its own deductive consequences. No interpretation of a language will be neutral between all theories. For example, when a predicate 'F' is assigned to a subset of the domain D, it will either be assigned to the null set or to a set with some member(s). Thus, the interpretation will either provide that '-(:lx)Fx' E L or it will provide that '(:lx)Fx' E L. In short the semantic rules which interpret the non-logical vocabulary cannot be thought of as belonging to the language. Rather they belong to a theory expressed in the language. Again, if two names or individual constants 'a' and 'b' are interpreted, then the interpretation must provide that 10 (' a')= 10 (' b'), if' a=b' E L, and if 'a=b' is ~ L, then the interpretation will not assign the same element of D to both 'a' and 'b' .21 Similar points hold for the interpretation of any two predicates 'F' and 'G'. Interpretations of the two predicates must be correlated with systems which either do or do not logically imply such sentences as '(x)(Fx only if Gx)', '(x)(Gx only if Fx)', and '(x)(Fx if and only if

21. More strictly, the interpretation should be thought of as concerned with the purported reference and co-reference of expressions of a theory. For instance, if 'a=b' is a logical implication of a theory T, then the theory purports that a and b are the same object, but T and 'a=b' may (in fact) be false. On the distinction between reference and purported reference see my "Reference, Variables, and the Empty Universe," 1979, pp. 85-98. Chomsky makes use of a somewhat similar distinction between co-reference and intended co-reference.



Gx)' .The general point is that an interpretation in standard semantics reflects the details of a theory. Thus, for example, if we reject '(x)(Fx only if Gx)' and make use of the same vocabulary in our new theory then, obvi-

ously, we must also reject any interpretation of this vocabulary according to which l 0 ('F') is a subset ofl 0 ('G'). Given this conception of semantics it is difficult to see how anyone could hold that the goal of semantic theory is to characterize the semantic rules of a language, independent of matters of fact and belief. Any interpretation rather elucidates a theory, or range of theories, in the syntactically specified language. From this perspective it is easy to see why many cast doubt on the distinction between truth based on language alone and matters of fact and belief. In standard semantics rules of interpretation are adjusted to agree with our theories and our theories must, ultimately, face the test of observation. However, semantic theory in linguistics makes little use of standard model-theoretic semantics. It will be useful to consider, therefore, the consequences of adding something to the semantics of an artificial language corresponding to the linguists' rules of lexical decomposition. The result is a set of constraints upon the interpretation of the non-logical vocabulary, serving to eliminate some interpretations. Kempson, for instance, suggests the following analysis of'wife'. wife: [Married)x -[Male]x


An interpretation in accordance with this proposal will be constrained to assign 'wife' to a subset of the set to which 'married' is assigned, and 'wife' to no (non-empty) subset of the set to which 'male' is assigned. Thus, acceptance of this rule amounts to restricting ourselves to interpretations required by theories whose axioms (or logical implications) include 'All wives are married' and 'No wife is male.' In the logician's usage of 'theory' semantics is once again directly relevant to a theory in a language rather than to language in abstraction from the various theories a language may be used to express. One important question then is whether this sense of 'theory' is fully relevant to the linguist's semantic investigations. Does it make sense to speak of 'theories' which reject 'All wives are married' or 'If I persuade you today is Tuesday, then you believe today is Tuesday,' and other theories which accept such sentences?

22. Kempson 1977, Semantic Theory, p. 92.



Obviously, from the point of view of the semantic work under consideration, a language is something which carries with it a certain conceptual system. Part of the objective of semantic investigation is to characterize the conceptual system of the language under study. Such a conceptual system is reflected in the speakers' intuitions regarding sentence relations. But is there a valid distinction between conceptual system and matters of fact and belief which will do the work which linguists want such a distinction to do? It is not to the purpose here to question the notion that the linguist can identify implicit rules of inference governing the discourse of a specific speech community. The question is whether the import of such rules can be sharply distinguished from matters of fact and belief, and whether they will suffice to distinguish a set of truths based on language alone. Arguments against such a distinction will find some support in Chomskian methodological writings. For instance, in the following passage, Justin Leiber points out that definitions may be given up in the face of contrary empirical-theoretical considerations: As Chomsky pointed out in talking about the vocabulary of structuralist linguistics, a science or scholarly discipline, makes the vital terms of its vocabulary theory-laden. Definitions of these terms, whether tacit or explicit cannot be regarded either as mere stipulations or as matters of ordinary usage; rather such definitions contain ... empirical presuppositions .... 23

Now if the definitions of theoretical terms in science are theory-laden, reflecting empirical presuppositions, why, we must ask, is not the same true of the semantic rules postulated as underlying the use of natural languages? There appears to be no grounds for insisting that such rules do not have their own empirical presuppositions. The only difference, it seems, is that we are inclined to take the conceptual system of a natural language, particularly our own language, so much for granted. In learning our own language, as Quine puts it, we learn "the lore of our fathers." Consider Chomsky's paradigms which make use of such terms as 'intend' and 'belief. Behaviorist critics of mentalism have long regarded such intentional language as superfluous and misleading. Yet, if it is coherent or in any sense empirically possible to give up the ordinary vocabulary of mentalistic psychological descriptions and psychological explanation, then Chomsky is merely begging the question by insisting upon truths based on

23. Justin Leiber 1975, Noam Chomsky: A Philosophical Overview, p. 16 I.



semantic rules governing the use of such expressions. My point is merely that such semantic rules can be given up, that there is no sharp distinction between conceptual system and matters of fact and belief. I need not hold that these rules governing intentional language are best given up. The reply may be made that there is no reason, on a Chomskian view, to regard the import of particular semantic rules as beyond fallibility. The semantic theorist is accounting for intuitions of speakers of a particular language as spoken at a particular time. Thus, if certain rules are given up, this is a matter for another study of another language. This reply also begs the question. It is premised on a distinction between languages which is drawn in accordance with the very distinction between conceptual system and matters of fact and belief which is under discussion. What counts as a change of language depends, in this reply, on what counts as a change in conceptual system. In short, we come to the question of differentiating between change in meaning and change of belief. Indeed this, in a way, has been the question all along. The linguist F. R. Palmer draws the conclusion which is dictated by our considerations thus far: There is, then, no such thing in semantics as linguistic ability that is unrelated to knowledge of the world. These are essentially one and the same thing. That does not mean, of course, that we cannot in some way limit our area of study, but it is a mistake to think that we can limit it to what is 'purely' linguistic. 24

The relevance of conceptual system to language is established in that we cannot learn a natural language without learning some conceptual system. This follows from the fact that a natural language is never entirely uninterpreted. But the line between conceptual or analytic truth and matters of fact and belief is inherently unclear and shifting. Further, the linguist has no need of such a distinction if the elucidation of idealized theories is taken as the objective of semantic work in linguistics.

3. Theory and Semantic Dialect To regard a sentence as analytic is to hold it constant in the rearrangements of belief which are consequences of the growth of empirical knowledge. Even if it is stipulated that we have an intuitive grasp of what to hold

24. Palmer 1976, Semantics, p. 46.



constant and what should be allowed to vary, this intuitive distinction is surely a function of the theoretical context in which we find ourselves. Thus, we draw the line, however, unclearly, in light of our estimates regarding the relative reliability of what are essentially theoretical alternatives. If no alternative denying a certain sentence appears or appears even minimally viable at a given time, the sentence will be intuitively a constant in all our thinking. Nor should we regret this since there is little real alternative to accepting those theories which we have the most reason to accept. Conservatism regarding the most central and explanatory definitions and postulates is a justly honored methodological principle. But viability is a matter of degree, and no methodology allows for an infallible selection of candidate theories. Problems appear when conservatism comes into conflict with other, equally honored, methodological principles. A good historical illustration is Galileo's claim that the earth moves. At least part of the reaction to this claim was based upon the notion that the earth, being at the center of the universe, could not be in motion; 'standing motionless' was part of the meaning of 'earth.' But in this case conservatism finally gave way to the grater simplicity of the Copernican system. Even with regard to the conceptual system of a natural language, then, there is no way to draw a sharp distinction between meaning and belief. Further, the intuitive distinction may shift as a function of empirical-theoretical innovations and discoveries. Semantic theory concerns the relationships of purported extensions of expressions. 25 Thus we should fully expect that any account of such relationships, even the minimal constraints imposed by theories of lexical decomposition, will only be viable relative to some empirical-theoretical account of the world, namely the account of the speech community under study. This should be particularly evident, if linguistics is regarded as a division of psychology in accordance with Chomsky's claim. If so, we can only take the object of semantics to be the elucidation of a linguistic representation of the world. Relations between expressions and a specified domain of objects are only established in so far as it is also established that the relevant representation is accurate. But it is not within the scope of linguistics to decide which of our theories are true, and by the same token,

25. This is a minimal claim. Since concern with purported reference and co-reference cuts across the traditional distinction between meaning and belief, it is reasonable to expect the notion of purported reference to do at least some of the work of the notion of meaning.



it is not within the scope of linguistics to provide a list of sentences true in virtue oflanguage. The appropriate conclusion is not that semantics is impossible as an empirical discipline. Rather, interest in the basic conceptual system and inference rules of a particular linguistic community may be expected to provide an explanatory account of both linguistic and non-linguistic behavior. But if there is no clear line to be drawn between conceptual elements and matters of fact and belief, then, strictly, every difference is belief between speakers may be an appropriate dividing line for a separate semantic theory. How far differences in belief between speakers should be followed out into separate semantic theories is then an essentially pragmatic question governed by the interest and usefulness of the resulting theories. This is essentially similar to the most prevalent attitude toward the fact that different speakers of the "same" language do not all have the same vocabulary at their command. Of course, from a strictly formal point of view such differences between speakers require different accounts of the syntactic component in transformational grammar, since differences in vocabulary must be accounted for by differences in the lexicon. But from a practical point of view, the linguist interests himself in some linguistic community sharing the same vocabulary. Strictly, then it is better to see the object of semantics as the elucidation of the basic theories expressed in human languages, especially when such theories are massively prevalent among the members of an important linguistic community. It should be noted, of course, that different theories may be expressed in the same language, syntactically identified. In so far as a single language, syntactically identified, is used to express contrary beliefs (and attitudes), differing intuitive relations will be found: for instance, differing patterns of consequence and paraphrase. Surely, not all differences in belief or attitude are equally of interest, but some such differences affect mutual intelligibility. Thus we should think of the object of semantic investigation as the elucidation of theories expressed in a given vocabulary with a given syntax. We cannot expect there to be no important semantic variations across all speakers of the same language in any intuitive sense of"same language."


In a series of interesting and influential papers on semantics, Hilary Putnam

has developed what he calls a "post-verificationist" theory of meaning. 1 As part of this work, and not I think the most important part, Putnam defends a limited version of the analytic-synthetic distinction. In this paper I will strrVey and evaluate Putnam's defense of analyticity and explore its relationship to broader concerns in semantics. Putnam's defense of analyticity ultimately fails, and I want to show here exactly why it fails. However, I will also argue that this very failure helps open the prospect of a new optimism concerning the theory of meaning, a theory of meaning finally liberated from the dead weight of the notions of analyticity and necessary truth. Putnam's work, in fact, makes valuable contributions to such a theory.

1. Trivial Analyticity To see Putnam's defense of analyticity in context, it is crucial to notice that he is largely sympathetic to Quine's attack on the analytic-synthetic distinction. To Putnam, Quine's criticisms in "Two Dogmas of Empiricism" were both "powerful and salutary." "The distinction," he says, "had grown into a veritable philosophical man-eater: analytic equaling necessary equaling unrevisable in principle equaling whatever truth the individual philosopher wishes to explain way." 2 The notion of analyticity needed to be pruned down to size, and Quine's criticism had an appropriate effect. Putnam agrees with Quine that the analytic-synthetic distinction does not have the massive epistemic significance once attached to it. In particular it will not provide "foundations" for knowledge by distinguishing purely factual matters from necessary truths

An early version of this paper was read at Seventh International Congress of Logic, Methodology and Philosophy of Science, Salzburg, Austria, June 1983. First published in Logique et Analyse, 109, March, 1985, pp. 41-60. Reprinted by permission of the editor. I. The crucial papers are collected in Putnam 1975, Philosophical Papers, Mind, Language and Reality, Vol. 2. 2. Putnam, Hilary 1975, "The Meaning of 'Meaning' " in Putnam 1975, Philosophical Papers, Vol. 2, p. 253. *



or linguistic rules-to function as an unalterable framework for knowledge. Putnam writes of "the silliness of regarding mathematics as consisting in some sense of 'rules of language' ."3 He argues for a refutation of conventionalism4 and holds that "there are no a priori truths."5 So much, then, for the linguistic theory of the a priori. These points fit quite nicely into a Quinean perspective, but Putnam's work also departs from Quine's views in quite important ways. Thus, Putnam is a critic of behaviorism, a defender of his own version of the analyticsynthetic distinction, and he expresses, at times, a constructive attitude concerning the theory of meaning which is a central and formative departure from Quine's views. In "The Meaning of 'Meaning"' (1975) Putnam repeats his contention that Quine's attack on analyticity "went too far." On the view expressed, "some limited class of analytic sentences can be saved."6 This claim reiterates those found in "The Analytic and the Synthetic" (1962), published thirteen years earlier. "Quine is wrong," Putnam had written; "There are analytic statements: 'All bachelors are unmarried' is one of them." But though Putnam claims that "there is an analytic-synthetic distinction," he also says that it is a "rather trivial one."7 That is to say, it will not do important epistemic work. Understanding why Putnam makes these claims will require a close look at the connection between Putnam's work on 'meaning' and his notion of analyticity. One key to Putnam's views on meaning and analyticity is the notion of revisability. Those who reject the notion of analyticity tend to substitute talk of conservatism as a methodological principle or criterion of theory acceptability. Much that once was reckoned "true in virtue of meaning alone" will now be characterized as only relatively immune to revision due to some crucial role in accepted theory. But Putnam retains the notion of unrevisability (1962, 1975). Analytic truths are, he says, unrevisable, but this unrevisability of analytic truths is not simply a function of there being linguistic rules to the contrary. "Linguistic obligatoriness," he writes, "is not supposed to be an index of unrevisability or even of truth." In consequence, "we can hold that 'tigers are stripped' is part of the meaning of 'tiger'-without being 3. Putnam, Hilary I 962, "The Analytic and the Synthetic." Reprinted in Putnam 1975, Philosophical Papers, Vol. 2, see p. 42. 4. Putnam, Hilary 1975, "The Refutation of Conventionalism," reprinted in Putnam 1975, Philosophical Papers Vol. 2. 5. Putnam 1975, p. xvii. 6. Putnam 1975, p. 253. 7. Putnam 1962, "The Analytic and the Synthetic," p. 36.



trapped into the problem of analyticity."8 Though feature F is part of the meaning of a word X, it does not follow, for Putnam, that 'All X's are F' is analytic---or even true. Not everything that is a matter of meaning is therefore true in virtue of meaning. Analyticity is a rather special characteristic, arising as a consequence of the meanings of words, but only in special circumstances. Thus, we cannot dispute Putnam's claim that 'tigers are striped' is part of the meaning of 'tiger' by pointing to albino tigers or the possibility of genetic mutations among tigers. Rather than being linked in the usual way to 'meaning', 'analyticity' is directly linked to 'unrevisability'. A decisive question, then, concerns the grounds for the claim that some matters of meaning are unrevisable. Answering this question will bring up very central elements of Putnam's views on semantics. Before proceeding, however, I want to comment briefly on some points which have already emerged. There is deep significance in Putnam's weakening of the traditional connection, in the theory of meaning, between 'meaning' and 'truth'. This point will become somewhat clearer in the following section where Putnam's notion of linguistic stereotypes in discussed. For now, certain general remarks are called for to provide some perspective. 'Truth' is closely tied to 'extension' by Tarski's theory where 'truth' (of a sentence in a particular language) is defined in terms of the basic semantic notion of satisfaction. The extension of 'tiger', for instance, is just the class of objects which satisfy 'tiger', i.e., the class of objects of which 'tiger' is true. Further, there is good reason to maintain, with Putnam, that when we ask for the meaning of a word like 'tiger' we are partly concerned to know the extension, or reference, of the word. However, as Putnam has emphasized in criticism of Davidson's truth-based semantics, there is also good reason to think that any story concerning extension alone is not the full story concerning meaning. "For many words," Putnam argues, "an extensionally correct truth definition can be given which is in no sense a theory of meaning of the word." Consider, for instance, the following, which might appear as part of a Davidsonian semantics for English: 'Water' is true of x, if and only if x is H2O

Concerning a pre-scientific community, it is reasonable to suppose that "most speakers don't know that water is H2O," Thus, "this formula in no way tells us anything about the meaning of the word 'water' ."9 In giving a characterization of the meaning of 'water', part of what we want is to capture an 8. Putnam 1975, p. 256. 9. Putnam 1975, p. 259.



element of the speaker's linguistic competence-those crucial elements of the overall belief system which encapsulate how the community speaks (or thinks) about water. Once it is recognized that what people commonly believe-as a matter of the meaning of their word~ay yet be false, then it becomes extremely implausible to hold that a theory of meaning need only proceed in terms of the actual extensions of expressions. Although extension is crucial to meaning, what is more to the point when it comes to speaker's competence is purported reference. 10 Putnam's weakening of the traditional connection between 'meaning' and 'truth' is an important advance. Matters of meaning, for instance 'tigers are stripped' need not be true. (Though, of course, they may be true.) Yet, the importance of this point leaves it all the more problematic why Putnam regards some statements as analytic. We need to look further and see what this "trivial analyticity" consists in for Putnam.

2. Meaning and Stereotype Putnam introduced the notion of linguistic stereotype in lectures during 1968. But the fullest development of the notion is found in "The Meaning of 'Meaning'." There Putnam states that "the theoretical account of what it is to be a stereotype proceeds in terms of the notion of linguistic obligation." What it means to say that being striped is part of the (linguistic) stereotype of 'tiger' is that it is obligatory to acquire the information that stereotypical tigers are striped if one acquires 'tiger', in the same sense of 'obligatory'

10. The concept of purported reference is developed in Callaway 1981, with particular attention to names. Purported reference of names, in relation to a theory, is there defined within the confines of extensional or referential semantics, in simulation ofFrege's notion of sense. Putnam's stereotypes can be understood as characterizations of the purported reference of expressions, if we treat the stereotype as a standard social theory (approximating to an axiom system) concerning the domain. Linguistic competence, conceived of in terms of syntactic competence plus knowledge of stereotypes can thus be thought of as a minimal socially shared "theory" used for purposes of communications and held in relative isolation from scientific developments and other information regarded as inessential for purposes of communications. A problem will remain as to how to individuate such a "standard social theory" (i.e., the distinction between "meaning" and "matters of fact and belief." But I do not see this problem as fatal to semantic theory. See the discussion in Callaway 1981, "Semantic Theory and Language," above.



in which it is obligatory to indicate whether one is speaking of lions in the plural or lions in the singular when speaking of lions in English. 11

ft is clearly possible that the rules of English pluralization may change, or have exceptions (in certain dialects?). Consider 'Lion is quite prevalent in this country' or 'There was no lion to be seen'. We can imagine the frequency of such usage increasing in some community to the final exclusion of the standard alternative. ln the same way, we can imagine the 'obligatory' character of 'tigers are striped' might well fade away under some circumstances, or not exist among certain speakers of English. It will be objected in some quarters that what we are imagining here is simply a language other than standard English, or perhaps a dialect of English-with some other obligatory features. A standing difficulty with this position is that it merely shifts the problem. At first concerned with what is genuinely obligatory, we are now left wondering about the identity conditions of 'languages' and 'dialects.' The temptation will be to distinguish genuine linguistic obligatoriness by simply multiplying dialects ad hoe. While this is an important kind of problem, it is not peculiar to Putnam's semantics, and I do not wish to dwell on it here. Putnam has some interesting suggestions regarding the problem in his paper, "Some Issues in the Theory of Grammar," (1961) where he criticizes the view that no distinction can be made between 'deviant' and 'non-deviant' statements. 12 However, if one hopes to characterize the analytic-synthetic distinction in terms of 'unrevisable linguistic obligation' it is best to notice the problematic character of 'linguistic obligation' at the outset. Even if linguistic obligatoriness is not a direct indication of analyticity, on Putnam's view, still it is presumably a necessary condition. Putnam's major proposal toward defining 'meaning' is to specify a "normal form" for the description of word meaning. The normal form description of the meaning of a word should be a finite sequence or 'vector', whose components should certainly include the following ... : ( l) the syntactic markers that apply to the word, e.g., 'noun' ; (2) the semantic markers that apply to the word, e.g., 'animal,' 'period of time,'; (3) a description of the additional features of the stereotype, if any; (4) a description of the extension. 13

11. Putnam 1975 , p. 256. 12. Reprinted in Putnam 1975, pp. 85-106. 13. Putnam 1975, p. 269.



Each of the components of the vector "represents a hypothesis about the individual speaker's competence, except the extension." Putnam provides the following example: 'WATER' SYNTACTIC MARKERS




mass noun; concrete

natural kind liquid

colorless; H20 (give or take impurities transparent; tasteless; thirst-quench; etc.

We should think of the stereotype as given under the two middle headings, and including perhaps elements under the first heading, but not including the description of the extension. A hypothesis concerning the extension of a word though required for the account of word meaning, is not part of the task of specifying the stereotype associated with the word. The description of the extension does not pertain to the ordinary speaker's linguistic competence, though it may agree with the opinion of experts in the community under study. While Putnam borrows the terrn 'semantic marker' from Katz and Fodor, his usage is importantly different. In particular, Putnam's stereotype is not supposed to give necessary and sufficient conditions for the extension of the word., while Katz and Fodor did aim for this in terrns of the combined force of their categories of 'markers' and 'distinguishers'. 14 On Putnam's account of it, meaning does deterrnine extension-though stereotype does not, or usually does not. Trivially, the four kinds of entries determine the extension of the word since the entry under 'extension' does so independently. But the first three kinds of entries are not required to determine the entry under 'extension'. Thus, for Putnam, to give the meaning ofa word (or the meaning of this kind of word) we must indicate the stereotype, and also the actual extension of the word. If Putnam's 'linguistic stereotype' is compared with Frege's notion of 'sense,' then a crucial point of difference with the Fregean tradition might be put as the claim that 'sense' does not determine reference --or at least that sense does not completely determine reference. Although some Fregeans are willing to stipulate the point, the larger tendency has been to ignore it.

14. See Katz, J. J. and Jerry Fodor 1963, "The Structure of Semantic Theory."



Even ifwe consider everything we know about water, making no distinction at all between matters of meaning (or stereotype, i.e., the ordinary speaker's competence) and matters of fact (for instance the tests which experts use to fix the extension of 'water'}-still considering even all this information-it is possible to reinterpret the predicates of this theory over a distinct domain. Essentially this is Quine's point where he argues for "the inscrutability ofreference." 15 But Putnam (along with Quine) does not reject the notion of reference or extension on these grounds. Instead, in the end, we must acquiesce in "the mother tongue"--or at least in our own best theoryindicating the extension of a word from the perspective provided. It is, of course, a pointless exercise to merely imagine distinctions our own best theory does not make, or possible reinterpretations of our theory, and on such grounds remain agnostic concerning the reference of our words. Putnam carries this point over into the general case where we are interpreting the words of others. Especially if we make distinctions that they do not make, it is perfectly in order to take notice of this by indicating the actual extensions of their words-by our best lights. It is to be counted as an advantage, then, that Putnam's theory does not require that stereotype determine extension. For, only given this point do we come to be able to indicate both the purported reference and the actual reference ofa speaker's words. As Putnam has said, "concepts which are not strictly true of anything may yet refer to something." 16 The same kind of point may be made where the stereotype is correct as far as it goes, but only determinately specifies an extension (i.e., only determinately specifies an extension relative to the way that our best theory determinately specifies an extension). On Putnam's view, this is the normal situation we find ourselves in when describing the meaning of English words in English. Nor need underdetermination of extension by stereotype (or even underdetermination of reference by theory) count against the very possibility of a theory of meaning-as on Quine's view where he argues from the "inscrutability of reference" to the indeterminacy of translation. For though linguist systems and theories serve to specify the extensions of the expressions employed only determinately (again, determinately relative to some new system of distinctions which may still be uncovered), still such a level or degree of discrimination of the system is something objective which can be

15. Quine, W.V. 1969, "Ontological Relativity," in Quine 1969, Ontological Relativity and Other Essays, p. 51. 16. Putnam, Hilary 1973, "Explanation and Reference," in Putnam 1975, p. 197.



characterized-relative to our best theory. Or, at least, the "inscrutability of reference" provides no conclusive grounds to the contrary. If the meaning of a word is given by its normal form description, then we must expect such normal form descriptions to tell us something about word synonymy-and analyticity-on Putnam's view. Although the obligatory features entering into the normal form description of the meaning of a word do not themselves guarantee analytic connections, or even truth, still if there is to be any analytic truth on Putnam's account, it must be because some of what is obligatory is also unrevisable. Could this be so, given Putnam's theory? As we saw above, the specification of the extension of 'water' as H 20 "does not mean that knowledge of the fact that water is H 20 is being imputed to the individual speaker or even to the society. It means that (we say) that the extension of the term 'water' as they (the speakers in question) use it is in 17 fact H 2 0." It does not follow, then, that if any other word 'X' has the same meaning as 'water' that the extension of 'X' must be specified by the use of the expression 'H 20'. For 'X' to have the same meaning as 'water' it is only necessary that the stereotypes agree and that the extension of 'X' be in fact the same as the extension of 'water'. Even supposing that the meanings of 'water' and of 'X' are unrevisable, and that 'X is water' is analytic, there is absolutely nothing in Putnam's account that will guarantee that anyone will ever know that 'X is water' is analytic. For there is nothing in Putnam's account to guarantee that the coextensiveness of'water' and 'X' is known, or will ever be discovered. Thus, the synonymy of the two words need never be recognized. Putnam's theory will not support anything very close to the traditional analytic-synthetic distinction, because he rejects the category of a priori truth. This kind of conclusion concerning whether one will know that 'X is water' is analytic holds, moreover, even in cases where a single speaker has these two words in his vocabulary. 18 In short, Putnam's theory allows for a "false negative" result in the detection of analyticity. The theory also allows the possibility of a "false positive" result in the detection of analyticity. To see why, suppose that the normal form descriptions of two words X and Y and identical, except for the descriptions of the extensions of the two words, and suppose further that the descriptions of the extensions are believed to be coextensive. We would then conclude that the words X and Y are synonymous. Could this synonymy claim be unrevisably correct? The answer must be "no," given the considerations traced so far.

17. Putnam 1975, p. 269. 18. Cf. Putnam 1975, p. 270.



From the fact that X and Y are believed to be co-extensional (in accordance with our best current theory) it does not follow of course, that the two words are co-extensional. We may imagine, for example, a linguistic group which treats the words 'mass' and 'weight' as synonymous. If we characterize the meaning of their words, being ourselves ignorant of the difference between mass and weight Gust imagine we are ignorant of Newtonian physics), then, in such a condition of ignorance, it will be reasonable to hold that the words 'mass' and 'weight' are actually synonymous and co-extensional. But physical theory since Newton tells us that mass and weight are quite distinct. The consequence of these considerations is that the linguist's judgment concerning synonymy (and analyticity) are open to revision in the light of any theoretical or empirical development-including those taking place entirely outside linguistics. Since we are to decide on the extensions of a speaker's words in light of our own best theories, any hypothesis concerning extensions -and thus synonymy-is open to revision on the grounds governing the development of our overall theory of nature. Putnam's account of word meaning creates so many problems for the analytic-synthetic distinction that it is difficult to see how an unrevisable synonymy could ever be reasonably ascribed. Putnam envisages only a scaled down version of the analytic-synthetic distinction, linked to a fundamentally revised notion of linguistic meaning. But the problems sketched in this section serve to suggest another theoretical option: a concept of meaning completely divorced from any version of the analytic-synthetic distinction. Such a concept of meaning deserves serious investigation if only because it is an option to which semantic research has so often been blind. Quine's critique of 'meaning', though motivated and seemingly forced by his rejection of the analytic-synthetic distinction, may have been aimed too widely. 19 If so, we certainly we certainly need to take serious notice of this point. Putnam's distinction between stereotype and extension, that is, his weakening of the traditional connection between 'meaning' and 'truth', could become an essential element of such an approach to semantic theory. The prospect is that the analytic-synthetic distinction may finally be allowed to die. If matters of meaning are thought of as elements of ordinary speaker's competence, then the distinction between 'matters of meaning' and 'matters of fact and belief stands some chance of

19. See Quine, W.V. 1974, "First General Discussion Session," Synthese 27, p. 493: "What I've really been concerned with or motivated by in this stuff about translation and indeterminacy hasn't been primarily translation but cognitive meaning and analyticity and the like."



an empirical resolution: meaning becomes the least common denominator of belief. This is a possibility if matters of meaning (stereotype) may be false, and stereotype is not assumed to determine extension. But on such an account there is little reason to think that matters of meaning ever determine unrevisable truth. For truth does depend upon extension.

3. Law-cluster Concepts In "The Analytic and the Synthetic" (1962) Putnam defended his limited form of the distinction in terms of a further distinction between "arbitrary stipulation" or "linguistic convention" and "systematic import." Speaking of many borderline cases between the analytic and the synthetic, he characterizes them as follows: What these statements reveal is something like different degrees of something like convention, and different kinds of systematic import. In the case of' All bachelors are unmarried' we have the highestdegree of linguistic convention and the minimum degree of systematicimport. In the case of the statement 'There is a past' we have an overwhelmingamount of systematic import-so much that we can barely conceive of a conceptualsystem which did not include the idea of a past.20 Putnam disputes the claim that the existence of the past is analytic, that is, he will not agree that it is self-contradictory to hold that the earth came into existence five minutes ago, complete with memory traces and etc. But he is willing to say that such hypotheses are "more than empirically false." "It is not empirically false," he says, if one means by 'empirically false statement' a statement which can be confuted by isolated experiments." 21 The point here and Putnam's concept of systematic import hark back to Quine's thesis that "our statements about the external world face the tribunal of sense experience not individually but only as a corporate body." 22 Thus, 'There is a past' is not particularly relevant to isolated experiments or observations, but plays a very important role in systematizing our overall theory of nature. To revised this statement would require changes throughout scientific theory and common belief. Thus we justifiably resist such a revision on grounds of conservatism. A statement P which cannot reasonably be revised on the basis of isolated experiments or observations, may yet be revised if we develop a rival theory 20. Putnam 1962, "The Analytic and the Synthetic," p. 39. 21. Ibid., p. 37. 22. Quine 1951, "Two Dogmas of Empiricism," p. 41.



which gives a better explanation. The alternative theory might be simpler, more comprehensive, or integrate better with other theories. In general, competitive criteria of acceptability need to be invoked to overcome the conservative attachment to the established theory. Acceptance of an alternative theory containing not-P, or not containing P, may well lead to a revision of the truth-value of P. As examples of such revisions, Putnam discusses "definitional" principles in Newtonian physics and the changes brought about by relativity theory; he also discusses the changes in the status of geometrical principles which followed the development of non-Euclidean geometry. "Einstein was to revise," he says, "principles that had traditionally been regarded as definitional in character." Einstein as we all know, changed the definition of 'kinetic energy'. That is to say, he replaced the law 'e = ½mv2 ' by a more complicated law.23

Before Einstein, 'e = ½mv2 ' had been used to provide countless predictions in experiments where the definition was not itself treated as being at risk. The definition had figures as a "framework principle." In order to effectively challenge it, to understand how it could be thought false, an alternative framework-relativity theory-was needed. Hume regarded it as impossible to imagine straight lines except as conforming to the principles of Euclidean geometry. That parallel straight lines do not meet was, for Hume, a consequence of the "relations of ideas." Thus physical space could only be thought of as being Euclidean. This concept of space as governed by the principles of Euclidean geometry was, obviously, not something that could be easily tested by isolated experiments. An entirely new framework of principles had to be developed including, first of all, the mathematics of non-Euclidean geometry. What these examples tend to show is that "conceivability" is context relative. The development of knowledge and theory may render conceivable what was heretofore inconceivable. Paradoxically (given his advocacy of the analytic-synthetic distinction) Putnam has pressed this kind of point eloquently and quite forcefully. In contrast with such examples, and the conclusion they suggest, Putnam maintains that 'all bachelors are unmarried' is an analytic statement. His argument depends upon his thesis that the kinds of changes which affected the energy definition and the postulate of parallels were

23. Putnam 1962, "The Analytic and the Synthetic," p. 44.



only possible because the relevant concepts are "law-cluster concepts." The idea is explained as follows: In analogy with the idea of a cluster concept, I should like to introduce the notion of a law-cluster. Law-cluster concepts are constituted not by a bundle of properties ... but by a cluster of laws which, as it were, determine the identity of the concept. The concept 'energy' is an excellent example. 24

Given the character of such law-cluster concepts, analytic principles involving them are "difficult." The reason it is difficult to have an analytic relationship between lawcluster concepts is that such a relationship would be one more law. But, in general, any one law can be abandoned without destroying the identity of the law-cluster concept involved ... 25

What is crucial here is the way that Putnam's view of the revisability of scientific definitions is tied up with the notion of concept identity. Scientific definitions may be revised, without a change in meaning of the relevant concept, because the identity of such a concept is constituted by an entire cluster of laws, with no single law being necessary for concept identity or synonymy. Given such a notion of concept identity, Putnam has an effective reply to the charge that a principle, such as the definition of kinetic energy, was and remains analytic, and that contemporary science simply makes use of an entirely different concept. This "law-cluster" account of concept identity contrasts, to some degree, with Putnam's later account of concept identity in terms of stereotype and normal form meaning description. It is difficult to imagine, for instance, that the everyday stereotype associated with 'energy' involves any laws at all. But it should also be noticed that Putnam's account of synonymy, in terms of words having the same stereotype, and the same extension, is not totally inconsistent with this law-cluster account of concept identity. For it is only reasonable to hold that Newtonian 'energy' and Einsteinian 'energy' are co-extensional because of a considerable continuity in the laws employing the concept of energy. Given either notion of concept identity, Putnam can allow a revision of definitional principles without a change in meaning of the concept involved. So it is clear that he has an effective criticism of the stronger, traditional version of the analytic-synthetic distinction. Law-

24. Ibid., p. 52. 25. Ibid., p. 52.



cluster concepts are insulated from meaning change by their "systematic import." Since Putnam regards 'bachelor' as a "one criterion word" it follows, on his account, that 'All bachelors are unmarried' is analytic. If being an unmarried man is the only criterion of the application of the word 'bachelor', then to reject the analyticity of' All bachelors are unmarried' one must change the meaning of the word. 26 Since 'bachelor' is a "one criterion word" it, unlike 'energy' lacks "systematic import." The truth of 'All bachelors are unmarried' is thus a matter of stipulation. The crucial premise of Putnam's argument is his claim that "if we ask what the meaning of the word 'bachelor' is, we can only say that 'bachelor' means 'unmarried man,' whereas if we ask for the meaning of the term 'energy' we can do much more than give a definition." 27 Holding 'All bachelors are unmarried' immune from revision "can do no harm" Putnam says, "because bachelor is not a law-cluster concept ... it is not independently 'defined' by standard examples, which might only contingently be unmarried men." 28 Though it is conceivable that 'bachelor' might become a law-cluster concept, in the absence of reasons to believe that it will become a law-cluster concept, Putnam maintains that we may hold 'All bachelors are unmarried men' immune from revision. The exact position here is rather subtle, and it is best to quote Putnam directly: It is perfectly rational to make stipulations to the effect that certain statements are never to be given up, and those stipulations remain stipulations to that effect, notwithstanding the fact that under certain circumstances the stipulations might be given up. 29

Surely, stipulations to the effect that certain statements are never to be given up remain such, even if they are given up. But given that such stipulations may have to be given up (that is, given that we are never in a position to guarantee that a given statement will not need to be given up), why it is "perfectly rational" to make such stipulations in the first place? Putnam says that it is rational to hold 'All bachelors are unmarried' immune from revision on the basis of the empirical finding that "there are no exceptionless laws containing the word 'bachelor' ."30 But, if it is

26. 27. 28. 29. 30.

Ibid., p. 68. Ibid, p. 53. Ibid., p. 59. Ibid., p. 60. Ibid., p. 59.



conceivable (even if unlikely) that such laws may be found, then surely it is more reasonable to simply hold 'All bachelors are unmarried' true. Why should anything more be either helpful or necessary? Putnam gives no sufficient answer to this question, and this, I think, is the essential reason why his defense of analyticity fails. Without the epistemic guarantee once thought to be provided by doctrines or theories of the a priori, the notion of analyticity is pointless. Putnam lists two advantages of the analytic-synthetic distinction and "strict synonymy" within a language: brevity and intelligibility. 31 But the advantage of brevity is provided by any definition of a longer expression by a shorter one, so that the definition need not be regarded as immune from revision to provide the advantage of brevity. It is worth pointing out that it is perfectly consistent to be extremely conservative concerning a particular definition (say, 'p & q' for '-(-p v -q)') without ever invoking the notion of analyticity. So, where supposed analyticity leads to definitions with the advantage of brevity, so does theoretical conservatism. As for intelligibility, a similar point holds. If some of the statements of a language are immune from revision, then says Putnam, "different speakers of the same language can to a large extent understand each other better because they can predict in advance at least some of the uses of the other speaker." Conservatism concerning definitions and central principles will serve to provide this advantage as well. Nor should we ask for more than a defeasible conservatism, if there is a chance that a particular definition or statement may need to be given up. From the stand point of methodology, I repeat, the notion of analyticity is pointless. Putnam claims that analytic statements are "true because accepted as true." 32 But he also says, concerning the stipulation that certain statements are to be held immune from revision, that "under certain circumstances the stipulations might themselves be given up." 33 He even allows that statements such as 'All bachelors are unmarried" might be given up in the lights of new empirical-theoretical developments (i.e., in the unlikely case that laws making use of 'bachelor' are eventually found). But, if so, then it is hardly consistent to hold that the statement is analytic, that is "true because it is accepted as true." Surely, given what we have seen so far, it is much more reasonable to simply hold the statement true, subject to unlikely revision.

31. Ibid, p. 56. 32. Ibid, pp. 68-69. 33. Ibid., p. 60.



What Putnam has apparently pointed out is an important distinction between purportedly analytic statements forming part of a larger systemwhere the relevant concept has systematic import-and purportedly analytic statements which lack such systematic importance. But it is not clear that this distinction will effectively support even Putnam's trivial notion of analyticity. Suppose that Putnam is correct in his claim that there are no exceptionless laws including the word 'bachelor'. It does not follow that "we can only say that 'bachelor' means 'unmarried man'." On the contrary, it can be plausibly maintained that ordinary speakers of English have a quite elaborate stereotype associated with the word 'bachelor'. For example, bachelors are often suitors, and they are commonly quite active socially. Moreover, bachelors are genuinely available for marriage and for dating "with good intentions." This is true even of "confirmed bachelors." Though perhaps not very interested, they are, in spite of that, still genuinely available. To say of someone that he is a bachelor is commonly a way of indicating that person's availability, and this fact, combined with social taboos against infidelity and aversions to promiscuity, goes a long way toward explaining the resistance to revision shown by the statement 'All bachelors are unmarried.' (In places, Putnam says that 'bachelor' means 'a man who has never been married,' but it is doubtful that such a definition accords very well with American usage-except in areas which are very conservative concerning the sanctity of marriage. Elsewhere, once divorced a man may again claim his bachelorhood-without really misrepresenting himself. To use Wittgenstein's phrase, the word 'bachelor' is enmeshed in an elaborate language game, associated with courtship practices, pairing, moral sensibilities, and assurances of genuine availability. Thus, to imagine a revision of 'All bachelors are unmarried' is to imagine some significant change in the associated social practices. Conservatism regarding 'All bachelors are unmarried' is conditioned by that conservatism of mores and morals which tells us that only the unmarried are genuinely available. The conclusion I want to draw from these considerations is that 'bachelor' is not a 'one criterion word' any more than 'energy' is. The word 'bachelor' also has systematic import, although in this case, we are concerned with a framework of moral notions and social practices. Even though this framework is not of the scientific sort which Putnam's papers lead one to expect or to look for, there does appear to be a framework which is highly relevant to the meaning of 'bachelor' in English. Once having identified this framework we come into a position of being able to imagine an alterna-



tive-where 'All bachelors are unmarried' might be revised. Gilbert Harman has remarked, "in this era of unstable marriages there are many bachelors who are still technically married." 34 Where conservatism concerning marriage appears a lost cause, we might expect bachelors to be identified by their behavior and their availability. This conclusion casts serious doubt on Putnam's notion of "one criterion words." Where there is serious resistance to revision (Bordering, apparently, on immunity from revision) we have every reason to expect that such resistance to revision on the part of a statement is due to the role the statement has in some framework or other. But, if so, a change in the relevant framework may recommend considerably less conservatism regarding the statement. Putnam has not, then, demonstrated that there are cases where it is reasonable to hold a statement totally immune from revision.

4. Semantics and the Contextually a priori Putnam's defense of the analytic-synthetic distinction flounders on the difficulties of defending absolute immunity from revision-a kind of difficulty Putnam has himself eloquently helped to demonstrate. What remains, then, is a relative distinction. There are some statements we can quite easily imagine revising, and others that we can imagine revising only with great difficulty. Such characterizations are viable only in relation to a given state of knowledge and theory development. In general terms, since we cannot predict with any confidence how knowledge and theory will develop (or how social practices will change), the grounds for insisting on absolute immunity from revision appear entirely too slim. In more recent papers, Putnam has apparently come to recognize this point. In "Realism and Reason," published only shortly after "The Meaning of 'Meaning'," Putnam speaks of absolute unrevisability as a mere idealization: In the foregoing, I used the idea of an absolutely 'unrevisable' truth as an idealization. Of course, I agree with Quine that this is an unattainable limit. Any statement can be 'revised'. But what is often overlooked, although Quine stresses it again and again, is that the revisability of the laws

34. Harman, Gilbert 1973, Thought, p. I 05.



of Euclid's geometry, or the laws of classical logic, does not make them mere 'empirical' statements. This is why I call them contextually a priori. 35

To speak of a given statement as "contextually a priori," we must assume, is to comment on the relative difficulty of revising the statementrelative to some assumed context. Since Putnam here agrees with Quine that absolute unrevisability is unattainable, obviously then, he has surrendered his notion of "trivial analyticity" as found in his papers from the sixties and seventies-through "The Meaning of 'Meaning'." Putnam has come to realize that the category of 'one criterion words" has too many problems to support the notion of unrevisable truth. He has moved on to a more thoroughgoing fallibilism. Even in "The Meaning of 'Meaning'," there are signs of this development. "It seems," he said in that paper, "that there is a strong tendency for words which are introduced as 'one criterion words' to develop a 'natural kind' sense, with all the concomitant rigidity and indexicality." 36 Words having a "natural kind" sense, on Putnam's view, are paradigms of words not synonymous with one specific description, and which do not enter into analytic statements. Thus his recognition of a tendency for "one criterion words" to develop into multi-criteria words is a step in the direction defended above-a step toward rejecting the notion of analyticity. The crucial step was taken, however, where Putnam's theory of word meaning cut the traditional links between meaning on the one hand (stereotype) and truth on the other. Since he rejects the assumption that stereotype (or speaker's competence) determines extension, the notion of analyticity, according to which meaning determines truth, could only be a vestige. Having rejected the notion of the a priori, so that even "necessary truth" is not known a priori for Putnam, there is little alternative to the final rejection of the notion of analyticity. The notion of necessary truth also seems to be a mere vestige of classical modem epistemology. It has no genuine role to play. If ascription of necessary truth to a statement is dependent on a specific context of knowledge and theory, and we cannot be certain exactly how knowledge and theory will develop, then there is little point to retaining the notion of necessary truth at all. For necessary truth could only be ascribed to a statement subject to revision. 37 35. Putnam, Hilary 1976, "Realism and Reason," Presidential Address to the Eastern Division of the American Philosophical Association, p. 138. 36. Putnam 1975, p. 244. 37. Interestingly, there is no discussion of"necessary truth" in Putnam's Meaning and the Moral Sciences 1978. But Putnam is at pains there to distinguish his



Given this point, contemporary philosophy seems on the way to a genuinely post-Quinean theory of meaning-meaning without analyticity or necessity. In Putnam's terms we can say that the theory of meaning is a branch of empirical linguistic investigation which explores the "contextually a priori."

realism from "metaphysical realism"-and this cnt1c1sm of "metaphysical realism" might also be read as a criticism of Kripke's notion of "metaphysical necessity."


Theories of linguistic meaning have been a major influence in twentieth century philosophy. This is due, in part, to the assumption that meaning is the crucial and interesting thing about language. To know the meaning of an expression is to understand it, and since understanding is central to philosophy in many different ways, it should be no surprise that the notion of meaning has often taken center stage. The aim of this paper is to briefly explore some influential views concerning linguistic meaning. The final objective will be to demonstrate some alternatives which are open to theory with respect to this notion----forthere are those who have wanted to ban talk of meaning from serious scientific discourse. 1 The point is that many of the disadvantages of traditional notions of meaning are avoidable-in particular, they are avoidable along a path which starts from Frege and moves on via Tarski and Davidson.

1. Frege's Argument from Identity Frege's paper, "On Sense and Reference," 2 opens with two puzzles concerning identity. First he asks whether identity is a relation between objects or a relation between signs of objects. In his earlier work Frege had held that a sentence such as 'a=b' expresses a relation between the two signs 'a' and 'b'. Now he rejects that view. If' a=b' expressed a relation between the two signs 'a' and 'b', then it could only mean that these two signs name the same object. This cannot be the correct analysis, Frege argues, because the fact that

The background of this paper is the second chapter of my Context for Meaning and Analysis 1993. The themes are up-dated and expanded in light of more recent contributions. I. Quine, W.V. 1960, Word and Object, p. 206, wrote, regarding the prospect of meanings as identical propositions that "The very question of conditions for identity of propositions presents not so much an unsolved problem as a mistaken ideal." 2. Frege, Gottlob 1892, "Uber Sinn und Bedeutung," translated as "On Sense and Reference," in Geach, Peter and Max Black (eds.) 3rd. Ed. 1980, Translations from the Philosophical Writings of Gottlob Frege, pp. 56-78. *



'a' and 'b' name the same object is the result of purely arbitrary conventions--a result of mere stipulation. This point demonstrates the error of the analysis, Frege holds, because a sentence of the form 'a=b' may convey "actual knowledge." For example, the discovery that the morning star is identical with the evening star was an empirical discovery. The sentence tells us something about the heavens and not merely something about the words. While Frege's conclusion here is correct, it is of some importance to see that his argument is faulty or at least questionable. The conclusion is correct simply because sentences such as 'The morning star=the evening star' are not claims about linguistic expressions. As Frege emphasizes, this is a claim about a certain observable astronomical object: an object which, as it happens is the planet Venus. Still, there is a flaw in Frege's argument. The astronomical sentence 'The morning star=the evening star' is true just in case a corresponding sentence about the names is true. Generally, if a=b, then 'a' is a name of b and 'b' is a name of a. Likewise, if' a' is a name of b and 'b' is a name of a, then a=b. Yet, in his argument, Frege assumes both that a sentence of the form 'a=b' may be informative and that the corresponding sentence about the names, perhaps "There is something x which 'a' and 'b' both name", must be trivial or stipulative. My point is that the sentence which is about the names is demonstrably no less trivial than the corresponding sentence of the form 'a=b'. The mistake come of thinking that semantic claims---regarding the relationship between names and objects named, for instance, must be trivial because they depend upon linguistic conventions. Put into general terms, this assumption is simply false. It overlooks the empirical element in semantic claims. To know that two names have been given to an object, we must know that it is the same object which has been named twice, but recognizing that we have the same object again sometimes involves considerable empirical uncertainty-as Frege himself emphasizes. Frege's oversight on this point will prove to be of some considerable importance. Frege's second puzzle concerning identity is the more important of the two, and the extensive influence of "On Sense and Reference" is closely tied to the solution Frege offers to this puzzle. This second puzzle depends upon the fact that there are two sorts of identity sentences. Those of the form 'a=b' are informative, and may represent significant empirical discoveries, those of the form 'a=a', on the other hand, are not so informative. Frege remarks that on Kant's view such sentences count as analytic. The problem is to explain this important difference between the two forms of identity sentences. Frege argues that we cannot explain the difference between these two forms of identity sentences unless his distinction between sense (or meaning)



and reference is brought into the account. Thus, suppose contrary to Frege's conclusion, that the meaning of a name just is its reference, that is, the object named. Given this assumption, if' a=b' is true, then 'a' and 'b' have the same meaning, since they do have the same reference. But if 'a' and 'b' have the same meaning, then 'a=b' must be as trivial as 'a=a'. This conclusion is obviously false, and in order to avoid it, Frege urges, we must distinguish between the reference of a name and its sense or meaning. According to Frege, the sense of an expression is "grasped by everybody who is sufficiently familiar with the language." The sense contains the "mode of presentation"-the way in which the sign gives us the object.3 Though having the same object as referent, two signs may still differ in sense, and it is only because of this that a sentences such as 'The morning star=the evening star' conveys "actual knowledge." Frege is surely correct to insist that sentences of the form 'a=b' are more informative than are sentences of the form 'a=a'. Moreover, he is also correct to reject the identification of meaning and referent. Still, these points do not amount to an acceptance of Frege's notion of sense or meaning. One reaction to Frege's argument from identity might be to reject the identification of meaning and referent, simply on grounds of a rejection of the notion of meaning in toto --given any traditional understanding of "meaning." This is a kind of view which W. V. Quine has often seemed to foreshadow in his arguments against traditional conceptions of linguistic meaning. It might be held, then, that traditional notions of meaning--such as Frege's notion of sense, have no place in properly scientific theory.4 Certainly, the problems connected with traditional conceptions of linguistic meaning are notorious. Yet, on the other hand, it is difficult not to appreciate the great influence of Frege's argument from identity, and it is difficult not to see considerable significance in it. We will return to the argument from identity in the following section. A closer look will be required in order to sensibly evaluate Frege's conclusions. It is perhaps sufficient for present, however, to see that Frege's argument, though extremely plausible, is not an unquestionable deductive demonstration. Rather, Frege offers an explanation of the difference between two sorts of identity sentences. He does not show that no alternative explanation is possible, nor does he show that no alternative explanation is preferable. This will be the point of departure for an evaluation of Frege's notion of sense, or meaning, and the argument from identity.

