USF Libraries
USF Digital Collections

Truth and judgment


Material Information

Truth and judgment
Physical Description:
Kelly, Jeremy J
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:


Subjects / Keywords:
Natural language semantics
Philosophy of language
Dissertations, Academic -- Philosophy -- Doctoral -- USF   ( lcsh )
non-fiction   ( marcgt )


ABSTRACT: I examine the difficulties that several philosophers of language are liable to encounter in their attempts to provide an account of the connection between truth and assertion. I then attempt to provide an account of this connection. The analysis is concerned chiefly with difficulties which consist in elucidating the conceptual connection between truth and assertion in a way that respects certain linguistic intuitions while at the same time rendering the concept of truth amenable to a semantic interpretation. The proposed view suggests one way in which we might go about meeting the theoretical demands implicit in addressing this concern, among others, demonstrating the extent to which a theory of truth should be regarded as belonging to the province of epistemology. Insofar as semantical considerations figure into such a theory, a more systematic investigation of the interface between epistemology and natural language semantics is recommended. The solution to many problems at this interface, I argue, lay in an analysis of judgment.
Dissertation (Ph.D.)--University of South Florida, 2009.
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Jeremy J. Kelly.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 146 pages.
General Note:
Includes vita.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002029677
oclc - 437049556
usfldc doi - E14-SFE0002956
usfldc handle - e14.2956
System ID:

This item is only available as the following downloads:

Full Text


Truth And Judgment by Jeremy J. Kelly A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Philosophy College of Arts and Sciences University of South Florida Co-Major Professor: Kwasi Wiredu, Ph.D. Co-Major Professor: Stephen Turner, Ph.D. Alex Levine, Ph.D. Richard Manning, Ph.D. Date of Approval: April 26, 2009 Keywords: assertion, propositions, natural la nguage semantics, philosophy of language, epistemology Copyright 2009 Jeremy J. Kelly


i Table of Contents Introduction 1 Chapter I. Frege on Functions 6 Chapter II. The Relation of the Theory of Functions to the Theory of Assertion 16 Chapter III. Frege and Russell on Assertion 40 Chapter IV. Theories of Truth 58 Russell on the Nature of Truth and Falsehood 58 Dewey on the Nature of Truth and Ju dgment; the Dewey-Frege connection 61 Chapter V. Austin and Strawson and the Vagaries of Correspondence 71 Chapter VI. Judgment and Propositions: Some Raditional Problems Revisited 99 Chapter VII. A Semantic Treatment of Relative Truth 112 Chapter VIII. Moore’s Paradox and the Logic of Assertion 127 References Cited 138


ii List of Figures Figure 1. Components of a decl arative sentence 32 Figure 2. Semantic content 34 Figure 3. Relation of semantic content to truth 36 Figure 4. Two levels of truth 37


iii Truth and Judgment Jeremy J. Kelly ABSTRACT I examine the difficulties that several philosophe rs of language are liable to encounter in their attempts to provide an account of the connection between truth and assertion. I then attempt to provide an account of this connect ion. The analysis is concerned chiefly with difficulties which consist in elucidating th e conceptual connecti on between truth and assertion in a way that respects certain lingu istic intuitions while at the same time rendering the concept of truth amenable to a semantic interpretation. The proposed view suggests one way in which we might go about meeting the theoreti cal demands implicit in addressing this concern, among others, demo nstrating the extent to which a theory of truth should be regarded as belonging to the province of epistemology. Insofar as semantical considerations figure into such a theory, a more systematic investigation of the interface between epistemology and natu ral language semantics is recommended. The solution to many problems at this interface, I ar gue, lay in an analysis of judgment.


1 Preface In what follows, I argue for an epistemic vi ew of truth that involves an analysis of the conceptual relations betw een truth and judgment. On th is analysis I propose a view that might be taken as a basis for a theory of logical form – a basis from which the problem of propositional unity may be trea ted among other kindred problems. The view is also intended to provide an epistemo logical basis from which we may address a concern which sometimes takes the form of an objection to Tarski -style disquotational theories of truth. The concern is that these theo ries do not tell us what it is that definitions of truth for interpreted languages have in common. The upshot of the analysis is the suggestion that our ordinary use of the concept of truth is an act that is explainable under a theory of truth which has something to say about the point of that use. The proposed view suggests one way in which we might go about meeting the theoretical demands implicit in addressing th is concern, demonstrating the extent to which a theory of truth is in a broad sense, an epistemological enterprise. I attempt to show how this is so in the final chap ter of the dissertation, where I offer an epistemological treatment of some problems that are customarily understood to be the province of philosophy of language. The claim that a theory of truth should be conceived more as an epistemological undertaking than otherwise is a point that may be admitted without prejudice to the contention that solu tions to many of the traditional problems surrounding the problem of truth depend crucia lly upon considerations having to do with


2 logical grammar. In view of these considera tions I have set out to examine several of those traditional problems that are now cen tral to many recent studies in the philosophy of language and natural language semantics. These problems concern propositional unity, predication, several issues rais ed by the notion of truth-makers, and the problem of what has come recently to be known as ‘Moore’s paradox’. It is customary to construe the bounda ries between epistemology and semantics as being sharply distinct, a tendency among so me analytic philosophers that has perhaps had more undesirable consequences than w ould be the case were the interface between these two fields more carefully and systemati cally explored. The bri dge that links these two domains of philosophical inquiry is one that should aim at a kind of conceptual elucidation, the business of whic h, so I argue, it is the business of a theory of assertion to provide. The aim of the first chapter is chiefly di agnostic. I examine the difficulties that philosophers of language are liable to encounter in their attempts to provide an account of the connection between truth and assertion. Wher e there is work bei ng done to provide an account of this connection, the difficulties consist in elucidating the conceptual connection between truth and assertion in a way that respects certain linguistic intuitions while at the same time rendering the con cept of truth amenable to a semantic interpretation. Such difficulties, I argue, have their origins ve ry early in developments in analysis and the philosophy of language, and in the work of Gottlob Frege, in particular. The second and third chapters are concerned with certain semantical considerations in Frege’s work. In this discu ssion I aim to provide an analysis of Frege’s theory of functions, and in particular Frege’ s treatment of identity in terms of functions


3 and values. I argue that some standard ways of treating sent ences have, in fact, emerged from Frege’s function-theoretic analysis a nd have had far-reaching implications with respect to the way in which the nature of judgment has come to be understood within the mainstream of philosophy of language. From the point of view of judgment, th e manner in which sentences are treated in Frege’s semantics has had, in turn, further und esirable consequences, in particular his identification of the semantic role of indica tive sentences with that of names and referring singular terms. The second and third chapters are in the main an endeavor to trace these consequences from Frege’s later semantical vi ews back to his theory of functions – i.e. what I regard as the source of the problem. I there examine two widely accepted inte rpretations of Frege’ s views on assertion as they arise out of his later semantical vi ews and theory of functions. After showing how these accounts are connected with Frege’s theory of functions, I then attempt to show that the locus of what is problematic in thes e accounts is a theoretical consequence of construing sentences as names. Because of th is fact, I suggest, th e interpretations of Frege’s theory of assertion ha ve supported what I take to be radically non-epistemic conceptions of truth. Such non-epistemic con ceptions have won many adherents over the course of a century thanks to Frege’s influence and have come into especial prominence since Tarski’s article “The sema ntic conception of truth” in 1944. Further developments of Frege’s ideas, as they appear in Russell’s early philosophy (1903), and later as they appear in the work of Alonzo Church (1951), have tended to support the two wide ly accepted interpretations of assertion. Russell’s own influential view of assertion is then examin ed, first in its relation to Frege’s view, and


4 then later in relation to the views of John Dewey (1938). Th e concept of truth is primarily epistemic, on Dewey’s view. In his Logic he argues for in his Logic a more useful distinction between sentences and propositions on the basis of which he is then able to maintain the connection between truth and i nquiry – twin notions that become divorced in the somewhat platonistically orie nted views of Frege and Russell. The fourth and fifth chapters are in larg e part a discussion of some traditional theories of truth. I begin with the reception of Frege’s theories of judgment and reference via Russell’s early philosophy ( 1903). There I argue that via Russell’s interpretation of Frege and Frege’s peculiar brand of realism may be seen to e ffectively rehabilitate the old problem of propositional unity. In order to show this, I atte mpt first an examination of Russellian propositions and then discuss Aus tin’s “purified” corre spondence theory of truth in relation to Strawson’s ‘performativ e-redundancy’ view. I then argue that the difficulties attending to traditi onal accounts of facts, specifically to th eir role as truthmakers, derive from those aspects of Frege’s realism which were investigated in the first chapter and which were there taken to be a consequence of his theory of judgment. The difficulties derive also from Russell’s early ‘a bsolute realism’, for different, but closely related, reasons. In chapters six and seven a distinction between primary and secondary concept of truth-value is introduced in connection with Strawson’s “performative-redundancy” view. I introduce on the basis of this distinction the concept of point of view which, an epistemic notion of truth argued in Wiredu ( 1975) This concept relativizes primary and secondary occurrences of truth-value to assert ions in a manner that avoids some of the pitfalls commonly associated with truth-re lativism. It is further argued that the