3. Frege 1892, "On Sense and Reference," p. 57. 4. Cf. Quine 1960, Word and Object, p. 206.



However, it will be important to fust see something of how Frege developed the sense-reference distinction. In order to properly evaluate Frege's distinction, one needs to know a bit more about it.

2. Frege on Sense and Reference It is a serious mistake, Frege holds, in The Foundations of Arithmetic, to ask

after the meaning of a word in isolation. Rather, "it is only in the context of a sentence that a word has a meaning."5 It seems that a version of this dictum was subsequently applied both to sense and to reference. Under certain circumstances--in certain sorts of sentences--a word or other expression does not have its customary reference. Frege distinguishes the customary reference of an expression from its "indirect reference," and he equates the indirect reference ofan expression with its customary sense. For example, the customary reference of 'the morning star' is the morning star, i.e., the planet Venus, and we may think of the customary sense of this expression as its usual meaning. A sense, or meaning, is for Frege a special sort of abstract entity, which is "grasped," when the corresponding expression is understood. In certain linguistic contexts, then, Frege's view is that 'the morning star' takes its indirect reference, it does not refer to the planet Venus, but instead refers to its own customary sense. These distinctions are employed in Frege's attempt to explain apparent failures of Leibniz' law. Given Leibniz' law, logicians have adopted the principle that expressions having the same reference may be substituted for each other salve veritate, i.e., preserving truth. Clearly, if x=y, then whatever is true of x is also true of y and vice versa, since x and y are the very same object. The way in which an object is referred to makes no difference to the truth or falsity of a sentence involving reference to that object. Frege's aim is to show that apparent counter-examples to this mode of inference can be explained-without amending the logical principle. Problems arise in so called "intensional contexts," for instance, linguistic contexts created by modal expressions such as 'necessarily' and 'possibly' and also in contexts created by verbs of "propositional attitudes" such as 'believes', 'knows', 'desires' and other psychological verbs. Consider the following sort of argument: (I) John believes that the morning star is a planet. (2) The morning star=the evening star.

5. Frege, Gottlob 1953, The Foundations of Arithmetic, p. x.



(3) John believes that the evening star is a planet.

The logician's principle seems to fail here, for assuming that (1) and (2) are true, (3) may yet be false. For example, John may not believe premise (2), and may believe instead that the evening star is actually a star, and not a planet at all. But, even if John does not believe premise (2), the sentence is true, and this is the point. The logical principle seems to fail because though (1) and (2) are true, the conclusion (3) may still be false. How can this be? For Frege, there is no genuine failure of the substitutivity principle. He maintains that co-referential expressions may safely be substituted for each other; and he maintains, further, that in the problematic cases such as the argument above, the substituted expressions are not genuinely co-referential. We cannot, on Frege's view, substitute 'the evening star' for 'the morning star' in (1 )--because, in the context created by 'believes', the phrase 'the morning star', does not have its customary reference, and hence is not co-referential with the phrase 'the evening star'. Instead, in such intensional or non-extensional contexts (as they have subsequently come to be called) the relevant expression refers to its own customary sense, that is, in Fregean terms, it takes its "indirect reference." Because 'the morning star' does not refer to the morning star in the context created by 'John believes that...', it is a misapplication of the substitutivity principle to bring in 'the evening star' in the conclusion (3). On Frege's view, there is a special ambiguity which affects the expression 'the morning star' in the case considered, and this is the source of the problem as he sees it. If someone argues from the premises 'Smith put his money in a bank' and 'A bank is the side of a river' to the conclusion that 'Smith put his money in the side of a river', we will not wonder long at the failure of the argument. Assuming the premises are both true, the conclusion does not follow, because 'bank' has neither the same meaning or reference in the two premises. Likewise, Frege argues, we cannot substitute 'the evening star' for 'the morning star' in 'John believe that the morning star is a planet' because the phrase 'the morning star' does not have the same meaning or reference in the first premise as it has in the second premise, i.e., in the identity sentence 'The morning star=the evening star'. According to Frege, again, in the context created by the expression 'John believes that...' 'the morning star' refers to its own customary sense and not its customary reference. This account of the problem may appear to add to the appeal of Frege's sense-reference distinction in that Frege provides further philosophical work for his notion of sense or meaning. Frege's solution to the above problem is especially appealing given the radical character of standard alternatives. It



appears that the only alternatives to adopting Frege's sense-reference distinction are either (1) to hold that the substitutivity principle allows of exceptions, or (2) to ban all intensional (and intentional) idiom from serious scientific discourse. But neither of these two alternatives has any great initial appeal; rather they seem to merely emphasize the attractions of Frege's solution given in terms of the sense-reference distinction. ln spite of all this, however, it has often been held that Frege's notion of sense is not sufficiently clear. One wants to know more, and in particular, one wants to know more about the identity conditions for Fregean senses. How can we know whether two distinct expressions have the same or different senses? Frege did not answer this crucial question, and it has proved to be the single most important problem facing the theory of meaning. It is this problem, more than any other, which has led philosophers to despair concerning the prospects for the theory of meaning. Frege also applies his sense-reference distinction to entire sentences. He calls the sense of a sentence a "thought." The German word here is 'Gedanke' which is also sometimes translated as 'proposition'. Frege' s usage might be more literally captured by a phrase such as 'that which is thought', since he is quite emphatic in his claim that these thoughts or propositions are not psychological entities. Like other senses, Frege's "thoughts" are abstract, objective entities existing independent of mind and language. Frege assumes, initially, that a thought or proposition exists corresponding to each sentence-either as the sense of the sentence or as its reference. He then argues that the thought cannot be regarded as the reference of a sentence, and consequently concludes that we must consider a thought to be the sense of a sentence. Consider, for example, the following sentence: (I) The morning star is a body illuminated by the sun.

ln accordance with Leibniz' law, we may substitute co-referential expressions, preserving truth. Substituting 'the evening star' for 'the morning star' in (1), the result is (2). (2) The evening star is a body illuminated by the sun.

Now, (2) differs from (1) only by the substitution of co-referential expressions. Therefore, Frege argues, the substitution could have no effect upon the reference of the two sentences. If we assume, with Frege, that a sentence has a reference, then in view of the relationship between (1) and (2), they could only have the same reference. Yet, (I) and (2) express different thoughts. A



person might believe (1) without believing (2) for instance. In consequence, Frege concludes that a thought cannot be regarded as the reference of a sentence and must be considered the sense of a sentence. Frege holds that the reference of a sentence is its truth value, and he also conceives of the truth values as abstract entities: a sentence either refers to the True or to the False. Although this view may seem quite artificial, it is in spite of that dictated in part by the above argument. Since the reference of a sentence must remain constant under the substitution of co-referential expressions, and by Leibniz' law, truth value does remain constant, Frege identifies the reference of a sentence with its truth value. Further support for Frege's notion of the truth values as referents of sentences is derived from his analysis of expressions into functional expressions and object expressions or names. Given a mathematical fonnula such as x+3=y, the value of y is said to be a function of x. Substituting '2' for 'x' , 2 is said to be an argument of the function, and each argument, 2, 3, 4, and so on, detennines a value of y. For example, if x=2, then y=(2+3)=5. In a similar way, Frege views sentences as being composed of function names and argument names. For example, we can think of 'Tom runs' as composed of the predicate '_ runs' (the name of a function), and the object name 'Tom'. Other names could equally well fill the blank space in '_runs' and each such name Frege regards as providing an argument to the function. To follow through the mathematical analogy, each such argument must detennine a value of the function. Obviously, such values are not numbers. Still, like the values of mathematical functions, Frege regards the values of sentential functions as objects-the two truth values. For each argument, the function will detennine a value of the function: either the True or the False. Although content to avoid Frege's view of truth values as objects, the subsequent philosophical tradition has generally placed a very high value upon Frege's analysis of sentences into function expressions and argument expressions. Now, to return to the question of identity conditions of senses, if Fregean senses are to be entities of some sort, we want to know what is to count as one sense and what is to count as more than one. The problem can also be put in slightly different tenns: if we are going to have a theory of meaning at all, then such a theory must tell us what it is for two expressions to have the same or different meanings. Neither Frege, nor the subsequent philosophical tradition, has provided any fully adequate answer to such questions, and as remarked above, this is the chief reason that many philosophers, Quine included, have sought to avoid employment ofFrege's theory of senses or any



similar theory. Lacking a clear account of sameness and difference of meaning, we are likely to get into conceptual difficulties on the assumption that two expressions have the same meaning or that they have different meanings. In order to develop this point adequately, it will be important to return to Frege's argument from identity. For, Frege's sense-reference distinction was introduced into philosophy by means of this argument, and the plausibility of the distinction has often been made to rest upon Frege's proposed solution to the puzzle concerning the two forms of identity sentences.

3. Modes of Presentation The argument from identity, as we saw, offers Frege's sense-reference distinction in explanation of the difference between the two forms of identity sentences: 'a=a' and 'a=b'. The latter form is the more informative, providing as Frege has it, "very valuable extensions of our knowledge," which "cannot always be established a priori." 6 Sentences of the form 'a=a' are, obviously, not so informative. Frege is certainly correct to point out this difference which any semantic theory must account for in some fashion. The problem, then, is how best to account for the difference. Interestingly, Frege cannot say-at the outset-that sentences like 'a=a' and 'a=b' differ in meaning or sense. This is his conclusion, but he cannot characterize the difference is this fashion initially without begging the question in favor of the sense-reference distinction, which he wants to introduce. Since Frege seeks to demonstrate the need for his notion of sense or meaning, he can only observe, at the start, that the two forms of sentences differ in their informative value. This much is beyond reasonable dispute. Since Frege's sense-reference distinction has lately looked so appealing, it is worth taking notice of the fact that a semantic theory invoking Frege's notion of sense also has some quite unattractive features. Frege's theory of sense is by no means a simple theory, and its complexities involve considerable ontological expenses. To claim that 'a=b' is more informative than 'a=a' because of a difference in sense between the names 'a' and 'b' involves acceptance of Frege's ontology of senses. Even if we restrict ourselves to consideration of the senses of proper names, the ontological proliferation is quite considerable. First, as we have already seen, Frege's theory invokes for every name, such as 'Venus' a corresponding sense. In addition, such a name, will, in intensional contexts, refer to its ordinary sense 6. Frege 1892, "On Sense and Reference," p. 56.



and thus take on a second sense-its "indirect sense." Further, the same word can be used to refer to its indirect sense, in which case, Frege hold that the word takes on a third sense. This will be so in iterated intensional contexts such as we find in the following sentence: Jones believes that Smith believes that Venus is a planet.

The sense of 'Venus' in this sentence is distinct from the sense it takes in 'Smith believes that Venus is a planet', which is distinct in turn from the sense the word has in 'Venus is a planet'. Since there is no end, in principle, to the complexity of such iterated intensional contexts, it follows that on Frege's view there is no end to the number of distinct senses a single name has. 7 What seems desirable, then, is to solve Frege's puzzle in a more economical way, and without introducing problems concerning sameness and difference of sense or meaning. In fact, this can be done, without need to introduce the sense-reference distinction at all, in terms of the wellunderstood concept of logical implication. As will be argued, this points the way toward a replacement for the traditional notion of linguistic meaning. First, it should be noted that there are sentences which logically follow from 'a=b' but which do not follow from 'a=a'. Thus, if' a=b' is true, then so are the sentences '(:lx)(a=x)'-something is identical with a-, '(:lx) (x=b)'-something is identical with b---, and '(:lx)(a=x & x=b)'-there is something which is identical with both a and b. But the latter two sentences do not follow from 'a=a'. Thus, in terms of these and further differences in the logical implications of the two sorts of sentences we can understand how sentences of the form 'a=b' are more informative. Further similar points arise if we consider the logical implications of the two sorts of sentences taken in conjunction with other sentences. Indeed, it is this sort of case which most easily explains the intuitive sense in which we want to say that the one sort of sentence is more informative. Thus, a sentence of the form 'a=b' is more informative than one of the form 'a=a' in that 'a=b & Fa' logically implies 'Fb', while 'a=a & Fa' does not logically imply 'Fb'. Given the substitutivity principle and 'a=b', whatever we know about a, for instance that Fa or that Rae, we get a further interesting implication concerning b, e.g., Fb or Rbc. But given 'a=a' in place of 'a=b' such logical implications are not forthcoming. Sentences of the form 'a=b' force rearrangements of our pre-existing knowledge in ways that the

7. Cf. Linsky, Leonard I 967, Referring, p. 46.



corresponding sentences of the form 'a=a' do not. Once such points are noted, we have an explanation of Frege's puzzle concerning the two sorts of identity sentences drawing upon the notion of logical implication. Frege identifies the sense of a name with its "mode of presentation," the manner in which a name represents its referent. In his argument from identity, the central contention is that we cannot identify the mode of presentation of a name with the mode of presentation of another, co-referential, name simply on the ground that the two names are co-referential. For, if we do so, then "the cognitive value of 'a=a' becomes essentially equal to that of 'a=b' provided that 'a=b' is true." 8 What is required then, in order to avoid such a conclusion is an alternative conception of "mode of presentation"----one we can make use of to deepen our explanation of the fact that the two sorts of identity sentences differ in informative value. The central requirement which Frege lays down is that the mode of presentation of 'a' not be the same as the mode of presentation of' b' simply because 'a=b' is true. Yet, this requirement is quite easily met without adopting the specifics ofFrege's theory of sense. Thus, let the mode of presentation of 'a' be a function of all the sentences we accept (at a given time) including 'a' and the mode of presentation of 'b' be a function of all the sentences we accept including 'b'. Now, since two names can be given to the same object without anyone knowing or believing that this is so (i.e., contrary to Frege's implicit assumption when concerned with his first puzzle concerning identity, "(:3.x)('a' names x & 'b' names x)" may be true without being trivial or obvious), in view of this, it follows that the modes of presentation of 'a' and 'b' are not the same, on this approach, simply in virtue of the fact that 'a=b' is true. Instead, on this approach, the modes of presentation of two names, 'a' and 'b' are the same only if, at a given time, we accept a sentence including 'a' for every corresponding sentence we accept with 'b' for 'a ' and vice versa. Something quite similar to this alternative conception of mode of presentation is already implicit in the standard semantics of logical theory. (This, by the way, derives in some large part from Frege's own work, but also from that of Tarski.) For instance, the standard semantics of logical theory will show that (1) below is a logical truth while (2) is not. (1) (Fa & a=b) only ifFb. (2) (Fa & a=a) only ifFb.

8. Frege 1892, "On Sense and Reference," p. 57.



The manner in which 'b' must be interpreted in (1) in order that the antecedent 'Fa & a=b' turns out true, rules out the negation '~Fb'. Thus 'Fb' is true on any interpretation which renders 'Fa & a=b' true-{l) is a logical truth. However, there are interpretations of 'b' consistent with the truth of 'Fa & a=a' on which 'Fb' is false-that is (2) is not a logical truth. The notion of interpretation in standard semantics can thus be looked upon as an explication or reconstruction of Frege's notion of mode of presentationwhere the mode of presentation, or interpretation, of a name is a function of interpretations given to a set of sentences (the premises of an argument or the axioms of a theory), employing that name, treated as though they are true. The difference between (1) and (2) has little to do with whether 'a=b' is actually true. Rather the difference arises as a matter of the different tentative assumptions we make in trying to decide if sentences such as (1) or (2) are logical truths. Given 'a=b' as a premise (or tentative assumption) the interpretations of' a' and 'b' must be the same. However, given only 'a=a' as an assumption, the interpretations of 'a' and 'b' are allowed to differ. It is not sameness and difference of reference which is crucial to logical validity, but rather sameness and difference of purported reference; and this is something which is capable of strict defmition within the confines of standard semantics.9 The conclusion is, then, that Frege's puzzle concerning the two sorts of identity sentences in open to a solution which does not invoke Frege's ontology of senses. The sense-reference distinction can be avoided in favor of an account of the puzzle which proceeds in terms of the well-understood notions of standard referential semantics. This point serves to return our attention to the general problem presented by Frege's semantics of sense and reference-the problem of the identity conditions of meanings or senses.

4. Meaning and Identity The point is that meanings or senses, on traditional accounts, like attributes and unlike sets, lack clear conditions of identity. Frege gives us little help with this problem, since he provides few examples of different expressions having the same sense; no examples at all are provided in "On Sense and Reference." Rather, Frege's notion of meaning, is tied to the analyticsynthetic distinction, and he attempts to give a more rigorous account of the distinction between meaning and reference without any fundamental depar-

9. Cf. Callaway, H.G. 1982, "Sense, Reference, and Purported Reference."



ture from traditional modem epistemological classifications of sentences as empirical or a priori. 10 The general character of this problem of a lack of clarity, concerning identity conditions for meanings, can be illustrated within the confines of Frege's semantics by considering, once again, arguments employing the substitutivity principle. For, while Frege's theory attempts to deal with problems involving the substitutivity of co-referential expressions within intensional contexts, such problems reappear within Frege's theory. Thus, premise (2) below is a consequence of Frege' s use of the sense-reference distinction to explain apparent failures of the substitutivity principle. (1) John believesthat Venus is a planet. (2) The referenceof 'Venus' in (I )=the customarysense of the word 'Venus'. Therefore, (3) John believethat the customarysense of the word 'Venus' is a planet. Obviously, sameness or difference of sense remains problematic here. Given (1) as an assumption, (3) is a consequence ofFrege's semantics of sense and reference. The problem is that (3) appears to be either false or nonsense. Problems associated with identity of meanings continue to haunt Frege's theory and all subsequent reconstructions of it. In view of this, and the possibility of avoiding the sense-reference distinction, as argued above, I conclude that Frege's notion of sense is best dispensed with entirely. The problem of substitutivity in "intensional" contexts is not solved by Frege's theory, and it has, of course, continued to attract much interest. The problem presented by Frege's arguments in "On Sense and Reference," has been the difficulty of imagining how we can do without the sense-reference distinction. Once this begins to become clear, there is every reason to tum away from traditional notions of meaning and look for substitutes. Arguments presented in the prior two section foreshadow the direction in which to look for such substitutes. 11 For, we found that it is possible to provide a solution to Frege's puzzle concerning identity sentences within the confines of the standard referential semantics of logical theory. The chief point in this is that the notion of interpretation available within

10. Cf. Resnick, Michael 1980, Frege and the Philosophy of Mathematics, pp. 179-80. 11. Cf. also Thiel, Christian 1965,Sense and Reference in Frege 's Logic, pp. 160-61 for a brief account of Frege's unpublishedview on identity of sense in terms of logicalequivalenceof sentences.



standard referential semantics allows for a paraphrase or reconstruction of Frege's notion of the mode of presentation ofan expression. Such recourse to standard referential semantics also serves to direct attention toward Davidson' s efforts to reconstruct the theory of meaning within the Tarskian theory of referential notions. A theory of meaning within the theory of reference is not, by that token a "referential" theory of meaning. 12 Such attempts are not without their own problems. However, in the contextualization of the interpretations of expressions to object-language theory, there is some hope of a notion of meaning or interpretation keyed to clear conditions of identity, viz., the identity conditions of theories viewed as deductively ordered syntactic systems. The point may be expressed by a development of Frege's "context principle:" for it is only in the context of a theory that expressions of a language are given interpretation in standard referential semantics.

5. Davidson on Truth and 'Inscrutability' of Reference A theory of meaning, or the aim of elaborating a theory of meaning, given in terms of the vocabulary of the Tarskian theory of referential notions has claims upon the realist tradition. The point is evident in Tarski's work, since he put forward his semantic conception of truth as continuous with the classical Aristotelian-Platonic conception of truth, according to which "To say of what is that it is not, or of what is not that it is, is false, while to say of what is that it is, and of what is not that it is not, is true." 13 This is the conception of truth which came to be called the correspondence theory. In the scholastic tradition, the Aristotelian doctrine is expressed in the formula that truth is "adequatio rei et intellectus" (the "adequation" or adequateness of thought to thing), and the "adequateness" is then expressed as a matter of "correspondence." But emphasis upon "correspondence," has often produced more questions than answers concerning what "correspondence" means, and what the things are to which statements are to correspond. 14

12. Cf. e.g., Davidson, Donald 1967, "Truth and Meaning." 13. Aristotle, Metaphysics r, 7, 27; and Cf. Tarski 1956, "The Concept of Truth in Formalized Languages." 14. See in this connection Callaway, H.G. 1993, Context for Meaning and Analysis, Chapter Three, "Meaning and Analysis," and especially §12 on the struggles of Bertrand Russell, early in the last century, attempting to elucidate the "ret' as a matter of "facts," as special entities, to which statements are to correspond.



In an important sense, Tarski's work goes around these traditional questions and puzzles. Aristotle speaks of "saying," and this may be interpreted by reference to "things said," statements or sentences. From that perspective, Tarski's innovation may be viewed as a matter of considering all the distinct forms of statements or sentences, in particular formalized languages, and then recursively defming truth by reference to all these types of "sayings." As a first step of paraphrase of the Aristotelian dictum, one might well be tempted to eliminate the reference to "what is not." Doing so, and putting 'true' in first position, we get something like the following: "To say of what is, that it is what it is (has the character it does), that is truth, and to say of what is that it is not what it is, that is falsity." Next we need to consider the various ways in which we might say of what is, that it is what it is (or fail to do so). This step Tarski accomplishes by reference to the rules of formation of the formalized languages of his concern. Though technical in its development, the basic idea is simple. So, if we suppose that we have a sentence 'a is an F', then this sentence is true if and only if the object named by 'a' is among the things to which 'F' applies. Given that the language in question has a fmite vocabulary and clear rules of formation which allow us to trace the truth conditions of complex or compound sentences (i.e., truth-functional and quantificational compounds) back to the truth conditions of simple sentences, then this allows us to defme 'true' for the language in question, by recursively specifying conditions under which each sentence of the language is true. Such a definition of truth for a language is not offered as a means of telling or fmding out whether the particular sentences are true, except as this is facilitated by making clear their logical relations. For instance, if we are told that a disjunction 'Fa or Fb' is true, if and only if one or both of the disjuncts 'Fa' or 'F b' is true, then we only know whether the compound or disjunction is true if we already know whether one or both of the disjuncts are true. Likewise, if we are told that a conditional sentence 'Fa only ifFb' is true if and only ifit is not the case that the antecedent is true and the consequent false, then evaluating the conditional sentence is made to depend on some means of establishing the truthvalues of the components. At the base of the recursion, we have it that sentences such as 'Fa' are true if and only if what is named by 'a' is included in the denotation of the predicate 'F', but the truth defmition itself does not tell us how to establish whether this is so. All of this is, instead, very much in the spirit of the Aristotelian original. We may safely agree that "to say of what is that it is, is true," but this does not even appear to carry the intention of telling us how to find out what is or what exists. Similarly, with Tarski, 'true-in-L' is defined for a particular language by providing a statement of



truth conditions for each sentence of the language, based on a semantic characterization of the finite vocabulary of the language. The statements of truth conditions take the form of the Tarskian T-sentences. If we are then asked about truth in general, we are driven back to something like the general Aristotelian dictum, which makes no mention of particular languages. Davidson has correctly emphasized that Tarski's approach to the definition of 'true' for a language, proceeding as it does in a meta-language rich enough to include the means of referring to all the expressions of the object language, and to the objects of reference of the object language, succeeds only because it implicitly draws on a concept of translation. For instance, we can say in the meta-language that a sentence 'Fa' of the object language is true if and only if the object named by 'a' is included in the denotation of the predicate 'F', only if we assume that expressions of the meta-language have the same meaning and/or reference as expressions of the object language. The point is most clearly evident in the Tarskian Tsentences, such as, 'Snow is white' is true-in-L if and only if snow is white.

Tarski assumed, as a condition of adequacy of a theory of truth for the language L, that the theory be able to generate a similar statement of truth conditions for each sentence of the object language. Yet this sort of Tsentence only retains its plausibility, as a statement of truth conditions, if we also assume that the sentence mentioned on the left is semantically indistinguishable from the sentence used on the right to give the specific truthconditions of the sentence mentioned. Likewise, we are assuming that, say, 'snow' in the object language is semantically indistinguishable from 'snow' in the meta-language. In short, the formalism of Tarski's approach to truth silently rests on a homophonic translation between the object language and the meta-language. Notice, too, that it is not merely sameness ofreference which is of interest for expressions of the meta-language and expressions of the object language. For suppose that snow just is crystals of sublimated water vapor, though the phrase 'crystals of sublimated water vapor' is not included in the vocabulary of L. Given these assumptions, if 'crystals of sublimated water vapor' is included in the meta-language, then we have an apparent choice of statements of truth conditions for expressions ofL: (I) (2)

'Snow is white' is true-in-L if and only if snow is white. 'Snow is white' is true-in-L if and only if crystals of sublimated water vapor are white.



Tarski's condition of adequacy for a theory of truth requires that (1) but not (2) be generated by the truth-theory for L. This helps make the point that a Tarskian truth theory for a language L is specifically oriented to, and depends upon, the (homophonic) semantic equivalence of expressions in the object language and expressions in the meta-language. The Tarskian formalism of a truth theory for L has no means of detecting empirical sameness of reference of differing expressions. The truth of (2) is something we know only on the basis of a sophisticated scientific theory about snow, and there is nothing to guarantee that this knowledge is present or expressed in the object language which we may wish to interpret. 15 Making use of the Tarski-style formalism for the purposes of an empirical theory of meaning, Davidson must reverse the order of derivation so that starting from empirically certified translations between particular sentences of an object language and a meta-language, stated as T-sentences, we can work our way back to a semantic characterization of the basic vocabulary of the object language from which the certified T-sentences can be generated. The semantic characterization of the vocabulary of the language under interpretation then functions to explain the certified T-sentences. Setting aside, for present purposes, the question of how we can establish T-sentences on an empirical basis, and without relying upon a homophonic translation between object language and meta-language, I want to focus on the consistency between this approach to meaning and Davidson's endorsement of the Quinean doctrine of inscrutability of reference. The central idea to be emphasize is that it is possible to elucidate the "mode of presentation" of an item of vocabulary, to use the Fregean expression, by reference to a theory which makes use of the expression: that this is implicated in the Tarskian conception of interpretation. Since, for instance, different theories, making use of the same vocabulary and syntax, may differ on the truth-value of sentences such as 'a=b', the appropriate interpretations of the names 'a' and 'b' may differ from theory to theory-that is, some particular theory in the object language may be regarded as entering into the account of the mode of presentation of the names. Much the same can be said about pairs of predicates, 'F' and 'G' since the Tarskian interpretation of the predicates, suited to elaborate the relation of logical implication among sentences of the theory, will differ depending on whether or not the given theory contains sentences such as '(x)(Fx only if Gx)', '(x)(Gx only if Fx)' and '(x)(Fx if and only if Gx)'. The point is not that such interpretations,

15. See the related discussionin my "SemanticCompetenceand Truth Conditional Semantics."



differing from theory to theory, or the particular kinds of conditionals and identity statements recently mentioned, give the meaning of the relevant expressions in an intuitive sense. It is more that they need to be considered, since they constrain the meaning of particular expressions, or help define the appropriate usage of the expressions, from the standpoint of the theory. Intuitively, the meaning of the relevant expressions sums up or explains the relevant usage, and usage is something we may hope to ascertain on an empirical basis. By focusing on a particular theory in the language under interpretation, and its interpretation, as contrasted with the language, taken as neutral between all alternative theories, we limit the range of alternatives regarding the meaning associated with a particular expression of the theory. We look to usage as an expression and indication of meaning, in the sense that usage may be summarized, preliminarily, in statements of truth conditions for particular sentences, regarded as constraints on the theory (or belief-system) of the speakers. Following in the Tarskian and Davidsonian tracks, the statements of truth conditions are generalized by some particular Tarskian formalism, which characterizes the constituent vocabulary in semantic terms and is capable of generating all the attested T-sentences. Any such formalism, suited to generate the empirically attested T-sentences will also generate further T-sentences which allow for the possibility of empirical testing. All of this may seem quite congenial to Davidson's approach to meaning, and in many ways it is, but in order to bring out some differences, it will be important to bring my emphasis upon interpreting theories in a language into detailed connection with some of what Davidson has to say about inscrutability of reference. The following passage comes from Davidson's "Indeterminism and Anti-realism" (1997). Davidson explains that the thesis of the "inscrutability of reference" amounts to the claim that "there is no way to tell which way of connecting words with things is the right way." From a technical point of view, this means that for the standard satisfaction relation (satisfaction is a sophisticated form ofreference) we can substitute endless other relations without altering the truth conditions of any sentence or the logical relations among sentences. Since all the evidence for interpreting language must come at the sentential level (for only sentences have a use in communication), the result is that there can be no evidence that one of the satisfaction (or reference) relations is the right one. 16

16. Davidson, Donald 1997, "Indeterminacy and Antirealism," p. 78.



Basically Davidson here makes a claim concerning what l have been calling the Tarskian formalism, or its axioms, which characterizes the vocabulary of a language in semantic terms in a meta-language and which is suited to generate all the T-sentences. 'Satisfaction' is the technical term which Tarski employed in the semantic characterization of the vocabulary, and the intuitive idea of satisfaction as a semantic term can be thought of as a matter of predicates ( of the language under interpretation) being satisfied by objects to which they apply. So, at an intuitive or common-sense level, we can say in the meta-language, that (x) ('F' is satisfied by x if and only if Fx). Similarly, regarding a dyadic predicate 'Rxy', an ordered pair of objects, satisfies 'Rxy' if and only if Rab. Satisfaction is the basic semantic concept employed in the Tarskian formalism to characterize the vocabulary of interest. So, according to Davidson, what it means to endorse the "inscrutability of reference" is that if we can find one semantic characterization of the vocabulary suited to generate all the T-sentences for L, one way of relating words to things which will do the job, then "we can substitute endless other relations without altering the truth conditions of any sentence or the logical relations among sentences." Or, equivalently, "There is no way to tell which way of connecting words with things is the right way." Davidson is basically following Quine's doctrine, and does not depart from Quine's "Ontological Relativity," in any way which will affect the 17 account here. Davidson continues: Here is an example. Suppose satisfaction relation s maps the word 'Rome' onto Rome, and the predicate 'is a city in Italy' onto cities in Italy. Then the truth definition will show that the sentence 'Rome is a city in Italy' is true if an only if Rome is a city in Italy. Now consider another satisfaction relation s· which maps the word 'Rome' onto an area 100 miles south of Rome, and the predicate 'is a city in Italy' onto areas I 00 miles south of cities in Italy. The truth definition will now say that the sentence 'Rome is a city in Italy' is true if and only if the area 100 miles south of Rome is an area 100 miles south of a city in Italy. The truth conditions are clearly equivalent. The thesis of the inscrutability of reference contends that that there can be no evidence that s is any better than s' for interpreting the sentence 'Rome is a

17. Cf. Quine, W.V. 1969, "Ontological Relativity," e.g., p. 50. "The relativistic thesis to which we have come is this, to repeat: it makes no sense to say what the objects of a theory are, beyond saying how to interpret or reinterpret that theory in an other." But Quine does not distinguish between 'interpretation' or 're-interpretation' and paraphrase for the sake of improvements. See the discussion in Callaway 1993, Context for Meaning and Analysis, pp. 53-54.



city in Italy'. There is no telling what a sentence is 'about', or what someone is thinking about. 18

Now, as a formal or mathematical possibility, I think there can be little doubt that the kind of re-interpretation proposed here can be carried through. Instead of assigning the name 'Rome' to Rome, we could assign it to something else, and make systematic adjustments elsewhere, so that all the sentences true on one interpretation tum out true on the alternative interpretation. The chief question concerns the semantic significance or interest of this formal possibility, both in Quine and in Davidson. Davidson draws very strong conclusions here, not unlike Quine's: Neither s nor s' can be regarded as providing a more correct interpretation of 'Rome is a city in Italy,' and to the objection that "'Rome' doesn't mean 'an area 100 miles south of Rome'," Davidson replies that "individual words don't have meanings. They have a role in determining the truth conditions of sentences, and this role is captured bys' as surely as it is by s." 19 While this is perhaps an appropriate corollary of the Quinean inscrutability of reference, "ontological relativity," and indeterminacy of translation-all doctrines which Quine has supported by similar arguments-, for those who are not already committed to the "inscrutability" of reference, these conclusions may well have the look of a reductio ad absurdum. In particular, we can view the conclusion in the light of an expanded Fregean "context principle," to the effect that words don't have meaning in isolation (not merely from a sentence , but) from a theory which employs those words. The appropriate view to take of all this, as I will argue, is that Davidson has departed from the "mode of presentation" of the (common-sense belief-system or) "theory" in the object language with which these words are normally associated. (I use "scare" quotes around "theory" to indicate that when we regard common-sense systems of belief evidenced in assent to sentences, as a theory, this involves some degree of idealization.) It is just this sense of departure from the ordinary usage of words which is the reasonable source of the objection Davidson is quoted as considering above. Since he rejects the objection that "'Rome' does not mean 'an area 100 miles south of Rome'," as a basis for questioning the proposed reinterpretation of expressions of the object language, no distinction is being made between 'interpretation' and Quinean paraphrase by means of proxy function.20 18. Davidson 1997, p. 78. 19. Ibid., p. 79. 20. See Quine, W.V. 1969, "Ontological Relativity," pp. 55-62.



Although there is no formal or purely logical and mathematical objection we can make to the prospect of a truth theory which generates Tsentences by interpreting 'Rome' as mapped onto an area 100 miles south of Rome (in the usual usage of that word) and which maps 'is a city in Italy' onto areas 100 miles south of cities in Italy (again, given the usually usage of those words); and though it is true, that we might get an "equivalent" production of T-sentences, in a purely formal sense of equivalence, since we generate a T-sentence for each sentence of L, the point is that this alternative interpretation of the related vocabulary has no genuine empirical motivation or interest. It is, instead, a sort of exercise which we might better regard as purely speculative paraphrase or re-interpretation. One way to see this point is to consider what changes to the meta-language would have to be made in order to generate the alternative Tsentences suggested by Davidson. So, our alternative truth theory is to generate the following: 'Rome is a city in Italy' is true-in-L if and only if the area 100 miles south of Rome is an area I 00 miles south of a city in Italy.

This requires in the basis of the truth theory, that 'Rome' maps onto an area I 00 miles south of Rome, and generally, that 'is a city in Italy' is satisfied by areas 100 miles south of cities in Italy. The problem is not simply that this violates a homophonic translation of the terms, between object language and meta-language, since it would beg the question to insist on the homophonic translation. The more basic problem is to understand, in empirical terms, how we could ever arrive at this proposed alternative T-sentence, given the actual usage of words in the object language. There seem to be just two basic possibilities regarding the empirical evidence. Supposing, with Davidson that we understand 'is an area I 00 miles south of a city in Italy' along with 'Rome' and 'is a city in Italy,' (I) either the phrase 'is an area 100 miles south of a city in Italy' has an empirical usage in the object language under interpretation which contrasts with 'is a city in Italy' or (2) it does not. We have to consider these alternatives, because that is part of what is involved in consulting usage in order to arrive at an interpretation and because the appropriate interpretations of these words in the meta-language will reasonably differ depending upon this point of usage. Basically, we want to know whether the interpretation departs from constraints based in common usage. Notice, then, that on the first assumption (which agrees with common sense) 'is a city in Italy' contrasts in its usage with 'is an area l 00 miles south of a city in Italy'. It is not that these terms are mutually exclusive in



their usage, so that people would generally assent to 'Something is a city in Italy if and only if it is not an area 100 miles south of a city in Italy', since there might in some cases be another city 100 miles south of an Italian city we select for consideration. It is more that people will not generally agree, as we may put the point for the sake of generality, that for variable x, x is a city in Italy if and only if x is an arealO0 miles south of a city in Italysurely Italy does not always extend 100 miles south of some of its cities, (and the area 100 miles south of Siracusa is no city at all but only the open Mediterranean Sea). So, we see that on the assumption that 'is a city in Italy' and 'is an area 100 miles south of a city in Italy' have contrasting usage in the object language, we are not going to get empirical support for the proposed alternative T-sentence to illustrate the doctrine of "inscrutability of reference." The other possibility is that there is no contrasting usage in the object language between 'is a city in Italy' and 'is an area 100 miles south of a city in Italy'. This might be because 'is an area 100 miles south ofa city in Italy' cannot be formulated in the object language, perhaps because some of the vocabulary is missing. That the vocabulary is not present in the object language may be implausible in connection with actual natural languages is clear, but we are considering a formal possibility and the language under interpretation has been specified only in a very sketchy way. We could still formulate the alternative T-sentence if 'is an area 100 miles south of a city in Italy' does not occur in the object language, since it only occurs on the right of the proposed alternative T-sentence. We only need to have it in the meta-language for purposes of the T-sentence. But, then, what empirical motivation, drawing on object language usage, could we possibly have to make use of 'is an area 100 miles south of a city in Italy' to state the truth conditions of 'Rome is a city in Italy'? By hypothesis, if the phrase does not occur in the object language, then the object language usage does not directly motivate this alternative T-sentence. It must be instead that the phrase is one we have some use for in the metalanguage, but it is not easy to imagine what this might be. In a way, that is just the rub, since the phrase seems to be an imposition of the proposed alternative interpretation. Yet, on the other hand, some relevancy of the phrase to usage in the object language appears to be implied, and we see this by the fact that it is quite natural to assume that the phrase would be found in the object language-though, as we would naturally assume, with some contrast to the usage of 'is a city in Italy'.



The basic problem with this argument, then, is that Davidson depends on this plausible assumption of relevancy while also explicitly denying it, since the alternative T-sentence tells us in no uncertain terms that "'Rome is a city in Italy' is true if and only if the area 100 miles south of Rome is an area 100 miles south of a city in Italy." More generally, given the alternative truth theory generated from s', it is going to tum out that x satisfies 'is a city in Italy' if and only if x is an area 100 miles south of a city in Italy. But we are offered nothing from the object language to motivate this interpretation. The formal possibility arises only because we are offered nothing from the object language which will clearly rule out this interpretation; and that is exactly why one may justly regard all such examples as offering mere speculative paraphrase. The mere fact that we could reinterpret the relevant language in some outlandish way is no evidence that such speculative paraphrase need be considered as a genuine alternative. Instead we quite properly wonder at the outlandish character of the proposal and at the lack of empirical motivation. Davidson's argument is at best inconclusive-and the conclusion doubtful.

6. Truth and Correspondence In conclusion, a few points can be drawn together, connecting the themes of truth and correspondence with the truth-conditional approach to meaning which runs from Frege through the Tarski-inspired work of Davidson. Davidson accepts the Quinean indeterminacy of translation in a modified form, as is evident above, and this indeterminacy, or "inscrutability" of reference casts some doubts on the prospects for Davidson's approach to linguistic or cognitive meaning, here viewed as a reconstruction ofFrege's concept of "mode of presentation." The meaning of a word, we reasonably assume, is not indifferent to the specific truth claims it is used to express, and since different truth claims may be made, using the same language and the same expressions, syntactically and homophonically identified, we will get no appropriate account of meaning which does not attend to the complex of claims typically made, in relation to distinct domains of discourse, and the conclusions there expressed. When we are at first ill-equipped, in our own language, to state truth-conditions or semantic characterizations of some foreign idiom, due to some divergence in belief, then this implies that the representation of the meaning of the speech of others will require some elaboration development, or even revision of our own language, as used to describe the expressions under interpretation. This is to say that in learning to describe the



discourse of others, we may learn not only something which is specific to the semantics of the language under interpretation, we may also learn something about the domain that language describes. To illustrate, ifwe do not ourselves understand a particular biological or chemical theory, and we set out the describe the semantics of this theory in our own prior language, then this will plausibly require an expansion of our prior language capable of expressing, in meta-linguistic form, the biological or chemical claims that we set out to interpret. We will have learned some biology or chemistry along with the semantics. Although there is much that is quite appealing in Davidson's critical approach to anti-realism, this criticism is limited in ways which suggest related limitations in his approach to meaning. In particular, Davidson accepts the typical anti-realist conception of "correspondence" as itself involving a means of deciding what is true and what is not. "The correct objection to correspondence" theories of truth, according to Davidson, is "that such theories fail to provide entities to which truth vehicles (whether we take these to be statements, sentences, or utterances) can be said to correspond." 21 Nor is this a defect that any correspondence theory of truth can remedy, on Davidson's view, since he explicitly claims that "Nothing, no thing, makes our statements true." 22 And in consequence, he goes on to reject the concept of representation. "If we reject facts as entities that make sentences true, we ought to give up representations at the same time, for the legitimacy of each depends on the legitimacy of the other." 23 In contrast with all this, I have argued that the semantic conception of truth, given its relation to classical correspondence theories of a realistic type, allows us to get around this kind of objection. We can say, following Davidson and his citation of P. F. Strawson that "a statement corresponds to (fits, is borne out by, agrees with) the facts" and that "this is merely a variant on saying that it is true." 24 We require no specially crafted objects with which true statements are to correspond, if the reference of true statements are regarded as consequent upon the reference which component lexical expressions of those statements must have, if the statements are to be regarded as true. This is to say, e.g., that if 'Fa' is true, then what makes it true, what the true statement corresponds to, is that the object x, which 'a' names is satisfied by 'F'. Of course, we know this only on the assump-

21. 22. 23. 24.

Davidson, Donald 1988, "Epistemology and Truth," p. 184. Ibid. Ibid. Ibid.



tion that Fa. But, in a way, this is just the kind of point I have been elaborating. The semantics for expressions applied to a particular domain of discourse cannot be sensibly developed, except on the basis of what the dominant belief-systems and consequent usage tell us regarding the domain in question. Conflicting beliefs sometimes constrain what we may count as commonalities of meaning, though elsewhere they bring differences of meaning in their wake. There is little we can say in semantics which is not parasitical, in related ways, upon the claims made in the object language. The only substantial conception of correspondence we require, does not define truth in terms of correspondence, unless correspondence is viewed in a trivial Aristotelian-Tarskian way, instead it views judgments of correspondence between language and ordinary objects as a consequence of the judgments of truth values, as these are subject to the continuity of inquiry. It is basically only (object-language) inquiry itself which allows us to judge, in our fallible ways, of either truth or correspondence; and meaning itself is subject to revision in object-language inquiry.


Quine's aim in this slim book is to "update, sum up and clarify variously intersecting views on cognitive meaning, objective reference, and the grounds of knowledge." Only nine pages had previously appeared as the book came to print. It is based largely on unpublished lectures and informal discussions of the past ten years back to the Immanuel Kant Lectures given at Stanford in 1980. It does not, then, duplicate Leonelli's Italian translation of the Kant lectures, La Scienza E I Dati di Senso, 1 which appeared in 1987. The focus is on "interrelating" thoughts and "firming up" an occasional faulty joint among themes of the last decade-sometimes made in less formal settings. The book should be read in connection with recent collections on Quine's work, most importantly Barrett and Gibson's Perspectives on Quine2 which will also be cited here. Quine divides his book into five chapters devoted to "Evidence," "Reference," "Meaning," "Intension," and "Truth." Readers familiar with his recent work will find significant innovations here, and subtle replies in on-going discussions-generally serving to emphasize Quine's empiricism and the goals of naturalized epistemology. It is difficult to resist turning back and forth between the chapters for comparisons. Those brought up at the knee (metaphorically or not) of America's greatest living philosopher will be grateful for the signposts as the generations change. The quotations at the start from Plato (and house paint makers Sherwin-Williams) announces a concern to save the phenomena, and one is invited to think of this in terms of sensory stimulation, stimulus meanings and the need to account for how we get from these to our theory of the world. "New vistas" suggested by neurology, psychology, psycholinguistics and other fields are certainly of interest, but the focus in the book is


Originally published in Dialectica, Vol. 45, Fasc. 4, 1991, pp. 317-22. Cf W.V. Quine, Pursuit of Truth, Cambridge, MA: Harvard University Press, 1990; second ed., 1991. Reprinted by permission. I. W.V. Quine 1987, La Scienza E I Dati di Senso, translated by Michele Leonelli. 2. Barrett, R. B. and Roger F. Gibson 1990.



more narrowly on what mere logical analysis can reveal about the passage from stimulus to science. Least one think that prediction is the defining purpose of science for Quine, he also lists "understanding" and "control and manipulation of the environment" as "major purposes." 3 Still, prediction is the test of theory, whatever its purposes. A predictionless science of understanding is, then, not expected; though this is not to say that a the study of understanding must forever do without predictions-such as might be forthcoming from new vistas in neurology or psycholinguistics for instance, but also such as are involved in the theory and accounts of evidence for translation. For the latter certainly belong to that chapter of psychology which our author terms "naturalized epistemology." Since this version of epistemology is to be a chapter of science, it also depends upon evidence and prediction. A major background question concerns how, and to what degree, cognitive meaning and Quine's own treatment of ontology (the notion of ontological commitment, for instance) can themselves be naturalized.

1. Evidence and Meaning Quine's fallibilism and knocks at the "Cartesian dream" of foundations for scientific certitude persist here through a rejection of any overall analysis of the concept of evidence. Rather, we get a new treatment of "observation sentence" (partly referring forward to the chapter on "Meaning") which involves "a new blend of reification with observation," 4 and a rejection of the heretofore "distinctive factuality" of observation sentences "by disavowal of shared stimulus meaning." 5 Along with Quine's "focused observation sentences," one also finds under the heading of "Evidence" a treatment of the theory laden-character of observation, observation categoricals (needed to integrate observation with theory), a discussion of Popper on falsification and discussions ofholism and empirical content (meaning). In defining "the empirical content of a testable sentence or set of sentences" by reference to "a set of synthetic observational categoricals" logically implied, Quine is well on the way to a contextualization of (empirical) meaning to object language theory. Talk of the analytic and the synthetic has lost its former epistemic burdens here, and it is to be expected that meaning might be liberated in some fashion as well. The 3. Quine 1990, p.2.