5 employment of the concept of point of view affords a solution to ‘Moore’s paradox’. The significance of the notion, and its application in treating th is so-called paradox, lies less in the results of the application and more in its general implications for the theory of truth. I suggest that what is philosophically signif icant about ‘Moore’s paradox’ is that its solution confronts us with those implications in a way that disclo ses the necessity of approaching the theory of truth as an analysis of the way truth and assertion are related.


6 Chapter I: Frege’s Theory of Functions Among the first expressions that Frege us es in ‘Function and Concept’ to name a mathematical function, (1) ‘2x + x’ is presen ted as an example of what “people who use the word ‘function’ ordinarily have in mind.” (Frege 1891, 24).What distinguishes (1) from other expressions which do not name functions is, inter alia that (1) “ indefinitely indicates a number.” As such, (1) is incomplete It is this incompleteness, which we have yet to adequately characterize, that is the essential property of a ll functions. What is instructive about the appearance of mathematical expressions such as (1) is that they reveal the manner in which certain linguist ic expressions possessing similar syntactic form exhibit a similar functionality. It is s uggested in a careful study by Rulon Wells that the capacity of mathematical e xpressions to exhibit functionality in just this way is in large measure a consequence of the “ method of generalization ,” (henceforth MG) one among several commitments which underwrite the programmatic aims of early Logicism. According to Wells, MG was a technique that had allowed Frege to explain concepts and relations as a species of function, a technique without which Frege would not have arrived at a functional de finition of concepts. It is interesting to note that the notions of function a nd concept are not (typically) discussed in connection with the more familia r stable of semantic notions that recent philosophers of language have inherited from Frege. This much was recognized as early as 1951. As R. S. Wells observes, “in spite of the current interest in Frege, the basic distinction between Function and Object, hi s basic distinction, has received scant


7 attention, the reason being that it is not so directly releva nt as the notion of sense, denotation, sense, proposition, and truth-valu e to the semantic problems with which discussions of Frege have been most concerned.”1 There are two general points about MG that are worth making: historically, it may be seen as an example of “cultural lag”.2 With respect to semantics, the method turned out to be, in Wells’ phrase, “too productive,” since it yielde d “some products one would sooner not have had.” Wells does not mention Frege’s treatment of identity in connection with this – as Wells otherw ise puts it – “uncontrolled met hod of discovery.” It is my contention that the problem of identity is for Frege is a consequence of this “uncontrolled” application of MG The identity problem is in some sense independent of the complications that the MG reveals with respect to his th eory of sense and the problem of cognitive significance. I argue, however, th at the latter complications stem largely from the former. The specific treatment th at identity receives in “On Sense and Reference” is the principal example of the MG principle yielding those sorts of results one would sooner not have had. Wells explains MG as follows: [Before] Frege it would have b een said that some but not all things that have a meaning additional to their sense have a denotation. Names (noun phr ases) have denotation but not truth; propositions have truth but not denotation. By generalizing the concept of de notation to subsume truth as a variety, Frege makes exceptionless, and therefore completely regular, the propos ition “All objective meaning other than sense is de notation” (Wells 1951, 395) An instance of this is seen (on p.31) wh ere Frege analyzes the sentence ‘Caesar conquered Gaul’ into the two pa rts, ‘Caesar’ and ‘conquered Ga ul’, where the latter is


8 said to express a one-place f unction and the former the argum ent for the function. In the passage that immediately follows, we fi nd the following linguistic analogue: ‘____ conquered Gaul’ It is said that a “complete sense” is obtain ed when the empty place is filled up with a proper name. Frege then goes on to introduce other kinds of mathematical expression, also said to name functions but differing fr om other designating expressions (explain) insofar as they contain the signs for equality, inequality, etc. (‘=’, ‘>,’ and ‘<’). As an instance of Frege’s MG all of this perhaps familiar, if not uncontroversial. What follows these examples is a passage in which Frege reflects on the ways in which the meaning of the word ‘function’ has been “stretched by the progress of science,” and the ways in which he believes hi mself to have contribut ed to its so-called stretching. I suggest that the f unction-theoretic treatment of concepts advanced within the first four pages of ‘ Function and Concept ’ (see also Grundgesetze for a parallel discussion) may be seen to ge nerate several difficulties in light of Frege’s theory of assertion. Moreover, it is suggested that the crux of Frege’s difficulties lay in his first attempts to extend the function-theoretic account so as to include the concepts of equality and inequality. In the first place, the field of mathematical operations that serve for constructing functions has been extended. Besides addition, multiplication, exponentiation, and their converses, the various means of transition to the limit have been introduced – to be sure, people have not always been clearly aware that they we re thus adopting something essentially new…Secondly, the field of possible arguments