4. Ibid., p. viii. 5. Ibid., p. 43.



question of a more realistic treatment of meaning "intentional realism" (and the related questions of indeterminacy of meaning and ontology) appears to be linked here to problems concerning the status and role of stimulus meanings in Quine's work and in his disputes with Davidson and others.6 Claims for inter-linguistic (but not intra-linguistic) stimulus synonymy are dropped in view of difficulties. Stimulus meaning as it figures in the evidential support of theory is thus more separated from stimulus meaning as it figures in theories of translation. Regarding evidence, "observation sentences," with their stimulus meanings are, "the vehicle of scientific evidence," though no definition of evidence is ventured. The absence of a definition is of some interest, since it bears on the Quinean thesis of the underdetermination of theory by evidence. Without an effective and theory-independent definition of evidence, one comes to think that what counts as evidence will depend in part on the perspectives of evolving theory (as in 'This is H2O' or 'This is H2 S' which may come to be observation sentences in light of known theory). This point weakens Quine's assumption that we will always have insufficient evidence for deciding between "empirically equivalent theories." Theories will be merely empirically equivalent, so far as we know and subject to future accounts of evidence. Where once Quine had argued from underdetermination of theory by all possible evidence to inscrutability of reference and indeterminacy of translation, the argument is here reversed-from ontological relativity to the existence of empirically equivalent but ontologically distinct theories. 7 Proxy functions remain crucial to indeterminacy of translation, reference, and meaning. "Focused observation sentences," allow Quine a kind of perceptual simulation of reference at the observational level. In contrast to a conjunction of observation sentences 'Lo a pebble and lo blue' the predication 'This pebble is blue' has a distinct stimulus meaning. For, while the conjunction merely requires that "the stimulation shows each of the component observation sentences to be fulfilled somewhere in the scenethus white pebble here blue flower over there," in the case of the predication "The blue must encompass the pebble." 8 Quine denies, of course, that observation sentences are all learned as unstructured wholes by direct conditioning to stimulation. Given that various compounds are included

6. Cf. Quine 1990, p.44. 7. Cf. Ibid., p. 96. 8. Ibid., p.4.



among the observation sentences, there are just too many for them all to be learned by direct conditioning. Rather, many of them are learned "by subsequent construction from sophisticated vocabulary." Observationality of the sentence only requires "direct correspondence to ranges of stimulation."9 It is the fact that a sentence could be learned in the direct way which renders it observational. Observation sentences are theory-laden for Quine in that, e.g., 'The mixture is at 1800 C' and 'Hydrogen sulfide is escaping' are observational though "relative to one or another limited community, rather than the whole speech community." 10 This specialized community could, in most cases fall back upon community-wide observationality, except perhaps in the case of "the indescribable smell of some uncommon gas." But, more importantly, there is a further sense in which all observation sentences are theory-laden, though in another sense none are. At the most primitive level observation sentences are associated "as wholes" to stimulation. In the sense that any observation sentence could be so learned, no observation sentence is theory-laden. But component words (though at first merely "component syllables") come to figure in complex theory in the course of time, thus providing the connection between theory and observation. "Retrospectively, those once innocent observation sentences are theoryladen indeed." Yet they retain their observationality for all that. 11 One important question in all this is whether, and in what way, the acquired theory-laden character of observation sentences is an element of the meaning of such sentences-beyond the level of stimulus meanings. For example, it might be plausibly maintained that 'The sun is rising' though observational, differs in meaning as it appears in the context of Ptolemaic astronomy-as against the interpretation given to the sentence within the context of the Copernican theory. After all, such sentences (interpreted behavioristically as unstructured wholes) might be thought to support either theory, but the Copernican must reject the literal interpretation put upon it by the Ptolemaic theory. The argument is, then, that the observation sentence the Copernican rejects as false differs in meaning (or interpretation) when reinterpreted within his own theory. But Quine does not pause here to examine the notion of meaning beyond stimulus meaning. Rather he goes on to a very interesting discussion of how observation relates to theory.

9. Ibid., p.5. 10. Ibid., p.6. 11. Ibid., p.7.



Of special interest to this theme is a particular kind of compound of observation sentences-"Whenever this, that"-involving an "irreducible generality prior to any objective reference," and which Quine calls "observational categoricals." Although compounded of observation sentences, these are standing sentences "and hence fair game for implication by scientific theory." These are not to be confused with the observation conditional of Theories and Things-which are formed from a pair of standing sentences. By way of example, Quine provides a "focal observation categorical" which "generalizes on a predicational observation sentence" rather than generalizing on some other compound of observation sentences. (1) When a willow grows at the water's edge, it leans over the water.

Thus, (1) generalizes on the predicational observation sentence 'This riverine willow leans over the water,' and it is suited to be logically implied by more or less elaborate theory involving the hypothesis that willow roots nourish chiefly their own side of the tree, that roots get more nourishment from wetter grounds and so on. Why regard such sentences as involving "irreducible generality prior to any objective reference"? One might argue instead that where we have cross-reference as in (1) between the pronoun 'it' and the phrase 'a willow' reference and individuation are also present. It might also be thought that the predicational observation sentence also involves a kind of reference: akin to 'This is a willow, it is riverine, and it leans over the water.' But Quine does not allow reference to enter at the level of observation sentences-it would conflict with the empirical inscrutability of reference. But even if reference does not enter with the early acquisition of observation sentences, and even if Quine's predicational observation sentences could be learned as unstructured wholes, it seems clear that they might subsequently come to involve reference and individuation of the kind Quine seems to resist here. There is perhaps a background story to Quine's predicational observation sentences to be found among the contributions to the Barrett and Gibson volume. For one finds there Hintikka insisting upon the ease of learning number words-a point which would appear to strongly conflict with Quine's story in Word and Object (§§24-25) according to which



identity, individuation and objective reference enter only at the level of "analytic hypotheses" and beyond the factuality of stimulus meanings. 12 Thus suppose that we can tell when the natives are cooking and even help out, while ourselves not yet beyond the level of stimulus meanings. If we are also able to use number words at this level of understanding, then it seems that we will be in a position to decide between various "analytical hypotheses" in our interpretation of 'gavagai'. If it is clear that the natives ask for one gavagai in order to cook it, then it will also be clear that they will not be satisfied with either a temporal rabbit stage (as when a child asks 'May I hold your rabbit?') nor with an undetached rabbit part. (May I hold your rabbit's leg?) Rather, if we see that a gavagai is wanted for cooking, then we expect an objection if a mere rabbit stage or an undetached rabbit part is offered. If the natives are dissatisfied where the rabbit is taken back, then this is evidence that 'one gavagai' does not mean the same as 'one rabbit stage', and if they are not satisfied with holding onto any of various undetached parts (while the investigator retains the rest) then this is evidence that 'one gavagai' does not mean the same as 'one undetached rabbit part.' The existence of such evidence for and against alternative "analytical hypotheses" is certainly a problem for the account of meaning in Word and Object, and one might therefore be led to think that Quine hopes to encompass the learning of a perceptual form of quasi-individuation and crossreference, e.g., 'This is the same rabbit, but not the same rabbit stage' without conceding empirical evidence for the translation of the full referential and indivduative apparatus of a given linguistic community. The plan would be to allow a kind of perceptual individuation of bodies, via the "focused" character of predicative observational compounds, involving a supplement to the notion of innate quality spacing (or inborn perceptual similarity), and to deny that full blown objective reference enters at this level. But if this is the strategy, there is reason to think it will not work. For, however, "objective reference" and individuation might be conceived, if we can have empirical evidence for elements of this complex, then it stands to reason that we might also have empirical evidence for further elements as well. In short, though reference and individuation within a language under interpretation might not count as observationally evident characteristics of expressions, this point would not prevent us from conceiving of them as fully factual characteristics to be assigned in light of various and sundry empirical evidence.

12. See Hintikka, Jaakko 1990 in Barret and Gibson 1990, p.169.



2. Intention and lntension Most of the discussion in the fourth chapter might better be labeled "intention" rather than "intension," since Quine is chiefly concerned with sentences reporting perception, their role in language acquisition, and sentences reporting belief-to generalize, the chief focus is on propositional attitudes 13-which culminates in an acquiescence in Davidson's anomalous monism: 14 a token physicalism where mentalistic idiom amounts to "ways of grouping" neurological phenomena. The account represents a shift of emphasis in Quine's views: "Brentano was right about the irreducibility of intensional discourse," and further, departing somewhat from Word and Object, it seems, "there is no dismissing it." For, mentalistic idiom "implements vital communications and harbors indispensable lore about human activity and motivation ...we have no substitute." In particular, knowing when a student sees that so-and-so is counted as crucial to learning and teaching the application of"so-and-so." But to forestall premature celebration among friends of intentional realism, Quine has it that "there is good reason not to try to weave it into our scientific theory of the world to make a more comprehensive system." The grounds for this are an expressed preference for "the crystalline purity of extensionality" in science. Thus, Quine's hesitation on the mentalistic idiom appears to stand as firmly as before on the one hand, though "efforts to reclaim territory from the intensional side by dint of discoveries and reconceptualizations ...are to be encouraged and watched" on the other. Encouragement for scientifically motivated deviations from the master who otherwise lumps the intentional and the intensional into one category? Theory formalization within the extensionalistic framework while not sufficient for full intelligibility, is "pretty nearly necessary." What else is required? Clearly, a science of the mental (or extensionalistic semantics in empirical linguistics) must also involve successful prediction. Where this is lacking (and a full-blooded anomalous monism seems to forbid the needed law-like connections among mentalistic predicates) we have at most bad science. Thus a dilemma. On the one hand Quine appears to rule out a nonpredictive science of understanding (leastwise as suitable for integration with our overall theory of nature) but on the other hand since naturalized epistemology is a chapter of science for Quine (centrally concerned with 13. Quine 1990,pp. 60-73. 14. Ibid., p.71.



understanding, i.e., the progress from stimulus to science and the relation of the one to the other), and since the test of science is prediction, then either we must expect genuine scientific developments, or we must give up on naturalized epistemology as a chapter of natural science. Anomalous monism will be sustained only if such scientific developments are restricted to neurology and we get no genuine predictive science from (mentalistic) psycholinguistics or (intentional) developments (contra Fodor) in empirical semantics. The thesis or hypothesis of indeterminacy of reference and meaning is thus a projection (or hypothesis) concerning how naturalized epistemology will develop. Turning to modality in §30, Quine develops his account of the contrary-to-fact conditional in terms similar to his familiar (and convincing) account of 'necessarily' as involving the "second-order annotation" that a sentence "is deemed true by all concerned, at least for the sake and space of an argument." Usage of 'possible' is treated in a similar fashionits contextual relation to the state of a discussion is recognized. At the end of the chapter, we return to the intentional where doubts are expressed concerning its animistic origins. However, seeing animism in the mentalistic idiom appears to be a kind of reversal and anachronism. For, some mentalistic notions "would seem to be as old as language," being involved in the teaching and learning of language at the observational level. Given this point, animism would then be a later elaboration. We cannot, in justice, always visit the sins of the child upon the father.

3. Truth and Reference Section 12 on "Indifference of Ontology" introduces the theme of ontological relativity, now clarified as "indeterminacy of reference": "Reference and ontology recede thus to the status of mere auxiliaries." For "what particular objects there may be is indifferent to the truth of observation sentences, indifferent to the support they lend to the theoretical sentences, indifferent to the success of the theory in its predictions." If Quine is granted this much, along with its illustration by means of "our freedom with proxy functions," then it seems that Quine's characteristic and problematic semantic theses must also be granted. One senses a suppression of earlier (relative) joy in Tarski-inspired extensional semantics in favor of a more austere empiricism liberated from ontology in some fashion. Rudolf Carnap's doctrine of ontological questions as "external" and instrumental seems to reappear in the form of a notion of a practical necessity to acquiesce in the ontological talk of the home language/theory. Thus,



in reply to Barry Stroud, Quine says: "As an individuative general term 'rabbit' denotes each rabbit. Such is reference in the home language, relative to the usual or homophonic 'manual' of translation. These paradigms are on a par with Tarski's familiar paradigm for truth." 15 All things considered, this point appears to place the factuality of truth in question along with that of reference. Having operated on a given theory by means of proxy functions, "we leave all the sentences as they were, ...merely reinterpreting. The observation sentences remain associated with the same sensory stimulations as before, the logical interconnections remain intact. Yet the objects of the theory have been supplanted as drastically as you please." 16 Thus, there is no fact of the matter as to which objects we refer to or purport to refer to, and since choice of a manual of translation in not a factual manner, there can be little comfort to the factuality of (scientific) ontology in Quine's clarification of the notion of ontological relativity as relativity to a manual of translation. 17 Indeterminacy of reference undercuts scientific realism, and thus it will also undercut Quine's physicalism-like a snake devouring its own tail. Still, these point serve to suggest an alternative position: intentional realism in semantics is needed to support scientific realism in the philosophy of science. Thus, there is room for doubt on the conclusion that ontology or reference is always a matter of indifference. Taking referential semantics seriously, one must insist that a reinterpretation by means of proxy functions involves a change of theory-though this is not to say that every such alternative theory generated by proxy functions will be of any serious interest. Where we have changed theory, then, from a semantic perspective we no longer have "the same sentences"-and we no longer have the same observation sentences in particular. Does every similar reinterpretation (or shift in objects of reference) leave us with no empirically accessible difference? (That some do is not sufficient to support Quine's claims regarding "indifference of ontology," and his focus on "empirical meaning" construed in terms of ontologically innocent observation sentences. (Contrast the essays in Quine's The Ways of Paradox, where we read "That the ontology should be relatively definite, pending revision, is required by the mere presence of quantifiers in the language ofscience ..." 18)

15. Barrett and Gibson 1990, p. 334-335. 16. Quine 1990, p. 32. 17. Ibid., p.51. 18. Quine, W.V. 1976, The Ways of Paradox, p. 245.



Can we agree that "the truth of observation sentences" is always indifferent to ontology? This is to say that "observation sentences are to be taken holophrastically from the standpoint of evidence," ("and analytically from the standpoint oftheory.") 19 But it has been argued that we must look to theory to even know what to count as evidence, i.e., that we must look at observation sentences from a particular theoretical perspective in order to integrate them with theory. It is only when so interpreted that they lend support, or serve to disconfirm, particular theories. Viewed holophrastically, or as unstructured wholes, they are indeed indifferent to ontology, but indifferent to theory as well. For example, Galileo and his more conservative inquisitors can agree on the holophrastic rendering of 'The sun rises in the morning and moves through the sky'. To this day we recognize the practicality of acquiescing in this common mode of speech. Still, as interpreted to support the Ptolemaic system, Galileo must reject the sentence as false. For he rejects the premise of such interpretation, i.e., that the sun is in motion around the earth. Moreover, it is only as so interpreted that the sentence serves to support the Ptolemaic-Aristotelian astronomy. Given our present theoretical perspective, we see that the sentence, as so interpreted, is in fact false. Thus, it seems false too that our ontology "is indifferent to the truth of observation sentences." A reinterpretation is required in order to render observation sentences once used to support the Ptolemaic astronomy suitable to support the Copernican alternative. The point is that there is a difference in meaning (or interpretation) between the two astronomical theories and that this difference is relevant to science. Thus, it cannot be that there is "no fact to the matter" between the two interpretations. Whether a given person (or culture) holds to the one theory or the other appears to be as factual as things come. Failing developments in neurology or non-intentional psycho-linguistics which would allow us to paraphrase out of commitment to such difference, Quine's indeterminacy theses remain unproved. But they continue to challenge empirical semantics. Quine interprets the correspondence theory of truth by reference to a version of the disquotation theory, 20 though 'truth' is also needed for semantic assent both within logic and elsewhere. One must wonder if there is not some deep conflict between Quine on truth and Quine on reference and ontology. If reference and ontology are indeterminate, the argument might go, then so is truth. Yet, since a truth predicate is needed for seman-

19. Quine 1990, p. 26. 20. Ibid., p. 80.



tic assent and for generalized statement of the logical truths as when "We say all sentences ofthe form 'Ifp then p' are true." 21 Quine seems to want to have his cake and eat it too. What are true, Quine holds, are sentences; but the truth of a sentence " ...consists in the world's being as the sentence says,"-which seems a paradigmatic intentional claim. Carry this over to quantified sentences, and you get some ontology. For example, the truth of '(:3x)Fx' consists of the world being as the sentence says-containing at least one object which is an F. But if all such ontological claims, and claims concerning ontological commitments, are radically indeterminate and unfactual, and if ontology is "indifferent" to science, then so are truth claims indifferent to science. Our mere acquiescence in the home language, where we say "There are F's" does not remove this indeterminacy according to Quine, (since radical translation begins at home), and thus we expect that the notion of truth must be similarly afflicted. That Quine carries on with talk of truth in spite of this is surely a point which introduces a considerable tension in his views, and it is a point which reflects back upon his title. If truth and truth claims are as radically indeterminate and unfactual as ontology, reference and meaning for Quine, then, what significance can he assign to a mere acquiescence in scientific inquiry and the pursuit of truth? Still, all things considered, Quine's work is a paradigm of the pursuit of truth; indeterminacy of reference and ontology retain their anomaly. Such are the ways of paradox. Younger philosophers try to solve problems. Older philosophers try to formulate them anew. Quine in particular seems intent upon reformulating the problem of the relationship between scientific objectivity and the ontological claims of the sciences. We should be grateful for the earnestness of intent at the roots of Quine's ever impressive accomplishments, even while recognizing that the intent arises because the problem is one within Quine's own views.

21. Ibid., p.81.



Reconciliation of semantic ho Iism with interpretation of individual expressions is advanced here by means of a interpretation of sentence meaning in relation to object-language theories viewed as idealizations of belief-systems. Fodor's view of the autonomy of the special sciences is emphasized and this is combined with detailed replies to his recent criticisms of meaning holism. The argument is that the need for empirical evidence requires a holistic approach to meaning. Thus, semantic realism requires semantic holism. This paper aims at a principled defense of the contemporary explication of linguistic meaning contextualized to theory or belief-system (moderate or pluralistic "meaning holism").Various criticisms and doubts concerning this conception of meaning will be considered here with the aim of showing that such a contextualization of meaning in relation to belief-system is capable of holding its own. I sketch here key ideas in Fodor's attempts to breath life into the representational theory of mind and intentional realism. Having established a context, problems in this program will be addressed which are more specifically semantic in character. Fodor's work serves to illustrate problems here, but also some progress, in escaping behaviorism while holding on to a basically Quinean conception of the goals and requirements of science. I argue in the end that Fodor has yet to devote sufficient attention to the provision of empirical evidence relevant to the individuation of contents or meanings. Doing so, I urge, requires departures from Fodor's opposition to meaning holism. Specifically, reconciliation of semantic holism with interpretation of individual sentences, will be advanced by means of an account of sentence meaning in relation to object language theories viewed as idealizations of belief-systems.

1. The Need for Meanings Fodor wants us to take intentional content seriously. "It appears increasingly," he says in the Preface to Psychosemantics, "that the main joint


Slightly revised for clarity, from Dialectica, 46, Vol. 1, 1992, pp. 41-59. Reprinted by permission.



business of the philosophy of language and the philosophy of mind is the problem of representation itself." At the center of this problem we find a quite fundamental question: "How can anything manage to be about anything; and why is it that only thoughts and symbols succeed?"' In order to answer such questions, Fodor advances a "naturalized theory of meaning." Such an approach, appealingly Quinean and anti-Quinean at one stroke, will be crucial to the possibility of a genuinely scientific vindication of commonsense psychological explanations given in terms of desires, beliefs and other propositional attitudes-as proposed in Fodor's Representational Theory of Mind. The propositional attitudes, then, are to have a place in scientific psychology. "Holding onto the attitudes-vindicating common-sense psychology-means showing how you could have (or, at a minimum, showing that you could have) a respectable science whose ontology explicitly acknowledges states that exhibit the sorts of properties that common sense attributes to the attitudes." 2 Such states are "semantically evaluable," i.e. they have semantic content, but they are also to have causal powers, and "the implicit generalizations of commonsense belief/desire psychology are largely true of them." 3 How much of a vindication of common-sense psychology to expect will depend here upon which implicit generalizations tum out true. This is more or less how we would expect things to go, given the aim of placing common-sense psychology on a scientific footing. Fodor approaches fundamental problems of semantics from an orientation in the philosophy of mind. In a sense, what we find in his work is the attempton behalf of the computer model of mind and contemporary cognitive science-to draw out corollaries in semantic theory and to come to grips with the problem of meaning as bequeathed by Quine. "Computers," Fodor argues, "show us how to connect semantical with causal properties for symbols. So, if having a propositional attitude involves tokening a symbol, then we can get some leverage on connecting semantical properties with causal ones for thoughts." 4 The idea here is that the causal powers of symbols are dependent upon their syntax (and as part of this that syntax is somehow physically realized); and further that there is a correspondence between syntax and semantics-"we know from modem logic that certain of the semantic relations among symbols can be, as it

I. Fodor, Jerry I 987, Psychosemantics, p. xi. 2. Ibid., p.10. 3. Ibid. 4. Ibid., p. 18.



were, 'mimicked' by their syntactic relations." This points in the direction of a theory of how "there could be non-arbitrary content relations among causally related thoughts." 5 This is wanted in order to explain, among much else, how a causally determined mental process could be structured in terms of meanings. Providing such an explanation will require us to quantify over mental contents (Fodor poses the problem in these Quinean terms), and this in tum sets the goal of making sense out of the great battle ground of the theory of meaning. Fodor's RTM is partly a semantic thesis. It requires that we settle on identity conditions for meanings. In consequence, Fodor picks up the Quinean gauntlet. If we are to have meanings, then, these must be theoretically well-behaved. There is little sign of that old comfortable reliance upon the analyticsynthetic distinction among the Chomskians here, though this is far from saying that the Quinean objections to meaning and meanings in science are about to quickly fall away. Rather, I will argue that closer attention to Quinean views on meaning will benefit Fodor's program of intentional realism. But, on the other hand attention to Fodor's work on special sciences serves to ameliorate Quine's skepticism regarding intentional realism.

2. Special Sciences Fodor has continually emphasized that common-sense psychology does work remarkably well, and that this alone argues against the familiar philosophical view that the generalizations employed must be either trivial or false. Still we need to account for the possibility of exceptions to psychological generalizations, and we will also need to reconcile such exceptions with genuine predictive power. Following Fodor, this is to be done by portraying psychology as one among other "special sciences." (Geology figures as a paradigm here.) Exceptions to generalizations of these special sciences, or ceteris paribus clauses attaching to them, are to be accounted for by going outside the vocabulary of the science in question to see how its governing idealizations have been violated. The ontology of intentional psychology, then, is to be no more suspect in relation to that of physiology, biology or chemistry-or physics-than is the ontology of geology suspect in relation to that of more basic sciences. Fodor's work on the notion of special sciences is certainly among his most interesting.

5. Ibid., p. 19.



The chief point in this work appears to be a generalization of an insight associated with later-day identity theories of the mind-body relation. For Fodor endorses the generality of physics in the sense that "all events that fall under the laws of any science are physical events and hence fall under the laws of physics." 6 Still this is far from saying that there are only physical laws or that laws of any other natural science must reduce to physical laws. The question of whether such reductionism, involving "bridge laws," is true is an empirical question according to Fodor, and a positive answer is not to be inferred from the mere truth of "token physicalism"-according to which the events or entities of any natural science must be physical events or entities. Thus, consider Fodor's example of Gresham's law. The generality of physics implies that any events involved in monetary exchanges will be covered by some physical law or other. However, there is little ground to think that all monetary exchanges have any interesting physical characteristics in common. Rather, "banal considerations suggest that a physical description which covers all such events must be wildly disjunctive." Although monetary exchanges do have interesting things in common, what is interesting about them "is surely not their commonalities under physical descriptions,'' 7 for we fully expect that monetary exchanges will take many different physical forms-some yet to be invented perhaps. The generalization that bad money drives good money out of circulation is thus extremely unlikely to correlate with physical laws in any reductivist scheme. Nor is the prospect of such reduction a useful constraint upon all economic theories and generalizations-and the same goes for all other special sciences-whether natural sciences such as biology, or psychology, or social sciences such as economics. Generalizations of the special sciences allow for the possibility of exceptions, then, because of the very physical diversity underlying them: We allow the generalizations of the special sciences to have exceptions, thus preserving the kinds to which the generalizations apply. But since we know that the physical descriptions of the members of these kinds may be quite heterogeneous, and since we know that the physical mechanisms ... may be equally diverse, we both expect that there will be exceptions to the generalizations and that these will be explained away .... 8

6. Fodor, Jerry 1974, "Special Sciences," p. 127; Cf. Fodor, Jerry 1979, The Language of Thought, Chp. I. 7. Fodor 1974, p. 134. 8. Ibid., p. 143.



In contrast, physics is basic in that any exceptions to its laws "had better be random." For in this case there is no more basic science to which we can appeal in order to explain such exceptions. Fodor is led to a realism concerning the objects of the special sciences in view of the possibility (and the existence) of laws, "important, counterfactual supporting generalizations," involving kinds which correspond to no physical kinds. He brings this realist perspective to the search for psychological laws involving generalizations over psychological states and psychological processes involving the same contents. The lesson is that intentional realism is only very doubtfully resisted on grounds of physicalism. In particular, we might thus argue against Quine's behaviorism, and his resistance to meanings, in so far as this is based on purportedly better prospects for the integration of behavioristic psychology into physical theory. 9 Still, substantial support for mentalistic psychology (or non-behavioristic semantics) must depend, of course, upon results-rather than the mere failure of particular arguments favoring behaviorism. This consideration will serve to direct attention below to the kinds of evidence which might be made available to support semantic hypotheses. Yet, before undertaking such an investigation greater clarity on the nature of semantic hypotheses will be an important preliminary consideration.

3. Twin Earth and Narrow Contents In the second chapter of Psychosemantics, Fodor aims to give a solution to Putnam's Twin-Earth problem and other similar puzzles: a solution consistent with the realistic principle that the psychological is supervenient upon the physiological. This is to say that brains must differ whenever minds do. "Mind/brain supervenience (and/or mind/brain identity) is," Fodor remarks, "the best idea that anyone has had so far about how mental causation is possible." 1 For to suppose that mental causation is possible is to suppose that mental differences are, in general, physically efficacious. But it is difficult to see how this could be so, unless there is a rather direct correlation or identity between mental states (or processes) and physiological states and processes. This desideratum leads Fodor to depart quite sharply from the treatment of the Twin-Earth puzzles originally suggested by Putnam.


9. Cf. Callaway, H.G. 1988, "Review of Gochet, Ascent to Truth: A Critical Examination of Quine 's Philosophy," pp. 53. 10. Fodor 1987, Psychosemantics, p. 30.



Suppose, then, that Twin-Earth is "just like here except that they've got XYZ where we've got H2O." Fodor's twin, e.g., is just like Fodor, down to his neurological microstructure. Still, it would seem that 'water is wet' means something different in the language used on Twin-Earth-simply because the extension of 'water' is different there, even though everything else is the same. Are the mental contents of Fodor and Twin-Fodor the same or different? Such stories essentially raise some difficult questions about the identity conditions of meanings or contents and about the relation between meaning and referential notions. How, then, are we to individuate contents and answer the Twin-Earth puzzles? Many see in these puzzles a demonstration of the thesis that commonsense individuation of contents violates supervenience ("meanings are not in the head" 11); but Fodor disputes this. He argues rather convincingly that "the considerations that militate for the non-relational individuation of mental states (hence, for preserving supervenience) are no different from the ones that militate for the non-relational individuation of brain states, molecular states and such." 12 Note that settling on "non-relational individuation" here implies that the differences between Earth and Twin-Earth do not require us to recognize a difference in meaning (regarding expressions such as 'water' or 'water is wet') simply on the basis of the hypothesis of a difference in reference. Fodor makes a good case. The reasoning is so convincing, in part, because it depends very little on anything specifically semantic. Whatever reasons we have to think that living on Twin-Earth makes no difference to brain states is equally a good reason to think it makes no difference to mental states, if mental states are to be involved in causal processes in a law-governed way. The differences that do obtain on Twin-Earth ('water' turns out to be XYZ rather than H2O) are "irrelevant to ....causal powers; hence, irrelevant to scientific taxonomy." 13 Given that Twin-Fodor cannot even distinguish H2O and XYZ, this is not the kind of difference which is going to lead him to act (or think) in one way (rather than another)--in view of the fact that he is daily confronted with XYZ rather than H2O. These are not the kind of differences which will lead us to distinguish different varieties of brain states, and any viable individuation of mental states (and thus of mental contents) must follow this lead.

11. Cf. Putnam, Hilary 1975, "The Meaning of'Meaning'," 12. Fodor 1987, Psychosemantics, p. 32. 13. Ibid., p. 34.

p. 227.



The Twin-Earth puzzles create serious difficulties in semantics, because they appear to require a rejection of the traditional principle that content determines extension. Lacking this principle, or a near relative, we are at a loss to say anything on the individuation of contents. "It was a test for the identity of contents that the extensions had to come out the same. And that was the best test we had." 14 Fodor's conclusion in the end is that "the Twin-Earth examples don't break the connection between content and extension; they just relativize it to context." Given a liberal interpretation of this, it will be difficult to disagree. However, Fodor as is his wont, becomes more specific. He introduces a distinction between narrow and broad contents, where broad content is relativized to environmental context. In terms of this distinction, "my Twin's 'water' thoughts are intentionally identical to my water thoughts; they have the same narrow contents even though, since their contexts are de facto different, they differ, de facto, in their truth conditions." However, "the 'broad content' of a thought, by contrast, is what you can semantically evaluate, it's what you get when you specify a narrow content and fix a context." 15 This distinction is quite problematic, however, in part because it remains unclear what it means to specify a narrow content. Assuming that the results Fodor wants from a solution are in fact desirable-preserving supervenience in particular-(and in spite of the popularity of this sort of distinction between narrow and broad content), it is not difficult to be left wondering what makes Fodor's narrow contents semantic at all. According to Fodor, "narrow contents aren't semantically evaluable; only wide contents have conditions of satisfaction." 16 Yet, if we accept this conclusion, then the original desiderata of vindicating common-sense psychology appear to be seriously compromised. In particular, it seems unclear, on this dual content theory, that "if you know what the content of a belief is, then you know what it is about the world that determines the semantic evaluation of this belief." 17 This stressed claim from the first chapter of Psychosemantics now appears ambiguous. Moreover, we want to know how we could specify a narrow content without use of some semantic vocabulary. It seems that we could not. Yet, use of such semantic vocabulary will bring along with it the possibility of some sort of semantic evaluation.

14. Ibid., p. 46. 15. Ibid., p. 48. 16. Ibid., p. 83. 17. Ibid., p. 11.



Consider on the other hand the (broad) contents which are semantically evaluable on Fodor's account of these matters. Apparently their individuation need not agree with the causal taxonomy. But this is a very difficult conclusion to accept within the confines of Fodor's program, for it seems to suggest the further conclusion that we may have to give up commonsense psychology after all. (If "real semantics" concerns relations to actually existing objects of reference.) Fodor has not so much solved the Twin-Earth puzzles, then, he seems rather to have shifted them elsewhere. The problems started as Fodor became more specific. So, I propose to return to that point in the story and to attempt to pick up the threads. Fodor is correct to attempt to preserve some form of the principle that content determines extensions, and I think that it is also correct to look for some sort of contextualization of this. However, it is crucial to see that the TwinEarth puzzles are essentially a version of Quine's doctrine of ontological relativity (or of the "indeterminacy of reference"). The puzzles presuppose an objective difference between H2O and XYZ, although this difference is not reflected in the context of knowledge of the speakers on Earth or on Twin Earth. Thus, these speakers, or those in particular who attract our attention in a given version of the puzzle, are not in a position to distinguish the meaning or reference of 'water' as used here or as used on TwinEarth. It is this point which leads us to the conclusion that the meaning of 'water' is the same although the reference varies. But recall that this conclusion is essentially dictated by the stipulation that Twin-Earth water is XYZ. Although such a stipulation is not ruled out by what we know about theoretical identities and their discovery, I want to urge that it is, nevertheless, out of place as a general constraint upon semantic theory. The puzzles arise basically from the superimposition of distinct contexts of knowledge: that of the speakers and that of the teller of the story. This is problematic because we must in the end judge of the reference of our words on the basis of our best theory concerning what there is in the world, and thus on the basis of our best theory of what there is for our words to refer to. But what is the best theory here? The Twin-Earth parables involve a built-in ambiguity regarding such a question, and this is essentially the source of the puzzles. We are provided with a God's eye view of the differences between Earth and Twin Earth and in consequence our more ordinary means for settling questions concerning meaning and reference (which depend upon accepting a single approved context of knowledge-however tentatively) are deprived of meaningful contextual constraint.



It is of course, possible (in the sense that such a discovery appears similar to some that have actually taken place in the history of science) that we should discover that what is called 'water' in one place is not the same substance that is called 'water' in another place. Or, to put the point in a somewhat different way, given the context of knowledge of our ancestors, before they discovered that water is H2 O, there was no way to predict that water would not have turned out to have some other composition, say XYZ. So, there is clearly a sense to insisting that meaning does not determine extension. For, 'water' was not less meaningful when our ancestors did not know that water is H2O rather than XYZ. What Putnam did in coming up with the Twin-Earth puzzles was essentially to give some imaginative clothing to a well appreciated point in formal semantics, viz., that it is always possible to reinterpret the vocabulary of a formal theory, so that all the sentences remain true while the objects of reference are changed. Though in this case, of course, it is only the extension of 'water' which is imaginatively changed. This is, of course, a quite impressive point in many ways, but it should not lead us to some sort of skepticism concerning whether the objects of our best theories really are what we take them to be rather than something else inscrutably different. For we can only talk of the reference of our words in relation to our own context of knowledge (and not, as it were, relative to everything that might turn out true). But, if so, then we must reject the "God's eye view" presupposed as valid in the Twin-Earth puzzles. This point amounts to a kind of deconstruction of the problems posed by these puzzles. But it also amounts to a kind of deconstruction of the proposed distinction between narrow and wide contents. That distinction appears to be required in order to account for the similarities between Earth and Twin-Earth, to maintain the principle of the supervenience of the mental upon the physical, to allow that meaning somehow determines reference or extension, and lastly to accommodate the (merely possible) stipulation that Twin-Earth water really is XYZ. However, this problematic distinction between narrow and wide contents will be avoided, and the problem of the reference of Twin-Earth 'water' obviated, if we accept any contextualization of the principle that meaning determines reference. Meaning or content, I would urge, determines reference in relation to the context of knowledge within which the words in question are embedded. (This is quite similar to holding, with Quine, that ontological commitments belong primarily to claims made (or theories). One can make no sense of ontology except in relation to what is being



claimed to be true. 18 In light of a broader context of knowledge, we might be able to judge that, e.g., Einstein's notion of force is a generalization of Newton's and thus that there are common objects of reference. But questions of reference cannot always be so answered given the task of comparing two arbitrarily selected viewpoints. There is no other (generally) sensible basis on which to judge of the reference of our words other than that provided by what our best theories tells us is there to refer to. What is wrong with the Twin-Earth puzzles is that they invite us to ignore this contextual nature of judgment and answer questions in semantics on some unexplained absolutist basis: in terms of constraints which determine what the reference of words "really are," independent of the relevant contexts of knowledge. Once this point is fully appreciated, it strongly suggests a more tolerant approach to the notion of meaning holism. For it suggests that meaning and meanings cannot be approach except in relation to an assumed context of knowledge: that matters of meaning are not independent of matters of fact and belief. If we are to recognize meanings at all, then what must be finally and definitively surrendered is the notion that questions of meaning can be settled independent of our knowledge of the world. Rather, meanings of expressions are so crafted as to allow us to represent the world in accordance with our best over-all theories. A viable "naturalized" theory of meaning must, contrary to Fodor, be more thoroughly Quinean. We do indeed ascribe meanings in light of overall patterns ofbehavior, and moreover, we change meanings of expressions on occasion in the light of overall empirical-theoretical developments. A justification of intentional realism must appeal to meanings in explanation of generalized and systematic regularities in linguistic (and non-linguistic) behavior.

4. Meaning Holism "Meaning holism," on Fodor's interpretation, "is the idea that the identity-specifically, the intentional content-of a propositional attitude is determined by the totality of its epistemic liaisons." Further, to sketch out the interpretation of meaning holism at work here, we must observe that P is an epistemic liaison of Q, "when an intentional system takes the semantic value of P to be relevant to the semantic evaluation ofQ (for that system at

18. Cf. Callaway, H.G. 1982, "Sense, Reference and Purported Reference," p. 96ff.



that time)." 19 Fodor sees great danger to the prospects for intentional psychology in the doctrine or hypothesis of meaning holism. He fears that in accordance with meaning holism, "no two people will ever get subsumed by the same intentional generalizations." For, "people quite generally differ in their estimates of epistemic relevance, and if we follow meaning holism and individuate intentional states by the totality of their epistemic liaisons, it's going to turn out that no two people ...ever are in the same intentional states." 20 But, ifno two people are ever in the same intentional states, then intentional psychology will have no real predictive power. A natural response to this criticism is to treat identity of contents as an idealization suited to render generalizations more useful the more nearly the idealization of a community with uniform beliefs is approximated. Yet, according to Fodor, this will not work. It will not work because, a view based on meaning holism cannot even say "what it is like to believe that P in the ideal case." 21 Supposing that people do generally differ in their estimates of epistemic relevance, it does seem clear that such a version of meaning holism, with individuation based on epistemic liaisons, slices things too thin. For, we have no hint as to how we should pick out one set of epistemic liaisons to serve as a standard set. Yet, contrary to Fodor's treatment of the theme, this also serve to suggest a need to investigate some more reasonable versions of meaning holism. Unfortunately, Fodor has shown little interest in such a project. He seems generally content to criticize arguments for meaning holism without a great concern to account for the appeal of the notion. Instead, there seems to be a rather singleminded concern to block arguments from meaning holism to anti-realist conclusions regarding intentional contents. However, it is not clear at all that every version of meaning holism will lead us on to such anti-realist conclusion. The central concern of this section is to show that Fodor's chief argument for against meaning holism does not serve to establish such a conclusion. This is Fodor's argument against making use of the idealization of a linguistic community of uniform believers. "The problem isn't that Meaning Holism forces us to scientific idealizations," he argues, "it's rather that Meaning Holism makes any old idealization seem just about as good as

19. Fodor 1987, Psychosemantics, p.56. 20. Ibid., p. 57 21. Ibid., p. 59.



any other." 22 In this objection, there is no way for the meaning holist to choose one idealization over any alternative: The function of a scientific idealization is (inter alia) to tell you how the observed values should vary as experimental conditions move toward the asymptote. But idealizations can't serve that function unless the theory say where the asymptote is. (E.g., at asymptote the planes are frictionless, the molecules are infinitely elastic, the chemical samples are pure, ...etc.) The application of this methodological principle .... seems clear. Such a theory can idealize to a homogeneous community of believers-that-P only if it can say what it is like to believe that Pin the ideal case. 23

The comparison here which seems most inviting, in crafting a reply, involves the idealization in chemistry-that one is dealing with chemically pure samples. For, we have reason to hold that samples (of reagents and materials made use of in qualitative or quantitative analysis, etc.) ever are totally pure, and this introduces a potential for doubt into every experiment or experimental determination. Nor is it clearly true, in this case, that the value of a scientific idealization must depend upon its ability to specify, "how the observed values should vary as experimental conditions move toward the asymptote." Rather, in qualitative and quantitative analysis, if the results go wrong, it is always possible to suggest that the problem arose from some impurity in the reagents used in the test. Where such doubts seem at all realistic, the chemist simply throws out his old supply of reagent and starts again with a new batch, because such contamination can cause the results to go wrong in totally unpredictable ways. (But it is regarded as extremely unlikely that the second batch will also be contaminated.) Part of the reason that this particular comparison is so inviting is that the notion of 'chemically pure samples' has so much chemistry built in. It presupposes that we can, in principle, distinguish any two chemical elements or compounds where one might be regarded as a contamination of the other in some circumstances or other. Of course, as a practical matter in particular situations, this could be very difficult to carry out; and leastwise when it comes to organic chemistry, the chemists no longer have pretensions of being able to distinguish any two compounds, given our present knowledge and technology (leastwise considering chemical methods alone). There are simply too many such compounds.

22. Ibid., p. 58. 23. Ibid., p. 59.



Unlike the case of frictionless planes, e.g., the idealization of chemically pure samples, as invoked in predictions of experimental results in chemistry, does not strictly determine "how the observed values should vary as experimental conditions move toward the asymptote." For ifwe do not know how a sample has been contaminated, then we have no way of saying how the results will go wrong. Even a little bit of certain sorts of contaminants could make the result unrecognizable (e.g., if it happened to serve as a catalyst). If such a contaminant were, therefore reduced by a small amount, there is no way of knowing how the results would change. Contamination of chemical samples is a qualitative matter as well as being quantitative. We cannot always depend upon some smooth asymptotic curve. The problems involved in the empirical demarcation of dialects (or idiolects) are somewhat similar to this, and this holds true whether or not we view meaning in relation to context of knowledge (i.e., to theory or belief-system). Consider, for example a syntactic case: that of double auxiliary verbs. Certain varieties of Southern speech in the U.S. are characterized by such double auxiliaries. However, if we are doing an empirical study of Northern speech and we find an informant who produces a double auxiliary in the middle of Philadelphia, then this does present a kind of problem. Nor can we say, independent of observation, what syntactic construction might tum up in Philadelphia-deriving perhaps from Indian English or some variety of African English to be found among the many students, and so on. Still, in spite of such problems the linguist will manage to idealize to a uniform speech community. It is the preponderance of evidence which directs us to exclude certain responses in moving toward the idealization of a uniform speech-community. Similarly with semantic cases. The isolated individual who tells us that he has parked his car on the third story in the parking lot will not thereby prevent us from characterizing the meaning of 'parking lot' so as to distinguish it from that of 'parking garage'. This example of usage will simply be thrown out unless it can be repeatedly found among some recognizable group of speakers. (Just as reagents may be thrown out in routine chemistry, if results go wrong in unexpected ways.) Though, of course, if we find an entire community in which this contrast is simply absent, then we have some grounds for distinguishing a dialect. Viewed simply as a system of rules there is, of course, no principled difference between a dialect and a standard language. But no one takes this point alone to imply that we should devote as much attention to every obscure dialect and idiolect as we do to major standard languages.



From a purely formal standpoint, there is no ground to study standardized languages in preference to dialects or idiolects. Yet, in characterizing a standard language, we ignore not only dialects but also the "regional coloring" which tends to creep in from the influences of localized dialects. The very notion of standard language, viewed in reference to empirical evidence, is a kind of idealization which ignores much of interest. Yet, no one is tempted to object to formal grammar, as an empirical theory, on the grounds that a formal conception of language or grammar does not tell us how to choose a particular idealization of a uniform speech community. The reply in defense of a formal conception of language is that we can study any dialect or idiolect, standard or deviant, that the principles involved in doing so will equally apply in any case we choose to consider. Moreover, we will need to exclude certain deviant examples of usage regardless of which idealization we choose. But selecting an idealization amounts to selecting a speech-community in which the favored usage predominates. In a sense, knowing what to count as favored usage, as usage predominant within the speech-community under investigation, depends in the extreme case upon knowing every other speech-community which might be confused with the speech-community under consideration. Our evidence is always under threat of unforeseen contamination. But none of this, I take it, represents a principled objection to empirical investigations within the scope of formal grammar. By the same token, we must not imagine that a theory which idealizes to a homogeneous community ofbelievers-that-P must first, and independently, be able to tell us "what it is like to believe that P in the ideal case." If, as proposed above, the meaning of 'P' is contextualized by reference to an embedding belief-system, then an interpretation of 'P' will depend upon, or be made in concert with, an interpretation of the selected beliefsystem. Further, we must not imagine that our idealization of a speechcommunity, homogeneous in its beliefs, is a mere abstraction. Rather, we will depend (basically) upon a uniform avowal of beliefs to identify the speech-community in question. It will be the preponderance of evidencewithin the selected speech-community-which dictates which system of beliefs to settle on; and in a similar way, the same evidence will be made use ofto settle on an interpretation of 'P' in particular. Language is a social art; and in consequence, we expect to find a large and principled difference between language on the one hand and matters which are purely particular or idiosyncratic on the other. It is surely part of the force (and appeal) of the oft repeated claim that semantic rules are con-



ventional (and stipulative) that it allows the advocate of this viewpoint to insist upon the social character of semantic rules. (They are valid because our society says they are valid, and the analytic truths are true because we agree that they are, and so forth. Moreover, our language teachers of all ages are bound to sympathize, since insisting on a clear right and wrong with respect to usage simplifies their work-as does an acceptance of the authority of their example. Though in sympathizing with our language teachers, we stand in danger of importing attitude appropriate to elementary education into theoretical debates where they are clearly out of place.) It might be thought, however, that to the degree that meaning is contextualized by reference to theory or belief-system, this social character of language will be lost from the account. For every individual has his own particular beliefs and the same observation might plausibly be made concerning any sub-group in society. Does it follow, then, that if meaning is contextualized by reference to theory and belief-system there will be no unified social aspect to meaning as we are led to expect? Essentially, this follows only if there is no social aspect to theory or belief-system. This, in tum, is a perspective which I dispute. The plausibility of the idea that we will in fact find speech-communities homogeneous with respect to belief-system is essentially the same as that of the assumption that the members of speech-communities normally understand each other-something which no one wants to deny. It is no special entitlement of theories of meaning tied to traditional conceptions to make such an assumption. This point should be especially evident when one considers, that purported analytic truths will certainly figure quite prominently in the characterization of any semantic theory for any speechcommunity.24 Although the problem of individuating the relevant speechcommunity is vastly simplified by considering a single speaker along with his semantic idiolect, the simplicity of this case should not mislead us into supposing that where meaning is contextualized by reference to beliefsystem we can only study idiolects or dialects. Rather, the social aspect of meaning, the fact that we speak a language of largely shared meanings is now located as arising from the fact that we largely share the same beliefsystems-leastwise where our normal presumption of mutual understanding is plausible. Nor is our attachment to shared belief-systems, (though, of course, we share different belief-systems with different social groups) and shared meanings, any less compelling merely in virtue of the fact that the

24. Cf. Callaway, H.G. 1988, "Semantic Competence and Truth-conditional Semantics," pp.18-22.



rejection of the analytic-synthetic distinction lets us see this in a slightly different (more fallibilistic) light. On the contrary, this point allows one to see the virtue of conservatism in theorization (or the principle of minimum mutilation as Quine sometimes puts it) as a practical necessity.

5. Meaning Holism, Evidence, and Extensionality In the last chapter of Psychosemantics, on "Meaning and the World Order," taking up the task of sketching a "naturalized theory of meaning," Fodor focuses on the semantic properties of mental states or mental representations. As he puts it, "its the interpretation of the primitive non-logical vocabulary ofMentalese that's at the bottom of the pile ..." But the account sketched is markedly similar to accounts of the semantics of linguistic expressions. Given an interpretation of the primitive non-logical vocabulary, we "can proceed by means which, though certainly not unproblematic, are at least familiar: viz., by the construction of a truth-definition. 25 My primary aim in considering this discussion is to show how such a theory might benefit, by way of the assembly of evidence, from a greater attention to the appeals of moderate or pluralistic meaning holism. I will (like Fodor) generally ignore complications introduced by the focus on the interpretation of mental entities. Interpretation of primitive non-logical vocabulary becomes, in Fodor's hands, a matter of stating causal conditions relating properties (or instantiations of properties) and tokenings of mental symbols. What underlies Fodor's "Crude Causal Theory" is the intuition that semantic interpretations of mental symbols are determined by law-like causal relations. Errors in this preliminary account have to do with misrepresentation, and the interesting discussion of such problems serves to remind us that Fodor is, after all, centrally concerned with a version of the classical problem of intentionality. On Fodor's account, the intentional cannot be reduced to the non-intentional, and contra Word and Objeci26 we are going to get an autonomous theory-i.e. a special science. Refinements lead on to a "Slightly Less Crude Causal Theory" of content.27 The account is surely complex and sophisticated: one would not expect classical objections against causal theories to be telling-leastwise not in any very simple way. Still, we seem to have a causal theory of 25. Fodor 1987, Psychosemantics, p. 98. 26. Cf. Quine 1960, p. 219-21. 27. Fodor 1987, p.126.



referential relations here, and there is certainly room to doubt that talk of reference can easily be accounted for in any causal terms. Although Fodor allows that theories "mediate symbol/world connections," this does not lead him on to holism. Instead, on his account, the content of the theory "does not determine the meanings of the terms whose connections to the world the theory mediates." Rather, their meanings depend upon "which things in the world the theory connects them to." 28 But it is important to notice that Fodor has a proliferation of "things in the world" (properties}-which are called in by his semantic theory: "For example, you fix a context for tokenings of the (Mentalese) expression 'this is water' by specifying-inter alia-that in the context in question the symbol 'water' expresses the property H 20, or the property XYZ, or whatever."29 Clearly, something does not get to be a water concept merely by "covarying with water instances" as Fodor sometimes seems to suggest.30 Presumably, merely co-varying with water instances could equally make something an H20 concept. Thus, the differing properties appear crucial here. Otherwise there will be no difference between a water concept and an H20 concept. Water just is H20, and we are not going to find any difference between the two such concepts merely by noting covariance with their common extension. Still we cannot doubt that there is a difference between a water concept and an H20 concept. Whether a given culture contains enough chemistry to justify our interpretation of a native expression as meaning 'H 20' rather than 'water' is not a matter to which there is no fact. However, in order for Fodor's semantics to be able to take such points into consideration, it appears to require reference to properties. Yet it seems a reasonable principle that a semantic theory for a given bit of (object level) theory or discourse should only minimally extend the ontology of the object language theory or discourse. In any referential semantics, one must talk about expressions and have vocabulary for relating the expressions to the objects of the object language theory/discourse. But one wants to avoid adding to the ontology of the object level much beyond these basics. Thus, if our talk (or thought) is not about properties, then our talk about such talk (or thought) should not call them in. This seems especially important here because of a lurking suspicion that a denotational theory of meaning, such as Fodor proposes, will be only too quick

28. Ibid., p. 125. 29. Ibid., p. 98. 30. Ibid., p. 126.



to populate the world with denotata to make its wheels tum-when such disorderly elements are otherwise neither welcome nor desired. This is to suggest clinging more closely to extensionalistic semantics, as suggested by much of Davidson's work, in order to provide a needed discipline. If, and to the extent that, a denotational theory of contents attempts to simulate, or reconstruct talk of intentional content, within standard referential semantics, such efforts are to be welcomed indeed. But Fodor's use of properties as denotata blurs the picture. What, after all, are the identity conditions of properties? The lesson in this is that the holistic relevance of object language theory to the interpretation of terms runs deeper. In the end it is our theories about the world which tell us what there is for our terms to denote, so that interpretation of terms seems only to make sense in relation to a theory within which such terms are embedded. 31 Since different theories have different ontologies, this seems to require differing interpretations of terms as employed in different theories. What are we to make of the idea that observational and theoretical vocabulary are "theory laden?" How could it be, as Fodor has it, that the interpretation of a term-its meaning -is independent of what particular theories say making use of the term? Let us suppose, for example, that rain, i.e., precipitation of liquid H2O from the atmosphere, causes tokenings of 'rain,' that this relationship is "nomologically necessary," that Fodor's conditions are satisfied. It would appear, then, that 'rain' means rain, as expected and desired. But since rain just is precipitation of liquid H2O from the atmosphere, it would appear that 'rain' means (or expresses) precipitation of liquid H2O from the atmosphere. But clearly we can imagine a community of speakers who understand 'rain' and yet have no ideas concerning 'precipitation' and 'H 2O'-so, 'rain' does not mean precipitation of liquid H2 O from the atmosphere in this community. (It makes no essential difference to this example that the one term is complex while the other is not, since we could as well imagine that we have a single short word which means the same as 'precipitation of liquid H2O from the atmosphere.') The point is that there seems to be no way to distinguish the meanings of 'rain' and 'precipitation of liquid H2O from the atmosphere' merely in terms of denotata-so long as corresponding properties are left out of the account. Moreover, the problem could well be posed as an empirical problem concerning a newly discovered linguistic community-a problem of radical translation. Seeing the problem so, it seems obvious that we are not going to be able to decide

31. Cf. Callaway 1981, "Semantic Theory and Language," pp. 64ff.


I 03

between two such interpretations merely by taking note of the observational situations in which the expressions are used or applied. Instead, we need to consider the complex of sentences held true making use of the term in question. For example, evidence of the absence of chemical theory in the native culture, would rule out an interpretation involving talk of H2O. Given that we are not going to be able to distinguish what properties might be involved in the tokenings of 'rain' by noting the observational conditions or situations which cause such tokenings, it seems clear that some further source of evidence must be called in. This additional evidence will require us to look closely at what else is held true making use of the term under interpretation. Clearly, then, the differing properties called for in Fodor's account, are merely (unnecessary) posits (perhaps totally alien to the native conceptual scheme) reflecting the different sentences held true. The general point is that any initial interpretation of a term (based on its use or application in observable situation) must be regarded as subject to revision in light of the full range of beliefs held true. Expressions have one meaning rather than another only in relation to a system of beliefs, distinctively expressed in usage,-which can reasonably be idealized as a theory. In order to make the point quite clear and more specific, suppose that the native expression in question is 'aqua'. Note that both of the following characterizations of the expression are extensional and suited to be included in the finite base of a truth theory: (l) 'Aqua' is true of x (in L) iff x is liquid & x is transparent & x is thirst quenching & ... (2) 'Aqua' is true ofx (in L) iffx is composed of2 parts hydrogen and one part oxygen.

Whether we should include something like (2) in the truth theory for the language of a given community is a question most clearly answered by attending to the complex of community claims involving 'Aqua'. Do they recognize hydrogen and oxygen at all? Non-linguistic evidence might also be highly relevant. Are they able to produce and manipulate chemical elements and compounds? If not, any evidence that they talk about them would come under doubt. Or, is talk of aqua in this community confined to something more like the ordinary language of 'water'? The investigator must consider and decide such questions in order to make any reasonable decision between (1) and (2). For answers to such questions will constitute empirical evidence for an appropriate interpretation of 'acqua'. The detailed investigation of how to collect and evaluate such evidence is itself



a worthy topic of further research. 32 Yet I think there can be no reasonable doubt that these are the kinds of questions we need to answer. Talk of properties only disguises the nature of the empirical problem and might well tend to block appropriate research. Fodor has amplified doubts, in his criticisms of semantic holism, focusing on the problems presented to a workable notion of identity conditions for propositional contents. He finds insoluble problems in the suggestion that a contextualized conception of identity of content (or meaning) could be employed as a scientific idealization--one which will most clearly find application to the degree that we are dealing with a community of homogeneous believers. I have argued here, however, that Fodor's doubts concerning meaning holism are not justified, and that a version of meaning holism as sketched above stands a reasonable chance of comprehending both some chief contemporary problems in the notion of cognitive meaning and specific problems which have appeared in Fodor's approach to semantic topics. In "Two Dogmas," Quine lamented the persistence of the notion that "each statement taken in isolation from its fellows, can admit of confirmation or information at all." 33 He concludes that "it is misleading to speak of the empirical content of an individual statement-especially if it is a statement at all remote from the experiential periphery." 34 If, in spite of such points, one persists with the aim of interpreting sentences and other simpler expressions, then having surrendered the notion of analyticity, this will only be viable if such interpretations are contextualized to a theory expressible in the object language. Considered apart from theory, sentences have no meaning to call their own.

32. Cf. Callaway 1988, "Semantic Competence and Truth-conditional Semantics," pp. l 8ff. 33. Quine 1951, p. 41. 34. Quine 1951, p.43.


A chief aim of this paper is to provide common ground for discussion of outstanding issues between defenders of classical logic and contemporary advocates of intuitionistic logic. In this spirit, I draw upon (and reconstruct) here the relationship between dialogue and evidence as emphasized in German constructivist authors. My approach depends upon developments in the methodology of empirical linguistics. As a preliminary to saying how one might decide between these two versions of logic (this issue is most closely approached in Section 5. discussing the constructivist approach), it is well worth the effort to look closely at how logic is (or might be) learned and at questions concerning logic in translation, i.e., the question of how we might detect the variety oflogic actually employed in a given speech community. These latter questions are, of course, worth investigating in their own right and not merely because of their relevance to disputes in logic. However, the two sides in the logical dispute have in fact employed arguments or characterized their positions based on views of how logic is learned or might be translated. Thus, issues in logic, language acquisition and semantic theory tend to converge here. Looked at more systematically, this is partly because versions of the "use" theory of meaning are relevant to each of these three areas of study. Though I will argue that the logical issue must be viewed as chiefly epistemic rather than primarily semantic (as advocates of intuitionistic logic have it), this will require development of neutral ground for discussion of the logical issue. Preliminaries, drawing on Quine's work on verdict tables, are crucial to isolating fundamental elements of the logical issue and to expanding accounts of how logic is learned as theory. In the fmal section some consequences of these investigations for the issues of semantic realism and indeterminacy of translation will be emphasized.


Research for this paper was partly supported under a grant from the Alexander von Humboldt Foundation. First published in Erkenntnis 37, No. 1, July, 1992, pp. 65-92. Stylistically revised. Reprinted by permission of Springer NL.



1. Translation and Participation Making use of "verdict tables," Quine argues in Roots of Reference 1 that observation alone will not allow discrimination between a linguistic community making use of classical logic and one making use of intuitionistic logic. For, he holds that while verdict functions "can be learned by induction from observation of verdictive behavior," genuine truth functions (whether construed in accordance with the principle of bivalence or not) cannot be so learned. Verdict functions are "more primitive" than truth functions; and "they are independent of our parochial two-valued logic, and independent of other truth-value logics." Classical bivalent logic is not learned directly from observation of speakers' behavior. Rather, "like other theory," it is learned "in indirect ways." On this account, the same goes for intuitionistic logic. Details of the differences between verdict functions and truth functions will be important below. This will provide a key to Quine's account of how logic is learned, an approach to the development of these views, and a relatively neutral evidential base for discussions of the logical dispute. For immediate purposes, however, it is sufficient to note that on the account to be found in The Roots of Reference, learning logical constants (with the exception of negation) is a matter of learning theory and not something to be treated merely on the level of stimulus meanings. In consequence, it will also be important to attend to the question of how theory is learned. As Gochet has recently emphasized, the account of learning the logical constants in Roots of Reference contrasts in some important ways with that to be found in Word and Object. For, Quine had held, in Word and Object, that the truth-functional connectives, unlike the quantifiers, escaped the effects of the indeterminacy of translation.2 The contrast is of interest here partly because the Word and Object account of learning the truth-functional connectives appeared to force an extreme and untenable conservatism upon the question of the revisability of truth-functional logic. It required, or at least appeared to require, an imposition of classical logic onto every translation of a foreign speech community. "Wanton translation," Quine had said, "can make natives sound as queer as one pleases. Better translation imposes our logic upon them ..." 3 But arguably, if in translating (or understanding) the foreigner's words we must impose classical truth-functional logic, then it will

I. Quine, W.V 1974, The Roots of Reference, pp. 75ff. 2. Cf. Gochet, Paul 1986, Ascent to Truth, p. 147 and Quine 1960, Word and Object, pp. 57ff. 3. Quine, W.V. 1960, Word and Object, p. 58.



make little sense to think of ourselves as taking up alternatives to our own logic and considering whether to adopt them or not. This implication (or apparent implication) of Quine's earlier view, while never explicitly embraced, led to much discussion of a lack of consistency between Quine's view of translating the truth-functional connectives and his rejection of the notion of analyticity. For it appeared, to some at least, that Quine had a "linguistic" theory of truth-functional truth-where any proposed revision amounts to either changing the subject of discussion or changing the notation in which the standard classical truth-functional truths are expressed. 4 Thus, the shift or development in Roots of Reference serves to facilitate discussion of intuitionistic logic from a Quinean perspective-at least by clearing away a somewhat distracting discussion of whether it is possible to revise truth-functional logic. The earlier view threatened to beg the question against intuitionistic logic by refusing to even recognize it. But Quine's view in Roots of Reference also facilitates discussion of the issues between classical and intuitionistic logic in other ways, since this account of learning the truth-functional connectives raises interesting points of contact and contrast with views to be found among the advocates of intuitionistic logic. For Quine maintains not only that classical logic is learned as theory "in indirect ways," but also that intuitionistic logic stands at a similar disadvantage. Those employing an intuitionistic logic have also adopted a theory, on Quine's account. It follows, for Quine, that we cannot learn an intuitionistic logic directly from observation of speakers' behavior. Moreover, I believe that Quine's view here is correct-given the notion of observation he employs. As we shall see, this point will serve to place some strain upon semantic arguments favoring an intuitionistic logic-those based upon versions of the "use" theory of meaning. Any viable notion of"use" will turn out to be something which is not observational in Quine's sense of the latter term. Usage turns out to be theory-laden. Clearly, if Quine's views in Roots of Reference are substantially correct, then we cannot determine, by means of observation alone, if a given linguistic community employs classical logic or an intuitionistic alternative to classical logic. As a matter of the methodology of empirical semantics, then, one wants to know how such a determination might in fact be made, and this question will be a chief point of interest below. But it will be helpful to take 4. Cf. also Quine, W.V. 1970, Philosophy of Logic, p. 82ff., Stroud, Barry 1969, "Conventionalism and the Indeterminacy of Translation," p. 86, and the discussion in Fretloh, Sigrid 1989, Relativismus versus Universalismus, Zur Kontroverse iiber Verstehen und Obersetzen in der angelsachsischen Sprachphilosophie, pp. 167-68.



up this question by means of an examination of some areas where Quine's views in Roots of Reference seem to conflict with those prevalent among the advocates of intuitionistic logic. In very general terms, advocates of intuitionistic logic favor such alternatives to classical logic on grounds that such logic is, in one way or another, clearer or more intuitively evident. Ultimately, this can be traced back to a notion of constructivity in mathematics-in terms of which it is argued that proofs making use of classical logic do not correspond to any possible intuitive constructs. In contemporary discussions, however, advocates of intuitionistic logic have frequently attempted to support their views by evoking versions of the "use" or "pragmatic" theory of meaning. The argument here would be, roughly, that the meaning of a connective, 'or' for instance, or the meaning of a mathematical statement employing the connective, must be something that we can detect in actual usage. For, it might be argued, what cannot be detected in actual usage cannot be learned--or at least it cannot be learned in the way we normally learn a language. As will be argued, however, such claims are quite problematic once it is understood that usage is not directly observational. For instance, in his Elements of lntuitionism, 5 the author argues that the "Platonist" or non-intuitionist "is compelled to repudiate the principle that meaning is use, although he is bound to admit that it is only from a training in the use of expressions in any given range that we derive a grasp of their meaning." 6 If we only suppose, as seems reasonable, that the use of an expression, i.e., the way it is used, is something directly open to observation -a matter of "training"-, then a direct conflict is evident between Dummett's view and what we have seen of Quine's view of language learning in Roots of Reference. However, if we suppose on the other hand, that usage is not in general learned by means of observation but requires the mastery of complex theory, then it will be considerably less plausible to insist that intuitionistic logic has any great advantage with respect to learnablity. Further, Dummett would have us believe that intuitionistic logic can be understood by reference to its use in a way that classical logic cannot be understood. But, if intuitionistic logic is fully expressed in its usage, then, contrary to Quine, there should be some crucial contrast in how it is or could be learned. One might suppose, for instance, that on Dummett's account we should be able to learn the intuitionistic connectives, and intuitionistic logic,

5. Dummett, Michael 1977, Elements of lntuitionism. 6. Ibid., p. 377.



merely from observation of speakers' overt behavior. But Quine argues against this. Views developed in Roots of Reference also present interesting points of contact with contemporary Gennan constructivism. In a brief article on constructivism, Thiel describes this general approach in the following terms: Der andere Ansatz des Konstruktivismus zu einer allgemeinen Wissenschaftstheorie und Philosophie wurde von der Erlanger (und Konstanzer) Schute entwickelt. Grundidee ist ein Aufbau von der Pointe, nicht nur Syntax und Semantik einer Wissenschaftssprachenach dem Vorbild einer interpretiertenformalen Sprache zur Verfiigung zu stellen, sondem aus deren Pragmatik Semantikund Syntax erst zu entwickeln.7 The approach in question gives pragmatics, rather than syntax or semantics a more central role. It is related to both Peirce's conception of pragmatism and to Wittgenstein's notion oflanguage games. In a properly scientific language or Orthosprache, all elements are to be introduced in a step by step manner and thereby grounded or justified by reference to pre-existing scientific practice. This is essentially a matter of establishing the usage of expressions. Teaching-learning situations get a quite important role in various versions of this approach, and in particular, logical particles are to be introduced in such a way that they can be understood by the student as means or tools to be used in dialogue games of argumentation. The constructivist emphasis upon dialogue as context of usage will be seen below to make contact with, in fact tending to support, Quine's proposal to regard the learning of logical constants as involving the learning of theory. In spite of this, the theoretical character of what is so learned seems not to have been well appreciated by some constructivist authors. Thus, there is a need to emphasize the distinction between acquisition and justification. In a general article on philosophy of language, Lorenz writes that "Spre-

chen und Handeln sind derart miteinander verflochten, daJ3 eine angemessene Behandlung des einen Bereichs den Bezug auf den anderen nicht

7. Thiel, Christian 1984, "Konstruktivismus," p. 451. "The other approach of constructivism to a general theory of science and to philosophy was developed by the Erlangen (and Constance) school. The basic idea is to build up a language of science ... with the point of making available, not merely the syntax and semantics, on the model of an interpreted formal language, but to first develop semantics and syntax from the pragmatics."



entbehren kann. " 8 In reference to understanding or teaching the logical constants and truth-functional logic, this thesis of inseparability of treatments of speech and action might be read as suggesting that participation in certain sorts of language-games, i.e., certain sorts of linguistic actions, are both necessary and sufficient for learning logic. The suggestion is supported by Lorenzen's approach to instruction of logic in his Normative Logic and Ethics.9 However, beyond the point concerning how logic is learned, there is also the question as to whether reference to existing practices will serve to justify one sort of logic as opposed to another. In particular, one must raise the question as to whether such a pragmatic conception of language and language learning will serve to justify the constructivists' insistence upon intuitionistic rather than classical logic. For, generally, constructivists object to classical logic, and the law of excluded middle in particular, on grounds that it represents an unjustified leap where good methodology requires a step by step progress in the introduction of the elements of a scientific language. The view is, then, that introduction of classical logic or the classically defined truth-functional connectives cannot be justified by methodological canons, explicable in terms of usage, which nevertheless do serve to justify and make clear, the corresponding elements of intuitionistic logic-and render it acceptable. We want to know if this kind of justification ofintuitionistic logic can be sustained. For the constructivists, semantics is part of the theory of linguistic action and thus part of pragmatics. What we learn of language and logic is learned through certain sorts of training. Thinking of language learning as a kind of training, the picture here might seem to fit in well with Quine's views on how to go about the first steps of radical translation. However, the constructivist emphasis upon the importance of participation in linguistic and non-linguistic action and the emphasis placed upon learning the behavior of others also points in a quite distinct direction-away from the Quinean stress upon language as keyed to observational elements of the physical environment. The Quinean translator observes linguistic behavior and its relationship to observational elements of the physical environment. The methodology of query and assent, is of course, an active one, so that Quine does not present the translator as involved in merely passive observation. However, it is arguable

8. Lorenz, Kuno 1980, "Sprachphilosophie," p. 12. "Speaking and action are so intimately interrelated that an appropriate treatment of the one area cannot do without reference to the other." 9. Cf. Lorenzen, Paul 1984, Normative Logic and Ethics, pp. 2lff.



that someone who learns the language of a given linguistic community by means of participation in on-going, linguistic and non-linguistic activities within the community has better access to what is to be learned than does someone who attends exclusively to the relationship between language and its application to the physical environment. In principle, Quine can, I think, agree with this point. It is, after all a perspective which approximates more closely to that of a normal child learning a first language. Moreover, the emphasis upon the relationship between language learning and the broader cultural domain of action is congenial to Davidson's emphasis upon the need for a "unified theory of meaning and action." 10 At the very least, it is clear that the interpretation of language and the interpretation of action serve to constrain each other in important ways. The prior question here, however, concerns the objectivity of translation or interpretation and the relationship of translations or interpretations to evidence we might come to by attending to linguistic usage as relevant to cultural practices, action, and argumentation in particular. My own view is that objective evidence for translations and interpretations is not exhausted by considering native assent and dissent from sentences in the face of environmental stimuli. The problem of detecting native logic is, from this perspective, a kind of test case. For we have every reason to hold that differences in logic between distinct linguistic communities are objective and not merely a matter of distinct "analytic hypotheses," as the account of translation in Word and Object leads one to suppose. But if we can distinguish between classical and intuitionistic logic on the basis of evidence which exceeds or extends Quine's account of the evidence available for translation, then this will strongly suggest that Quine's thesis of the indeterminacy of translation is seriously defective in its account of the proffered distinction between translations which are and are not objectively distinct. An examination of Quine's treatment of the learning of truth-functional connectives proves a good place to start-both with respect to the question of the objectivity of our translation or interpretation of the logic of a given community and also regarding the kinds of evidence that will be required. For he poses, in quite considerable detail, questions which the advocates of intuitionistic logic have also tried to answer-concerning the meanings of the connectives and the way or ways in which logic can be learned. Especially since Quine concludes that the differences between classical and intuitionistic logic are not detectable by merely observational means, Quine puts us in a

10. Davidson, Donald 1980, "Toward a Unified Theory of Meaning and Action."



good position to ask about how such differences might in fact be detected or learned. This in turn will serve to cast light upon the evidence available for semantic hypotheses and semantic arguments favoring intuitionistic logic.

2. Verdict Tables and Observationality Verdict tables, originating in Quine's work, differ from truth tables in not being bivalent. Rather than depending upon the truth or falsity of sentences, verdict tables make use of a three-way distinction between assent, abstention and dissent. This is the essential reason why Quine holds that verdict tables are more observational. "A compound sentence is a verdict function of its components," Quine says, "if a verdict to the compound is determined for each assignment of verdicts to the components." 11 By this standard, the tables Quine provides to illustrate his discussion in The Roots of Reference are not, strictly, verdict tables-the tables have gaps. But part of his point is that we do not in fact treat words such as 'and' and 'or' as verdict-functional connectives. Rather, the tables Quine provides more closely reflect actual usage. Consider, for instance, the table provided for 'and.' And




assent abstain dissent

abstain ? dissent

dissent dissent dissent


The component sentences here may be either occasion sentences or standing sentences. One notes first that dissent from either component (or both) is sufficient for dissent from the compound. This point covers the bottom row and the column on the right. It leaves us with a square in the upper left comer: four possibilities. Clearly, assent to both components leads to assent to the compound, and assent to either component combined with abstention from the other leads to abstention from the compound. The position at the center of the table is left open, and it constitutes the most interesting case. Sometimes, though we abstain from both components, we will still dissent from the compound: viz. where the two components are intuitively incompatible. Quine gives the example where P = 'It's a mouse' and Q = 'It's a chipmunk.' Obviously, we will dissent from calling something both a mouse

11. Quine, W.V. 1974, The Roots of Reference, p. 77.



and a chipmunk, even when it is unseen. In other cases, where P and Q are not incompatible, abstaining from each of the two components, we will also abstain from the compound. Quine remarks that by "specifying abstention at 12 the center, the word 'and' will be construed as a genuine verdict function." But, as we have seen, this step would be somewhat artificial. The conclusion that observation cannot distinguish between classical and intuitionistic logic derives directly from Quine's treatment of truth-functional connectives as more theoretical than verdict functions. To help see this point, a thought experiment in radical translation suggests itself. Imagine, then, that we are working to understand some radically foreign linguistic community. So far, we have succeeded in identifying some occasion sentences and some standing sentences. Next, we take the further step of interpreting some native expression as corresponding to the English 'or.' Our evidence for this interpretation is that native. components and compound sentences, making use of this expression, fit the verdict table for 'or' as shown in the following matrix borrowed from Roots of Reference.13 ?







assent assent assent

assent ? abstain

assent abstain dissent

We note that where this expression is used to compound sentences, assent to either component, or both, brings with it assent to the compound. So, our evidence fits the left hand column and the row at the top. Further evidence favoring our translation hypothesis is found in the fact that the natives dissent from a compound employing the expression in question whenever they dissent from each of the two components separately. We find, too, that when the natives dissent from one component and abstain regarding the other, they also abstain from the compound. The position at the center is once again left open. Clearly, in some cases where we abstain from the two components of a disjunction we will also abstain from the disjunction. However, for other choices of P and Q, though we abstain from both of the components separately, we will nevertheless assent to the compound-because we view P and Q as exhausting the possibilities. An example in English is a pair of sentences making use of the

12. Ibid., p. 78. 13. Ibid., p.77.



predicates 'is animate' and 'is inanimate' respectively. (I assume that Schrodinger's cat is not yet effecting our community-wide intuitions.) Though we remain ignorant of whether some particular object is animate or instead inanimate, we will (quantum physicists aside) agree that it is either animate or inanimate. Let us assume that our evidence concerning the native expression fits in, as well, with these further constraints on translating an expression by means of our word 'or'. Will this observational evidence also allow us to decide whether we are dealing here with a classical 'or' rather than an intuitionistic version? The evidence might at first seem to better fit the hypothesis of an intuitionistic version of disjunction. For, the two sorts of cases which lead us to leave the center position in the table open seem to provide such evidence. But, this is misleading. To take our evidence relevant to the center of the table as evidence of an intuitionistic version of disjunction is really to construe the evidence in terms of a further hypothesis. To see an intuitionistic version of disjunction, given only the evidence so far assembled, involves attributing a certain explanation for the difference between the two sorts of cases relevant to the center of the matrix. One needs to assume that the natives assent to the compound, though abstaining from each component, only when they have, or believe that they have, a certain sort of proof or proof procedure suited to show that 'P' is true if 'Q' is false and that 'Q' is true if 'P' is false. While our evidence is consistent with the hypothesis that the natives employ an intuitionistic logic, it is also consistent with the hypothesis that the natives employ classical logic. Though abstaining from both 'P' and from 'Q', we might assume, they sometimes accept the disjunction because they view the two components as exhausting the possibilities-by classical lights. Quine's claim that observation alone cannot distinguish classical from intuitionistic logic, in this case, comes down to the claim that the kind of observational evidence considered cannot distinguish the classical from the intuitionistic theory of when two disjuncts serve to exhaust all possibilitiesso that one or the other must be regarded as true even though the natives cannot say which component is true. Quine's conclusion here appears both true and important. However, what is perhaps even more important is that this conclusion raises the question of how we are able to distinguish between classical and intuitionistic logic. What is most relevant to this question in Quine's discussion is the distinction he employs between learning language through observation and learning language through learning theory. But it will be best to approach this topic via an objection raised by Nozick. Nozick's comments in his "Experience, Theory and Language," suggest how to distinguish classical from intuitionistic logic at the level of observa-



tion-in spite of the kind of argument sketched above. The idea is to focus on compounds of the form 'P or not-P.' In any linguistic community employing a classical logic, we might suppose, learning a word corresponding to our 'or' involves learning the law of excluded middle. By the same token, learning a word corresponding to our 'and' might be thought of as partly a matter of learning the law of non-contradiction. Thus suppose that we have already identified a native expression as corresponding to 'not.' This assumption is unproblematic here, since Quine has it that negation is both a verdict function and a truth function. 14 To identify the classical 'or' in translation, Nozick suggests looking for the following pattern of evidence: Or


Q (= not-P) ASSENT












The point is that if the natives employ classical disjunction, then we will expect that they will assent to every disjunction of a sentence and its own negation. However, if the natives employ an intuitionistic logic, then they will sometimes assent to such disjunctions and sometimes not. Nozick notes concerning such a verdict table for 'P or not-P', that there are "six cases that don't arise." 15 This point explains the x's in the above matrix. For example, we do not expect that anyone will assent to both a sentence and its negation. Thus, the question of what will be said about the corresponding disjunction simply does not arise. Rather than assenting to every disjunction of the form 'P or not-P' (where they abstain from each of the two disjuncts separately), we expect that in a community employing an intuitionistic logic, they will sometimes assent and sometimes abstain---depending upon what sentence is substituted for 'P'. This is to be expected because they will sometimes allow of proofs that the two disjuncts exhaust all the possibilities and sometimes not. Thus, it seems that if we are able to collect evidence of the matrix above on the basis of observation alone, then observation does allow us to distinguish classical and intuitionistic logic-this is the argument against Quine's view.

14. Ibid., p. 77. 15. Nozick, Robert 1986, "Experience, Theory and Language," p. 361.



In reply to Nozick, Quine comments that "if the learning of 'or' involves learning, among others, the table for 'P or not-P', then, contrary to Roots of Reference, the law of excluded middle is indeed clinched in the learning." Beyond this, he expresses doubt that "cases of 'P or not-P' would figure among the sample from which the child learns 'or."' 16 These points, while plausible, do not seem to help with the present question very much. Quine's reply focuses on early language learning rather than the question of how classical and intuitionistic logics might be distinguished among mature populations. But these questions are quite distinct. Has Nozick shown how to distinguish classical from intuitionistic logic on the basis of Quinean observation alone? The answer to this question is negative, as I will urge. Moreover, in showing why we require a negative answer, we will also see much more deeply into the limitations of learning language or logic by means of attention to Quinean observation and stimulus meanings alone.

3. Learning and Observationality Part of the appeal of Quine's notion of stimulus meaning is the appeal of thinking of language learning as a kind of training. (Although Quine has come to place considerable emphasis upon language learning as a matter of learning theory.) Experience tells us that language learning is largely a matter of drills and repetition and rote memorization. In Quine's work we get an interpretation of this as a matter of our being conditioned to assent or dissent from sentences in the face of environmental stimuli. The picture Quine draws of the field linguist's work of radical translation thus appears as an idealization of language learning-highlighting the question of how empirical evidence (or the lack of such evidence) relates to translation proposals. In addition, the notion of stimulus meaning, or something quite similar, appears to be needed in empirical semantics in order to express basic relationships between linguistic expressions and the physical environment to which they are applied. It is very difficult to imagine how else one might get started in radical translation. In so far as we think of linguistic meaning as a matter of the use of expressions in application to observable objects and situations, then we can also think of the notion of stimulus meaning as a behavioristic explication of the notion of meaning as use. Still, however plausible it may be to think of observational traits of language in such terms, it is considerably less plausible to think of meaning and 16. Quine, W.V. 1986, "Reply to Robert Nozick," p. 365.



language learning in such terms alone. Moreover, Quine has not hesitated to emphasize such points himself. In "Ontological Relativity," for instance, he wrote that for "words not directly ascribing observable traits to things, the learning process is increasingly complex and obscure." Further, he holds that this "obscurity is the breeding place ofmentalistic semantics." 17 But the point cuts both ways. One will look quite naturally to a non-behavioristic alternative in so far as behavioristic accounts appear incapable of illuminating the acquisition of non-observational elements of language. More recently, reiterating a point crucial to formalistic theories of language, Quine emphasizes that "the learning of language consists necessarily in an internalization of recursive procedures, and the genesis of language was necessarily a genesis of recursive procedures." 18 One thinks of syntax at first for an example of such recursive procedures. However, there is also an obvious connection between recursive procedures and theory, since we can think of a theory as recursively generated from its axioms. Along similar lines, Davidson's work on the semantics of natural language invites us to think of the interpretations of sentences of a language as recursively generated from interpretations of the basic vocabulary. Thinking of meaning and language acquisition in any similar terms, we depart from a purely behavioristic account of semantics in terms of stimulus meanings. Looking at sentences in terms of stimulus meanings, they are considered as unstructured wholes. Yet, we are naturally somewhat uncomfortable with the picture of attributed formal structures as merely a matter of analytic hypotheses-where there is no fact to the matter as to which formal structure is present. For, recursive procedures are thought to play a role in our ability to generate or understand indefinite or potentially infinite numbers of sentences. Moreover, the notion of formal (recursively generated) structure-as merely attributed-is particularly uncomfortable where we are concerned with formal structures related to the attribution of a particular logic. Surely language must have such formal structure, and there is little reason to think of such formal structure as observationally detectable-in the manner of stimulus meanings. But if formal structures are not observational, then they must be a matter of theoretical hypotheses; and if we cannot avoid such hypotheses in our account of formal structures (which must be present given the generative or recursive nature of language), then it is clear that some particular formal structure must be present in any given case. It will not

17. Quine, W.V. 1969, "Ontological Relativity," p. 28. 18. Quine 1986, "Reply to Nozick," p. 366.



do to hold that any of a variety of observationally equivalent sets of rules will be adequate in our account of language-there being no fact to the matter in deciding between such alternatives. For, such a view yields no explanation of the actual process of production or understanding. This point casts very serious doubt upon Quine's hypothesis of the indeterminacy of translation beyond the level of stimulus meanings. Thus, the question of how language is learned beyond the level of stimulus meanings remains open and appealing-along with similar and related questions regarding logic. To return to these matters more directly at hand, the point is to develop an account of language learning with specific application to the learning oflogic. The crucial point here is that if we assume a range of component sentences conditioned to environmental stimulation, it by no means follows that truth-functional compounds of these sentences must also be conditioned to environmental stimulation-i.e., semantic structures relevant to truthfunctional logic are not generally observational in Quine's sense. Otherwise, translation of truth-functional logic would be determinate at the level of stimulus meanings. We will do best learning to understand a language, or learning to translate, by means of methods similar to those Quine describes in Word and Object, where language itself is most observational. However, where language is less observational, methods suited to the study of the observational base of a language become less appropriate and more problematic. Quine's hypothesis of translational indeterminacy beyond the level of stimulus meaning-or the mere plausibility of this hypothesis-is a reflection of this point. Moreover, it constitutes a challenge to tighten up our accounts of the evidence available to justify translations or interpretations. How, then, are we to think of language learning beyond the level of stimulus meanings? Given Nozick's suggestion of seeking out the native reaction to the law of excluded middle itself, it becomes clear that either both classical and intuitionistic logic are detectable on the basis methods suited to detecting stimulus meanings, as he proposes, or that neither is. There is no asymmetry here in the observational situation to favor the learning of one sort of logic over the other. For, if the evidence we collect corresponding to the verdict table for 'P or not-P' is observational, then both classical and intuitionistic logic will be observationally detectable; but if such evidence does not count as observational, then neither classical nor intuitionistic logic will count as observationally detectable-or, at least, there will be no reason to think of them as being so. Thus, the Quinean approach to logic acquisition in terms of



verdict tables has already proved quite useful as a counter to the notion that one sort oflogic is more readily accessible. But clearly, if we determine whether a given linguistic community employs a certain sort of logic, then we are learning, or learning to translate, native theory. Moreover, there is little temptation to count the distinctively logical vocabulary as somehow directly conditioned to environmental stimuli. Still, seeing only this much, we do not yet see the full force of the distinction between learning language by observational methods and learning theory. Why exactly should we count native logical theory to the nonobservational part of language learning in spite of the possibility of a verdict table for 'P or not-P' allowing us to determine whether the natives employ classical as against intuitionistic logic? The answer comes when we see the consequences of the fact that logical compounds, and disjunctions in particular, are not generally observational. In Quinean terms, they are not generally conditioned to observational (nonverbal) stimuli. The most important reason for thinking this is so, is simply that there are too many logical compounds, even too many truth-functional compounds, for them all to be learned one by one. In principle, the number of truth-functional compounds which can be understood by a native speaker of a natural language is unlimited. Thus, assent or dissent from such a compound cannot be thought of as a matter of the compound being conditioned to environmental stimuli. It is, of course possible for selected truth-functional compounds to be directly conditioned to environmental stimuli-perhaps, for instance 'This is ham and eggs.' But, this is not generally the case. Thus, even where the components of a disjunction are themselves observational, it by no means follows that the disjunction must be directly conditioned to environmental stimuli. In this sense, the evidence we can collect concerning the native verdict on instances of the law of excluded middle is also not observational. In constructing a verdict table for the native equivalent of 'P or not-P' we must suppose that the natives have some recursive procedures to determine a decision concerning compound sentences, given their decisions concerning the sentences which figure as components of these compounds. Thus, though we observe whether the natives assent, dissent, or abstain from certain disjunctions, such observations only allow us to detect the native logical system given certain auxiliary hypotheses. In particular, one will need to assume, for instance, that the native decision concerning 'P or not-P' is controlled by their decision concerning 'P' together with their usual ways with the equivalents of 'or' and 'not.' We can also collect evidence concerning the native equivalents of 'or' and 'not', but when we tum from this to consider



the native reaction to instances of the law of excluded middle, then we can confirm a hypothesis concerning their logic only assuming that they bring all their relevant constraints to bear on their decisions concerning instances of the law of excluded middle. In short, when we do detect native logic, the information used to fill in a verdict table will count as evidence confirming or disconfirming hypotheses concerning the native logical system. Clearly, we could find evidence sufficient to confrrm such hypotheses beyond any reasonable doubt, though the hypotheses will not count as observational in the sense that Quine's stimulus meanings are observational. But, this is reason enough to doubt the claim that beyond stimulus meanings there are only "analytic hypotheses" and no facts to the matter to enable us to decide between analytic hypotheses. For, this would be to return to a view according to which we can only attribute our own logic when translating and where, presumably, it makes no objective sense to think of ourselves as examining alternative logics with an eye toward their evaluation-a too onerous conservatism concerning the revisability of truth-functional logic. Beyond the level of observational features of language, and beyond the limits of the Quinean notions of stimulus meaning and stimulus synonymy, then, sentences and other expressions of a language must be interpreted in relation to each other and not one by one. To understand sentences beyond the limits of stimulus meaning, one must find the place of any such sentence in a system. It is such points which normally require the role of an active participant on the part of one who learns a language. Argumentation is normally an important element in such an active role. For it is certainly inviting to think of theory as being learned by means of participation in the practice of argumentation. To learn theory is to learn what other sentences to accept, given the axioms. One must learn to apply rules of inference consistently and to bring the needed axioms into play. But there is simply no way to check on whether someone is doing all this except by looking at an entire system of sentences-though, of course, we can only do this a step at a time. Knowing our place in an argument or dialogue helps to keep the steps identified in terms of their place in a system. This provides a kind of evidence for interpretations-evidence drawn from inferences made-quite distinct from the evidence involved in interpretation at the level of stimulus meanings. We are able to acquire linguistic competence involving



semantic structures which are non-observational-by interpreting individual sentences in relation to each other. 19 Although we have been most directly concerned with learning and detecting the logic of a particular speech community, the contrast between direct conditioning and learning theory has more general application. In particular, since there is reason to think of both syntax and meaning or interpretation as matters dependent upon some recursive procedures, there are grounds for thinking of most language learning as a matter of learning theory-and as dependent upon argumentation. Drawing upon argumentation as a source of evidence for interpretations, in tum, will plausibly overturn the notion that interpretation is only objective to the level ofQuine's stimulus meanings.

4. Learning Logic as Learning Theory The native reaction to disjunctions, and other logical compounds, is mediated by theory. To learn to interpret native compound sentences, either for translation or simply to understand them, requires more than noting which native sentences are conditioned to specific observational stimuli. We must learn native logical theory as expressed in usage. How do we learn the native theory? We can begin to answer this question by taking note of how we learn our own theories and by considering the state of a person who is at first without, or substantially without, the theory in question. Obviously, if we must learn logic by learning theory, then it must be possible to be without it. In a sense, this point has already been made by means of the Quinean description of language at the level of stimulus meaning. The point is also implicit in Quine's claim that it is "unlikely that cases of 'P or not-P' would figure among the samples from which the child learns 'or'." Imagine, then, a child who has not learned sufficient logic to assent to all instances of 'P or not-P'. We may assume, consistent with this, that the child has come to assent to or dissent from various observation sentences in the face of appropriate stimuli, and even that the child has learned to dissent from 'not-P' where he or she assents to 'P'. Leaming to assent to every instance of 'P or not-P' represents a further step here. Though this is an easy step into theory, it is still a step into theory. It is an easy step because it represents a generalization along an instinctively prepared similarity gradient: we must suppose that we have a built-in quality space reflecting the similariI 9. Cf. Callaway, H.G. 1988, "Semantic Competence and Truth-conditional Semantics," for closely related points concerning interpretation of the quantifiers.



ties among humanly producible vocalizations. What sentences of the form 'P or not-P' all have in common is a complex phonetic characteristic (or plausibly, depends upon some set of complex phonetic characteristics). Leaming this much logical theory amounts to learning to generalize assent along lines of recognizable phonetic similarities of sentences. (There is, of course, little reason to think that this much logic is involved in acquiring a rudimentary usage of the word 'or'.) Taking note of such small steps into theory, we can also imagine their incomplete or defective realization. Contrary to Quine in Word and Object,20 the description of a pre-logical mentality (or partly pre-logical mentality) could be given in similar terms. Remaining pretty much at the level of sentences directly conditioned to non-verbal stimuli, for instance, it is not too difficult to imagine that a child might be conditioned to assent to S under stimulus s when queried by one person and simultaneously conditioned to dissent from S when queried by another. Such is one unfortunate effect of authority on learning. Thus, such a "pre-logical mentality" would not be a great surprise in traditional (pre-scientific) cultures where the force of authority is stronger. Inconsistency, even where conscious, might seem a small price to pay for acceptance by conflicting authorities. Defective or incomplete learning of logical relations among sentences is possible in various ways. Moreover, it is not difficult to imagine that where practice is limited some logical principles might be more or less strongly entrenched-in terms of how likely it is for a person to respond negatively to utterances which involve their violation. It is crucial to note that all such stages on the way to acquisition of logical competence involve various reactions to sentences rather than reactions to the things the sentences might be about. Thus, one will expect such defects to be exposed in the learner's reactions to sentences-i.e., in dialogue or argumentation. By the same token, detection of an alternative logic in a particular community will depend upon recognizing similar-but systematic-differences in the community reaction to sentences-as suggested by Nozick. We see here one sure sign of the limits of Quine's notion of stimulus meaning, since this concept restricts attention to cases where assent or dissent from a sentence S is directly conditioned to non-verbal stimulus. Quine mentions, in reply to Nozick, that he was aware that "exclusion of verbal stimuli has the undesirable effect of crippling the stimulus meaning of certain sentences in an arbitrary way." He notes that "stimulus meanings do not carry

20. Quine, W.V. 1960, Word and Object, pp. 58ff.



us far in any event." 21 In particular, stimulus meaning will not carry us so far as an account of the logic ofa given community. It is crucial to detecting the logic of a given community to attend to reactions to sentences. Logic is a matter of sentence relations, and evidence of the logic of a community must therefore depend upon evidence of inference between sentences. Moreover, without attention to such relations, we will also have to forego explanation of the "obviousness" of logical truths-in terms of a built-in phonetic quality space.

5. Semantic Competence and Argumentation Attention to conversation, dialogue, argumentation and participation in ongoing community practices will carry us further in accounts of language acquisition. It puts us in a position to take notice of reactions to linguistic utterances, and the practices in which they are embedded. Much in the constructivist approach to language is suggestive of this, though also linked to a quite restrictive conception of justification or validity-a conception which will be disputed here. Where I part company with the constructivists is in their tendency to equate validity with community acceptability. In his Normative Logic and Ethics, Lorenzen equates a sentence having a meaning with the condition that "it is determinate how to defend it in a dialogue." 22 In a more or less corresponding way, he defines 'truth' as "defensibility in all cases." 23 My basic objection is that there may be true and meaningful sentences that we do not know how to defend and that 'defensibility' is not a well defined term. What we count as an adequate defense of a view evolves with the progress of science and the development of theory. 'Defensibility' (or 'justification') is not suited to the task of defining 'truth'. It was perhaps that old, unreconstructed yearning after "absolute" justifications which first lent appeal to intuitionistic logic. What is perhaps of more interest is to emphasize the role that dialogue and argumentation may play in someone's learning his own language and its (presently) accepted semantic rules. Lorenzen claims, in a somewhat Wittgensteinian spirit that "to understand a sentence very often means to know how to use it..." 24 The aim is to base the theory of meaning within the theory of usage, i.e., pragmatics. Teaching dialogue rules is then thought of

21. 22. 23. 24.