9 and values for functions has been extended by the admission of complex number s…In both directions I go still further. I begin by adding to the signs +, -, etc., which serve for constructing a func tional expression, also signs such as =, >, <, so that I can speak, e.g., of the function x = 1, where x takes the place of the argument as before. The first question that arises here is what the values of this function are for different arguments. (Frege 1892, 28) We then find the following examples of the function x = 1 having been completed by the successive replacement of x with arguments -1, 0, 1, and 2: (2) (-1) = 1 0 = 1 1 = 1 2 = 1 It is evident in this passage that Frege finds it unnecessa ry to further comment upon the difference between expressions containing ‘=’ and expressions which do not (such as (1)) once the examples are introduced.3 In light of this fact, it ma y be suggested that such a difference is inessential the theory of functi ons and its intended app lication; but I suggest that failure to acknowledge the import of the difference is responsib le for much of the subsequent and widespread incomprehension of Frege’s theory of assertion among his most prominent interpreters. The suggestion should be unsurprising given what scant attention Frege gives to the con cept of judgment in the early Begriffsschrift One way to get at this difference is to raise the question of whether or not there is a unique kind of incompleteness that characteri zes functions, a question which Frege seems not to have considered in the earlier work on the subject.4 The suggestion is that in discovering there to be an ambiguity in the notion of incompleteness (in the grammatical


10 sense) we are then in a better position to s ee how it is useful to distinguish between the two types of expression used to name functions Further, this difference may be seen to have some interesting implications for Freg e’s semantic theory in general, which I discuss in a later sec tion. Let us look at the followi ng two expressions of distinct functions: (i) x ( x – 2) (ii) x > 0 Given ‘2’ as the argument in (i) we obtain ‘0’ as the value of the function. This may be expressed by the following sentence: (i*) ‘2 4 = 0 However, if for (ii) we take ‘1’ as the argum ent, then, given that the value yielded by the function is not a number, the value must be one of the two truth values. A not uninteresting question at this point is – how is this best to be expressed? It may be thought that we could write, (ii*) 1 > 0 We know this to be incorrect on Frege’s view since it fails to expr ess the idea that the function yields a truth-value (in this ca se, ‘the True’), even though we know the expression to be correct just insofar as it is syntactically complete (This is expressed by


11 the declarative sentence: ‘One is greater than zero.’). Therefore, th e occurrence of ‘1’ in (ii*) would seem to satisfy the function (ii) insofar as the resultant expression becomes thereby complete. Such an interpretation woul d also seem to be a consequence of what Frege has given as a definiti on of a function in general ( Begriff. ; Grundgesetze ). Rather, we might express the same idea without loss of clarity by introducing the following phraseology: (ii**) ‘1 > 0 has the value ‘the True’, where – if (ii**) is to avail us in drawing the analogy betwee n (1*) and (ii*) – we are to infer that “has the value” and “indicates” are to be read in a way an alogous to the way in which “=” is read in (i*). But if this is correct, then to say that “‘2(2 – 2) = 0’ indicates ‘the True’” amounts to saying that the senten ce indicates a truth-va lue once the value for the function – given ‘2’ as argument – has alr eady obtained. That is, in this instance ‘the True’ in (i) is indicated once the numerical va lue ‘0’ is yielded for the function (given ‘2’ as argument). I have here attempted to expr ess a point made by Freg e in a different way, stressing how the treatment of mathematical equality gives rise to the semantical categories of Frege’s system. Frege claims: I now say: the value of our function is a truth-value and distinguish between the truth-va lues of what is true and what is false. I call the first, for short, the True; and the second, the False. Consequentl y, e.g., what ‘2 = 4’ means is the True as, say, ‘2’ means 4. And ‘2 = 1’ means the False. (Frege 1891, 28-29)