Quine 1986, "Reply to Nozick," p. 365. Lorenzen 1984, Normative Logic and Ethics, p. 47. Ibid., p. 27. Ibid., p. 22.



as a matter of teaching (a particular) use oflanguage. However, my aim is to adapt general features of this approach while detaching the quasi-rationalist epistemology. Meaning is not to be equated with usage in argumentation, and actual usage is defeasible as justification, but such usage must be treated as evidence for meanings. There can be little real temptation to equate meaning outright with usage. For, many examples of usage have little to do with linguistic meaning in any plausible rendering of that term. What we "do say," and what we "don't say" often has more to do with the complex of non-linguistic goals which are expressed in our use of language. What is more reasonable is to insist that usage provides access to the meanings of expressions and that we in fact learn language, in large part, by means of our participation in human interactions having both linguistic and non-linguistic elements. It is more plausible, then, to regard language acquisition as a matter dependent upon usage. Usage is the crucial evidence of meaning, though meaning is more theoretical in the sense that it serves to explain (more or less) systematic usage. My view is that interpretations or meanings of expressions are transmitted between members ofa linguistic community by means of their joint participation in linguistic and non-linguistic activities. We learn our own language (especially in its higher reaches and as relevant to semantics) by learning and disputing the practices in which the usage is embedded and the theories or belief-systems that the language is used to express. Leaming theory, or belief-system, is sufficient for learning the meanings of the relevant expressions, and such theories or belief-systems are expressed in relevant usage. Argumentation or dialogue between members of a linguistic community is especially important in the process by means of which interpretations are transmitted, and in consequence we should expect that attention to such argumentation, dialogue and the inferences made will be especially important to an understanding of a linguistic system. Meanings are not to be identified with usage, and nor is one form of logic more evident in usage than another; rather meanings are posited to explain systematic usage (most directly evident in patterns of inference) and to distinguish it from usage which is not systematic-e.g., that linked to idiosyncratic purposes of individuals or of particular groups. One will expect, then, that attention to linguistic and non-linguistic practices in a given community will provide evidence for the interpretation of language. Where the practices in question are bound to a specific human culture, we should expect as well that the theory or belief-system, along with the related semantic rules, will also be culture-bound. However, where the



relevant practices have a better claim to universal validity (as in the natural sciences, for instance), then we should also expect the theory or belief-system to have a better claim to universal validity. There is plenty of room in the view sketched for skepticism on the idea that a pragmatic foundation for semantics serves to justify one belief-system over another or one set of semantic rules over an alternative. Classical and intuitionistic accounts of logical truth, for example, can be viewed as based upon different versions of good practice in mathematics. Thus, to justify one system of logic over the other would require that we justify the intuitionist's practice in mathematics (or alternatively classical practices in mathematics) in order to justify the corresponding logic and the corresponding theory of mathematical proof. This kind of project would involve investigations in mathematics, general epistemology or philosophy of science, and some account of the relationship of logic as employed in non-mathematical sciences (and general discourse) and its employment in mathematics. For, even supposing that good mathematics requires an intuitionistic logic, we would want to know if this might force the adoption of intuitionistic logic for general purposes. Merely showing that a given logic fits the practices of a certain group of language users (say intuitionistic mathematicians) tells us little about how to justify that logic-either for mathematics or for general purposes. Rather, we will merely have identified the logic conventionally employed in a certain group. Yet, this point conflicts with the general neoWittgensteinian tendency to be found among constructivist authors. In his Elemente der Sprachkritik, for instance, Lorenz refers to the "bloj3 deskriptive Aufgabe der Philosophie," 25 and he acquiesces in Wittgenstein's distinction (and attitude) concerning description of language in its relation to language games and forms oflife. 26 The basic problem is that this descriptive conception of the role of philosophy allows very little room indeed for a reasoned decision between different varieties of usage. We want to know what would allow us to decide between classical and intuitionistic versions of logic or mathematics in a principled way, but such questions become very problematic in the constructivist tradition. Thus Thiel gives the following defmition of 'Geltung' in a short Encyclopedia article: "haufig gleichbedeutend mit intersubjektiver Verbindlichkeit, terminologisch die (solche Verbindlichkeit alterers! ermoglichende) objektive Grundlagen des Anerkanntseins van Satzen, Gesetzen, Normen, Werten etc."27 How,

25. Lorenz, Kuno 1970, Elemente der Sprachkritik, p. 115. 26. Ibid., pp. 115-119. 27. Thiel, Christian 1980, "Geltung," p. 729.



then, are we to show that there is an "objective ground" for one version of logic? What is it that provides for the intersubjective validity of our logical laws or mathematical proof procedures? Lorenz has it (in his comprehensive article on "Sprachphilosophie" in the Lexikon der germanistischen Linguistik) that "Die Geltung der Aussage ..., ihre Verlaj3lichkeit,wird als Erfilllbarkeit ihres Sinns behauptet," while "es ist... zulassig, den Sinn einer Aussage, ihre Verstandlichkeit, mit dem Sinn desjenigen Artikulators zu identifizieren, der diejenige Situation artikuliert, die herbeigefilhrt werden muj3, um den Geltungsanspruch der Aussage einzulosen." 28 The 'Geltung' or objective validity of a statement depends upon the fulfillment of its sense. But since the sense or meanings of statements are regarded as being established by their use (and the use in practical situations and teaching-learning situations is of particular importance in this), there is a danger of arguing from established or accepted standards to objective validity-when, in fact, this would tend to ignore the ways in which the evolution of scientific (theoretical-empirical) investigations tend to change the meanings of terms, the standards of acceptability, the methods of testing and even the advisability of accepted practices. On the other hand, if such openness in our notion of objective validity (or truth) is allowed, then the door is also open to a very unconstructive realism. More generally, it should be noted that though the Wittgensteinian usage of expressions is rooted in "forms of life," these forms of life are also subject to change and revision-changes often induced by scientific and technological advances. With Quine, I would emphasize the defeasible character of established or accepted knowledge (and usage) in the holistic evaluation of new developments. By the same token, there is plenty of room for skepticism of the idea that consulting practices will allow us to justify certain meaning assignments (or semantic rules) rather than others. Basically, an evaluation of meaning assignments (defmitions for instance) will only be possible as part of the general undertaking by means of which we attempt to justify (and criticize) existing belief-systems. Scientific defmitions get changed when we have reason to reject one scientific theory in favor of another. Paradigmatically, one thinks of Einstein's rejection of Newton's defmition of 'force'. We change other sorts of defmitions when we change the underlying social practices. Thus, conservative practices connected with marriage, dating and morals dictated the defmition of 'bachelor' (explicit in dictionaries from the 30's) as a man who has never been married. More liberal practices of later

28. Lorenz 1980, "Sprachphilosophie,"p.14.



years are connected with the definition of 'bachelor' as a man who is not married, i.e., 'an unmarried man'. But merely seeing the connections between such definitions and related practices does not serve to determine which definitions (or practices) to accept. Instead, by seeing this connection, we merely come to understand the reasons behind certain linguistic practices and the relationship between conventional definitions and the on-going life of a community. We get a justification of the meaning assignments (and the related belief-system) relative to the practices. Nothing in this will prevent a (reasoned) reform of the practices along with all the rest. Thus reference to practices and methodology give no final, unquestionable justification to semantic rules erected on such a basis. It is to the point to recall here that the basic meaning of 'conventional' is 'sanctioned by general custom'. To be sure, definitions can be fairly regarded as conventional in this sense. They are sanctioned not only by custom regarding the way words are used, but also by the underlying (scientific or non-scientific) belief-systems and related activities and ways of life. But to say that definitions, or meaning assignments, within a linguistic community, are conventional in this sense is very far indeed from saying that they are stipulative or arbitrary. Thus, to say that meaning and usage arise from forms of life and particular practices should not lead us to regard them as fixed and unquestionable. (Rather the problem is often to see how forms of life, and embedded linguistic practices, might best evolve-in the face of scientific and technological change for instance, or to accommodate relationships to other peoples.) However, what does appear reasonable, especially for present purposes, is to treat what we can see ofa community's form oflife as clues to the interpretation of its linguistic systems. The aspect of this which will prove to be of most particular importance for semantic purposes, is the practice of argumentation. To sum up on choice between logical systems, it cannot plausibly be claimed that the practices of classical mathematics cannot be learned or do not show in their usage of language. The question the defense of intuitionism should be addressing is rather whether the practices and usage associated with proofs in classical mathematics can be justified. As we have seen, this will involve an evaluation of the usage of language in argumentation. The central question will be whether certain forms of argumentation employed in classical mathematics are good ones or not. For, once having seen that the notion of usage is so multifarious as to extend from training to highly developed forms of argumentation, there will be little temptation to think that any



argument from linguistic usage alone will serve to answer more substantial questions.29

6. Semantic Realism and Alternative Conceptual Schemes We have seen above Quine's retreat from the position of Word and Object, according to which translation of truth-functional connectives was determinate (and apparently inevitably classical) at the level of stimulus meanings. Given the contrast between truth-functions and verdict-functions in Roots of Reference, the logic of a given community was no longer determinate at the level of stimulus meanings. Leaming logic was accounted a matter of learning theory. I have argued that an account of learning classical as against intuitionistic logic, as a matter of learning theory, serves to emphasize the role of determinate recursive procedures involved in informants' assent or dissent to truth-functional compounds. A difference in logical theory between two linguistic communities is then empirically detectable as a matter of their dispositions to assent or dissent from systems of sentences: a difference in recursive dispositions, one might say, relative to other sentences already accepted or rejected. While such differences will not become evident if we only consider the informants dispositions regarding isolated sentences, I have argued that decisive evidence will be forthcoming in argumentation or dialogue-where we go through a system of sentences, with the informants, in a step by step manner. These points strongly suggest the inadequacy of Quine's account of the observational evidence available to support translation proposals or interpretations, and in consequence they also suggest the inadequacy of arguments for Quine' s thesis of translational indeterminacy in so far as these arguments presuppose Quine's limited conception of linguistic dispositions. The complex recursive dispositions which become evident in argumentation, I have urged, are crucial to the interpretation of particular sentences which form part of a system. The unit of interpretation is not the isolated sentence but rather systematically related sets of sentences.30 Here I want to briefly generalize on this point and provide some perspective by reference to recent debates. Any account of the evidence for interpretations which concentrates exclusively upon dispositions to assent or dissent to isolated sentences (particularly where there is too much concentra29. Cf. Quine, W.V. 1981, "What Price Bivalence?" 30. Cf. Callaway 1988, "Semantic Competence and Truth-conditional Semantics," and Callaway 1981, "Semantic Theory and Language," above.



tion upon observation sentences) will be inadequate-it will exclude crucial evidence. In consequence, Quine's account of evidence for translation of the logical connectives is inadequate, and since there is every reason to believe that there is an objective difference between classical and intuitionistic logic, and it appears that the evidence available from consideration of argumentation will be sufficient to distinguish the logic of a given linguistic community, it is clear that there are determinate translations beyond the level of Quine's stimulus meanings. However, the hypothesis of the indeterminacy of translation has always depended upon the claim that evidence for stimulus meanings exhausts the evidence for translations. Consulting the evidence available from argumentation, we should expect, in general, to be able to distinguish different theories, and different meanings of embedded expression, whether or not such theories issue in all the same observational consequences-as defined by reference to stimulus meanings in Quine's sense. Such considerations also cast doubt upon Quine's argument from proxy functions. 31 Though this cannot be considered here in any detail. 32 By considering the translation of the logical system, we have come to see that there is evidence available for the interpretation of a system of sentences in relation to each other which does not become available considering the stimulus meanings of sentences in isolation. The general point is that we need not depend exclusively upon the stimulus meanings of the observational consequences of a theory when it comes to interpreting the theory as a whole. To put the same point in a more Quinean fashion, we have found reason to hold that there is an objective difference, detectable from evidence available in argumentation, for differences in conceptual system between different linguistic communities-though such differing conceptual systems be used only to systematize the very same evidence. If Davidson is wrong in rejecting "the dualism of scheme and content, of organizing system and something waiting to be organized," as Quine has it,33 then Quine was certainly wrong to rule out the possibility of finding other logics in translation (since a difference in logic is a paradigmatic difference in conceptual scheme) and more generally wrong to reject the possibility of an objective semantic difference between empirically equivalent theories or theory formulations. There is no reason to expect difference in the observa-

31. Cf. Quine, W.V. 1990, Pursuit of Truth, p. 96. 32. Cf. Callaway, H.G. 1988, "Review ofGochet," p. 55. 33. Cf. Quine, W.V. 1981, Theories and Things, p. 38ff.



tional consequences of two object language theories merely on grounds of their employing distinct logics. Thus, if such distinct logics are detectable, then there must be evidence available to establish an objective semantic difference between object language systems even where there is no observational difference in the logical consequences of the systems with respect to observation sentences. If we are to make sense of differing conceptual schemes, then I think we will do so in terms of the possibility of semantically distinct, but empirically equivalent, theories or systems ofbelief. 34 One source of failure to see such points, as we have seen, is an overly narrow account of the evidence available in empirical semantics-linked to the behavioristic semantics of Word and Object. A further source is an overly robust versions of the "principle of charity." In seeking areas of agreement as a basis for translation and mutual understanding, one must not be so "generous" as to force the assimilation of alien conceptual systems to that of the translators. Too much generosity or charity in interpretation can amount to a veiled demand-out of ignorance-that others assimilate their thought to our own. "Given the underlying methodology of interpretation," Davidson has held, "we could not be in a position to judge that others had concepts or beliefs radically different from our own." 35 In Davidson's critique of the distinction between conceptual system and content, finding an alternative scheme is made to depend upon "two kinds of cases that might be expected to arise: complete and partial failure oftranslatability."36 But while Davidson is correct to reject the notion of untranslatable languages, and the positions he criticizes (in Worf, Kuhn and Feyerabend) do appear to involve such claims, Davidson has construed the target of his criticism too broadly. His version of the principle of charity appears to generalize the approach to translating truth-functional connectives (from Word and Object) which Quine later gave up in Roots of Reference. Yet, insisting, with Quine's later view, that we can detect conceptual differences (whether in logic or more generally) in translation presents a challenge to the hypothesis of translation indeterminacy. Taking that hypothesis quite

34. Cf. Callaway 1988, "Review of Gochet." 35. Davidson, Donald 1974, "On the Very Idea ofa Conceptual Scheme," p. 197. 36. Ibid., p. 185.



literally, one would expect to find, as Davidson argued, that there is no fact to the matter regarding mere conceptual difference.


This book follows Christopher Hookway's work on Quine in the Stanford University Press series on "Key Contemporary Thinkers." As such it recommends itself, since Hookway provides one of the best critical introductions to Quine. Evnine attempts a similar service on the complex challenge ofDavidson's writings. The strength of this book is its accomplishments in comprehending both theory of action and theory of interpretation in Davidsonincluding crucial tensions. The weakness is that the reader must struggle to separate the author's own convictions-a task stressed here. The book claims accessibility "to readers with little philosophical background." Thus, technical aspects have been "restricted to 4 sections, which the reader can omit if necessary." However, these deal with intensionality and extensionality, Tarski's theory of truth in relation to semantics for natural language, and the relationship between holism and indeterminacy. Thus, the reader who omits them will also miss crucial aspects of Davidson. Though the book is generally clear and well-written, claims for accessibility are overstated. The bibliography takes us up to 1990--though not including Davidson' s Dewey Lectures. Tracing the background of Davidson's work in the positivists' philosophical emigration of the 30's and in Quine, Evnine's "Introduction" offers a "map of the terrain to be covered" which stresses the "rationalistic" character ofDavidson's views on holism and rationality. Thus, "his main philosophical concerns ... language, the mental and action ...are the ingredients of a philosophical anthropology." In spite of Quinean roots, the view is that "Davidson has now wholly removed himself, philosophically speaking, from the empiricist tradition."' The result: a "rationalism, ...a genuine, non-empiricist philosophical vision." 2 Though appealing to those of a "rationalist" leaning, this theme seems to generalize Davidson's criticisms ofQuine's behaviorism


Review of Donald Davidson by Simon Evnine, 1991. This review was originally published in Philosophical Quarterly, (Scotland), 43, October, 1993, pp. 555-560. The present version briefly expands on a few points. Reprinted by permission of the editor. I. Evnine, Simon 1991, Donald Davidson, p. 4. 2. Ibid., p. 6.



as a lapse from empiricism. Often, it arises from interpretive gloss just where the tough-minded reader seeks quotation.

1. Rationalist Idealism? The argument rests, in part, upon Davidson's rejection of the "third dogma of empiricism." "In rejecting this dogma, along with the first two, Davidson rightly sees that there is nothing distinctively empiricist about his position anymore." 3 Quine once criticized Davidson in similar terms,4 but a quote from Davidson is lacking. The plausibility of the interpretation is connected with the fact that the disjunction between rationalism and empiricism has grown less important. If we think of Quine as "relative empiricist," for instance, urging that our hypotheses should go no further from the evidence than necessary, then we might also be tempted to see him, ignoring his protests, as equally a "relative rationalist:" urging that we adopt the most modest "rational" principles which will encompass relevant evidence-though adoption of general principles is in no sense dictated by evidence alone. But, what this comes to is just Quine's pragmatism and fallibilism. Or, looking to Popper's "critical rationalism," we might take the noun seriously, but stress how Popper's viewpoint differs from traditional rationalism: not by maintaining that knowledge must "arise from" the senses, but rather in maintaining that insight and hypothesis are not self-certifying: they stand in need of constant, strenuous, empirical tests. There is some plausibility to Evnine's overall interpretation. However, if such a "rationalist idealism"5 seriously departs from emphasis upon empirical evidence, as in Quine, Popper and others, this will be a weak position-standing in dire need of an in-depth reevaluation or expansions. In particular, one expects no viable semantics for natural language lacking considerable emphasis upon empirical evidence. Evnine leaves room for doubt. On facing pages in the section which introduces the term "rationalist idealism," we find some apparently conflicting points on the role of epistemology in Davidson's views. On the one hand, given the rejection of the dualism of scheme and content, "Davidson has

3. Ibid., p. 147. 4. Quine comments that by abandoning "what he calls the third dogma," Davidson thereby parts "the last mooring of empiricism." See, Quine 1981, "The Very Idea of a Third Dogma," p. 38. 5. Evnine 1991, Donald Davidson, p. 147.



changed philosophy." For, "epistemology ceases to be the study of some sort of intermediary;" and "in a sense one could say that for Davidson, epistemology is no longer a real issue." 6 We are to reject traditional empiricist notions of the mind "as a passive receiver of sensory information;" and the notion that the mind has "a way of processing that information in terms of concepts or theory," also "no longer has a place." The rejection of the "third dogma" is of one piece, on this interpretation, with the rejection of interposed ideas or sense-data mediating between knower and known. But though Davidson does reject such epistemic intermediaries, there seems room to think that the real target in the rejection of the "third dogma" is (untranslatable) conceptual differences, "incommensurability" and conceptual relativism. Davidson continues to talk of both concepts and theories. We read on the facing page of Davidson's claim "that given a correct epistemology, we can be realists in all departments." 7 Epistemology, is still an issue in some sense. What that sense is we are not told in these passages. Our author seems to suggests a "Copernican revolution" (though some would rather think of it as "Ptolemaic reaction") in Davidson's realism. While Davidson says that " ... knowledge is of an objective world independent of our thought or language," this point is "secured," according to Evnine, by means of the thesis that "mental states are constituted directly by the ideal of rationality (the Principle of Charity) and the way the world is (the veridical nature of belief)." According to Evnine, "Davidson is an idealist, about propositional contents." For, "what people actually believe, desire and mean is, at least in part, constituted by what it is ideally rational for them to believe, desire and mean." 8 Heady stuff. Is Davidson forced to such conclusions to avoid an (intentional) realist "night" in which all concepts are black? We can fmd Davidson explaining, in reply to Rorty's probings, that he "ought not to have called (his) view a coherence theory" if only because it does not lead him "to conclude that reality and truth are constructs of thought." 9 Moreover, Davidson "pretty much concurs," with Rorty that his view of truth amounts to a rejection of both classical coherence theories and classical correspondence theories; it "should be properly classed as belonging to the pragmatist tradition," though Davidson rejects both James and Peirce on truth. The resemblance seems to be of a more general character. If

6. 7. 8. 9.

Ibid., p. 148. Cited in Evnine, p. 149. Evnine 1991, p. 149. Davidson, Donald 1987, "Afterthoughts, 1987" p. 155.



doubt presupposes a problem in established belief, the traditional pragmatist argument goes, then no genuine doubt can be global. Davidson seems to extend a similar perspective to interpretation, linking this to the claim that belief is "intrinsically veridical." While Evnine emphasizes Davidson's "A Coherence Theory of Truth and Knowledge," 10 equal stress on the corrections is lacking. Davidson follows Quine in holding that "We have no choice, but to read our own logic into the thoughts of a speaker." Quine says this for the sentential calculus, and Davidson "would add the same for the first-order quantification theory." But Davidson might regard this as a-necessary but preliminary-methodological step, requiring some (relatively) small correction-for a community of intuitionist mathematicians for instance. Such a correction would not be inconsistent with the interpreter finding this community consistent and correct "on the whole" in its beliefs. Davidson does allow room for the interpreter to find error among the beliefs of a speaker-and why not non-global conceptual differences as well? Might the owl of Minerva begin modem daylight excursions?

2. Anomaly of the Mental In the first chapter, interpretation of Davidson' s thesis of the anomaly of the mental is highlighted by texts from Ryle and Wittgenstein. This is not, then, a scientific perspective, treating mental states as theoretical entities; rather it is important to sympathize with Ryle's behavioristic disdain for reference to "occult episodes," and to appreciate Wittgenstein's view that "inner goings on" are not essential. Yet, one deviation from this orthodox "ordinary language" critique of "the ghost in the machine" is that "Davidson defmitely believes that mental states are the causes of overt acts and utterances..." 11 This point is crucial in Davidson's "Actions Reasons and Causes." Evnine provides a reasonable and clear interpretation of Davidson on the anomaly of the mental, and there is interesting stress on the tensions it is wont to create in Davidson's views. For, to put it directly, Davidson was correct to argue in "Actions Reasons and Causes" for the compatibility of a causal role for mental states and the "rationalizing" and justificatory role of our attributions of mental states. However, fuller justification of particular causal claims depends upon noting causal regularities (to distinguishing them from accidents or happenstance). Thus, giving a causal "role" to mental 10. Davidson, Donald 1983. 11. Evnine 1991, Donald Davidson, p. l 0.



states, we also seem to be committed to causal regularities involving mental states; but this is contrary to the thesis of the anomaly of the mental. Mere "psychological generalization" which are not law-like, (cf. the discussion on p. 20) will not secure Davidson's view on the causal role of the mental. There is perhaps a bit too much of the perspective of Ryle and Wittgenstein here. Davidson, in contrast, seems to stand somewhere between a purely "rationalizing" conception of the attribution of mental states and a position more like Fodor's-where we do expect to find causal laws. Perhaps Davidson will move in this direction under pressure from Fodor's arguments -of the sort rehearsed on pp. 22ff. For, if reasons are causes, then we will expect that "there is also mental reality which underlies any attempt to interpret it." 12 Moreover, Davidson holds, in "The Conditions of Thought" (1989) for instance, that "in the simplest cases the events and objects that cause a belief also determine the contents of that belief." 13 The point is closely connected with the idea of innate quality spacing or inborn standards of perceptual similarity. Is this not a law-like causal regularity involving the mental? One can "legalistically" agree with the author's reply to Fodor's arguments-i.e., that "there is no underlying mental reality whose laws we can study in abstraction from the normative and holistic perspective of interpretation." 14 But this might be read as consistent with causal regularities and a mental reality, if we maintain that interpretation is in fact holistic (or semiholistic), and requires recognition of normative principles people actually employ in thought-principles not merely attributed to them. A distinctive logic, in a community of intuitionist mathematicians, seems a realistic case in point-and a paradigmatic conceptual difference. Moreover, one might reasonably expect genuinely predictive generalizations here. As Evnine concisely explains, by reference to Dennett's instrumentalist approach to the intentional, "By making ever stronger assumptions about something's functioning, we can gain even simpler and more compact methods of dealing with it." 15 For instance, imposing an idealized conception of rationality, in interpretation and prediction of behavior, we can neglect a great deal of detail, e.g., regarding the specific program of a chess-playing computer or its functioning as a physical mechanisms. Instead, we can predict its next move by considering that which it "ought to do." The computer

12. Ibid., p. 23. 13. Davidson, Donald 1989, "The Conditions of Thought," p. 195. 14. Evnine 1991, Donald Davidson, p.23. 15. Ibid., p. 158.



will make the move it is "rational" to make; though there is a "possibility that the program will be inadequate to find the best move for the computer." 16 In spite of some resemblance to idealist themes, it seems more conservative to see traces of a Dennett-like instrumentalism in Davidson's approach to rationality and interpretation.

3. Rationality Naturalized? Davidson does reject "naturalized rationality," and, in his views, this point wars with the causal role of the mental. 17 This book is particularly good at drawing together and exhibiting the resulting tensions. It is a very useful book, leastwise for the specialist deeply concerned with the spectacle of competing metaphysical perspectives and the present situation in analytical philosophy-as it confronts traditional themes from continental sources. It seems that Davidson's work is uniquely placed to assist in mediating contemporary expressions of analytic and continental traditions: an important and difficult task. Gadamer's hermeneutics (which Evnine does not bring in) seems to be especially relevant to Davidson's work on interpretation: a point which is well illustrated by Davidson's recent Hegel Prize Lecture "Dialectic and Dialogue" (1993). Evnine assists in the overall task by providing both a concise overview of Davidson-and a somewhat continentally oriented perspective. Though we remain skeptical of a "rationalist idealism" in Davidson, there can be little doubt that the dark figure of Hegel is brooding in the background of contemporary developments. This book makes a start on bringing Hegel out of various closets. None of this is completely new of course: it is one of the remarkable facts of contemporary philosophy that we are faced with confrontations of realist, pragmatist and idealist sources and influences: much like the situation in the early part of the century. But we are not merely going in circles, for the analytic tradition has placed us in a position to look at these conflicts in a new and more intense light. Much of this comes to a head in questions and debates regarding the nature of rationality. The analytic tradition struggles with a scientific orientation running to get a glimpse of itself in the act. How else?-by evolving (thus changing) scientific means. On the other hand, continental traditions seem still unwilling to acknowledge that the "rationality" of traditional idealisms (where the Absolute slumbers on serenely in the face of the merest persistent Sittlichkeit) amounts to harkening to "reason," as 16. Ibid., p. 159. 17. Cf. /bid., p. 157.



Peirce put it, merely as "that which 'we' find ourselves inclined to believe:" -which tends toward mere acquiescence in narrowly culture-bound consensus however conservative or self-satisfied. Partly by scuttling the overly tough-minded behaviorism of his mentor Quine, Davidson helps us see the conflicts of these traditions more clearly. For instance, his work helps raise the question of what a non-behavioristic approach to semantics and the mental might be. While Davidson does not always bristle with science in his later work, he cannot disregard it. Thus, his work supports an inter-cultural dialogue which does not require us to lose sight of our intellectual roots. Davidson emphasizes, at the close of his "Dialectic and Dialogue" a point from Gregory Vlastos-which facilitates a focus on basic intellectual values, a focus surely needed for fruitful dialogue between diverse traditions. If, like Socrates, we practice the dialectic method we must also accept "the burden of freedom which is inherent in all significant communication." Like Socrates, we cannot expect to always end up with the most congenial opinions and positions. By helping us to see Davidson more clearly, this book, among other accomplishments, may help us to better appreciate this point as well.


This paper explicates and defends a social-naturalist conception of intentionality and intentions, where intentionality of scientific expressions is fundamental. Meanings of expressions are a function of their place in language-systems and of the relations of systems to object-level evidence and associated community activities-including deliberation and experiment. Naturalizing intentionality requires social-intellectual reconstruction exemplified by the scientific community at its best. This approach emphasizes normative elements of pragmatic conceptions of meaning and their function in orientation. It requires social conditions and intellectual practices making knowledge of intentions possible. Scientific ends, methods, and meanings, together, constitute culturally evolved instruments of adaptation to, and reconstruction of, physical and cultural environments.

1. Naturalism, Ends, and Means Max Frisch explores Homo Faber, man as maker: human beings self-perceived as merely makers. The engineer is a symbol. But Frisch's chief point is not a critique of technological imperatives. His concern is human relationships and how to better arrange them. Developing self-knowledge and normative self-identity, dramatized by Faber's compromised and stunning self-recognition, is the guiding theme. This theme has an Emersonian quality. Do not be seduced into disregard of central values to get on in the world. Technology need not lead us astray. It is defective human relations, human relations as mere technique, purposes taken as given-means in isolation as limiting. Frisch warns against self-instrumentalization and shallow instrumentalization of social relations. The consequences are severe: declin-


An earlier version of this paper appeared in Naturalism or Transcendentalism? Proceedings of the 1993 Bern Congress for Philosophy, Dialectica, 49, Fasc. 2/4, 1995, pp. 147-168. Reprinted by permission.



ing integrity and loss of basic orientation. Though Frisch was sometimes a critic, I invoke his themes to honor Swiss insight, courage, and normative guidance-of the sort I have seen exemplified by Henri Lauener. In his Reconstruction in Philosophy (1920) John Dewey traces talk of Home Faber back to Bergson. Dewey agrees, the human race "is ...the tool-making animal." 1 Having read Frisch, we wonder if there is a danger of definition per accidens. But the context is ameliorative. The rise of science conditions the plausibility of humankind as tool maker, and though the advance of science has left nature and human beings a pointless whirl of the atoms in the void for some, Dewey does not ignore creativity, drama, and purpose in the human undertaking. Ideas and values are tools, and tools are thereby raised in dignity. Modem science banished "ends and forms from the universe." 2 Yet Dewey is sanguine on the demise of essentialism and teleological forms. Values arise from human needs and purposes. Ending predetermined ends, we are freed for cooperative challengesbending nature to joint purposes. "It was not till ends were banished from nature," says Dewey, "that purposes became important as factors in human minds capable of reshaping existence." 3 After ancient metaphysics, we select our ends, and ifwe think of that metaphysics as an ideology of the purposes of the few, then Dewey's point has deep moral and political significance. Having escaped absolute and inherent purposes in nature, does naturalism leave us vulnerable to technique and defective forms of human relationships? Does it leave us subject to purposes of others, so that we can only accommodate, or rebel, with no way to select between better and worse? These ideas have some currency. What chiefly argues against them is the principle of continuity in the pragmatist tradition, including continuity of ends and means. But the anti-subjectivist and anti-relativist themes are long neglected. Dewey's naturalism provides a fundamental value orientation,4 centered on community democratically restructured for the sake of mutual self-realization. Difficulties in achieving this, beyond the intellectual problems, are a matter of effectively dealing with social rigidities and reshaping established habits for more reasonable goals. Dewey calls for intelligent integration not meek assimilation. The justifications of moral norms and of the methodological norms of the sciences are essentially similar in Dewey's work. Justification and mediation of human ends and means are no mere will to believe. It

1. Dewey 1920, Reconstruction in Philosophy, p. 71. 2. Ibid., p. 69. 3. Ibid., p. 70.

4. Cf. Callaway 1993,"Democracy, Value Inquiry, and Dewey's Metaphysics."



depends on achieving fuller communication: social conditions facilitating mediation of intentions. For Dewey, normative claims have cognitive force "only as matter-of-fact grounds are presented in support of what is advised, urged, recommended to be done." 5 Like theory, value claims depend on cooperative and factual inquiry. A history of misunderstandings has led to underestimations of pragmatism and the normative guidance of Dewey's naturalism. Effective justification of norms depends on willingness to clearly state and express purposes and intentions. This is a practical-methodological requirement for development of joint purposes, akin to the dependence of individual growth upon social intercourse. Dewey advances specific forms of sociality productive of integration, cooperation (this must include rules of competition), and collective intelligence.

2. Representation, Pluralism, and Possibility Peirce's "pragmatic maxim," points to relations between meaning and practice: Consider what effects, that might conceivably have practical bearings, we conceive the objects of our conception to have. Then, our conception of these effects is the whole of our conception of the object. 6

But this famous passage does not define the exact character of the relation. Consider the phrase "effects that might conceivably have practical bearings." Looking further in Peirce, the phrase is interchanged with talk of experiment: " ...if one can define accurately all the conceivable experimental phenomena which the affirmation or denial of a concept could imply, one will have therein a complete definition of the concept..." 7 But doubts on definability of "conceivable experimental phenomena" are not wanting, and we must consider Peirce's maxim, and "conceivable experimental phenomena" in connection with the continuity of inquiry. Since what we count as evidence depends upon the range of theories we entertain, and the development of methods of research and instrumentation, there is no once-and-for-all definition of evidence or experimental phenomena. We must consider the conception of"experimental phenomena" subject

5. Dewey 1945, "Ethical Subject Matter and Language," p. 686. 6. Peirce, Collected Papers, Harvard University Press, Vol. 5, paragraph 402. 7. Peirce, CP, 5:412.



to elaboration. This suggests a certain vagueness ofreference. We have reason to seek presently unspecified subcategories of our concept of experimental phenomena. Although there is use for a notion of meaning linked to present, and presently predicted, evidence for a theory, it is also important to see the openended character of evidence, as this progresses through theoretical and conceptual changes. Doing so, we approach Peirce not as an operationalist-instrurnentalist, but as providing for continuity between present science and broader developments. This does not break the pragmatic link of experimental practice to meaning. That link is strengthened by breaking the domination of purely conceptual, mathematical, or a priori conceptions of possibility and conceivability. Whatever the "analytic" or conceptual truths proposed, and whatever the present possibilities for reinterpretation, these are limited by present theory, and there is little reason to think that they exhaust what may be found in future theoretical-empirical developments. It is instructive to think of pure mathematics as limning the abstract possibilities of relations, including new discoveries. But with fruitful empirical applications, abstract possibilities come to elaborate our conception of possible evidence. Thus, accepting some list of logical, mathematical, and conceptual truths does not define the possible development of the concept of evidence. First of all, these truths are subject to revision. But further, holding these truths constant and allowing the truth of observation sentences to vary, within categories presently recognized, does not defme the possibilities of the development of science through conceptual change. There is a broader, more open-ended, conception of possibility related to the continuity of inquiry. It allows that categories of evidence will evolve. We need not envisage an end to inquiry nor a fmal list of methods to speak of continuity of inquiry. What is required is a thorough-going fallibilism, including commensurate respect for established doctrine. Dewey hesitated long on 'truth.' He was wary of James's treatments and wary of connotations of infallibility. But in reply to Russell he emphasized, both the provisional status of warranted truth claims and their value. While "the 'truth' of any present proposition is, by definition, subject to the outcome of continued inquiries," he says, "and its 'truth,' if the word must be used, is provisional," this is still "as near the truth as inquiry has as yet come." This status is "a matter determined not by a guess at some future belief, but by the care and pains with which inquiry has been conducted up to



the present time." 8 Truth claims are justified by methods successfully relating theory to experimental results. That science is possible is a fundamental desideratum. We reconstruct theory and belief to meet the demands of observation; and we reconstruct scientific practices to facilitate tests of theory. These possibilities are consistent with laws the sciences may uncover. I assume that scientific laws are not something within our control: they record regularities of nature. Some things we can change and others we cannot. We do not know a priori what we can change.9 Still we know that some things can be changed, since we have changed our scientific beliefs and practices. This latter idea is basic in Dewey's concept of reconstruction of experience. Experience is an interactions within nature which reveals natural potentialities, and these potentialities are generative. Given existing beliefs and practices in interaction with nature, we come to change our beliefs and practices, and this enables us to evoke further natural potentialities-to reveal further natural regularities. Accepted belief or established theory, and dependable methods, provide basic intellectual-practical orientation to the growth of knowledge. The problem of establishing appropriate contexts for judgment has a social element. Though intellectual problems are not to be reduced to social problems, we must proceed with an eye toward the social conditions for the possibility of knowledge and its growth. Ignoring the social element leads to the bedeviling dualism marked out by naturalist critics. As Dewey put it in Experience and Nature," ... philosophical dualism is but a formulated recognition of an impasse in life; an impotence in interaction, inability to make effective transition, limitation of power to regulate and thereby to understand."10 Failures at social regulation lead to failures of understandipg: appropriate regulation, method, is a pre-condition of understanding. The point culminates in methodological norms governing communication, deliberation, cooperation, and experimentation. Cooperative intelligence, exemplified by the scientific community at its best, has its own requirements. "Here, as in so many other things," Dewey argued, "the great evil lies in separating instrumental and final functions." Even intelligence is "partial and specialized" where "communication and participation are limited, sectarian, provincial, confined to a class, party, professional group." 11 Cooperative intelli8. Dewey 1939, "Experience, Knowledge and Value," p. 573. 9. Cf Hawking 1993, Black Holes and Baby Universes, pp. 92-95, on possible limits of prediction in quantum gravity theories. 10. Dewey 1929, Experience and Nature, p. 241-242. 11. Ibid., p. 205.



gence requires normative regulation of instrumental communication. Ends are implicit in our means, and defective deliberation, or unclear connection to experiment, action, and evidence lead on to defective judgment. Genuine openness to new insight requires social-institutional flexibility, since it may require attention, evaluation, and support. This point is both moral and methodological. It argues for an experimental pluralism of goals, methods and views. This pluralism contrasts both with absolutism and with value relativism. Whatever the valid perspectives of cognitive relativisms, normative and cultural relativism have become, in spite of pragmatist roots, a chief vehicle of collective egoism, based in and productive of intellectual balkanization. The danger is dualisms perpetuated or replaced by a multitude of proclaimedly incommensurate ideologies. The value of criticisms depends on their mutual consistency, their relation to a positive position advanced, and deeper understanding of alternatives. Significant criticism and deeper understanding of alternatives are not encouraged by the thesis that no rational decision is possible.

3. Continuity, Logic, and Dewey's Naturalism Dewey's naturalism explicates Peirce on continuity. In his Logic: Dewey says that "logic is a naturalistic theory," and 'naturalistic' "means ...there is no breach of continuity between operations of inquiry and biological operations and physical operations." Continuity means that "rational operations grow out of organic activities, without being identical with that from which they emerge." 12 Organic activities include biological functions, minimal instinctual behavior, and socially acquired habits. Rational activities develop from habit by correction and regulation; 13 and logic, Dewey's "theory of inquiry" is the theory of better and worse in rational activities. The continuity of inquiry substitutes for the principle of the uniformity of nature, and it is an anti-reductionist principle. It leads us to seek relations and underlines the possibility of inquiry, though not guaranteeing specific results or kinds of laws. Since it emphasizes relations and relatedness, it is holistic. Yet this is a holism open to moderate, pluralistic and empirical interpretations. Where suitable relations are not discovered, this leaves a domain in relative and provisional isolation.

12. Dewey 1938,Logic, pp. 18-19. 13. Cf. Dewey 1922, Human Nature and Conduct, pp. 15-16.



The meaning of 'continuity' Dewey says, "excludes complete rupture on one side and mere repetition of identities on the other; it precludes reduction of the 'higher' to the 'lower' just as it precludes complete breaks or gaps." For naturalistic epistemology, this implies the continuity of the grounds of judgment or criteria of knowledge with successful inquiry. Dewey rejects indubitable deliverances of sense or "unvarnished news." Perception and observation involve interpretation. He also rejects self-evident principles. Thus he rejects any absolute presuppositions of knowledge. In the fallibilism of the naturalist tradition we see methodology, and relationships between the theoretical and what is less theoretical, as a movement by hypotheses-which usually do not reduce to what they explain, predict, or comprehend. Autonomy of hypothesis with respect to evidence is the root of the autonomy of special sciences, since we may view related systems of hypotheses as conjunctive hypotheses. This point concerning autonomy is reminiscent of the "plausibly short leaps" Quine identified in his account of the Roots of Reference and language acquisition. 14 This is one root of moderate epistemic holism in Quine's epistemology. Adopting the normative injunction of relative empiricism: "Don't venture farther from sensory evidence than you need to," 15 we do not arrive directly at a unified theory of everything. 16 Still, autonomy of hypothesis with respect to evidence and autonomy of the special sciences is consistent with there being no change without a physical change. Dewey understands logic, meaning, and method as continuous with empirical science and common sense. "Application of the principle of the continuity of inquiry," he says, "enables an empirical account to be given of logical forms, ...while at the same time it provides that the interpretation of them as a priori is unnecessary." 17 The validity of logical forms, and hence aspects of meaning beyond the level of stimulus meanings, have a hypothetical character in this passage-a point fully consistent with later criticism of the analytic-synthetic distinction. This interpretation is supported by C.I. Lewis, who found no notion of analyticity in Dewey's work. He remarks in a 1941 paper that "it is doubtful whether Dewey would recognize ...analytic statements as occurring in actual human thinking or discourse." Generally, "amongst pragmatists, the distinction of empirical meaning from the signifi18 cance of analytic statements for the most part passes unremarked." That 14. Cf. Quine 1973, p. 101. 15. Ibid., p. 138. 16. Cf. Callaway 1993,"0pen Transcendentalism and the Normative Character of Methodology." 17. Dewey 1938, Logic, p. iii. 18. Lewis 1941, "Logical Positivism and Pragmatism," p. 93.



attitude is more plausible if we hold that it is "folly to seek a boundary between synthetic statements, which hold contingently on experience, and analytic statements, which hold come what may." 19 Lewis's claim is an understatement. Rejection of absolute claims for analyticity is congruent with the Hegelian roots of Dewey's thought, as opposed to the Kantian and neo-Kantian roots of the positivists' analytic-synthetic distinction.20 Richard Bernstein put it this way: "Dewey believes that while Kant had a profound insight in recognizing the activities of thought in constituting what we know, the sharp distinctions between the a priori and the a posteriori, the analytic and the synthetic, are untenable."21 Dewey's position traces back to Peirce.22 "Logical principles," says Dewey, "are necessary to conduct every successful inferential inquiry," and 'successful' "means operative in a manner that tends yield results that are either confirmed in further inquiry or that are corrected by use of the same procedures." 23 The grounds or epistemic status of logical principles are linked to their success in inquiry. They are "hypotheses about operations to be performed in all inquiries which lead to warranted conclusions ..." 24 "What is excluded by the postulate of continuity," says Dewey, "is the appearance upon the scene of a totally new outside force as a cause of changes that occur." 25 Dewey excludes the "absolute other." He excludes an absolute difference between the physical, the biological, the social-behavioral, and the mental. An absolute difference, a Cartesian mind for instance, would block the path of inquiry concerning the relationships of mind and body and violate the continuity of inquiry. Continuity implies relations and hence the possibility of inquiry. But Dewey does not claim that methods already exist for any given problem, and this openness is well supported by the developmental history of scientific methods. As Bernstein put it, continuity "is undoubtedly the most fundamental principle in Dewey," it "is the heart of his naturalism."26 Israel Scheffler makes a related point: "The interpretation of thought as intimately interwoven with action in a purposive context is stressed by pragmatism as

19. 20. 21. 22. 23. 24. 25. 26.

Quine 1951, "Two Dogmas of Empiricism," p. 43. Cf. Scheffler 1974, Four Pragmatists,p. 195. Bernstein 1967,John Dewey, p. 69. See Dewey 1938, Logic, pp. 13-14; Peirce CP Vol. III, pp. 154-68, Vol. V, pp. 365-370. Dewey 1938, Logic, p. 13. Ibid., p. 156. Ibid., p. 24. Bernstein 1967, John Dewey, p. 180.



indicating the continuity of mind and nature." 27 Continuity of mind and nature is naturalized intentionality. Lack of clarity on the principle is removed once we see it as both a result of science and a principle of normative guidance. Quine attributes a similar status to empiricism. "Empiricism ... is both a rule of scientific method and a scientific discovery." 28 Here the normative is a discovery of science. Perceived lack of clarity bespeaks dominance of the fact-value dichotomy. Dewey's naturalism and moderate holism arise from his aim to generalize on the success of the sciences-in their reliance on study of "a multitude of complicated, obscure, and minute relationships." 29 Since the facts of successful inquiry underlie the justification of methodological norms, the principle of continuity is both factual and normative.

4. Pragmatism as Normative The pragmatic maxim, relating ideas to consequences, also has a normative character. We can think or speculate out of meaningful relation to facts and consequences. Unverifiable metaphysics is not meaningless or impossible. It is undesirable. Following the model of science, we strive to reconstruct thought and theory to bring them into more meaningful relation to evidence and tests, but the speculations we may begin from are not meaningless before we tie them down. This points to a concept of linguistic meaning more independent of the relation of object language theory to evidence. It is a matter of broader interrelations of linguistic expressions, and this aspect of meaning is preserved in lexical definitions of well established science and its rigorously regimented language. The process of relating theory to evidence is part of naturalizing the intentional. However, since new problems and situations are always arising, and new proposals for solutions and action follow in their wake, speculation repeatedly escapes full verification, even if we strenuously work for it. We dare not be so scrupulous in relating meaning to evidence and consequences as to forbid semantic revisions, even though revisions renew the task of bringing abstract thought into connection with fruitful consequences. On the other hand, we dare not avoid relating theory to evidence as fully as possible. Metaphysics exists because of its social and intellectual functions. Unverified metaphysics, with no working relation to evidence, still creates 27. Scheffler 1974, Four Pragmatists, p. 8. 28. Quine 1990, "Comments on Lauener," p. 229. 29. Dewey 1922, Human Nature and Conduct, p. 41.



expectations and interpretations of experience.30 Its capability for creating expectations is central in its social functions. We will have metaphysics, the question is whether we will have better metaphysics, more reasonable, though always fallible, metaphysics, or metaphysics more doubtful in relation to the natural and social world. Holding to the ontology of the sciences is needed in making sense of metaphysics. There is a "blurring of the supposes boundary between speculative metaphysics and natural science" in the "shift toward pragmatism" announced in "Two Dogmas of Empiricism." 31 Rejecting the dogma of reductionism, we see that the meaning of theoretical statements is not exhausted in "logical constructs upon terms referring to immediate experience." In rejecting the first dogma of analyticity, we reject claims except as capable of mediation through empirical developments. This includes purely formal claims regarding conceptual possibilities and the possibility of reinterpretation in particular. Revision of conceptual systems or of the ontology of science requires specific intellectual motivation. The meanings of theoretical claims in science are not exhausted by evidence accumulated or foreseen, for they involve broader inter-theoretic relations. Nor are they exhausted in a theory-neutral account of observation. No concept of possible evidence can foresee all empirical developments, and we have no complete conception of possible evidence. Still, present theory, and established ontology, do create guiding expectations. They enter into the direction of research via a principle of defeasible conservatism. Though shared linguistic conventions are not required for communication under conditions of radical interpretation, defmitions in the sciences, and ontological claims, remain crucial instruments of communication and fundamental in cognition. The problematic status of metaphysics arises from purely social support for otherwise unsupported expectations, and this is a problem of social reconstruction which Dewey sought to solve through social unification of experience and a consequentialist element in the ethics of discourse, designed to make intentions available and joint projects possible.

5. Meaning and its Place in Nature To understand the relationship of nature and experience in Dewey, and the relationship between meaning and nature, in particular, (content and 30. Cf. Callaway 1992, "Does Language Determine Our Scientific Ideas?" above. 31. Quine 1951, "Two Dogmas of Empiricism," p. 20.



object, in traditional vocabulary), we must draw on the principle of continuity. To naturalize intentionality, and to naturalize linguistic meaning as the preeminent form of intentionality, is to show the continuity between nature and these particular aspects of culturally dependent experience. I assume that our talk of linguistic meaning partly explicates Dewey on experience. Linguistic meanings are in nature. What is primary are regularities of usage, and this is an empirical phenomenon. The phenomenon stands in need of normative guidance, but normative guidance is a systemization ofpreexisting usage. Habit precedes intelligent reconstruction or regimentation of habit, though normative reconstruction may subsequently become habitual. Meanings as Platonic entities are at best an idealization, systematizing facts of usage. For, "Meaning," according to Dewey, "is primarily a property of behavior," and as he continues, it is "secondarily a property of objects." 32 Objects have cognitive significance, because posited objects explain regularities of interpreted observation. Language is a fact in the world among other interrelated facts. That language is as plastic as human intelligence, culturally generated and culturally maintained, does not detract from its status as a natural phenomenon. The development of systems of meanings in science has real effects in the world. We can now do what we could not do before. Systematic observation and prediction on the basis of guiding theory is the preferred form of interaction developed in the sciences. Our meaning something in particular by our words is an actualization of a potentiality within nature and cultures, and in consequence, using our words meaningfully, we are able to say and do something about nature-anticipate and influence the course of events. Though meaning and meanings are in nature, there is another sense in which nature is in our meaningful discourse. What we know of nature is expressed in discourse. This is the intentional, or semantic, character of language. We mean things in nature, refer to them, by use of words, and these relations are not always superficial or subjective. What is subjective contrasts with what is intersubjectively agreed and confirmed. Greater determinacy of reference comes with regimentations of language, and we should not expect absolutes. We are aware of nature in depth. Meanings developed in science point deeply into nature, and there is no significant sense in which they merely reflect our own doings. Our doings, whether in the cognitive sphere, or

32. Dewey 1929,Experience and Nature, p. 141.



more broadly, are interactions within nature capable of revealing natural potentialities. In scientific discourse, we come to recognize deep regularities of nature. The significance of truth claims as applied to scientific statements has to do with growing control over nature, evident in experimental prediction and technology. Experiment resulting in successful predictions is, indeed, a special form of technology-technology for cognitive purposes. It is difficult to see how we can prevent a robust scientific confidence in the use of a truth predicate from inducing a similar confidence on the meaningfulness of language and the reality of our posits. Through scientific regimentation of language and its application, we are aware of the development of the universe over eons of time, and the existence of animals extinct for hundreds of millions of years. We are aware of the structure of sub-atomic matter. As Dewey puts the point, experience is in nature and also of nature. There is an intentional realism, expressed by this "of," and we should have no doubt that there is sufficient realism for an intentional realism. "Not all existence asks to be known," says Dewey in The Quest for Certainty, "and it certainly does not ask leave from thought to exist." 33 We do not have to know everything in order to know anything, and reference to specific kinds of objects does not require a finished catalogue of distinctions among referents. Fallibilism is not a reason to doubt particular claims. Dewey's instrumentalism contrasts with positivistic doctrines of theory as convenient fiction. Replying to Hans Reichenbach, he rejected reception of his views as "a non-realistic interpretation of scientific concepts." 34 He did not deny scientific objects in contrast to the objects of everyday life. This disputes the view that " ...Dewey viewed science as a conceptual shorthand for organizing observations," and that "Dewey's reality consisted of observable objects," 35 though such views are found in, or suggested by, Dewey's less mature work. Roots of a doctrine of "ontological parity" (to borrow Justus Buchler's term) are in Dewey's work as far back as 1929. He relates positing of objects to the thesis that "knowing is not the act of an outside spectator but of a participator inside the natural and social scene." He argues, in effect, against the notion of ontological priority and grades of being. "For on this basis there will be as many kinds of known objects as there are kinds of effectively con-

33. Dewey 1929, The Quest/or Certainty, p. 236. 34. Dewey 1939, "Experience, Knowledge and Value," pp. 535ff. 35. Quine 1981, "The Pragmatists' Place in Empiricism," pp. 33-34.


153 36

ducted operations of inquiry which result in the consequences intended." We recognize kinds of objects not kinds of being. Conclusions drawn in the name of ontological parity have sometimes tended toward proliferation. But the fact that our ontology responds to human and scientific purposes does not show that dependable regularities, even those encoded in the lexical semantics, are purely created rather than also discovered. Nor does it show that all human purposes are equally suited to sponsor ontic commitments. We understand Dewey on meaning, and his instrumentalism, by seeing concepts or meanings as tools. Evidence of meanings is found chiefly in joint experimental and problem solving activities facilitated by specific theories or concepts. Meanings become more evident in argumentation or discourse leading on to joint activities and experimental results. Because scientific language enables us to effectively deal with problematic situations, we regard some statements as true and systems of related statements as meaningful. This is the final court of appeal. Science makes a difference in nature, engendering potentialities-significant new possibilities for human action. Not all our meaningful linguistic doings are equally suited to revealing nature in depth. On the contrary, that some methods are better than others is the presupposition of systematic studies of methodology and of naturalized epistemology. Whether we regard meanings as posits of common sense or as objects of linguistic investigation, there is as little ground for anti-realism concerning meanings in a Deweyan philosophy of language as there is for anti-realism about corresponding objects of reference. Still, I am certain that Dewey would regard any theory of meaning as requiring empirical development and confirmation. Congruent with this point, semantic anti-realism is a skeptical challenge.

6. Contextual Holism and Conceptual Identity The continuity of inquiry is a methodological hypothesis or "leading principle" for Dewey, it leads us to look for relations, connections and explanations. In 1922 he explained continuity by reference to the success of science, which "began with the recognition that every natural object is in truth an event continuous in space and time with other events; and is to be known only by experimental inquiries which will exhibit a multitude of complicated, obscure, and minute relationships." 37 This view is explicitly opposed to the

36. Dewey 1929, The Quest for Certainty, p. 157. 3 7. Dewey 1922, Human Nature and Conduct, p. 41.



ancient notion of an internal or substantial form, the idea that "each object in the external world carried its nature stamped upon it as a form."38 Relations are constitutive of identity. In the origin of modem science, the quantifiable relationships identified by measurement are of particular importance, and in recent advances we see that the identity of meanings depends upon their recursive generation within a system. More generally, Dewey's naturalism developed out of Peirce's logic of relations. In light of Darwin's theory of evolution, giving a central role to the logic of relations meant that knowledge of things required a knowledge of their relations to a surrounding environment. The notion of biological species as adapted to their environments, and of their existence depending on specific relations to an environment, is empty without knowledge of relevant environments. Essence becomes dependent on details of fact. Darwin brought the "eternal forms" into the stream of becoming. As Dewey put it, Darwinism "conquered the phenomenon of life for the principle of transition, and thereby freed the new logic for application to mind and morals and life."39 Human life is not a static phenomenon existing in relation to an unchanging social and physical environment. Human beings adapt themselves, and they are capable of transforming their physical and social environments to make them more suitable. The postulate of continuity tells us it is idle to postulate objects apart from their relations. In particular, continuity of means and ends directs us to evaluate ends by reference to available means and ascertained consequences of established ends in practice. Through an examination of the relations of ends to means, viewing ends as subject to revision, we uncover the potentiality to reconstruct human ends. Our history of collective conflicts suggests much need of this. Continuity in nature implies continuity between common sense, custom, and habit on the one hand and science on the other, since these cultural phenomena are part of nature. Common sense, custom, and habit are the context from which knowledge developed and develops-they continue to provide a working and fallible constraint upon the growth and development of knowledge, while slowly evolving under pressure of scientific confirmations or disconfirmations. Departing from common sense, we have more explaining to do, but through their relation to science, departures from common sense may return to enrich it. Common sense and the habitual are not absolute presuppositions of science. But science does not render common-sense knowledge superfluous, unless it provides an

38. Ibid., p. 41. 39. Dewey 1910, "The Influence of Darwinism on Philosophy," pp. 7-8



adequate substitute. Our working belief-systems stand as the context of inquiry, except in so far as a problem arises to put some aspect of a system in question. But even in this situation, we identify a problem against the background of established belief or theory. The idea that statements or questions lack any determinate meaning, beyond the level of stimulus meanings, seems particularly unappealing for questions and problems. A negative result in experimental tests of a theory creates a problem for that theory. The statement of the problem must have some determinacy of meaning and reference, if it is to constitute a problem for the theory in question. Clearly, we cannot always interpret away substantial empirical problems. The possibility of alternative reinterpretations of problematic observations will not always save the theory dictating the interpretation which leads to tests and negative results. Since evidence must be interpreted in relation to some theory or other, in order to make contact with theory, and since theory can be effectively falsified, it follows that observation sentences have some determinacy of meaning, beyond the level of Quinean stimulus meanings.40 This point expresses continuity of the observational and the theoretical. Theoretical interpretation of observation statements is needed for falsification of the interpreting theory. Thus, I think that Quine's "principle of minimal mutilation" can be applied to scientific meanings and profitably viewed as one implication of Dewey's principle of continuity-representing the continuity of established science or belief with innovation. Dewey places special stress on how science is constrained by the background of common-sense belief and habit. More generally, a moderate epistemic holism is a consequence of the continuity of inquiry. But a defeasible conservatism in theoretical work is of special interest in understanding the contrast between Dewey's views and traditional talk of "transcendental" or once-and-for-all final presuppositions of the possibility of knowledge. Knowledge develops on the basis of pre-existing belief and problems encountered within a context established by pre-existing belief, though no pre-existing belief is an absolute presupposition. A defeasible conservatism, recognizing degrees of centrality and explanatory value, constitutes a fallibilized version of foundationalist notions. On this basis, there are grounds for embracing a concept of linguistic meaning, for investigation, focusing on scientific and common-sense meanings. For Peirce,41 natural phenomena display an interrelated and continuously developed series of dependent forms, though developed forms, such as

40. Cf Context for Meaning and Analysis, Callaway 1993. 41. Peirce 1955, "Synechism, Fallibilism, and Evolution," in Buchler, p. 356



human life or scientific inquiry, are capable of subsequently reconstructing the environments from which they arose. In a similar way, the activity of inquiry is a development from habits of action and discourse and depends on them for its working orientation. In the first instance, science is a development of truthfulness and cooperativeness in our less systematic practices. A defeasible conservatism in the theoretical sciences reflects the persistence of habit at a lower level, while rising above it in sophistication. The difference is that a defeasible conservatism allows the possibility of less conservative alternatives: those favored by simplicity or greater comprehension for instance. While the persistence of habit becomes an irrational force in human affairs, as a refusal to adapt to changing situations, at the level of our persistence in intelligent belief, we are prepared for needed adaptations. Still, this point should not blind us to problems which arise in consequence of the pace of change in modern life, and it does not eliminate the need for basic orientation. We regard some claims and values as more central. A defeasible conservatism is the appropriate methodological guardian of the saliencies of phenomena recorded in the systematization of established belief and theory. The Greeks could more plausibly view the eternal species and forms as static, perfect, and finished. Failing to develop the insight that living forms change and develop in their relations to a wider world, they saw less need for changing forms of language to reflect a changing world and expanding experience within it. "Meaning," as Quine put it, "is what essence becomes when it is divorced from the object of reference and wedded to the word." 42 The divorce was effected in order to more easily accommodate change in explanatory theory. But divorce is too strong a word. Change in the meaning of concepts, scientific concepts in particular, and the relation of conceptual change to empirical-theoretical developments, strongly suggests a normative and contextual view of meaning, where the meaning of an expression depends on its function in a given system. We may judge of better and worse among systems of meanings in relation to available evidence and the interrelations of belief-systems. In particular, meanings of observation sentences depend on their theoretical context. A defeasible conservatism in referential semantics opposes both rigid essentialism, and a Heraclitean flux of meanings which ignores the ontic commitments of established science and common-sense belief.

42. Quine 1951, "Two Dogmas ofEmpiricism," p. 22.



7. Context for Radical Interpretation To say that meanings, or cultures, emerge in nature implies causes beyond human control, though these causes do not uniquely determine the meanings 43 or cultures we adopt. The point acknowledges human contributions. Yet given that we successfully reconstruct nature and culture in light of our beliefs, and the success or failure of our actions, our beliefs and actions are not self-validating. But neither are they ineffectual. To say that there is a human contribution in linguistic meaning, something we try to take account of, on a meta-level, in our talk of linguistic convention and conceptual systems for instance, is not to say anything about the intentional or representational character of meanings. No "instrumentalist" anti-realism follows from this. To recognize a convention of language is not to patronize it. Signaling dissatisfaction with one convention, as obsolete, expressing merely, "the conventional wisdom," we cling to others to get leverage. Positing objects of certain sorts implies adopting a given concept or convention in our beliefsystem, so rejecting posits patronized, we more robustly accept corresponding concepts. Our conventions and concepts can be better or worse. They make a difference to our ability to represent the world as shown in prediction and reconstruction. There is no compelling reason to hedge the truth of claims we make because they involve or depend upon conventional regulation of language, except where relevant evidence is purposefully neglected. This is a fallibilistic and anti-absolutistic stance on truth claims. Fallibility no longer suggests doubtfulness. We are familiar with this attitude from Quine's 44 work, and a similar stance can be found in C.I. Lewis. 45 I urge that this "robust realism" concerning truth claims (if its not broke, don't try to fix it), is justified. It provides the context required for inquiry into problematic claims. It should be recognized as crucial for a naturalist theory of truth. The point has some bearing on the principle of charity. In the Introduction to his Inquiries into Truth and Interpretation, Davidson quotes Quine in Word and Object, providing an initial statement of what he means by charity in interpretation: " ...assertions startlingly false on the face of them

43. Cf. Callaway 1993, "Open Transcendentalism and the Normative Character of Methodology." 44. Quine 1960, Word and Object, p. 25. 45. Cf. Lewis 1955, "Realism or Phenomenalism?" p. 339.



are likely to tum on hidden differences of language."46 Few will have doubts on this claim. Davidson continues, remarking that he "cannot use Quine's notion of stimulus meaning as a basis for interpreting certain sentences," and this is the reason offered for applying the principle of charity "across the board." Applied more broadly, "it counsels us quite generally, to prefer theories of interpretation that minimize disagreement." 47 Here we need contextualization. Davidson claims that "massive error about the world is simply unintelligible," and that it is impossible to "correctly interpret someone as being massively mistaken." 48 But these claims have raised doubts.49 An anthropologist attributing a thorough-going animism is a good example of rightly interpreting a group as massively mistaken. This points toward contextualization of talk of"massive error." What is true is that attribution of error depends upon community of belief. Davidson is right to emphasize "that much community of belief is needed to provide a basis for communication or understanding," and that "without a vast common ground, there is no place for disputants to have their quarrel." 50 It does not follow that we must always interpret by minimizing disagreement, and I think Davidson will agree. Still, departing from the notion of charity as minimizing disagreement, we have a serious modification. We must establish a substantial community of belief to make sense of differences, and interpretations attributing error and disagreement depend upon established contexts of unproblematic interpretations. We look to areas of agreement with speakers we wish to interpret in order to establish a working theory of interpretation, covering as wide an area as possible. Once having established a dependable context for further interpretation, however, we may then tum to errors and disagreements. Indeed, open expression of disagreements is quite as important to mutual understanding as finding areas of agreement. It helps to orient us in joint inquiry. Where disagreements are more theoretical, we speak of conceptual differences. Contextualizing the principle of charity, we may begin with a comparison to the classical pragmatist doctrine of doubt and belief. This begins from 46. Davidson 1984, Inquiries into Truth and Interpretation, p. xvii; Quine 1960, Word and Object, p. 59. 47. Davidson 1984, p. xvii. 48. Davidson 1977, p. 20 I. 49. Cf. e.g., Fuller 1988, Chapter 6. 50. Davidson 1977, "The Method of Truth in Metaphysics," p. 200.



Peirce's rejection of Cartesian doubt. If doubt presupposes a problem in established belief, the argument goes, then no genuine doubt can be global.51 The application of this for present purposes is a version of the principle of charity. Genuine disagreement (rather than lack of understanding) cannot be global, since it presupposes a problem in the comparison of beliefs. Charity is required, at the beginning, in order to establish a working context for further interpretations. But it is not clear that we need a working context of global range. One might establish context for interpretations within relatively small and autonomous domains. It is reasonable to suppose that these can be identified by reference to joint activities: cooking, eating, fishing or hunting, gathering, child care, maintaining relations to relatives, friends and acquaintances, etc. If the interpreter joins in the group activities of the community, then this will obviate the problem of needing to know the background beliefs of the speakers's intended audiences. Having understood the sound system and syntax of a given community and the semantics associated with these concrete activities, we come into a position to notice that all this talk is pervaded by views about spirits: spirits in the fish and streams, spirits in the game and forest, spirits of the fire or hearth, spirits of growth and fertility, spirits of dead relatives etc: Massive error. Leastwise this is massive error from the perspective of a naturalistic ontology, though we suppose the community is mostly correct in what they say about how they catch fish, hunt in the forest, gather fruit, or care for children. Linguistic expressions may be interpreted at differing levels, and if we do not suppose the community under interpretation has things mostly correct at a less theoretical level of interpretation, and in some domains, then we could not get started in interpretation. Further, starting with only our own beliefs, it is reasonable to tentatively attribute these beliefs in situations where this is at all plausible. Still, we must beware of anything, resembling attributive anachronism, augmenting the context of knowledge of the community under interpretation.

8. Meaning and Anticipations of Experience In concluding, I return to meaning in natural science, where context and semantic reconstruction is directly relevant to scientific aims or projects. Emphasis on truth of unproblematic belief persists, as does emphasis on 51. Cf. Callaway 1993, "Review of Simon Evnine, Donald Davidson;" Chapter 8 above.



reflective self-definition and the need for basic normative orientation in scientific practice. Science orients itself through established belief and dependable methods. Seeking to comprehend new phenomena by prediction and explanation, science also seeks to preserve established doctrine. For example, Darwin sought to explain the origin of species. In doing so, he had to retain a reasonable account of the preservation of the species: the ancient knowledge of plants and animals reproducing after their kind. Explaining change and development by random mutations and the survival of those better suited to their environments, he was guided by semantic norms of classificatory systems worked out by generations of biologists. Darwin interpreted the ontology of traditional biology anew, creating expectations for further evidence of the evolution of the species. This expectation could not be elucidated purely in terms of the prior categories of evidence. Even if needed evidence was available in some meager sense, it was not properly conceptualized before Darwin. Instances of evolutionary change, had they been observed, would have likely been interpreted, say, as changes in population distributions. In spite of introducing new categories of evidence, such as species distributions on isolated islands, the basic orientation of Darwin' s work was provided by ascertained facts on the existence and differentiation of species. Like Newton, he stood on the shoulders of giants. Continuity with the results of prior biology was fundamental to the validity of his innovations. In providing a rational and empirically motivated reconstruction of biology, a defeasible conservatism regarding prior results was indispensable. In proposing how particular species might have arisen, in competition with others, the Darwinian theorist plausibly focused on current definitions. It was natural to focus on similarities and on characteristics which differentiate particular species. For these comparisons and contrasts could reveal combinations of characters rendering a species successful in its physical and biological environment. In other words, the meanings of biological terms, or species names, reflected results of prior observation and systematic comparisons, as well as experience in cultivation and breeding. Meanings, emergent in classification, marked saliencies which evolutionary theory had to explain and comprehend. The progress of science rests on prior results as systematized in a particular ontology or system of concepts. This point does not exclude arbitrary elements in classification. It argues that what is arbitrary, though subject to a defeasible conservatism, may easily be overturned in further inquiry.



No guarantee exists that prior results of science contain an adequate interpretive scheme for all possible evidence relevant to further investigations. This point blocks an inductive argument favoring underdeterrnination of theory by all possible evidence. While observation sentences must be capable of conditioning, it is not stimulus meanings, or observation sentences holophrastically interpreted which test theory. Observation sentences require interpretation in light of theory to make contact with it, to support or falsify theory, and the possibilities for the interpretation of observation sentences depend on evolving theory. The progress of science requires openness to evolution in relational categories of evidence. This argument suggests that the notion of "all possible evidence" is doubtful. It also suggests inadequacy in the project of modal logic as dependent on a supposed a priori conception of possible synthetic truths. Evolution of conceptual resources argues against a priori concepts of necessary truths and against a closed class of possible synthetic truths. We have no guarantee of interpretive resources needed for further development. Established belief or theory provides crucial, though defeasible, orientation to further experience. It provides context for understanding problems in established belief, and problems in established belief are openings for development. Further grounds exist, emphasized by Quine, for conservatism on logical and mathematical truths, or explanatory and broadly comprehensive principles and methods of the special sciences. Without this, resting on success in applications, we lack fundamental orientation. The robust realism of a defeasible conservatism on scientific concepts and posits is crucial in naturalizing the intentional.


Bjorn T. Rarnberg's book focuses on Davidson's work in the philosophy of language, published between 1984 and the appearance of the book. Recent papers provide the focus for an overview of Davidson's philosophy of language and its relations to broader debates and influences. Still, the reader is warned: the author "cannot claim" that the book "is in every detail a faithful representation or development of Davidson's own current theory." Instead, what we have is a "reconstruction" of Davidson on language and meaning, an account "Davidsonian in spirit and in all its fundamental features." 1 The result is a projection of Davidson's views, or important aspects of them, in a particular direction: Davidson and interpretation in process. The following critical discussion of main issues in Ramberg's book should not distract potential readers from this useful and thoughtful overview of Davidson on interpretation and meaning. The book is an "introductory" reconstruction of Davidson on interpretation-a claim to be taken with a grain of salt. Writing introductory books has become an idol of the tribe. This is a concise book and reflects much study. It has many virtues along with some flaws. Ramberg assembles themes and puzzles from Davidson into a more or less coherent viewpoint. A special virtue is the innovative treatment of incommensurability and of the relation of Davidson's work to hermeneutic themes. The weakness comes in a certain unevenness. While generally convincing and well written, the book has low points which may leave the reader confused or unconvinced. Davidson is the hero in this book, and our hero is sometimes over-idealized. The Introduction starts with a famous yet startling quote from Davidson's "A Nice Derangement of Epitaphs:" "There is no such thing as a language."2 The reader is hooked-leastwise those readers who were once gripped by the project of semantics for natural language, who have spent hours looking for

Review of Bjorn T. Ramberg, Donald Davidson 's Philosophy of Language, An Introduction, Basil Blackwell, Oxford and New York, 1989, l 53pp. This review originally appeared in Dialectica, Vol. 50, Fasc.l, 1996, pp. 63-71. Written with J. van Brake!. Reprinted by permission. I. Ramberg 1991, Donald Davidson 's Philosophy of Language, p. 3 2. Davidson 1986, "Nice Derangement," p. 446. *



the "latest," or had occasion to hope for dispensation in the mail from Berkeley. Emphasis on Davidson's "Nice Derangement" gives Ramberg's book a particular slant. To be sure, the aim is still to construct theories of meaning for languages, modeled on the structure of Tarskian truth theories, but the meaning of "theory" and "language" have shifted: "linguistic understanding arises only through an on-going process of theory construction and reconstruction." 3 How does Davidson's recent work relate to the project for semantics of natural language dating back (at least) to "Truth and Meaning" (1967) Ramberg argues that Davidson came to de-emphasize the notion of language, shifting attention to linguistic communication. "Though Davidson does write as if we can make good sense of the concept of language," this is to be regarded as provisional. "By taking for granted that there are languages, Davidson was able to articulate a theory of meaning." The concept of language became problematic, according to Ramberg, still the result "is not a theory which undercuts itself, but a comprehensive, coherent account of the phenomenon of linguistic communications" where "truth" and "satisfaction" are the crucial explanatory concepts.4 We need to take into account how "languages" change in communications-even in the communication required to interpret them. The approach is enlightening, though the process of interpretation threatens to swallow the products. The interpretation is interesting and plausible, all the more since it seems that Davidson has arrived, in a sense, at a conclusion pessimistically forecast by Chomsky. For Chomsky had argued, via a reductio ad absurdum, against the idea that belief or attitude are fully relevant to formal semantics in empirical linguistics, favoring a strict distinction between matters of meaning and matters of fact or belief. Thus, given Davidson's approach to meaning via radical interpretation, and the role of belief (or "holding true") in this, any de-emphasis or shift on the concept of language is of special interest. Chomsky argued that since speakers of the same language differ in beliefs and attitudes, if such differences are fully relevant to semantic theory, then from a semantic perspective, "language is a chaos which is not worth studying."5 But there may be ways around this pessimism.

3. Ramberg 1991, Donald Davidson's Philosophy of Language, p. 138. 4. Ibid., p. 34. 5. Cf. Chomsky 1977, Language and Responsibility, p 153; Callaway, 1981, "Semantic Theory and Language,"reprinted above, p. 21ff.



1. Language and Communication A chief point of interpretation is argued in Chapter 8, "What is a Language?" Ramberg seeks to "show how Davidson's articulation...of the explanatory function of the concept of truth fmally leads him to reject the very idea of a language as a semantically uninformative concept." 6 But there is some going back and forth here between "language, as that which a speaker makes use of when successfully communicating," which "cannot be given independently of the truth-theory which determines its structure,"7-this turns out here a rejected "reification"-and a broader, non-formalizable conception of language. In the latter, linguistic competence does not amount to "knowledge of Tarskian theories of truth," instead "the essential competence turns out to be the ability to continuously form and reform such theories by interpreting assertions as on the whole true." 8 Stress on a non-formal conception of language or communicative competence arises in connection with the point that "communications does not depend, as we naturally tend to suppose, on our speaking the same language."9 It arises too in connection with Davidson's emphasis upon innovative interpretation, "the passing theory" of an interpreter, and the fact that we derive such passing theory "by wit, luck, and wisdom." 1 For Ramberg, Davidson is wrong to reject the concept of language in semantic theory. For, "the move is precipitous," and while Davidson is right in rejecting explication of all interpretation in terms of knowing a (formalizable) language alone, "the concept of language can still be useful in our understanding of linguistic communications," if it is assigned a new explanatory function. Thus Ramberg emphasizes an informal conception oflanguage. In particular, talk of language plays a crucial role in the author's analysis of incommensurability in Chapter 9. Both the concept of truth and that of incommensurability turn out as essential elements "in a dialectic of critical, reflexive interpretation." In the short fmal chapter, emphasizing "a concept of meaning in which conventions are not intrinsic"-because based on the model of radical interpretation-, 11 Ramberg argues for crucial similarities between Davidson's work and Gadamer's. "What Gadamer says of the hermeneutically enlightened consciousness-opposing it to the unre-


6. Ramberg 1991, p. 5. 7. Ibid., p. 100. 8. Ibid., p. I 04. 9. Ibid., p. 105. 10. Ibid., p. 104; Cf. Davidson 1986, "Nice Derangement," p. 446. 11. Ramberg 1991, p. 139.



flexive critique of prejudice-is also a description of the process of radical interpretation." 12 Quoting Gadamer: 13 "it allows the foreign to become one's own explicating it within one's own horizons with one's own concepts ..." 14 In a similar way, Davidsonian radical interpretation allows us to overcome the conventions of our own language in encounter with the speech of another. One might better say that we come to be able to express the foreign conceptual system making use of our own grammar and vocabulary except, of course, that Davidson's famous (or infamous) rejection of the notion of conceptual system would not allow him this way of stating the matter. Instead there is talk of differing theories and the modification of interpreter's theories through communication. All well and good. The only sense to be made of differing conceptual systems is in terms of differing theories which embody them. 15 In order to follow out Ramberg's program, what is needed is a frank recognition that interpretation aims to elucidate a theory (or belief-system) expressed in the syntactically identified language. Interpretation may in fact be a matter of continuous process, with no complete and formalized product, as ordinarily practiced. However, a formalized conception of languagesystem would seem to be in order for a more scientific approach to semantics-even if empirical semantics usually only aims at a partial theory of speaker's (system-immanent) competence or the (idealized) semantic common denominators of a given community. In Ramberg there is danger that such products will be lost to the process, though this is not a reason to ignore the emphasis on the linguistic intelligence responsible for creative development of language and interpretation. Davidson seems to approximate this perspective in his article, "A Nice Derangement of Epitaphs," and thus, Ramberg's opening quotation from this paper might be regarded as an instance of Davidson following the "ways of paradox," making or suggesting a point which has already been suggested by others: the naive concept of language is not suited to scientific investigation in empirical semantics. Further, according to Davidson, "language," like "meaning" must take a back seat to "truth" and "belief"

12. Ibid., p. 140. 13. Ibid., p. 140f. 14. Gadamer 1976, "Semantics and Hermeneutics," p. 94. 15. Cf. Callaway 1988, "Semantic Competence and Truth-conditional Semantics," p. 20.



2. Truth and Reference The third chapter "Reference," is largely devoted to explication of Davidson on reference and satisfaction. Consistent with the thesis of Davidson's "Reality without Reference"(l 977), the notion of reference is turned over to various advocates of intensional meanings or the causal theory of reference. Some will find this disconcerting, since Quine, among others, has used the term "theory of reference" as a general designation of Tarskian semantics, including theory conducted in terms of "denotation," "designation," "satisfaction," "values of the variables," and so on. Moreover, it seems reasonable to view Davidson's essential innovation as a matter of simulating or reconstructing traditional concepts of meaning within the Tarskian theory of referential notions. Thinking ofQuine's "Notes on the Theory of Reference," or The Roots of Reference, we are reminded of extensionalist claims upon the vocabulary. Still, these claims have not gone uncontested, and the epistemic neutrality of Tarskian semantics has long been contested or misunderstood. So, there is this much excuse for Davidson's repudiating the notion of reference. Still the point does not go down easy. Ramberg supports Davidson on reference and satisfaction by arguing convincingly that we can have no general criterion or theory of reference with independent epistemic bite. Just as it is hopeless to look for a general criterion of truth, by means of which we could decide if an arbitrarily selected sentence is true or false, so there can be no general criterion of reference, by means of which we could decide if arbitrarily selected expressions refer to, or are true of, a given object or set of objects. Ramberg and Davidson rightly reject the conflation of epistemology and semantics involved in the search for general criteria of truth or reference. "This is the mistake of thinking that if we only stare at the concept of truth hard for a sufficiently long time, we will crack its riddle and thus recognize its mark on all true sentences." 16 The Cartesian dream of a certain mark of truth is not generally entertained in contemporary philosophy. Yet, the search for a general criterion or theory of reference has persisted, though this is equally a mythical philosopher's stone. For suppose we knew what reference is in the sense of being able to decide for a given name 'a' and a given predicate 'F' what object 'a' designates and what diverse objects 'F' is true of. Given that, the sentence '" a' designates an object of which 'F' is true," tells us, in the semantic metalanguage that 'Fa' is true, a theory which completely determined reference 16. Ramberg 1991,p. 9.



would seem to make any other inquiry unnecessary: it would amount to a "certain mark" of truth. This may seem plausible, if not explicitly stated, in light of the fact that expressions get linked to objects and classes of objects by means of linguistic conventions, or if we hope to uncover referential relations by tracing back the causal history of the use of expressions. But all of this is mistaken in so far as it assumes that semantics can independently answer epistemically substantial questions. Most answers are parasitic on theoretical-empirical developments in the object language. In standard Tarskian semantics (considering logical truth in terms of domains and interpretations, for instance), interpretations are given to expressions of a theory, regarded as true, or provisionally regarded as true. Our substantial knowledge about the reference of constituent expressions is normally derivative from, and dependent upon, the assumed truth of an objectlanguage theory (or its axioms), and it can have no greater or independent epistemic value-since there are no analytic truths valid independent of all empirical-theoretical revisions (or variations) of belief. The approach to referential notions which Davidson rejects involves the quest for independent epistemic import. "Davidson's strategy is the opposite, reflecting his inversion of the building block approach. In this theory, correspondence (in the form of satisfaction) derives whatever content it has from the part it plays in a theory which is testable ..." 17 To generate the theorems of a T-theory, we must invoke "a relation mapping expressions to objects. And 18 so it does stipulate a relation between words and objects." This is what is retained of the notion of correspondence. But this relation does no epistemic work in a T-theory. We "assign" expressions to objects in the axioms ofa Ttheory, so as to make the T-sentences come out right-where these are attested by empirical investigation of the environmental conditions under which sentences are held true. There can be no question of first knowing, independently, what objects satisfy which expressions and then deriving, on this basis, truth-conditions for object language sentences. From this perspective, there is no reason to expect that a causal theory of reference could, in general, tell us the referents of arbitrarily selected expressions. At the most, one expects a limited and fallibilistic account of reference in causal terms. We are ultimately dependent upon science or common belief to know what objects there are to stand in causal relations to expressions. Obviously, knowing all about what reference is, will not be a substitute for scientific or common-sense inquiry. Hence, we should not expect to totally

17. Ibid., pp. 43-44. 18. Ibid., p. 44.



substitute causal for referential notions. There will be no reduction of semantics to physical sciences. In Tarskian semantics, we have a kind of definition in use of referential notions (relative to a language-system), but this confers no independent answers upon questions about reference and referents. "Going on to explicate reference further in terms of causal connections would then add nothing to our understanding of what it is for words to mean." 19 (Better, causal-historical research can at best supplement, never fully replace, other theoretical research.) For any substantial answers we come to will be parasitic upon object-level language-systems and their epistemic status. Meanings (interpretations of expressions) are so crafted, and recrafted, as to allow us to say what our (object-level) empirical inquiries lead us to need and want to say. Thus the priority of "truth" over "reference" (or satisfaction) and of"theory" (or "belief-system") over "conceptual scheme."

3. Radical Interpretation Chapters 6 and 7 are devoted to Davidson on radical interpretation, the first focusing on the principle of charity and the second largely concerned with "anomalies and indeterminacies" of interpretation. This substantial focus on the theme of radical interpretation, at the center of the book, is fully in order. It is such thought experiments which allow an appropriate emphasis upon the empirical evidence required by a semantic theory for a linguistic community. Errors in interpreting Davidson's theory easily arise if this point is missed: "The process of radical interpretation is intended as a theoretical description of linguistic competence, a rationalization of the practice of interpreting speech, not as a description of an actual procedure such as the methods of translators." 20 It is by reference to the process of radical interpretation that we are to see what is involved in "getting the T-sentences right." To get the process of radical interpretation started a number of presuppositions have to be made. The radical interpreter has to assume that she's confronted by creatures with beliefs and intentions who assert sentences and "when they assert, they largely do so correctly."21 The latter part, the ill-named Principle of Charity, is the subject of an extensive literature and given due attention by Ramberg. But there are two, more hidden, assumptions which are at least as crucial. First, its assumed that the radical interpreter can identify sentences or assertions-truth-value bearing units amidst a 19. Ibid., p. 36. 20. Ibid., p. 66. 21. Ibid., p. 68.



mass of phatic communion, questions, orders, and other noise-without having already identified particular beliefs, meanings, desires, intentions, and so on.22 Second, the "concept of truth that underlies a theory of interpretation is a concept of absolute truth." 23 Ramberg mentions these two further assumptions but might have gone further. The first point raises a question of whether it would not be better to closely examine the interrelations of the ascription of beliefs, meanings, and grammar. We are tempted to think that "light dawns gradually over the whole." Moreover, it is not merely a matter of a set of beliefs and meanings. An entire form of life becomes involved, since action generally is relevant to interpretation. The second point, concerning "absolute" truth also warrants more attention, particularly where the attempt is made to relate Davidson to continental philosophers such as Gadamer. We need to ask about "absolute truth"-the notion of"truths-forlanguages as somehow the same"-the notion that "drives interpretation,"24 whether it amounts to or is presupposed by the claimed universality of hermeneutic rationality. What is clear on all sides is that our ability to ask about what is true always outruns our ability to answer related questions in compelling ways. Ramberg provides useful discussions of problems and errors in the understanding Davidson's theory of interpretation. Three points deserve particular emphasis. First, the Principle of Charity is "not a pragmatic constraint...but a precondition for interpretation. " 25 The temptation is to see Davidson as mistaken if he does not regard initial charity in the attribution of truth to particular speech acts as a defeasible way of getting started. Moreover, we might try to simply imitate what the natives say, while trying to help in what they do, focusing on imperatives and their execution rather than statements and truth. But it does seem clear that Davidson is not tempted, to this point, by a pragmatic view of the principle of charity. What is not clear is that his is the best approach. There is much resemblance to pragmatic themes, since problems in interpretation (where we are tempted to attribute explicable error) can only arise within the context of something understood as correctly interpreted-just as a problem in our own beliefs can only be understood against the background of our normal success in action based on unproblematic belief. If we generalize the traditional pragmatist treatment of belief and the problematic to problems in interpretation, the need to establish a rela-

22. 23. 24. 25.

Ibid., p. 68; p. 72. Ibid., p. 76. Ibid. Ibid., p. 77.



tively (and fallibilistically) unproblematic context stands out. Why claim more, and what more could we plausibly claim? Second, Ramberg insists that the procedure of radical interpretation is not a matter of guessing what, say, 'gavagai' means by giving its translation. The radical interpreter "is trying to formulate a sentence which states as specifically as possible the combination of features characterizing occasions when speakers of L utter 'gavagai'." 26 This is an approach to the problem of synonymy in Davidson's program--dating back to criticisms due to Putnam, and John Foster (see the discussion of"counterfeit T-theories" on p. 79). One problem with this approach is the talk of "features." If features are enough like properties to enable this account to make reasonable interpretations plausible, then it seems that features are just meanings-projected into the environment. Nor is it clear that Davidson must avoid "translation" with its implicit synonymy claims. For example, ifwe interpret the native's "Questa e acqua" as "this is water" rather than "this is H20," then we are apparently focusing on the "water feature" rather than the more esoteric chemical feature. But looking to the axioms of the truth theory, (I)

'acqua' is true ofx iff x is water

is no less extensional than (2) 'acqua' is true ofx iffx is H20

Whether (1) or (2) gives the meaning of 'acqua' seems not the kind of problem we could settle by postulation of distinct "features." The question concerns the interrelations of speech acts in which the word is used. To justify (1) as against (2) we need to consider the entire range of usage of the word, and that is basically how we would have to explicate the talk of features. Clearly we can aim to simulate talk of meanings without departing from extensionality. Third, Ramberg insists that the radical interpretation model is not a model of a static state of semantic competence, but a model of a process. Semantic understanding is essentially dynamic.27 Therefore it is a mistake to think "that individual sentences must have their own determinate meanings, that there is something that determines what language someone speaks prior to the empirical task of systematizing the semantic structure of someone's speech behavior ... What is indeterminate is not what a sentence of a language means,

26. Ibid., p. 66. 27. Ibid., p. 78.



but what language is being spoken . ... Since we never apply the exact same theory to any two speakers, or even to any one speaker at different points in time, this makes the identity of languages a matter of degree." 28 In view of this final point it might be speculated that Ramberg himself sometimes takes the radical interpretation procedure pragmatically, for example when he suggests that the radical interpreter has to assume "that she is dealing with one homogeneous language community."29 The major consequence of the radical-interpretation model, according to Ramberg, is that it provides us with a concept of meaning in which conventions are not intrinsic: As communicative exchange proceeds, as language is used in monologue or dialogue, what keeps an interpreter in the game is not any one theory, but the ability to come up with new ones. Even if it were granted that any given truth-theory captures a set of conventions, the continuous production of such theories cannot be described as a matter of conforming to conventions of meaning." 30

However, in connecting these ideas with those of Gadamer, a tension arises. "No actual speaker is able to achieve complete freedom from linguistic convention and still understand, or convey meaning."31 This leaves one wondering whether the non-actual radical interpreter isn't too bleak:32 If really all conventions and psychology-any "principle of humanity"-are dropped the task seems practically hopeless: What justification do we have of the assumption that the radical interpreter is faced with a human being? Of course Davidson doesn't deny, as e.g. Dummett33 stresses, that in practice conventions are important for communications, but they are "only" of pragmatic value; they are contingent with respect to the task of the radical interpreter. The difference between the truth-theory (or truth-theories) the

Ibid., p. 90. Ibid., p. 8 I; cf. p. 88. Ibid., p. 104. Ibid., p. 139. Cf. Blackbum 1984, Spreading the Word, p. 278; Blackbum 1996, "The Dispute on the Primacy of the Notion of Truth ...," p. 1017, regarding linguistic conventions. Also, see Gadamer, quoted in Ramberg, p. 139, "One of the fundamental structures of all speaking is that we are guided by preconceptions and anticipations in our talking in such a way that these continually remain hidden and that it takes a disruption in oneself of the intended meaning of what one is saying to become conscious of these prejudices as such." Gadamer 1976, p. 92. 33. Cf. Dummett 1986, "Some comments on Davidson and Hacking."

28. 29. 30. 31. 32.



radical interpreter has to construct and the formations of conventions is that 34 the first task is synchronic and the second a diachronic phenomenon. Critics like Dwnmett imply that to interpret an utterance is to "translate" it into your own idiolect; but "what already is expressed in your own idiolect is not interpreted, you somehow just get it."35 Instead the radical interpretation model reverses the logical priority of the concept of a language and the concept of interpretation 36 The salient contrast is not that between idiolects and languages, but between occasions of utterance, on the one hand, and abstractions, such as languages or idiolects, on the other." 37

4. Incommensurability Ramberg suggests that "the question of whether continuity of reference is what makes communication possible" 38 is at the core of the incommensurability debate. If this is the correct analysis of the incommensurability issue, then Davidson's theory might have something interesting to offer, because it appeals to a conception of semantics which "makes no essential use of reference." It follows directly from Davidson's model of radical interpretation that the thesis of incommensurability taken as absence of translatability is incoherent because, "Davidson equates a conceptual scheme with a language or a set of inter-translatable languages, and then goes on to show that the only such set we can conceive of is the set of all possible languages." 39From the point of view of the radical interpreter and her extensional semantics, incommensurability understood as impossibility of translation is not conceivable.40 If a radical interpreter cannot interpret a "language" then it is unclear how the speakers of this "language" would have learned it.41 One feels inclined to remark again that those speakers might have learned their language because they are allowed more resources than the radical interpreter, thus pointing to practical problems of access.

34. 35. 36. 37. 38. 39. 40. 41. 41.

Ramberg, 1991, p. 111. Ibid., p. 108. Ibid., p. 109. Ibid., p. 106. Ibid., p. 118. Ibid., p. 119. Ibid., p. 120. Ibid., p. 121. Ibid.



Ramberg suggests it is a mistake to identify incommensurability with intranslatability. It is a mistake related to the error of identifying a language with a set of conventions. lncommensurability is a "disruption in the ongoing interpretation-through-application of our linguistic conventions"42 But it does not follow that translation or communication is, therefore, not possible, "only that interpretation rather than reliance on convention, is required to a greater degree than usual."43 Kuhn is on the right track to suggest that "what is required in cases of incommensurability is precisely interpretation or translation."44 Probably both Kuhn and Feyerabend would agree with Ramberg's broad assessment of the communicative aspect of scientific revolutions and incommensurability. But Ramberg is wrong to think that that is all there is to the incommensurability issue. First, contrary to the impression Ramberg gives, in recent writings Kuhn (and also Feyerabend) make a clear distinction between learning a language (a theory) and translating one theory/language into another. For example Kuhn (1991) writes: "more recently, I've spoken of the historian's recovery of older meanings as a process of language learning rather like that undergone by the fictional anthropologist whom Quine misdescribes as a radical translator... . The ability to learn a language does not, I've emphasized, guarantee the ability to translate into or out of it." Second, Ramberg seems to miss the original worry about incommensurability when he says that it is "part of the semantic evolution of language."45 Although Kuhn himself appeals to evolutionary considerations, the original problem was precisely that no reasons or rational justification can be given why we should prefer one theory over another "incommensurable" theory. In Ramberg's solution, either this worry is dropped in favor of the salvation of the universal conversation of mankind, or the unjustified assumption is made that "evolution" means "progress" (of universal reason, toward absolute truth). It takes away the distinction between learning and translation by assuming that no restrictions are placed "on what can be said in a given language, because we are, of course, able to construct new sentences with new extensions"46-an idea which, interestingly, can be traced back to Edward Sapir (1949). But nothing follows as to whether we have to commit ourselves to one scientific theory or another, no matter how well we've

42. 43. 44. 45. 46.

Ibid., p. 130. Ibid., p. 131. Cf. Ibid., p. 125. Ibid., p. 129. Ibid., p. 67.



understood both of them by extending our language in the appropriate ways using the procedure of radical interpretation. Still Ramberg makes an important contribution by suggesting how incommensurability can be overcome by continuing inquiry. Kuhn's talk of Gestalt switches is inappropriate; the conversion is not one of seeing the world differently, but "is learning and using a new language."47 Differences in conventions between distinct linguistic communities may be overcome by formulating a joint or common language to meet outstanding problems between the groups. Ramberg applies the same reasoning to the interpretation of foreign cultures. Incommensurability will occur, not "because alien traditions as such could turn out to be incommensurable, but because of the immensity of the task of penetrating the prejudices that cause us to misinterpret what is foreign, and so to mistake the language being spoken."48 "Davidsonian semantics is entirely without epistemological ramifications."49 But it is precisely because Davidson's theory of interpretation does not cut "much anti-relativist ice" 50 that one wonders why Ramberg is confident that it can solve the problems of intercultural communication. Consider the following two quotations from Ramberg: The difference between the interpreter and the L-speakers is not a difference in what they see and feel, but in what they look for, and in what they deem relevant to something's being a particular something. This will depend on what sort of features of the world they find useful to call attention to, juxtapose, or ignore. 51 Just as the principle of charity has no implications about the similarities or dissimilarities of respective world views, so the ineliminable possibility of translation is no guarantee that we share criteria of truth or have a -common core of empirical knowledge or that our languages enable us to refer to some fundamental common set of references. 52

Surely, at the level of broad sweeping statements this is not very different from what we find in the writings of Kuhn or the odd cultural relativist. (So they might be said to have undermined the third dogma of empiricism as

47. 48. 49. 50. 51. 52.

Ibid., p. 130. Ibid., p. 133. Ibid., p. 124. Ibid. Ibid., p. 85. Ibid., p. 124.



well, because none of them has ever denied that, in practice, radical interpretation works.) On the contrary, if we read the quotations more carefully it would seem that a more thorough analysis is required of what exactly is shared and not shared, different and not different, or given and not given, and how words like "world" and "similarity" are to be taken.

5. Conclusion If we take seriously Davidson's suggestion that "communication does not depend, as we naturally tend to suppose, on our speaking the same language" 53 one would expect a more revolutionary change than Ramberg is offering, restricting his book to questions of truth, meaning, and belief. Ramberg briefly mentions 54 Davidson's long-term aim of providing a unified theory of meaning and action, but even "an interlocking account of the central cognitive and conative attitudes" showing that "truth rests in the end on belief, and even more ultimately on the affective attitudes" 55 might not be enough. In a truly unified (or holistic) theory of interpretation, ascription of desires, emotions, intentions, and so on, as well as the meaning of non-verbal actions, would seem to be prior to, or at least as important as, the ascription of meaning to truth-value bearing utterances.56 On the one hand, because more "data" are added, this might make it more plausible that radical interpretation is always possible. On the other, it would make theory construction more difficult, because we would have to give up semantics as a completely autonomous discipline, embedding it into a comprehensive theory of the mutual interpretation of human (inter-)action.

53. 54. 55. 56.

Ibid., p. 105. Ibid., p. 90f. Davidson 1990, p. 315, p. 326. J. van Brake! 1991, p. 239; Callaway 1993, Introduction and Chapter 6.


Editor Larry Hickman remarks in his Introduction that "this is the third edition of The Collected Works of John Dewey, 1882-1953," 1 and "primarily a critical reading text that offers unprecedented access to Dewey's work." It differs in several ways from the two printed predecessors. Unlike the first hardbound edition, edited at the Center for Dewey Studies under the direction of Jo Ann Boydston and published between 1967 and 1990 by Southern Illinois University Press in thirty-seven volumes, the electronic edition has no need for the title and subject indices to the Collected Works, or the indices of the individual volumes. But in contrast to the paperback edition, and following the first-edition volumes, the present electronic version contains textual commentaries, lists of variants, emendations, line-end hyphenations, and substantive variants in quotations, checklists of Dewey's references, and other critical materials. Practically the entire end-matter of the hardback edition has been reproduced. However, the appendices, containing essays and reviews written by authors to whom Dewey had replied or referred, are missing from the electronic edition. These important materials can be found only in one of the two print editions or in the original publications. This "electronic Dewey" is available in Windows and Macintosh formats. The presel)t edition "is based on the critical edition, but differs from it in significant ways," partly because "its text had to be completely rekeyed" and partly because "no machine-readable text was produced at the Center as a part of the editorial process." Thus, as Hickman testifies, "almost fifty million characters had to be re-keyed in producing the electronic version," and "it was checked electronically for typographical errors and other anomalies" and "sight checked by the Center's editors." The distinctive design in layout and typography of the Early, Middle, and Later Works has been preserved, "but some changes were required because of


Adapted from The Journal of Speculative Philosophy, Vol. XI, No. 3, Fall 1997, pp. 225-230. Reprinted by permission of Penn State Press. I. The Collected Works of John Dewey (1882-1953), The Electronic Edition, on CD-ROM, 63MB, Charlottesville, Va.: Intelex Corporation, Past Masters, 1997.



the unique ways in which electronic texts are used." Because it involved re-keying, as well as some re-editing and reproofing, the electronic version "is not identical to the edition that was inspected by the Committee on Scholarly Editions." The expectation is, as Hickman puts it, that "this edition will be used frequently as a supplement to the critical edition (and vice versa)" and with this point in mind, the chief objective of this review is to say something about how the electronic Dewey can be used. A central advantage, of course, is the ease of searching of37 volumes of text. It is worth looking at this in some detail to illustrate the special advantages of Intelex's Folio Views software. Suppose, for example, that the reader has an interest in Dewey's relationship to Santayana. One way to approach this is to search directly for the word 'Santayana,' through the entire 37 volumes, or one might view the tables of contents of each volume in turn, looking for some mention of Santayana or his writings. A more helpful alternative, provided by the software, is first to call up the "contents" listing (by making use of a special icon, and afterwards to query Santayana (by use of another icon). The result is, once the listing of the contents is expanded into its details, that the reader will get a display telling which volumes make reference to Santayana. Since the contents for any of the 37 volumes can also be expanded into details by a click of the mouse, one arrives at a number assigned to the occurrences of Santayana for each article, review, chapter, or other element of the text. Having expanded the query in this manner, and given a little experience, one can more easily estimate the potential importance of any given occurrence of the term queried. For instance, it is fairly easy to distinguish among substantial discussions of Santayana in Dewey's writings, mere bibliographical mention of his writings, or the often quite interesting discussions of Dewey's relationship to Santayana in the Textual Commentary of a given volume. Once the full detail of the contents is brought out, each actual text can be accessed at the endpoints of the displayed "tree." In the entire Collected Works, there are, as it happens, a total of 255 occurrences of the word "Santayana." (Searching for William James would be a bit more difficult, of course, since "James" is a common first name of many people, and because the influential philosopher is not always distinguished in the text by his full name.) There is no mention of Santayana in the 5 volumes of the Early Works (1882-98), suggesting that Dewey's acquaintance or concern with Santayana began after 1898. In the Middle Works (1899-1924), however, Santayana comes up 70 times, and the Later Works (1925-53) contain 185



references. Except for the way in which these points of reference can be effectively sorted out among the various volumes and internal divisions of the volumes, the task of looking into Dewey on Santayana would be considerably more daunting. As it happens, the first mention of Santayana comes in Sidney Hook's introduction to the second volume of the Middle Works (MW 2:xiv). Hook there remarks that Dewey "agreed with the oft-cited dictum of Santayana that those who forget the past are doomed to repeat it;" and beyond a warning from Hook against our dwelling in the past, there is little more of interest. However, volume 3 of the Middle Works (1903-6) contains 8 references to Santayana. In the front-matter, the table of contents includes mention of Dewey's 1906 review of the first 2 volumes of Santayana's The Life of Reason which review, interestingly enough, was published in the journal Science. Dewey's essays in volume 3 contain one further reference to Santayana and the text of the early review of Santayana is available to the reader, with 4 occurrences of Santayana marked by highlighting resulting from the query and each available in tum by clicking the icon for "next." Selections from the text, or entire essays and articles (or entire books) are easily copied into separate files for future use or inserted into ongoing work. The remaining 2 references are in the end-matter of the volume, specifically in Dewey's footnotes. Having found these various references to Santayana, the task remains of sorting through them for significance. One should expect that reading the larger-print texts on a computer screen is generally easier than reading the same text in a printed volume. In addition, the software provides for the possibility of jumping between various pre-linked portions of the text, including the linkage of footnote markers to footnotes, variant reading markers to variants, and cross-reference markers to cross referenced texts. All of this is a great help in keeping track of publication dates of texts compared and in digging out the context of particular claims or developments of Dewey's views. The software also provides for attaching "sticky notes" of your own composition, attaching bookmarks, and highlighting particular passages. These personal markers are stored in a "shadow file," on the hard disk or a floppy disk and accessed from the main menu. The CD-ROM version has the advantage that the user does not need to copy the entire 37 volumes of material onto the hard disk. What follows are some of the early passages of Dewey on Santayana that seemed particularly noteworthy and that illustrate the search described above. Commenting on his "description of the function of knowledge in its own terms and on its merits," Dewey remarks, in his 1906 essay "The



Experimental Theory of Knowledge," that it is a description that "in intention is realistic," if by "realistic we are content to mean naturalistic, a description undertaken on the basis of what Mr. Santayana has well called 'following the lead of the subject-matter"'(MW 3: 118). But though Dewey lauds Santayana's call to follow the subject matter itself, the following text complains of difficulties in the path of inquiry: "At present the attempts are not, for the most part, even listened to on their own account;" instead, "they are compared with undertakings of a wholly different nature, with an epistemological theory of knowledge, and the assumptions of this extraneous theory are taken as a ready-made standard by which to test their validity" (MW 3:118-19). As is often the case with Dewey's writings, we have to wonder if the conjunction of a mention of Santayana with complaints and criticisms is intended to aim the criticism at Santayana or perhaps to call it to his attention. Often the great problem in reading Dewey is to understand why he says what he says-to reconstruct the context that guides his development of themes. Knowing of the various 1903 Emerson celebrations to which both Santayana and James made contributions,2 mention of Emerson in Dewey's review may catch the eye. Commenting on Santayana's style and the underlying logic, Dewey says that "in the main, Emerson's demand for a logic, so long that it may remain unspoken, is fulfilled" (MW 3:321). Thus, Santayana is praised in Emersonian terms. This establishes a positive relationship of Dewey to Emerson and raises the question of Santayana's relationship to the sage of Concord. (At this point, one might well look into Dewey on Emerson. Following the same search procedure described above, one comes to 93 entries for "Emerson." But while Dewey's references to Santayana are skewed toward the Later Works, Dewey's mention of Emerson began very early and continued throughout his long career.) Dewey's text, following the Emersonian praise, again suggests a criticism. "Dr. Santayana forces too far the doctrine of the inherently chaotic or maniacal character of consciousness by itself," says Dewey in the following paragraph, and Santayana's relationship to Emerson is suggested by Dewey's claim that Santayana "underestimates the part played in the progress of mankind by the ventures and insistencies of just brute vital impulse, however uninformed; and that accordingly, at times, the pale cast 2. See George Santayana, "Emerson" in Interpretations of Poetry and Religion (New York, 1900); William James, "Address at the Emerson Centenary in Concord," (delivered 1903) published in Memories and Studies (New York, 1911), and John Dewey, "Ralph Waldo Emerson," (1903) republished in Middle Works Vol. 3.



of thought is too emphasized and the fear of individualistic assertion too acute" (MW 3 :321-22). This is an important criticism, since The Life of Reason is subtitled "The Phases of Human Progress." If we assume, as according to Dewey's 1903 address from the Emerson Memorial Meeting at the University of Chicago, that Emerson is justly regarded as "The Philosopher of Democracy" (MW 3:184-92), then the implied criticism is that Santayana is too little concerned with the need for democratic participation and too much concerned with precautions against "individual assertion." The question arises if this is the start of Dewey's long debate with Santayana and his philosophy, and we may wonder, too, if Dewey persisted in the critical perspective suggested by the text. These questions may take us beyond what can be answered by the work under review alone, though I think the present work certainly helps in raising and answering this and similar questions. If we return now to the problem Dewey addresses in "The Experimental Theory of Knowledge," recalling Dewey's oft repeated imperative of "following the lead of the subject-matter" (the wording seems to be borrowed from Santayana, as we have seen), and the complaint that "attempts at the development of knowledge and inquiry" are not listened to in their own terms, but instead subjected to an epistemic "higher" criticism of sorts, then we may see a typically Dewey point being developed in preliminary criticism of Santayana. There is a need for broad democratic participation in the development of knowledge and inquiry, which insures against academic or class self-insulation. What we have so far are chiefly informed hypotheses concerning Dewey's early perspective on Santayana, and it is certainly necessary to look beyond Dewey's own works to track the relationship with greater prospect of verifying or disconfirming such hypotheses. (An early and very critical work of Santayana's on Emerson from 1900 has some importance, since both Dewey's address of 1903 and James's piece on Emerson from the same year are easily viewed as criticisms ofSantayana's 1900 essay on Emerson.) Yet, having said this much, I will add that we should not underestimate the potential of the present work. Looking beyond volume 3 of the Middle Works, there are many points in the texts yet to be examined. Important for an understanding of the early interaction between Dewey and Santayana is volume 4 of the Middle Works, including Dewey's much lengthier review of the full 5 volumes of The Life of Reason. Themes now familiar recur in the longer review of 1907. Dewey's impression of a "naturalistic idealism" in Santayana's early work (MW 3 :319), does not reappear in his account of the full 5 volumes.



However, what Dewey says about Santayana on knowing tends to confirm the apparent direction of criticism as sketched above: "Mr. Santayana seems to me to be in the precarious position," says Dewey, "of at once throwing an immense burden upon 'vital impulse' and of then damning it in the light of its own ulterior products in the way of 'form"(MW 4:241). Although Dewey's review is generally very appreciative of the positive aspects of Santayana's work, and although he did say that The Life of Reason was "the most adequate contribution America has yet madealways excepting Emerson-to moral philosophy"(MW 4:241), there can be little doubt that Dewey had entered onto a criticism of Santayana whose general features would persist in Dewey's writings for many years. In a sense, Dewey continued, in his readings of Santayana, the defense of Emerson and democracy that seems to have been evoked by Santayana's critical 1900 essay. Signs of this abound in Dewey's further work. Beyond the major encounter occasioned by Santayana's review ofDewey's Experience and Nature, and his reply, the electronic edition makes readily available to the scholar a long list of Deweyan reviews and comments on Santayana. In 1916 Dewey published an insightful review of Santayana's Egoism in German Philosophy; in 1923 Scepticism and Animal Faith received Dewey's attentions; in 1927 he reviewed Santayana's The Realm of Essence; and in 1941 Dewey reviewed the Schilpp-Hahn volume, The Philosophy of George Santayana. All of this, and much more, are put at one's fingertips in the electronic edition. There can be no doubt that the CD-ROM version of John Dewey, The Collected Works will provide a vital new stimulus to further Dewey scholarship. The ease of access it provides facilitates a qualitative difference in speed and dexterity of scholarly searches and comparisons of the full range of the marvelous 37 volumes of text provided by Southern Illinois University Press and the long years of work of the editors at the Center for Dewey Studies. The new Folio Views software is a great improvement over what I had heretofore seen, and since there are other electronic editions of philosophical works available from the same firm, or in production (including Peirce's Collected Papers and the emerging Santayana edition), the range of potential comparisons and searches can and will be vastly expanded. No serious Dewey scholar will want to be without this electronic edition. We may hope that the hefty opening sales prices for the electronic Dewey will eventually come down.



This book arose from the author's recent dissertation written under the Gerhard Schonrich at Munich. It focuses on Peirce's theory of categories and his epistemology. According to Baltzer, what is distinctive in Peirce's theory of knowledge is that he reconstrues objects as "knots in networks of relations." The phrase may ring a bell. It suggests a structuralist interpretation of Peirce, influenced by the Munich environs. The study aims to shows how Peirce's theory of categories supports his theory of knowledge and how "question concerning a priori structures of knowledge" are transformed within this relational framework. A chief critical target is David Savan's semiotics, specifically the idea that "the multiplicity of development of the categories" is "conditioned by nothing but the indefiniteness of the categories." 1 But in contrast with this, if there is any indefiniteness in the categories, they cannot fully direct their own application, and this is to say regarding them "that our knowledge is never absolute but always swims, as it were, in a continuum ..." 2 If the doctrine of continuity applies to the categories, they also have a continuum to swim in. The books gives considerable attention to the relationship between Peirce's theory of categories and its precursor in Kant. The Introduction opens with this quote: "Now upon the table of categories philosophy is erected,-not merely metaphysic but the philosophy of religion, of morals, of law, and of every science." 3 A concern with foundationalist theme marks the central thesis of the work.. "It is my task in the present book to demonstrate [a]. ..dependence relationship," says Baltzer. This is a dependence of semiotic upon the theory of categories. "It is not, as Savan 4 would have us believe, that signs ground the categories. Instead it is much more

* Somewhat expanded from my review of Ulrich Baltzer, Erkenntnis als Relatio-

1. 2. 3. 4.

nengeflecht: Kategorien bei Charles S. Peirce, 1994, which first appeared in the Transactions of the CS. Peirce Society, Vol. XXXI., No. 2, Spring 1995, pp. 445-453. Reprinted by permission of Indiana University Press. Baltzer 1994, p. 13. C. S. Peirce, Collected Papers, Vol. 6, paragraph 175, = (CP6. I 75). Peirce 1866, cited in Baltzer 1994, p. 11. Cf. Savan, David. 1977, "Questions Concerning Certain Classifications Claimed for Signs," Semiotica 19, No. 3/4, pp. 179-195.



the categories which are the foundational element, having their (most prominent) expression in the triad of signs."5 The claim is perhaps overly strong. Given Peircean fallibilism, we may wonder if sufficient stress has been placed upon the reciprocal or mutual dependence between the theory of categories and the theory of signs and predicates. Baltzer's thesis requires comparison with Elisabeth Walter's, in her Introduction to the German edition of the Lectures on Pragmatism: "Since, however, everything at all which can be said or thought is said in signs, they are the foundation of every other domain."6 The book consists of an Introduction, three major divisions consisting of a total of 18 numbered chapters, and a brief concluding Overview or Prospect. Beyond the main text, there is also a listing of abbreviations, a bibliography, and an index of personal names. The strength of this book is the elaboration of the author's themes through discussions of substantial portions of Peirce's work and the secondary literature. This could help to open the discussion of Peirce to wider audiences. If there is a weakness, it is a narrowness on alternative interpretations of fundamental Peircean ideas. Part I, "The Design of the Theory of Categories in the 'New List' " concentrates on Peirce's early development of the theory of categories, starting with an exposition, in the first two chapters, of Kant's theory of categories and Peirce's departures from Kant. There follows an initial exposition of the Peircean theory, including the start of an interesting emphasis on the idea of levels of analysis in the application of the categories. Part II, "Distinctive Features of the Ramified [ausdifferenzierten] Theory of the Categories," looks at the Peircean categories more systematically, including consideration of the value and interrelations of the various definitions Peirce gives to the three categories, the irreducibility theses, the question of the completeness of the categories, and the functions of the distinction between genuine and degenerative cases. Part III concentrates on Peirce's later "phenomenological" elaboration of his theory, where the categories are brought into closest connection with the epistemological themes of grounds for knowledge and their relational character. Overall this book is a thorough and rigorous treatment of the topic from a somewhat Kantian, and structuralist point of view. It provides an expository basis for discussions of structuralist vs. anti-structuralist and

5. Baltzer 1994, p. 11. 6. Walther, Elizabeth 1991, Charles Sanders Peirce, Vorlesungen uber Pragmatismus, "Einleitung," p. xvii.



foundationalist vs. anti-foundationalist interpretations of Peirce's work, especially for the German literature, suggesting significant relationships between Kantian and structuralist themes. In the present review, I briefly sketch the author's approach while suggesting an alternative perspective.

1. Relational Structure "It will be shown," argues the Introduction, "that Peirce assumes, for each relatum-whether mental act, state of affairs, or object of the physical world--that it is representable as a more or less complex relational structure and that this constitutes its essential character." 7 Objects in their "essential character" are analyzable into relations. The interpretive claim is used in an account of the multiplicity of Peircean applications. Because each relation is to be brought back to a categorical model, the ubiquity and variety of the categories would be explained as the various forms of appearance of differently structured relational complexes. Such a process of reconstruction of a state of affairs by means of relational structures is to be called a "categorical ordering" (kategorische Verortung). A central concern of this work is to analyze Peirce's theory of categories in terms of categorical ordering as the basic principle of understanding. 8 The term "categorical ordering," and the relationship of this to Baltzer's theme of levels of analysis involve some innovation and an especially valuable points in the interpretation. (See CP5.223 for a related Peircean discussion.) The idea ofrelational structure as essential is more problematic. "From the fact that the categories are fundamental elements of knowing, there follows a particular conception of the knowable object." 9 "Categorically structured knowledge," argues Baltzer, "cannot do without reference to further objects in the construction of the object of knowledge." That is, categorically structured knowledge "necessarily involves the reference or relation of one object to another, because the categories only structure such relations among objects." Thus, "the objects in question are always only determined [bestimmt] relative to the relation." Citing Peirce in Volume II of Kloesel and Pape's Semiotische Schriften, the author urges that "in Peircean context, the external reference or rela-

7. Baltzer 1994, p. 14. 8. Ibid. 9. Ibid., p. 163.



tions [externe Bezug] of an element are of central interest and not its internal construction." 10 Insofar as these interpretive points reflect Peirce's deep and thoroughgoing emphasis upon the logic of relations, they stands beyond reasonable question. For example, "Peirce attempts to show the superiority of organization in terms of types of valences or connections over sortings in terms of internal characteristics, by means of the analogy with the periodic table." 11 But the contrast put forward between relational or structural characterizations and characterizations in terms of "internal construction" is problematic. In the first place, Peirce's logic of relations clearly encompasses traditional subject-predicate logic by including monadic predicates. So, holding as he does, with reference to chemistry, that the valences or types of connectivity of the elements are crucial in the period table, this would not forbid consideration of differences between elements of the same valence and nor would it forbid application of monadic predicates in the details of classification or individuation. Baltzer urges that according to Peirce, "it is not in terms of internal characteristics that we can deduce law-like generalizations in chemistry, but instead only (simply? [lediglich]) on the basis of external connections to other elements." 12 The proper Peircean point is surely that seeing things in terms of their relations is crucial to understanding them. But this does not show that we need never consider "internal characteristics," on Peirce's account. In fact the chemical theory of valences came to a further explanation by reference to atomic number and correlated shells of electrons. Knowing the atomic number of an element one can deduce, in chemical theory, the number of electrons in each shell and thus its valence. Would this show that monadic predicates, e.g., 'x has-theatomic-number-79' and "internal characteristics, inside the atom" are really the only considerations of importance? It seems not. But this is just to repeat the point that the author's distinction between "relational structure" and "internal characteristics" is not clearly suited to make the kind of point he seems to want to make. What is missing is emphasis upon the ways that elements themselves may be better understood in virtue of seeing them in terms of their relations. "Internal character," may be

10. Ibid. 11. Ibid., cf. CP 3.469f.; CPI.288. 12. Baltzer 1994, p. 163.



revealed or more deeply probed, by thorough consideration of the full range of relations. According to Baltzer, "Peirce's conception of the categories constitutes the central core ofa theory of knowledge which conceives of the objects of knowledge as the limit-value of an unending series of conclusions." The point implicitly relies upon Peirce's conceptions of truth and inquiry and their relationship to the infinite community of inquiry. In support of the idea of objects as the "limit-value" of an unending series of conclusions, appeal is made to the 1868 Journal of Speculative Philosophy Series. The Peircean point that "There is no immediate intuitive knowledge of states of affairs," 13 is, then, equated with the conception of objects of knowledge as the "limit-value of an unending series." Presumably, if our relational reconstruction or analysis is never fmished, we can only know objects "as they appear," i.e., as relational complexes, and we would only know the "object itself' if the limit-value were to be reached. Baltzer's Peirce seems very Kantian. This impression is reinforced by his juxtaposition of Peirce on indeterminacy of objects and vagueness, with a discussion of Peirce on the "absolute individual." 14 Following Peirce (CP 3.93), the author argues that "An absolute individual can neither be thought nor perceived. It can not be perceived, because each of our senses only covers a partial domain of perception, so that many aspects of the perceived are not determined." Further the absolute individual cannot be thought, "because its determination in thought presupposes the defmition of all conceivable, i.e., infinitely many predicates." 15 The problem is that Peirce's discussion is plausibly seen as grounds for rejecting anything very similar to Baltzer's quasi-Kantian distinction between relational structure and internal structure. Or, to put the point in a slightly different way, we need not know everything about an object in order to know anything about an object. Peirce takes a similar stance on this issue when he says, in his review of Frazer's edition of Berkeley's works, that for many questions, the final agreement is already reached." 16 The point is expanded on in Peirce's draft review of Royce (c. 1885) . ...upon innumerable questions, we have already reached the final opinion. How do we know that? Do we fancy ourselves infallible? Not at all; but

13. Peirce quoted in Baltzer 1994, p. 164. 14. Baltzer I 994, p. 169ff. 15. Ibid., p. 171. 16. Peirce 1871, CP 8.12.


MEANING WITHOUT ANALYTICITY throwing off as probably erroneous a thousandth or even a hundredth of all the beliefs established beyond present doubt, there must remain a vast 17 multitude in which the final opinion has been reached.

Neither infallibility nor complete knowledge is required in order to know something about particular objects or domains of objects. Maintaining this point is crucial if we are to make the Peircean distinction between what is problematic and what is not problematic, in avoiding versions of ontological agnosticism, and in the Peircean rejection of the Kantian "thing-in-itself."

2. Irreducibility of the Categories According to Baltzer, the foundational function of the categories "is to be grasped in the domain of philosophical semiotic." His argument for this is partly that the "sign is a paradigm for one of the three categories." But he seems not to allow for a dependence of the theory of categories on semiotic. Clearly, for Peirce the categories are metaphysical notions which play a methodological role in inquiry. It would be an error to view them as absolutely a priori instead of being epistemically supported via fruitful applications. For example, if we found no compelling application for triadic relations, in results of semantic inquiry, this would call the category of Thirdness into question. The point can be illustrated by reference to Baltzer on the irreducibility of triadic relations. This theme is central in the second Part of the book, running through two chapters on "The Strict Conception of the Irreducibility Proof," and "The Broad Conception of the Irreducibility Proof." The distinction depends on the point that aspects of the thesis of the irreducibility of Thirdness are built into the logical system which Peirce makes use of in the strict proof. Thus Baltzer uses a broader conception of the Peircean proof as the context appropriate for evaluating presuppositions of the "strict" proof. Though it is possible to express what could be said using three-placed predicates by means of two-placed predicates, in the standard predicate logic, this is accomplished by means of a bound variable with three occurrences within the analysis, something that Peirce's system does not allow. In Peirce's system, this use of the bound variable would be replaced by the

17. Peirce, CP 8.43.



explicitly three-placed relation of "teridentity." 18 To put the point in another way, thinking of Thirdness as a matter of plurality beyond two, the standard predicate logic confirms Peirce's emphasis upon plurality, since it implicitly allows identity between any number of elements of a domain, via the cross-reference of bound variables. But obviously, it does not favor 1, 2, and 3, in quite the way that Peirce does. I suspect that there is a question here about the usefulness of the standard predicate calculus in cases, where we are less confident of our understand of particular relational notions. A similar point arises in discussion of the broad conception of the irreducibility proof. 19 Though chiefly drawing on Peirce's reaction to the work of A. B. Kempe, 20 and replying to Christopherson and Johnstone's "Triadicity and Thirdness," 21 the argument also casts a critical light on Quine's "Reduction to A Dyadic Predicate." 22 The idea involves the analysis of three-placed relations or rhemata as ordered triples. The Peircean paradigm of 'A gives B to C' can then be analyzed in terms of orders pairs, either or . Thus, "the triple is dissolved into an ordered pair of an element and an ordered pair, and this, on first sight, seems not to presuppose any element ofThirdness." 23 Baltzer objects as follows: As against the situation at the start which works directly with the three indices A, B, and C, the representation with the help of ordered pairs requires the creation of an ens rationis, namely the ordered pair of B and C or of A and B. The result thus combines, in either case, an ens rationis and an index. This shows that in such a method of representation, we cannot do without Thirdness in an adequate analysis of the state of affairs. 24

The reduction would only go through, if one allows both that an ordered triple can be analyzed in terms of ordered pairs, and that the further analysis of ordered pairs into two elements is not relevant to the

18. See Baltzer 1994, p. 115. 19. Ibid., p. l l 6ff. 20. Peirce CP 3.424. 21. Christopherson, R. and H. W. Johnstone 1981. 22. Cf. Quine 1954, p. 224. Quine aims to prove that for any interpreted theory, formulated in the notation of quantification logic, with interpreted predicate letters, that the theory is translatable into another quantificational theory with only one predicate letter, "and it is a dyadic one." This implies the reduction of any theory containing triadic predication to another which does not. 23. Baltzer 1994, p. 120. 24. Ibid., p. 120.



irreducibility thesis. But Baltzer objects that this is "not an adequate analysis of the state of affairs." In Peirce's words, replying to Kempe, "The diagram fails to afford any formal representation of the manner in which this abstract idea (or ens rationis) is derived from the concrete ideas." 25 Thus, the irreducibility thesis seems to rest on an semiotic anti-nominalism on what is to count as "adequate analyses" of states of affairs. Yet this threatens to beg the question in favor of Thirdness, unless we are able to appeal to an independent assessment of the value of results in applications. This point allows a guiding function for the categories in inquiry. But it suggests that a guiding function or "fallibilistic foundation" ("a tarmac on the road of inquiry," in a phrase once suggested to me by Cathy Legg), is subject to a holist assessment in light of the fruitfulness of its applications; the categories cannot fully determine either what to count as fruitful results or their own ultimate validity. The suggestion is, then, that the standard predicate logic is better suited as a "logic of inquiry," (or logic ofresearch) as contrasted with a logic suited to the statement of established results of science or inquiry. Peirce's logic can be understood to suppose that (paradigmatic) triadic relations will never be adequately analyzed in lower terms, and that we will never need any relational predicates of n-places, (n>3) which cannot be adequately analyzed, or explained by use of predicates of I, 2, or 3 places. I suspect that the devil is in the details here, since "adequate analysis" is something we may only achieve after prolonged inquiry. Nor should we suppose that such inquiry will remain unaffected by empirical developments.

3. The Structure of Firstness Fallibility is a persistent theme in Peirce, just as Firstness and chance are persistent themes. Thus whatever the foundational element we may find in his work, it must be subjected to a fallibilist therapy. I emphasize this theme, though it is somewhat external to Baltzer's approach in his book, because it is especially worthy of attention at present. Baltzer's treatment of Firstness is particularly illuminating. For Firstness seems to most resist formal definition or any formalistic interpretation. The unexpected, that which breaks down our preconceptions and constrains our reconstructions, is to be expected in the world that Peirce envisages, and if the category of Firstness did not allow of a certain vagueness, and resistance to a priori or structuralist treatments, then it is difficult to see how there would be any plausibility at all in the view of Peirce as an 25. Peirce CP 3.424.



"evolutionary realist." 26 Yet in his discussion of the expressions or developments (Auspragungen) of Firstness, 27 Baltzer disputes the idea that "Peirce's categories are at once formal and material concepts." 28 This contrasts with Peirce who equates "Firstnesses," with positive internal properties of the object in itself."29 This point relates to the author's closing acknowledgment of the need for further work on Peirce's concept of reality in connection with the categories. "In regard to Peirce's conception of the structured character of all knowledge, states of affairs, and things of the physical world, ...the concept of reality possess quite obviously a fundamental (foundational?) character." 30 In his scholarly and clearly written work, Baltzer thus prepares the way for further discussions in the German literature and beyond.

Cf. Hausman, Carl 1993, C. S. Peirce's Evolutionary Philosophy. Baltzer 1994, p. 97ff. Ibid., p. 98. Peirce 1907, ms. 318, in Kloesel, Christian and Helmut Pape (eds) 1993, Semiotische Schriften, Vol. 3, p. 278. 30. Baltzer 1994, p. 226.

26. 27. 28. 29.


Susan Haack presents a striking and appealing figure in contemporary AngloAmerican philosophy. In spite of British birth and education, she appears to bridge the gap between analytic philosophy and American pragmatism, with its more diverse influences and sources. Well known for her writings in the philosophy of logic and epistemology, she fuses something of the hardheaded debunking empiricist style of a Bertrand Russell with a lively interest in Peirce, James, and Dewey. As Americans generally know, or ought to know, this is a quite volatile mixture. It crosses the internal boundaries of our Anglo-Saxon and "continental" European philosophical and cultural heritages. The moderation of the title is well suited to appeal to the scientifically oriented establishment of phllosophy in America, with its frequently anglophile tones, while the "passion" and Haack's knowledge of the pragmatists, promises broader perspectives. If we think of American pragmatism as having arisen from a crossfertilization between our Whiggish (and Scottish) "common-sense," realist, and empiricist establishment of the nineteenth century, and the German romantic and idealist inspiration to Emerson and the Transcendentalists, then given the current revival of American pragmatism, one might suspect that the Brits have more fully joined the party. Happily, the Americans are clearly on both sides. In full protean and pluralistic spirit, we do not stand to lose the debates. But do high rational and scientific standards combine with passionate, even if moderate, commitments on controversial social and academic questions? We are about to find out. The book is a collections of 11 essays and addresses from the past 5 years or so. There is also a short Preface, an Introduction, and the Acknowledgements at the end of the book detail the providence and history of the writings. Haack defends the "old pragmatists," C.S. Peirce, William James, and even John Dewey (G.H. Mead makes no appearance), against the latter-day "neo-

* My review of Susan Haack, Manifesto

of a Passionate Moderate, Unfashionable Essays. University of Chicago Press, Chicago and London, 1998, x+223pp., which first appeared in Erkenntnis, Vol. 53, 2000, pp. 407-414. Reprinted by permission of Springer NL.



pragmatists," exemplified by Richard Rorty; and she also takes on a varied collection of multiculturalists, relativists, radical feminists, anti-racists, and others, whose commonality consists in the idea that "'knowledge' is an expression of power [and] that the concepts of evidence, objectivity, truth are ideological humbug." 1 The manifesto is directed against the contemporary leftward challenge to the possibility of objective inquiry which arises, directly or indirectly, from that unholy trio of cultural pessimists Heidegger, Foucault, and Derridawith or without benefit ofRorty's "neo-pragmatist" appropriations and blessings. In amongst all this critical fireworks, a commitment to inquiry as the Peircean pursuit of truth shines through. Haack can also be found to explain, extend, or apply central theses of her "foundherentist" epistemology, as formulated in her prior book, Evidence and Inquiry, Toward Reconstruction in Epistemology (1993). In language and style, at least, she here departs from the classical pragmatists, who generally avoided the word "epistemology," though not the recognized themes, in favor of a Peircean or Deweyan "theory of inquiry." What are the various empiricists and pragmatists to make, then, of "reconstruction in epistemology," which both recalls Dewey's Reconstruction in Philosophy, and invites us to consider the pragmatists's criticisms of classical empiricism and epistemology generally? Will Haack's foundherentism ameliorate the pragmatists attachment to contextualism or redeem "epistemology" in their eyes? "Foundherentism" seems to follow Peirce's implicit advice, evident when he re-christened his philosophy from "pragmatism" to "pragmaticism"-pointing out that the word was ugly enough to be safe from kidnappers. A chief claim is that "the appearance of cultural divergence of epistemic standards is a kind of conceptual illusion." 2 Haack insists that "though quality of evidence is an objective matter, our judgments of the quality of evidence are perspectival, since they can only be made from the perspective of our (fallible) background beliefs." 3 This is to allow that background beliefs do influence our judgments of the quality of evidence used to support a claim or hypothesis. This is a contextualist theory of evidence. When our context of knowledge and belief expands, other standards of judgment may recommend themselves. What gets our "warranted assertion" in one context of knowledge may later turn out false. But Haack appears to miss this point. She is intent on

I. Haack 1998, p. ix. 2. Ibid., p. 145. 3. Ibid., p. 144.



denying that there is "real relativity of standards of evidence," 4 and she envisages no non-relativist versions of contextualism. She neglects to emphasize the distinction between "relativity" of standards on the one hand, and justification in relation to information assembled and standards achieved on the other. 5 The tension here reflects real difficulties and confusions in the pragmatist tradition. Perhaps they cannot be adequately addressed without a more historically oriented approach. Haack takes aim specifically at an "epistemological counter-culturalist" version of multiculturalism. It certainly agrees with the best in the pragmatist tradition to hold with Haack, that the question "whether the standards of this or that culture are better, makes sense."6 There is "better and worse" in inquiry, and it is the objective of the pragmatists's "theory of inquiry" to address these matters. But notice that this is quite distinct from saying that we are always (already) in a good position to judge whether the standards of "this or that culture" or perspective are better. Epistemic standards, by pragmatist lights, are rooted in evolving methods of first order inquiry, and can be revised in light of the success and failure of those inquiries. Making such a judgment might require us to learn something of an alien culture or subject-matter, and that may not be so easy. Yet, in such a situation, if we continue to accept the best standards we know, that would not by itself make of us anything akin to Haack's "tribalists." The "tribalists" agree with the "relativists" that standards differ from culture to culture, but insist that their own standards are best, somewhat in the chauvinistic spirit of "my standards, right or wrong!" Haack seems to both allow and ignore the crucial factor of the development and change of standards ofjudgment. It is allowed, since our judgments of evidence are regarded as "perspectival," suggesting they can change with changes of perspective. But she does not seem to make sufficient use of the point. Adequate standards may come to depend on interaction of diverse "cultures," disciplines, and perspectives. Briefly, that is how contextualism avoids the conflict of the "relativists" and the "tribalists." We can hold that the development of standards of evidence is objective, without thinking that we always have all that is required, ready at hand. To use other words, epistemic standards of good evidence are broadly based in successful practices of inquiry but not known a priori.

4. Ibid. 5. Cf. e.g., Sidney Hook 1961, The Questfor Being, pp. 196-208. 6. Haack 1998, p. 144.



The titles and style of the essays are provocative and forceful. We are offered at the start "Confessions of an Old-Fashioned Prig," which borrows the self-description from Richard Rorty's claim that "philosophers who think of themselves as seeking the truth," in Peircean style, are "lovably old-fashioned prigs" 7 Haack seldom resist the ample opportunity to take a well-deserved poke at Rorty, though others come in for similar treatment, such as Stephen Stich who wrote that "once we have a clear view of the matter, most ofus will not find any value having true beliefs," and Jane Heal, who thinks "there is no goddess, Truth." 8 Speaking as one "old-fashioned prig" to another, bravo, Susan Haack. The second essay, "'We Pragmatists...': Peirce and Rorty in Conversation" shows that if Rorty did not exist, Haack might have needed to invent him to fully debunk the current misappropriation of the pragmatist tradition. By means of adroitly selected quotations, C.S. Peirce and Richard Rorty have a face-off in this piece, with Haack ever so gently entering in, on occasion, as a helpful facilitator of the conflicting views. In reply to Rorty's doubts about priggish truth seeking, Peirce replies that "In order to reason well..., it is absolutely necessary to possess ...such virtues as intellectual honesty and sincerity and a real love of truth," and "[genuine inquiry consists] in diligent inquiry into truth for truth's sake." 9 Where Rorty is quoted to the effect that "the very idea of a 'fact of the matter' is one we would be better off without," Peirce replies that "he wishes his opinion to coincide with the fact;" while Rorty insists on his anti-representationalism and rejection of truth as correspondence, Peirce obviously thought otherwise: "Truth is the conformity of a representamen to its object" 10 In similar style, Haack, like others before her, 11 quite effectively disputes Rorty's claims on the pragmatist tradition. Generally, Rorty over-emphasizes the influence of romanticism, an element especially distorting of Peirce's contributions. Though it is sometimes more difficult to see the realist, anti-subjectivist, and anti-nominalist themes among the other pragmatists, Rorty simply ignores how the typical pragmatist concern with objectivity combines with fallibilistic and evolving standards of "better and worse."

7. Quoted in Haack 1998, p. 7. 8. Quoted in Haack 1998, p. 7. 9. Quoted in Haack 1998, p. 31. 10. Quoted in Haack 1998, p. 33. 11. Cf. Herman J. Saatkamp, editor, Rorty and Pragmatism, The Philosopher Responds to His Critics, 1995.



In the third essay, "Studying in a Literary Spirit," Haack provides further exposition of Peirce, and beyond that she notes the commonality of social and political concern between Rorty and Dewey. But she adds that Dewey's pragmatism still contrasts with Rorty, since Dewey "stressed the importance of using the method of science to discover what is really conductive to human flourishing," and "the need for inquiry as the basis of wise and effective social change;" 12 this helps make clear that the "literary spirit" of Rorty's writings indeed over-emphasizes romantic imagination and free creation, neglecting the pragmatic emphasis on inquiry and evidence. Like Peirce, Haack aims to "rescue the good ship Philosophy for the service of Science from the hands of lawless rovers of the sea of literature." 13 Still this is no denigration of literature or of imagination. It is rather that "'literature' does not refer to a kind of inquiry," 14 though inquiry cannot do without imagination. "When a man desires ardently to know the truth," say Haack quoting Peirce, "his first effort will be to imagine what the truth can be." 15 So part of the point is that we dare not confuse the first step with the entire process. In essay 7, "Knowledge and Propaganda: Reflections of an Old Feminist," the author again borrows the self-description from someone else's ~~~~~w~~~~~~~~~~

believes in the primary importance of the human being," in some contrast with the "New Feminism" which emphasizes the importance of"the women's point of view." 16 Greater opportunity for women in the professions is much in order, then, but Haack takes aim at the notion that "there is any such connection between feminism and epistemology as the rubric "feminist epistemology" requires. 17 She urges that "politics should be kept out of science." 18 The basic question in this essay is whether the social conditions and standing of women (and implicitly, the question extends to other groups) justify talk of "women's ways of knowing," with epistemological relevancy. What the reader may think of here is special social circumstances which result in a particular facility to see the "glass ceilings," so to speak.

12. Haack 1998, p. 62 13. Peirce CP 5.449, quoted in Haack 1998, p. 48. 14. Haack 1998, p. 59. 15. Ibid., p. 57. 16. Quoted in Haack 1998, p. 123. 17. Haack 1998, p. 123. 18. Ibid., p. 127.



Haack objects that if oppression creates special circumstances for particular groups, and "one of the ways in which oppressed people are oppressed" is that their "oppressors control the information that reaches them," 19 then this does not suggest an epistemic privilege of the disadvantaged. Moreover, if there is some epistemic privilege among those who are socially disadvantaged, this would seem to argue for the paradoxically strong claim of greater epistemic privilege for those more disadvantaged. The argument is ingenious but perhaps not fully compelling. The reply remains open that the special circumstances involved depend not only on social disadvantage but also upon social processes in which these disadvantages might be overcome: it is not the fact that one lives under a glass ceiling which is telling, but instead that it is sometimes possible to come to see it. That could well amount to a special vantage point on social problems. "Those who wear the shoes know where they pinch," as Sidney Hook liked to say. Yet there is nothing in this reply which suggests a particular female epistemic privilege, if, as seems plausible, the social disadvantages of differing groups are basically similar. So, we might still be left with something resembling Haack's "old feminism." More basically, on the assumption that members of socially disadvantaged groups do stand to gain special insight into the means by which they are disadvantaged, this argues for an epistemic privilege, of information and/ or methods regarding essentially social problems, and not something which would immediately create a new epistemology in general. It would remain to argue that scientific and scholarly disciplines are so afflicted with prejudice as to undermine their own purposes, perhaps strongly tending to favor the wrong theory or thesis merely because of the social standing of the advocates. While such an situation is not impossible in particular cases, and universalizing such claims would seem to produce incoherence, the particular problems might still tell us something about social inquiry. John Dewey said of women writing philosophy, "we cannot conceive that it will be the same in viewpoint or tenor as that composed from the standpoint of the different masculine experience of things." 20 The point is an important theme in Charlene Seigfried's Pragmatism and Feminism (1996), and comparisons with Haack on feminism and knowing might be fruitful regarding related points. The genuine feminist will doubtlessly decide for herself.

I 9. Ibid., p. 126. 20. Dewey 1919, "Philosophy and Democracy," Middle Works, Vol. 11, p. 45.



Rightly rejecting strong claims for the sociology of knowledge, and those cognitive and cultural relativism which insists that "reality is however some epistemic community determines it to be," 21 Haack returns to her plausible and pragmatic insistence that the better and worse of evidence is an objective matter. She maintains that "the epistemological significance of feminist criticisms of sexism in scientific theorizing, though real enough, is undramatic, and by no means revolutionary."22 The examples of Nazis and Soviet science do show us that we need to care for the social conditions of inquiry, and that poor social conditions can be disastrous for inquiry; but in making such arguments and judgments we depend on the idea that we can identify theories and evidence as "better or worse," that these judgments are not always or essentially political and contentious or socially determined. There is perhaps room here for closer consideration of one further essay, where the author takes up consideration of American programs of affirmative action. The topic is especially controversial, and we may hope for a bit more objectivity, given Haack's partially non-American background and education. Generally, Haack objects to the prevalent "This-or-Nothingism" concerning affirmative action, which ignores the details of its advantages and disadvantages and fails to consider alternatives. She confines herself to affirmative action in academia. Marking a "key distinction" along the lines of the differences between "procedural fairness policies" and "preferential hiring policies," she argues that though she is sure that discriminatory hiring is a bad thing, "I am not sure that preferential hiring is the best solution."23 Haack is a radical critic of academic hiring practices, and she doubts that preferential hiring will help either the obvious or the deeper flaws. What drives academic hiring, she says, is not the merit and quality of candidates so much as the internal and "corrupt" politics of academicians. "The hiring process is too often less a straightforward if ill-informed and clumsily conducted effort to identify the best candidate than an unseemly struggle of 24 greed and fear." Greed: we want someone who will improve the standing of the department, who has contacts from which we might benefit, who will willingly do the teaching we'd rather not do, who will publish enough so the tenure process will go smoothly. Fear: we don't want someone so brilliant or

21. 22. 23. 24.

Haack 1998, p. 130.

Ibid., p. 130. Ibid., p. 169. Ibid., p. 172.



energetic that they make the rest of us look bad, or compete too successfully for raises and summer money, or who will vote with our enemy on controversial issues. We look, in short, for someone who "fits in" (The foamrubber PR term is "collegiality.")25

The position is qualified and careful, overall, but basically looks at preferential hiring as something like applying a small bandage to a deadly and bleeding gash. The detailed American discussions, often quite polemical, concern whether this approach or that, one program or another, will succeed in finding the best candidates in spite of racial or gender prejudice, but the underlying problems center in a too prevalent lack of interest in seeking genuine merit in candidates. Haack remains "unpersuaded that preferential hiring has, overall improved quality." 26 She contends that the essential problem is being ignored "how to make the hiring process less disgracefully corrupt."27 She takes on some sacred cows of the academic establishment, and the arguments are deserving of attention. Overall, this book, including essays not mentioned here, provides a somewhat provocative introduction to the contemporary interplay between analytic and pragmatic influences in American and Anglo-American philosophy. The author is particularly persuasive in bringing forward virtues of C.S. Peirce and his long-neglected writings. But Haack also continues her debates and examinations of some chief figures of contemporary analytic philosophy, and these are sure to interest many readers. I close with a particular example of Haack on Davidson. In the 4th essay, "Dry Truth and Real Knowledge: Epistemologies of Metaphor and Metaphors of Epistemology," the general concern is with the place of metaphor in science and philosophy. Along the way, Haack effectively disputes a key claim from Davidson, and one which Rorty has employed as a hook on which to hang his hat, as it were. She points to the danger of"mistaking a figurative use for a literal" one, illustrating the related points with Davidson's criticism of "experientialist foundationalism," the "idea that some beliefs are justified, not by the support of other beliefs, but by the subject's experience." 28 Davidson objected that a "confrontation of belief and experience, would have us 'getting outside our skins' to compare belief and experience;" and the reply is made by Haack that it is hard not to suspect that Davidson "is taking advantage of the fact that we cannot literally get outside our skins,

25. 26. 27. 28.

Ibid., Ibid., Ibid., Ibid.,

p. 173.

p. 176. p. 177. pp. 84-85.



cannot literally confront our belief with experience, to ease the reader into accepting that experience can have no relevance to justification." 29 Experience does have its relevance to justification, and reading Haack on "experience," whether as empiricist or as pragmatist, the point seems quite significant.

29. Ibid., p. 85.

BIBLIOGRAPHY Albert, Hans (1991) Traktat über kritische Vernunft. Tübingen: J. C. B. Mohr. Axelrod, Robert (1984) The Evolution of Cooperation. New York: Basic Books. Baltzer, Ulrich (1994) Erkenntnis als Relationengeflecht: Kategorien bei Charles S. Peirce. Paderborn, Vienna, and Zurich: Schöningh. Barnes, Barry (1985) “Thomas Kuhn,” in Skinner, Quentin ed. (1985) The Return of Grand Theory in the Human Sciences, pp. 83-100. Barrett, R. B. und R. F. Gibson eds. (1990) Perspectives on Quine. Oxford: Blackwell. Bernays, Paul, “Überlegungen zu Ferdinand Gonseths Philosophie,” Dialectica, Vol. 31, 1977, No. 1-2, pp. 119-128. Bernstein, Richard J. (1967) John Dewey. New York: Washington Square Press. Bever, T. G., J. J. Katz, and D. T. Langendoen (1976) An Integrated Theory of Linguistic Ability. New York: Thomas Y. Crowell. Blackburn, Simon (1984) Spreading the Word. Oxford: Oxford University Press. . (1996) “The Dispute on the Primacy of the Notion of Truth in the Philosophy of Language,” in Dascal, M., D. Gerhardus, et al. Sprachphilosophie, Vol. 2., pp. 1012-1024. Callaway, H. G. (1993) Context for Meaning and Analysis. Amsterdam and Atlanta: Rodopi. . (1979) “Reference, Variables, and the Empty Universe,” Logique et Analyse 85-86, pp. 85-98. . (1981) “Semantic Theory and Language: A Perspective,” Proceedings of the Southwestern Philosophical Association/Philosophical Topics, summer issue, pp. 61-70. . (1982) “Sense, Reference, and Purported Reference,” Logique et Analyse, 97, pp. 93-103. . (1985) “Meaning without Analyticity,” Logique et Analyse, 109, pp. 41-60.



. (1988) “Semantic Competence and Truth-conditional Semantics,” Erkenntnis, 28, pp. 3-27. . (1988) “Paul Gochet, Ascent to Truth,” Dialectica, Vol. 42, No. 1, pp. 45-58. . (1990) “J. A. Fodor, Psychosemantics,” Erkenntnis, 33, No. 2, September, pp. 251-59. . (1991) “W. V. Quine, Pursuit of Truth,” Dialectica, Vol. 45, No. 4, p. 317-322. . (1992) “Logic Acquisition, Usage, and Semantic Realism,” Erkenntnis 35, July, pp. 65-92. . (1992) “Does Language Determine Our Scientific Ideas?” Dialectica, 46, No. 3/4, pp. 225-242. . (1993) “Democracy, Value Inquiry, and Dewey’s Metaphysics,” Journal of Value Inquiry, Vol. 27, pp. 13-27. . (1993) “Open Transcendentalism and the Normative Character of Methodology,” Grazer Philosophische Studien 43, July, pp. 1-24; Alex Burri and Juerg Freudiger, eds. (1993) Relativism and Contextualism, Essays in Honor of Henri Lauener. Amsterdam: Rodopi, pp. 1-24. . (1993) “Intentionality Naturalized,” in Naturalism or Transcendentalism? Proceedings of the Bern Congress for Philosophy, Dialectica, 49, No. 2/4, pp. 147-168. . (1993) “Simon Evnine, Donald Davidson,” Philosophical Quarterly (Scotland), Vol. 43, October, pp. 555-560. . (1994) “Liberalism and the Moral Significance of Individualism: A Deweyan View,” Reason Papers, 19, Fall, pp. 13-29. . (1996) “Synonymy and Analyticity,” in Dascal, M., D. Gerhardus, et al. Sprachphilosophie, Ein internationales Handbuch zeitgenössischer Forschung, Vol. 2, pp. 1250-1262. . (1996) “No Need to Speak the Same Language?” (with J. van Brakel) Review of Bjorn T. Ramberg, Donald Davidson’s Philosophy of Language. Dialectica, Vol. 50, No. 1, pp. 63-71. . (1999) “Intelligence, Community, and Cartesian Doubt,” Humanism Today, Vol. 13, pp. 31-48. Chomsky, Noam (1977) Language and Responsibility. New York: Pantheon Books. . (1977) Essays on Form and Interpretation. New York and Amsterdam: North Holland.



. (1980) Rules and Representations. New York: Columbia University Press. . (1993) Language and Thought. Wakefield, CT and London: Moyer Bell.  . (1995) The Minimalist Program. Cambridge and London: MIT Press. . (2000) New Horizons in the Study of Language and Mind. Cambridge: Cambridge University Press. Christopherson, R. and H. W. Johnstone (1981) “Triadicity and Thirdness,” Transactions of the C. S. Peirce Society, Vol. 17, pp. 241-246. Clark, Andy (1988) “Critical Notice,” of Fodor, Psychosemantics. Mind, Vol. XCVII, pp. 605-17. Dascal, M., D. Gerhardus, et al. (1992, 1996) Sprachphilosophie, Ein internationales Handbuch zeitgenössischer Forschung, 2 Vols. Berlin and New York: Walter de Gruyter. Davidson, Donald (1980) Essays on Action and Events. Oxford: Clarendon. . (1984, 2001) Inquiries into Truth and Interpretation. Oxford: Clarendon. . (2001) Subjective, Intersubjective, Objective. Oxford: Oxford University Press. . and Jaakko Hintikka (1969) Words and Objections. Dordrecht and Boston: D. Reidel. . and Gilbert Harman (1972) Semantics of Natural Language. Dordrecht: D. Reidel. . Davidson, Donald (1963) “Actions Reasons and Causes,” Journal of Philosophy, 60. Reprinted in Davidson (1980) Essays on Action and Events, pp. 3-20. . (1967) “Truth and Meaning,” Synthèse, 17, pp. 304-323. Reprinted in Davidson (1984) Inquiries into Truth and Interpretation, pp. 17-36. . (1967) “The Logical Form of Action Sentences,” reprinted in Davidson (1980) Essays on Action and Events, pp. 105-122. . (1968) “On Saying That,” Synthèse 19, pp. 130-146. Reprinted in Davidson (1984) Inquiries into Truth and Interpretation, pp. 93-108. . (1970) “Semantics for Natural Language,” in Linguaggi nella Società e nella Tecnica. Milano: Edizioni di Comunità. Reprinted in Davidson (1984) Inquiries into Truth and Interpretation, pp. 55-64.



. (1970) “Mental Events,” In Foster, L. and J. W Swanson (1970) Experience and Theory. Amherst: University of Massachusetts Press, pp. 79-101. Reprinted in Davidson (1980) Essays on Action and Events, pp. 207-227. . (1973) “Radical Interpretation,” Dialectica, 27, pp. 313-328. Reprinted in Davidson (1984) Inquiries into Truth and Interpretation, pp. 125140. . (1974) “On the Very Idea of a Conceptual Scheme,” Proceedings and Addresses of the American Philosophical Association, 47. Reprinted in Davidson (1984) Inquiries into Truth and Interpretation, pp. 183-198. . (1974) “Belief and the Basis of Meaning,” Synthèse 27, pp. 309-23. Reprinted in Davidson (1984) Inquiries into Truth and Interpretation, pp. 141-54. . (1974) “Replies to Lewis and Quine,” Synthèse 27, pp. 345-49. Reprinted in Davidson (2001) Subjective, Intersubjective, Objective, pp. 281-285. . (1976) “Reply to Foster,” in Evans, Gareth and John McDowell eds. (1976) Truth and Meaning: Essays in Semantics, pp. 33-41. Reprinted in Davidson (1984) Inquiries into Truth and Interpretation, pp. 171180. . (1977) “The Method of Truth in Metaphysics,” in French, P.A. et al eds. Midwest Studies in Philosophy, No. 2, pp. 294-304. Reprinted in Davidson (1984) Inquiries into Truth and Interpretation, pp. 199-214. . (1979) “Quotation,” Theory and Decision, 11, pp. 27-40. Reprinted in Davidson (1984) Inquiries into Truth and Interpretation, pp. 79-92. . (1980) “Toward a Unified Theory of Meaning and Action,” Grazer Philosophische Studien, pp. 1-12. . (1982) “Paradoxes of Irrationality,” in Wollheim, Richard and James Hopkins eds. (1982) Philosophical Essays on Freud. Cambridge: Cambridge University Press, pp. 289-305. . (1983) “A Coherence Theory of Truth and Knowledge,” reprinted in Davidson (2001), Subjective, Intersubjective, Objective, pp. 137-153. . (1985) “Incoherence and Irrationality,” Dialectica 39, pp. 345-54. . (1986) “A Nice Derangement of Epitaphs,” in LePore, E. ed. Truth and Interpretation. Oxford: Basil Blackwell, pp. 433-446. . (1987) “Afterthoughts, 1987,” reprinted in Davidson (2001), Subjective, Intersubjective, Objective, pp. 154-157.



. (1988) “Epistemology and Truth,” reprinted in Davidson (2001) Subjective, Intersubjective, Objective, pp. 177-191. . (1989) “The Conditions of Thought,” Le Cahier du Collège International de Philosophie, Paris: Éditions Osiris. Reprinted in Brandl, J. and W. Gombocz eds. (1989) The Mind of Donald Davidson. Amsterdam: Rodopi, Amsterdam, pp. 193-200. . (1990) “The Structure and Content of Truth,” Journal of Philosophy, 87, pp. 279-326. . (1991) “Epistemology Externalized,” Dialectica, Vol. 45, No. 2-3, pp. 191-202. . (1993) “Dialektik and Dialog:” Rede von Donald Davidson anläßlich der Verleihung des Hegel-Preises. Frankfurt: Suhrkamp Verlag. . (1997) “Indeterminacy and Antirealism,” reprinted in Davidson (2001) Subjective, Intersubjective, Objective, pp. 69-84. Dewey, John (1920, 1948) Reconstruction in Philosophy. Boston: Beacon Press. . (1922) Human Nature and Conduct .New York: Modern Library; (1988) Reprinted in John Dewey, The Middle Works, Vol. 14. Carbondale and Edwardsville: Southern Illinois University Press. . (1927) Experience and Nature, second ed. New York: Dover. Reprinted in John Dewey, The Later Works, Vol. 1. Carbondale and Edwardsville: Southern Illinois University Press. . (1929) The Quest for Certainty. New York: Minton, Balch. Reprinted in John Dewey, The Later Works, Vol. 4. Carbondale and Edwardsville: Southern Illinois University Press. . (1934) A Common Faith. New Haven: Yale University Press. Reprinted in John Dewey, The Later Works, Vol. 9. Carbondale and Edwardsville: Southern Illinois University Press, pp. 3-58. . (1938) Logic, The Theory of Inquiry. New York: Irvington Publishers. Reprinted in John Dewey, The Later Works, Vol. 12. Carbondale and Edwardsville: Southern Illinois University Press. . (1939) Theory of Valuation. Chicago: University of Chicago Press. Reprinted in John Dewey, The Later Works, Vol. 13. Carbondale and Edwardsville: Southern Illinois University Press , pp. 189-252. . (1903) “Ralph Waldo Emerson,” International Journal of Ethics, 13, pp. 405-413. Republished in John Dewey, The Middle Works, Vol. 3.



Carbondale and Edwardsville: Southern Illinois University Press, pp. 184-192. . (1910) “The Influence of Darwinism on Philosophy,” in Dewey, John (1910) The Influence of Darwinism on Philosophy. New York: Henry Holt, pp. 1-19. Reprinted in John Dewey, The Middle Works, Vol. 4. Carbondale and Edwardsville: Southern Illinois University Press, pp. 3-14. . (1910) “The Experimentalist Theory of knowledge,” in The Influence of Darwin on Philosophy, pp. 77-111. Reprinted in John Dewey, The Middle Works, Vol. 3. Carbondale and Edwardsville: Southern Illinois University Press, pp. 107-127. . (1939) “Experience, Knowledge and Value,” in Schilpp, P.A. ed. (1939) The Philosophy of John Dewey. Evanston and Chicago: Northwestern University Press, pp. 517-608. . (1945) “Ethical Subject Matter and Language,” The Journal of Philosophy, Vol. XLII, No. 26, December 20, issue. Reprinted in Morgenbesser, Sidney. ed. (1977) Dewey and His Critics. p. 686. Dewey, John, Sidney Hook and E. Nagel. (1945) “Are Naturalist Materialists?” Journal of Philosophy, 42, pp. 515-530. Dummett, Michael (1977) Elements of Intuitionism. Oxford: Clarendon Press. . (1978) Truth and Other Enigmas. Cambridge: Harvard University Press. . (1974) “What is a Theory of Meaning?” Part I. In Guttenplan (1974); Part II in Evans and McDowell (1976). . (1986) “Some comments on Davidson and Hacking,” in E. LePore ed. (1986) Truth and Interpretation. Oxford: Basil Blackwell, pp. 459-476. Edwards, Paul ed. (1967) The Encyclopedia of Philosophy. New York: Macmillan and The Free Press. Evans, Gareth and John McDowell (1976) Truth and Meaning: Essays in Semantics. Oxford: Clarendon Press. Evnine, Simon (1991) Donald Davidson. Oxford: Oxford University Press & Polity Press. Fodor, J. D. (1977) Semantics, Theories of Meaning in Generative Grammar. New York: The Harvester Press. Fodor, Jerry (1979) The Language of Thought. Cambridge: Harvard University Press. . (1987) Psychosemantics. Cambridge and London: MIT Press.



. (1974) “Special Sciences,” Synthèse, Vol. 28, pp. 77-115. Reprinted as Chapter 5 of Fodor (1981) RePresentations. Cambridge: MIT Press, pp. 127-45. Follesdal, Dagfinn (1973) “Indeterminacy of Translation and Underdetermination of Theory,” Dialectica 27, pp. 289-301. Foster, J. A. (1976) “Meaning and Truth Theory,” in Evans, G. and J. McDowell eds. (1976) Truth and Meaning: Essays in Semantics, pp. 132. Frege, Gottlob (1953) The Foundations of Arithmetic. Translated by Austin, J. L., second revised Ed. Oxford: Basil Blackwell. . (1977) Logical Investigations. New Haven: Yale University Press. . (1980) “Über Sinn und Bedeutung.” (1892) Translated as “On Sense and Reference,” in Geach, Peter and Max Black eds. (third Ed.) Translations from the Philosophical Writings of Gottlob Frege. Oxford: Basil Blackwell, pp. 56-78. Fretlöh, Sigrid (1989) Relativismus versus Universalismus, Zur Kontroverse über Verstehen und Übersetzen in der angelsächsischen Sprachphilosophie: Winch, Wittgenstein, Quine. Aachen: Alano Verlag. Frisch, Max (1985) Stichworte. Frankfurt: Suhrkamp Taschenbuch Verlag. Fuller, Steven (1988) Social Epistemology. Bloomington and Indianapolis: Indiana University Press. Gadamer, H.-G. (1971) “Semantics and Hermeneutics,” in Philosophical Hermeneutics. Berkeley: University of California Press. Gibson, Roger (1982) The Philosophy of W. V. Quine, An Expository Essay. Tampa: University Presses of South Florida. Gochet, Paul (1986) Ascent to Truth, A Critical Examination of Quine’s Philosophy. Munich and Vienna: Philosophia Verlag. Gonseth, Ferdinand (1947) “L’Idée de Dialectique,” Dialectica, Vol. 1, No. 1, pp.21-37. Guttenplan, Samuel ed. (1975) Mind and Language. Oxford: Clarendon Press. Haack, Susan (1978) Philosophy of Logics. Cambridge: Cambridge University Press, Cambridge. Harman, Gilbert (1973) Thought. Princeton: Princeton University Press. . (1967-68) “Quine on Meaning and Existence,” I & II Review of Metaphysics, 21, pp. 124-151; 343-367.



Hausman, Carl (1993) C. S. Peirce’s Evolutionary Philosophy. Cambridge: Cambridge University Press. Hawking, Stephen (1993) Black Holes and Baby Universes. New York: Bantam Books. Hintikka, Jaakko (1969) “Semantics for Propositional Attitudes,” in Hintikka, J. (1969) Models for Modalities. Dordrecht: D. Reidel. . (1990) “Quine as a Member of the Tradition of the Universality of Language,” in Barret and Gibson, pp. 159-175. Jackendorf, Ray (2002) Foundations of Language. Oxford: Oxford University Press. James, Susan (1985) “Louis Althusser,” in Skinner, Quentin ed. (1985) The Return of Grand Theory in the Human Sciences, pp. 141-58. James, William (1903) “Address at the Emerson Centenary in Concord,” in James, William (1911) Memories and Studies. New York: Longmans, Green. Janich, Peter et al. (1974) Wissenschaftstheorie als Wissenschaftskritique. Frankfurt: Aspekte Verlag. Kant, I. (1781, 1787) Kritik der reinen Vernunft. Riga: Johann Friedrich Hatknoch; (1987) Critique of Pure Reason. London: Macmillan. . (1783) Prolegomena zu einer jeden künftigen Metaphysik, die als Wissenschaft wird auftreten können; (1902, 1985) Prolegomena to any Future Metaphysic. La Salle, IL: Open Court. Katz, J. J. and Jerry Fodor. (1963) “The Structure of Semantic Theory,” Language, Vol. 39. Reprinted in Katz and Fodor (1964) The Structure of Language: Readings in the Philosophy of Language. Englewood Cliffs, New Jersey: Prentice Hall. Kempson, Ruth (1977) Semantic Theory. Cambridge: Cambridge University Press. Koppelberg, Dirk (1987) Die Aufhebung der analytischen Philosophie. Frankfurt: Suhrkamp Verlag. Kuhn, Thomas. (1963) “The Function of Dogma in Scientific Research,” in Crombie, C.A. ed. (1963) Scientific Change, London: Heinemann, pp. 347-369. . (1990) “The Road since Structure,” P. S. A., 2, pp. 3-16. Lauener, Henri (1977) “Ferdinand Gonseth 1890-1975,” Dialectica, Vol. 31, No. 1-2, pp. 113-18.



. (1982) “Method in Philosophy and Logic,” Dialectica, Vol. 36, No. 4, pp. 353-71. . (1984) “Ansätze zu einer Zeitgenössischen Transzendentalphilosophie,” in Kant a due secoli della “Critica,” Brescia, pp. 113-144. . (1987) “Philosophie als normative Tätigkeit (offener Transzendentalismus versus Naturalismus),” Dialectica, Vol. 41, No. 1-2, pp. 23-38. . (1990) “Transzendentale Argumente pragmatisch relativiert: über fundamentale Optionen in der Philosophie.” Erkenntnis, 33, pp. 223249. . (1990) “Holism and Naturalized Epistemology Confronted with the Problem of Truth,” in Barrett and Gibson (1990), pp. 213-228. Lewis, Clarence Irving (1941) “Logical Positivism and Pragmatism,” in Goheen, J.D. and J. L. Mothershead eds. (1970) Collected Papers of Clarence Irving Lewis. Stanford: Stanford University Press, pp. 99112. . (1955) “Realism or Phenomenalism?” in Goheen, J.D. and J. L. Mothershead eds. (1970) Collected Papers of Clarence Irving Lewis. Stanford: Stanford University Press, pp. 335-347. Leiber, Justin (1975) Noam Chomsky, A Philosophic Overview. New York: St. Martin’s Press. Linsky, Leonard (1967) Referring. New York and London: Humanities Press. Lorenz, Kuno (1970) Elemente der Sprachkritik. Frankfurt: Suhrkamp Verlag. . (1980) “Sprachphilosophie,” in Althaus, H. P. ed. (1980) Lexikon der germanistischen Linguistik. Tübingen: Max Niemeyer Verlag, Vol. 4, pp. 1-28. Lorenzen, Paul (1984) Normative Logic and Ethics. Mannheim, Vienna and Zurich: Bibliographisches Institut. Luce, R.D. and H. Raiffa (1957) Games and Decisions. New York: Wiley. Lyons, J. (1995) Linguistic Semantics, An Introduction. Cambridge: Cambridge University Press. Morgenbesser, Sidney, ed. (1977) Dewey and His Critics. New York: The Journal of Philosophy, Inc. Mulvaney, R. J. and P. M. Zeltner eds. (1981) Pragmatism Its Sources and Prospects, University of South Carolina Press, Columbia SC.



Nozick, Robert (1986) “Experience, Theory and Language,” in Hahn L. and P. Schlipp eds. (1986) The Philosophy of W. V. Quine. La Salle, IL: Open Court, pp. 339-63. Palmer, F. R. (1976) Semantics A New Outline. Cambridge: Cambridge University Press. Peirce, Charles Sanders (1931-35, 1958) Collected Papers, 6 Vols. Cambridge: Harvard University Press. . (1956) “Synechism, Fallibilism, and Evolution,” in Buchler, Justus ed. (1940, 1956) The Philosophy of Peirce. London: Routledge, Keagan and Paul, pp. 354-360. . (1993) Semiotische Schriften. Kloesel, Christian and Helmut Pape eds., Vol. 3. Frankfurt: Suhrkamp Verlag. Pinker, Steven (1999) Words and Rules. London: Phoenix. Putnam, Hilary (1975) Philosophical Papers, Vol. 1, Mathematics, Matter and Method. Cambridge: Cambridge University Press. . (1975) Philosophical Papers, Vol. 2, Mind Language and Reality. Cambridge: Cambridge University Press. . (1978) Meaning and the Moral Sciences. London: Routledge and Kegan Paul. . (1962) “The Analytic and the Synthetic,” in Feigl, Herbert and Grover Maxwell eds. (1962) Minnesota Studies in the Philosophy of Science, III. Minneapolis: University of Minnesota Press. Reprinted in Putnam. (1975), Philosophical Papers, Vol. 2, pp. 33-69. . (1968) “The Logic of Quantum Mechanics,” in Cohn, R. and Max Wartofsky eds. (1968) Boston Studies in the Philosophy of Science 5. Dordrecht: D. Reidel. Reprinted in Putnam (1975), Mathematics, Matter and Method, Philosophical Papers, Vol. I, pp. 174-197. . (1973) “Explanation and Reference,” in Pearce, G. and P. Maynard eds. (1973) Conceptual Change. Dordrecht: D. Reidel. Reprinted in Putnam (1975) Philosophical Papers, Vol. 2, pp. 196-214. . (1975) “The Refutation of Conventionalism,” in Munitz, M. ed. (1975) Semantics and Meaning. New York: New York University Press. Reprinted in Putnam. (1975) Philosophical Papers, Vol. 2, pp. 153191. . (1975) “The Meaning of ‘Meaning’,” in Gunderson, K. ed. (1975) Language, Mind, and Knowledge, Minnesota Studies in the Philosophy of



Science, VII. Minneapolis: University of Minnesota Press. Reprinted in Putnam (1975) Philosophical Papers, Vol. 2, pp. 215-271. . (1976) “Realism and Reason,” Presidential Address to the Eastern Division of the American Philosophical Association. Reprinted in Putnam (1978) Meaning and the Moral Sciences. London: Routledge and Kegan Paul, pp. 123-138. Quine, W. V. (1953) From A Logical Point of View. Cambridge: Harvard University Press. . (1960) Word and Object. Cambridge: MIT Press. . (1966) The Ways of Paradox and Other Essays. Cambridge: Harvard University Press. . (1966) Selected Logic Papers. New York: Random House. . (1969) Ontological Relativity and Other Essays. New York: Columbia University Press. . (1970, 1986) Philosophy of Logic. Englewood Cliffs, New Jersey: Prentice-Hall. . (1973) The Roots of Reference. La Salle, IL: Open Court. . (1976) The Ways of Paradox, 2nd Ed. New York: Random House. . (1981) Theories and Things. Cambridge: Harvard University Press. . (1987) La Scienza E I Dati di Senso. Translated by Michele Leonelli. Rome: Armando. . (1990) Pursuit of Truth. Cambridge: Harvard University Press. . (1995) From Stimulus to Science. Cambridge: Harvard University Press. . (2003) Wissenschaft und Empfindung, die Immanuel-Kant-Lectures. Translated and introduced by H. G. Callaway. Stuttgart: FrommannHolzboog. . and J. S. Ullian. (1978) The Web of Belief, 2nd ed. New York: Random House. Quine, W. V. (1948) “On What There Is,” Reprinted in Quine (1953) From a Logical Point of View. pp. 1-19. . (1951) “Two Dogmas of Empiricism,” Philosophical Review, 60, pp. 20-43. Reprinted in Quine (1953) From A Logical Point of View, pp. 2046. . (1953) “The Problem of Meaning in Linguistics,” in Quine (1953) From a Logical Point of View, pp. 47-64.



. (1953) “Notes on the Theory of Reference,” in Quine (1953) From a Logical Point of View, pp. 130-138. . (1954) “Reduction to A Dyadic Predicate,” Journal of Symbolic Logic, Vol. 19. Reprinted in Quine (1966) Selected Logic Papers, pp. 224226. . (1960) “Carnap and Logical Truth,” Synthèse 12, pp. 350-374. Reprinted in Quine (1966) The Ways of Paradox and Other Essays, pp. 107-32. . (1969) “Existence and Quantification,” in Quine (1969) Ontological Relativity and Other Essays, pp. 91-113. . (1969) “Ontological Relativity,” in Quine (1969) Ontological Relativity and Other Essays, pp. 26-68. . (1969) “Natural Kinds,” in Quine (1969) Ontological Relativity and Other Essays, pp. 114-138. . (1972) “Methodological Reflections on Current Linguistic Theory,” in Davidson, D. and G. Harman (1972) Semantics of Natural Language, pp. 442-454. . (1974) “First General Discussion Session,” Synthèse 27, pp. 467-470. . (1977) “Review of Evans and McDowell,” Journal of Philosophy 74, pp. 225-41. . (1981) “On the Very Idea of a Third Dogma,” in Quine (1981) Theories and Things, pp. 38-42. . (1981) “The Pragmatists’ Place in Empiricism,” in Mulvaney, R. J. and P. M. Zeltner eds. Pragmatism Its Sources and Prospects. Columbia, S.C.: University of South Carolina Press, pp. 23-39. . (1981) “Things and their Place in Theories,” in Quine (1981) Theories and Things, pp. 1-23. . (1981) “Five Milestones of Empiricism,” in Quine (1981) Theories and Things, pp. 67-72. . (1981) “What Price Bivalence?” in Quine (1981) Theories and Things, pp. 31-37. . (1986) “Reply to Robert Nozick,” in Hahn, L. and P. Schilpp eds. The Philosophy of W. V. Quine. La Salle, IL: Open Court. . (1990) “Comments on Lauener,” in Barrett and Gibson (1990) Perspectives on Quine. p. 229.



Ramberg, Bjorn T. (1989) Donald Davidson’s Philosophy of Language. Oxford: Basil Blackwell. Resnick, Michael D. (1980) Frege and the Philosophy of Mathematics. Ithaca and London: Cornell University Press. Root, Michael and John Wallace (1982) “Meaning and Interpretation,” Notre Dame Journal of Formal Logic 23, pp. 157-73. Santayana, George (1900) “Emerson,” in Interpretations of Poetry and Religion . New York: Scribner’s. . (1953) The Life of Reason. New York: Scribners. Sapir, Edward (1949) Edward Sapir, Selected Writings in Language, Culture, and Personality. Mandelbaum, D. G. ed. Berkeley: University of California Press. Savan, David (1977) “Questions Concerning Certain Classifications Claimed for Signs,” Semiotica 19, No. 3/4, pp. 179-195. Scheffler, Israel (1967) Science and Subjectivity. New York: Bobbs-Merrill. . (1974) Four Pragmatists. London and New York: Routledge and Kegan Paul. Schlick, M. (1936) “Meaning and Verification,” Philosophical Review, 45. Excerpted in Lehrer and Lehrer eds. (1970) Theory of Meaning. Englewood Cliffs, N.J.: Prentice-Hall, pp. 98-112. Skinner, Quentin ed. (1985) The Return of Grand Theory in the Human Sciences. Cambridge: Cambridge University Press. Smith, Neil (1999) Chomsky Ideas and Ideals. Cambridge: Cambridge University Press. Stegmüller, Wolfgang (1987) Hauptströmungen der Gegenwartsphilosophie. Eight Ed., Vol. II. Stuttgart: Kröner. Stich, Stephen (1975) “Logical Form and Natural Language,” Philosophical Studies, 28, pp. 397-418. Stroud, Barry (1969) “Conventionalism and the Indeterminacy of Translation,” in Davidson, D. and J. Hintikka. (1969) Words and Objections, pp. 82-96. . (1990) “Quine’s Physicalism,” in Barrett and Gibson (1990) pp. 321333. Tarski, Alfred (1956) “The Concept of Truth in Formalized Languages,” in Tarski, (1956) Logic, Semantics, Metamathematics. Oxford: Oxford University Press, pp. 152-278.



Thiel, Christian (1965) Sense and Reference in Frege’s Logic. Dordrecht: D. Reidel. . (1980) “Geltung,” in Enzyklopädie Philosophie und Wissenschaftstheorie, Vol. I, Mannheim, Vienna and Zurich: Bibliographisches Institut, p. 729. . (1984) “Konstruktivismus,” in Enzyklopädie Philosophie und Wissenschaftstheorie, Vol. II. Mannheim, Vienna, and Zurich: Bibliographisches Institut, pp. 449-53. van Brakel, J. (1991) “Meaning, Prototypes, and the Future of Cognitive Science,” Minds and Machines 1, pp. 233-257. Walther, Elizabeth ed. (1991) Charles Sanders Peirce, Vorlesungen über Pragmatismus. Hamburg: Felix Meiner Verlag. Wittgenstein, Ludwig (1922) Tractatus Logico-Philosophicus. London: Routledge and Keagan Paul. . (1961) Philosophical Notebooks (1914-1918) Oxford: Oxford University Press.


a posteriori, 148. a priori, 32, 38, 44, 46-48, 56, 60, 144-145, 147-148, 161, 183, 188, 190, 195. Albert, Hans, 3n, 18n. Althusser, Louis, 6, 6n. anachronism, 80, 159. analysis, 4, 14, 24, 26, 49-50, 49n, 55, 61n, 66n, 74, 96, 155n, 165, 173, 176, 184-185, 187190. analytic, 25, 28, 31-35, 32n, 3847, 40n, 41n, 50, 59, 74, 78, 87, 99-100, 111, 117, 120, 138, 144, 147-148, 168, 193, 200. analyticity, 31-35, 38-39, 39n, 4345, 47-48, 104, 107, 147-148, 150. analytic philosophy, 193, 200. analytic-synthetic distinction, 3132, 35, 39, 41-42, 44, 46, 59, 87, 100, 147-148. anti-nominalism, 190. argument, 1, 22, 41, 43, 49-53, 55-56, 58-59, 70, 75-76, 80, 82, 85, 95, 108, 115, 120, 128129, 134, 136, 169, 161, 188189, 198. argumentation, 1-2, 5-6, 18-19, 109, 111, 120-124, 127-129, 153. Aristotle, Aristotelian, 61n, 61-63, 72, 82. artificial language, 25-26. assimilation, 130, 142. astronomy, 76, 82.

authority, 101, 130. Axelrod, Robert, 1, ln, 12-18, 13n, 15n, 16n. background, 74, 77, 133, 138, 155, 159, 170, 194, 199. Baltzer, Ulrich, 183-191, 183n, 184n, 185n, 186n, 187n, 189n, 191n. Barnes, Barry, 3n. Barret, R.B., 73, 73n, 77, 78n, 81n. Beardsley, Monroe, 21n. behaviorism, 32, 85, 89, 133, 139. behavioristic psychology, 89. belief, 1-2, 4-8, l ln, 18, 21, 2324, 26-30, 29n, 34, 34n, 39-40, 67, 70, 79, 86, 91, 94, 130, 135-137, 144-145, 155-161, 164, 166, 168-170, 176, 188, 194,200. belief-systems, 6, 11, 65, 67, 72, 85, 97-99, 124-127, 155-156, 166, 169. Bernstein, Richard, 148, 148n. bivalence, 106, 128n. Blackbum, Simon, 172n. Brake!, J. van, 163n, 176n. Brentano, Franz, 79. Buchler, Justus, 152, 156n. Camap, Rudolf, 80. Cartesian doubt, 159. causal, causation, cause, 11, 11n, 23-24, 46, 86-87, 89, 90, 92, 96, 100-103, 136-138, 148, 157, 167-169, 175.



Chomsky, Noam, 21, 21n, 23-24, 22n, 23n, 24n, 25n, 27, 29, 164, 164n. classes, 171. cognition, 2-5, 156. commitment, 2, 4, 21, 74, 82-83, 94, 153, 157, 193-194. common sense, 11, 66-68, 86-87, 91-92, 147, 153-157, 168, 198. communication, 1-2, 4-5, 6-7, 15, 18, 34n, 65, 79, 139, 143, 145146, 150, 158, 164-166, 172176. communicative practices, 1, 3, 18. computers, 14, 86, 137-138, 179. conceptual identity, 153-156. conceptual schemes, 9n, 103, 128131, 130n, 169,173. conditionals, 62, 65, 70, 77, 80. conservativism, 29, 32, 40, 44-45, 100, 106, 120, 150, 155-156, 160-161. constructivism, German, 109, 109n. context, 14, 19, 31, 45, 51-55, 65, 74, 78, 85, 92, 94-96, 99, 104, 113, 144, 154, 162, 164-165, 167-170, 172, 176, 181-182, 188, 191. context of knowledge, 14, 52, 9496, 99, 169. context principle, 61, 67. contextualism, 194-195. contextualization, 65, 75, 85, 9495, 167-168. continuity, 42, 72, 141-144, 146149, 151, 153-155, 160, 173, 183. convention, 43, 166, 178, 180. conventionalism, 32. cooperation, 1, 13, 15-19, 21-22,

146, 149. critical rationalism, 134. criticism, I, 18, 31, 33, 42, 48n, 71, 85, 95, 104, 130, 133, 146147, 171, 180-182, 194, 199200. cultural naturalism, 7. culture, 7, 9-10, 15, 87, 104, 106, 134, 143, 166. Davidson, Donald, 12-14, 34, 49, 65, 68-69, 71-76, 77-79, 81, 105, 115, 123, 135-143, 165172, 174-176, 178-179, 182183. defeasible conservatism, 44, 150, 155-156, 160-161. definition, definitions, 16, 27, 29, 33, 41-45, 59, 62-63, 66, 75, 100, 125-127, 142-144, 149150, 160, 169, 184, 187, 190. democracy, 181-182. Dennett, Daniel, 137-138. description, descriptions, 12, 27, 35-36, 38, 42, 47, 88, 121-122, 125, 166, 169, 179-180, 196197. Dewey, John, 7, 7n, 133, 142-155, 142n, 143n, 145n, 146n, 147n, 148n, 149n, 151n, 152n, 153n, 154n, 177-182, 177n, 180n, 193-194, 197-198, 198n. dialects, 15, 28-30, 35, 97-98, 99. dialogue, 107, 113, 128, 131-133, 139, 143, 177. dictionaries, 24, 126. disjunction, 9, 62, 113-115, 119, 121, 134. dogmatism, 22. dualism, 140, 137, 149. Dummett, Michael, 108, 108n,


172, 172n, 173. education, 4, 18, 99, 193, 199. Einstein, Albert, 44, 96, 136. Emerson, Ralph Waldo, 182-184. empirical, empirically, 8-10, 2122, 27-29, 39-40, 43-44, 4748, 50, 60, 64-65, 68-70, 7475, 77-82, 85, 88, 94, 97-98, 102-105, 107, 116, 126, 128130, 134, 144, 146-147, 150151, 153, 155-156, 160, 164, 166, 168-169, 171, 175, 190. empiricism, 5, 73, 84, 136, 154, 182. enlightenment, 7. epistemology, 52, 73-74, 83, 133134, 137-138, 151-152, 160, 171, 185. Erlangen school, 109, 109n. essentialism, 144, 165. ethics of discourse, 157. evidence, 2, 5, 9-11, 19, 65-68, 70, 73-75, 78, 82, 85, 89, 9798, 100-104, 105, 111-112, 113-116, 118-121, 123-124, 128-130, 134, 141, 143-144, 146-147, 149-150, 153, 155157, 160-161, 169, 194-195, 197-199. Evnine, Simon, 133-138, 159n. excluded middle, 110, 115-116, 118-120. existence, 3, 40, 75, 78, 89, 142, 152, 154, 160. experience, 4, 7, 7n, 9n, 40, 114, 115n, 116, 145, 145n, 148, 150-152, 151n, 152n, 156, 160-161, 178, 182, 198, 200. explanation, 27, 37n, 41, 51, 56, 58, 86-87, 94, 114, 118, 123,


153,160,186. extensional, extensionality, 33, 34n, 39, 42, 53, 79-80, 100104, 133, 167, 171, 173. facts, factual, factuality, 3, 13-14, 21, 23-24, 26-28, 30-31, 3739, 43, 45, 49-50, 58, 70-71, 74, 78, 81-82, 91, 94, 99, 101, 112, 117-120, 124, 127, 131, 137, 143, 149, 151, 153, 160, 164, 166, 168, 185, 196, 198. fact-value dichotomy, 149. fallibilism, fallibility, 28, 51, 75, 137, 144, 148, 151, 157, 186, 188, 190. feminism, 197-198. Feyerabend, Paul, 3, 130, 174. firstness (Peirce), 190-191. Fodor, Jerry, 36, 80, 85-92, 86n, 88n, 89n, 94-95, 100-104, 137. forms of life, 125-127, 170. Foster, John, 171. foundations, foundational ism, foundationalist, 31, 52, 52n, 74, 125, 155, 183-185, 188, 190,191,200. freedom, 4, 80, 139, 172. Frege, Gottlob, 34n, 36, 49-61, 49n,5ln,52n,60n,64,67, 70. Fretloh, Sigrid, I 07n. Frisch, Max, 1-2, 2n, 141-142. functions, 9, 16, 55, 58-59, 62, 67, 75, 80-81, 106-107, 110-113, 115, 118-120, 128, 130; proxy functions, 67, 75, 80-81, 129; verdict functions, l 06, 112113, 115. Gadamer, Hans-Georg, 138, 165166, 166n, 170, 172, 172n.



Galileo Galilei, 29, 82. game theory, 10-12, 17. Gedanken, 54. geometry, 23, 41, 47. Gibson, Roger F., 73, 73n, 77, 78n, 81n. Gochet, Paul, 89n, 106, 106n, 129n, 130n. grammar, 21, 23, 25-26, 33, 100101, 169, 175. Haack, Susan, 193-201. Hacking, Ian, 172n. Harman, Gilbert, 46, 46n. Hausman, Carl, 191n. Hawking, Stephen, 145n. Hegel, G. W. F., 138, 148. Hickman, Larry, 177-178. Hiz, Henry, 21n. holism, 45, 85-104, 133, 146-147, 149, 153-157. holistic, 85, 102, 126, 137, 146, 176. Hook, Sidney, 179, 198. Hookway, Christopher, 133. Hume, David, 41. hypotheses, 4-6, 19, 23-24, 36, 39-40, 69, 77-78, 80, 89-90, 95, 111-114, 117-120, 129130, 134, 147-148, 153, 181, 194. idealism, 133-134, 138, 181. identity, 13, 36, 45-46, 49-52, 55, 57-58, 60-65, 70, 80, 87-88, 91-92, 97, 105, 108, 144, 161, 177, 191. identity conditions, 35, 54-55, 59, 60-61, 87, 90, 102, 104. ideology, 2-3, 23, 145. idiolect, 100, 102, 179.

incommensurability, 135, 163, 165, 173-176. indeterminacy of translation, 40, 74, 76-77, 108-109, 116, 124, 140. indirect sense, 59. information, 7, 9, 14-16, 19, 22, 35-36, 38, 108, 127, 138. innate ideas, 23. innate quality spacing, 78, 137. inquiry, 3-5, 79, 88, 145-146, 148, 150-154, 160-164, 167-168, 171, 173, 181-183, 189, 191, 193. inscrutability of reference, 37-38, 64-67, 69, 75, 77. instrumental ism, 138, 141-161, 152-153. intension, 73,79. intensional, intensionality, 52-54, 56-57,60,79, 133,167. intention, 15, 62, 79-80, 100, 180. intentional, intentionality, 27, 54, 75, 79-83, 85, 87, 89, 91, 9495, 100, 102, 135, 137, 141161, 149, 151, 155, 157, 161. interpretation 2, 7-8, 14-16, 2528, 62, 65, 69-70, 72-78, 8081, 85-86, 92, 97, 101, 103107, 115-116, 118, 122, 128129, 134-135, 137, 139-141, 136-142, 151-152, 154, 156, 159, 163, 165, 167-169, 167169, 171, 174-177, 179-183, 185, 188, 193. intuition, speaker's, 22, 27. intuitionistic logic, 9, 105-108, 110-116, 118-119, 123,125, 128-129. James, Susan, 6n.


James, William, 135, 144, 178, 180-181, 180n, 193. Janich, Peter, 3n. Johnstone, H.W., 189, 189n. justification, 3, 94, 109-110, 123124, 127, 136, 142-143, 149, 172, 174, 195, 200. Kant, Immanuel, 50, 73, 148, 183184, 187-188. Katz, J.J., 22n, 23-24, 36, 36n. Kempe, A.B., 192, 193. Kempson, Ruth, 22n, 24, 24n, 25, 26, 26n. Kloesel, Christian, 185, 191n. knowledge, I, 2, 4, 7, 11, 21, 23, 28, 31, 32, 34n, 38, 41, 46-47, 50-51, 56-57, 64, 73, 86, 9294, 96-97, 126, 134-136, 141, 145, 145n, 147, 154-155, 152n, 159-160, 165, 168, 175, 179-181, 183-185, 187-188, 191, 193-194, 197, 199-200. Kripke, Saul, 48n. Kuhn, Thomas, 3-4, 3n, 4n, 130, 174-175. language, 1-4, 6-12, 9n, 15, 1819, 21-30, 21n, 22n, 23n, 24n, 31 n, 32-33, 34n, 35, 44-45, 49, 51, 54, 61-72, 74, 78-81, 83, 85-86, 88n, 90, 97-99, 102n, 101-103, 105, 108-111, 109n, 114, 115n, 116-121, 123-125, 127, 128n, 130, 133-136, 141, 143n, 147, 149, 150n, 151153, 156-158, 163-169, 163n, 164n, 171-176, I 94; object language, 63-64, 67-70, 72, 74, 85, 101-102, 104, 130,


149, 168; meta-language, 6364, 66, 68-69, 167. language acquisition, 6, 10, 12, 79, 105, 117, 123-124, 147. language-games, 110. Lauener, Heni, 142, 149n. law, law-like, laws, 2, 7-9, 11, 4046, 47, 52, 54-55, 79, 88-90, 100,110, 115-116, 118-120, 126, 137, 145-146, 183, 186, 197. learning, language learning, 14, 27, 79-80, 106-111, 114-116, 116-121, 122-124, 126, 128, 174-175. Legg, Cathy, 190. Leibniz' law, 52, 54-55. Leonelli, Michele, 73, 73n. Lewis, C. I., 147-148, 157. lexical analysis, 24. linguistic competence, 11, 34, 34n, 36, 120, 165, 169. linguistics, 3, 9, 21, 23, 26-29, 39, 79, 82, 105, 164. Linsky, Leonard, 57n. literature, 1, 169, 184-185, 191, 197. logic, 11-13, 23, 51, 87, 107-117, 119-122, 124-132, 134-135, 138-139, 141-142, 150, 152, 161-162, 172, 182, 188, 191193. logical analysis, 74. logical theory, 61, 65, 126, 129, 130, 138. logical truth, 23, 61-62, 87, 131, 134, 172. Lorenz, Kuno, 109, 110n, 125126, 125n, 126n. Lorenzen, Paul, 110, 110n, 123.



mathematics, 32, 45, 111, 134135, 138, 147. Mead, George Herbert, 193. meaning, 3, 5-6, 11, l ln, 22-24, 22n, 28-29, 29n, 31-48, 3 ln, 34n, 39n, 47n, 49-57, 49n, 5961, 61n, 63-65, 66n, 67, 70-78, 80-83, 85-87, 90, 90n, 92-105, 107-108, 111, 111n, 116-118, 120-124, 126-127, 141, 143144, 147, 149-151, 153, 155159, 155n, 163-167, 171-172, 172n, 176. meaning holism, 85, 96- 98, 103, 108. meaning, theory of, 3, l ln, 23, 31-34, 37, 48, 54-55, 61, 64, 86-87, 94, 100-101, 105, 107108, 111, 123, 164, 176. Mentalese (Fodor), 100-10 I. mental states, 89-90, l 00, 135137. meta-language, 63-64, 66, 68-69, 167. metaphysics, 3-5, 23, 145, 155156. methodology, 29, 3 ln, 44, 105, 107, 110, 127, 130, 147, 147n, 153, 157n. mind, 56, 85-88, 91, 138, 154, 162, 180. minimum mutilation, 103. modes of presentation, 61 . multiculturalism, 195. names, naming, 25, 34n, 49-51, 55-59,62-64,67,71, 153,160, 167, 178, 184. natural language, 3, 22, 22n, 27, 28-29, 69, 117, 119, 133-134, 163-164.

natural science, 80, 88, 125, 150, 159. naturalism, 7, 141-143, 146-149, 154. nature, 18, 42, 44, 82, 90, 96, 108, 124, 138, 143-145, 148-149, 151, 154, 157-162, 166, 182. necessary truth, 31-32, 52, 172. necessity, 24, 52, 84, 103. negation, 12, 62, 109, 120-121. Newton, Sir Isaac, 39, 41-42, 94, 126, 160. non-contradiction, 115. normative, 141, 143-145, 149150, 152, 154-155, 157, 165, 170. notation, 110, 192. Nozick, Robert 12-13, 120-123, 125, 131. object language, 63-64, 67-70, 72, 74, 85, 101-102, 104, 130, 149, 168. observation, observational, 3-5, 9, 26, 40, 74-78, 80-82, 97, 99, 102-103, 106-111, 112-121, 128-130, 144-145, 147, 150152, 155-156, 160-161. observationality, 76, 112-121. observation sentences, 9, 74-77, 80-82, 121, 129-130, 144, 155-156, 161. ontological commitments, 74, 83, 94. ontological relativity, 74, 76, 8485, 94. ontology, 59, 74-75, 80-83, 86-87, 93, 101, 150, 153, 159, 160. ordinary language, 107, 140. Palmer, Frank R., 22, 22n, 28,


28n. Pape, Helmut, 185, 191n. paradox, 6, 10, 41, 81, 81n, 83, 166, 198. paraphrase, 33, 65-66, 73-75, 77, 87. participation, 5-6, 18-19, 106, 110-111, 120, 123-124, 145, 181. Peirce, Charles Sanders, 113, 139, 143, 146-147, 150, 153, 161, 164, 168, 185-189, 190-194. perception, 3, 79, 147, 187. philosophy of language, 86, 113, 160, 165. physicalism, 79, 81, 88-89. physics, 2, 23, 39, 41, 87-89. pluralism, 143, 146. Popper, Karl, 7, 74, 134. positivism, positivists, 3, 133, 148, 152. possibility, 1, 3, 10, 13, 18, 34, 39, 41-42, 64, 71, 73, 76-77, 86, 88-90, 93, 126, 141-142, 147-149, 151, 154, 156, 163164, 181-182. practice, practices, 1-4, 18, 45-46, 109-111, 120, 122-127, 139, 141, 143-145, 154, 156, 160, 166, 169, 172, 176, 195, 199. pragmatic, pragmatics, 6, 30, I 08110, 109n, 123, 125, 141, 143144, 149, 170, 172, 197, 199200. pragmatism, 109, 133-139, 143, 147n, 148, 149-150, 184, 184n, 193-194, 196n, 197-198. predicate logic, 188, 191, 193. predication, 77, 192. prediction, 74, 82-83, 141, 148, 158, 166, 170.


principle of charity, 13, 141-142, 167-168, 174-175, 182. prisoner's dilemma, I 0-14, 15. properties, 23-24, 42, 86, I 00104, 151, 171, 191. propositional attitudes, 52, 79, 86. propositions, 54, 144. proxy functions, 67, 75, 80-81, 129. psychology, 32, 74, 86, 88-90, 93, 97, 178. Putnam, Hilary, 31-48, 31n, 32n, 34n, 47n, 89-90, 90n, 93, 171. quantification, quantificational, 10, 62, 136, 189n. quantifiers, 81, I 06, 121n. Quine, W.V., 5, 9, 10, 21, 24, 27, 31-32, 37, 39-40, 39n, 46-47, 49n,51,55,66-67,66n, 73-83, 86-87, 89, 92-93, 100, 104118, 120-122, 126, 128-130, 133-134, 134n, 136, 139, 147, 149, 155-158, 161, 167, 174, 189, 189n. Quinean, 22, 32, 52, 70, 74, 7677, 85-88, 96, llO, 114, 122, 126, 128-129, 136, 140, 163. Ramberg, Bjorn T., 163-176. radical interpretation, 10, 1On, 150, 157-159, 164-166, 169173, 175-176. rational, rationality, 4, 43, 133135, 137-138, 138-139, 146, 156, 169, 170, 174, 193. realism, 5n, 10n, 46, 47n, 48n, 65, 65n, 71, 75, 79, 81, 85, 87, 89, 94, 105, 126, 128, 135, 152153, 157n, 161. reconstruction, 59, 60-61, 70, 141,



142, 145, 150-151, 157, 160, 163-164, 185, 187, 190, 194. reference, 8-10, 9n, 23, 25n, 29n, 33-34, 34n, 36-38, 37n, 49-57, 49n, 51n, 56n, 58n, 59-67, 59n,60n,69-71,73-75,77,78, 80-83, 90, 92-94, 94n, 98-99, 101, 106-110, 106n, 110n, 112-113, 112n, 116, 127-130, 136-137, 144, 147, 151-156, 159, 167-169, 173, 178-179, 185, 188-189. reference, theory of, 61, 167-168. referential semantics, 34n, 59-61, 81, 101-102, 157. relations, relationship, 2, 6, 10, 14, 18, 22, 24, 27, 29-31, 4142, 46, 49-50, 54, 62, 64-66, 70-71, 80, 83, 85-90, 92-94, 97, 100-103, 105, 110-111, 116, 120-123, 125, 127, 129, 133, 141-144, 146-151, 153156, 159, 161, 163, 168, 170171, 178, 180-181, 183-190, 195. relative empiricism, 147. relativity, 37n, 41, 66-67, 66n, 67n, 75, 80-81, 92,117, 117n, 194-195. Resnick, Michael, 60n. Rorty, Richard, 135, 193-194, 196-197, 196n, 200. Russell, Bertrand, 61n, 144, 193. Ryle, Gilbert, 136-137. Saatkamp, Herman J., 196n. Santayana, George I 78-182, 180n. Sapir, Edward, 174. satisfaction, 33, 65-66, 68, 70-71, 91, 164, 167-169.

Savan, David, 183, 183n. scholasticism, 18. Scheffler, Israel, 148, 148n, 149n. science, 1-4, In, 3n, 7-8, 8n, 1819, 22-23, 27, 31n, 42, 47n, 74, 79-83, 85-89, 88n, 93, 100, 109n, 123,125,139,142, 144145, 147, 149-156, 159-161, 168-169, 179, 183, 190, 197, 199-200. scientific belief, I, 145. scientific idealizations, 8, 21, 24, 46, 67, 85, 87, 95-98, 104, 116, 151. Seigfried, Charlene, 198. semantic competence, 11-12, 15, 21, 23, 64n, 99n, 123-128, 166n, 171. semantic meta-language, 167. semantics, 5-6, 21, 22n, 25-26, 28-31, 28n, 33, 34n, 35, 46, 58-59, 60-61, 64n, 71-72, 7982, 85-86, 86n, 89, 91-94, 100-102, 107, 109-110, I 16117, 124-125, 130, 133-134, 139, 153, 156, 159, 163-164, 166-169, 173, 175-176. sense-data, 135. sense-reference distinction, 52-56, 57, 59-60. sensory stimulation, 73, 75-76, 81, 118. similarity, 78, 121, 137, 176. simplicity, 15, 29, 99, 156. skepticism, 21-22, 87, 93, 125126. Skinner, Quentin, 3n, 6n. sociological determinism, 6, 19. sociology of belief, 2, 4, 5-8, 18. special sciences, 8, 85, 87-89, 100, 147, 161.



speech community, 27, 29, 76, 97-98, 105-106, 121. Stalinism, 6. stereotypes, 15-16, 18, 34n, 3540, 42, 45, 47. Stich, Stephen, 21 n, 22, 22n, 196. stimulus meaning, 73-76, 78, 106, I 16-118, 120-123, 128-129, 147, 155, 158, 161. stipulation, 27, 43-44, 50, 92-93. stipulative, 50, 99, 127. Strawson, Peter F., 71. Stroud, Barry, 81, I 07n. substitutivity, 53-54, 57, 60. synonymy, 24, 38-39, 42, 44, 75, 120,171. syntax, 6, 21, 30, 64, 86, 109, 109n, 117,121,159. synthetic, 31-32, 35, 38-42, 44, 46, 59, 74, 87, 100, 147-148, 161. systemization, 7, 13, 157. Tarski, Alfred, 34, 49, 61, 66-69, 72, 77, 84, 135. Tarski an semantics, 170, 172-173. teridentity (Peirce), 189. theoretical-empirical, 126, I 44, 168. theoretical entities, 140. theory-laden, 4, 6, 27, 76, 107. theory ofreference, 61, 167-168. Thiel, Christian, 64, 112, 135. tolerance, 2, 7, 24. translation, 6-7, 11, 13, 15, 42, 68-69, 73-76, 81, 84-85, 88, 106-107, 109, 114-116, 118, 120, 122, 125, 129, 138-142, 176, 179-182. truth, truthfulness, 10, 22-28, 3134, 38-41, 43, 46-47, 47n, 52,

54-55, 58-59, 61-69, 70-73, 80-83, 88, 91, 99, 100, 103, 106-107, 112, 117-119, 123, 125-126, 133, 135-135, 144145, 152-153, 156-157, 159, 161, 164-166, 167-169, 170, 171-172, 172n, 174-176, 187, 194, 196,197,200. truth conditions, 6, 11, 62-63, 64n,65-67,69,91, 168. truth functions, I 06, 115. truth-functional logic, I 06-107, 110, 118, 120. truth theories, 64, 68, 70, I 03, 171. truth values, 55, 72. T-sentences, 67-69, 71-72, 74-77, 172,174. Twin Earth, 89, 92. underdetermination, 39, 76, 171. understanding, 1, 5, 7-8, 9, 11, 32, 49, 51, 74, 78-80, 99, 106, 110, 118, 124, 130, 143, 145146, 155, 158-159, 161, 164165, 169-171, 181, 185, 186. usage, use, 1, 5-6, 5n, 10n, 11, 22, 26, 35-36, 45, 54, 65, 67-69, 72, 80, 97-99, 103, 105, 107112, 121-124, 126-128, 151, 171. v~ue~ 2, 4, 7, 13, 55, 62, 72, 9697, 139, 141-142, 156. values of the variables, 167. variables, 25n, 167, 189. verdict functions, 106, 112-113, 115. Walter, Elisabeth, 184, 184n. warranted assertion, 194.



Wittgenstein, Ludwig, 45, 109, 123, 125-126, 136-137. words, 6, 10-11, 15, 17, 22, 24, 33-39, 42-43, 45-47, 50-52, 54, 57, 60, 65-68, 70, 76-79, 92-94, 100, 102, 106, 111-115,

117-118, 122, 127-128, 130, 144, 151, 156-157, 160, 168-169, 171, 176, 178, 190, 194-195. word meanings, 35-36, 39, 47. Worf, Benjamin L, 3, 130.