12 However, we may easily see that this is not true of (ii): given that ‘1 > 0’ is the result of having completed the function expr essed by ‘x > 0’, there is good reason, prima facie not to proceed to explicitly name the value for the function of the argument ‘1’, (viz. ‘the True’) in the way that Frege has done To give a name for the value in this way suggests that the function is in further need of completion. This is suggested because the syntactical element (here recognized only implic itly) that is containe d in the meaning of the sign ‘>’ is articulated in (i*) by what we understand to be expressed by the functional expression ‘= 0’ – ‘ is 0’ – which forms the expression in (i) given ‘2’ as the argument. It is useful, then, to observe that in (i *), ‘=’ is a sign used in and for the construction of the complete expression. An expr ession containing two argument places, together with the relation-expression ‘ is greater than ’ is syntactically of the same form as is any expression containing those arguments together with th e function expressed by the words “is,” “is the same as” and “is identical to.” This much is evident given the finite form of the principal verb, “to be”. The point of exhi biting syntactical comp leteness, however, may be generalized to apply to th e occurrence of the principal fi nite verb in any declarative sentence. In the use of the terms “syntactical element” and “syntactical completeness” throughout, what is intended is a sense of an element or feature of language that is not wholly destitute of semantic significance. As concerns sentential structure, this element is to be taken as something that is essentia l to the rules of sentence formation for an interpreted language; it therefore pertains too in specific ways to specific rules governing semantic composition. It is assumed throughout th e discussion that th e notion of structure as pertains to sentences is not something con ceptually prior to certain semantic concepts; considerations as to structure in this sense are not taken necessarily to be something that


13 should be wholly abstracted from semantic investigations into natural language.5 The concept of linguistic meaning is, rather, assume d to be prior in some sense, but pursuing the question of precisely in what this priority consists would lead us too far astray from the subject of discussion. An obvious consequence of all these cons iderations as to logical composition is the very plausible sense in which functions ma y be characterized as being made complete ( saturated ) regardless of what numbers are dete rmined as arguments for them, and therefore regardless of the sens es of the sentences that obtain as a result of completing the function. A very similar point is made by K. Wiredu’s “Truth as a Logical Constant; with application to the Principl e of Excluded Middle”: To assign a truth value to a func tion is not to say of it that it is true or is false, for, of course, it can be neither; it is merely to supply it with a truth claim. A correlative observation is this: The type of truth value that is involved when an assertion is said to be true or to be false is not exactly identical with the type of truth value that may be assigned to a truth function, though as we shall see, there is a very close relation between them (Wiredu 1975) We may note that not only is it the case that the completion of the function in (ii) occurs in virtue of designating ‘1’ as argument, as expressed in (ii*); and by parity of reasoning, the same kind of completion is just as easily rendered by designating ‘-1’ as the argument. Analogously, (i*) may be taken to be the result of having completed the function as expressed by (i). (Function) ‘the capital of ’ ‘2 + ’ (Funct.+ argument) ‘the capital of India ’ ‘2(4) + (4)’ (Value) ‘New Delhi’ ‘132’


14 As a consequence of MG, ‘The capital of India’ indicates ‘New Delhi’ in the same way in which ‘2(4) + (4)’ indicates ‘132’. This is, as is evident by the explanations given in “Function and Concept,” the model on which extending the function is based. Frege’s extending the concept of function, and thereby the extending of the domain of values to include ‘=’, leads to the followi ng parallel expressions: ‘2 + = 132’ ‘____ is a man’ ‘2(4) + (4) = 132’ (‘4’ as argument) ‘ Socrates is a man’ ‘the True’ ‘the True’ Thus we have a doctrine according to which ‘S ocrates is a man’ is a complex designation, given that ‘2(4) + (4) = 132’ is a complex designation6. If we have followed Frege’s reasoning this far, then we ar e led without too much difficult y to the related doctrine that ‘the True’ and ‘the False’ are l ogical objects designated by sentences qua complete designating expressions. We find arguments in support of this doctrine in, e.g., “Function and Concept” (Frege 1891). A sentence is no t incomplete; therefore the sense expressed by a sentence is an object. This conclusion is also consistent with other of Frege’s methodological principles. Howard Jackson reports, having studie d one of Frege’s unpublished papers of 1891-1892, that Frege argues “that th e sense of an expression is an object and since, for Frege, objects and concepts are in every case to be distinguished (a distinction made consistently throughout his writi ngs), the sense of an expression is never to be confused


15 with a concept.”7 A fortiori what is denoted by the sense of a sentence is complete. Since the denotation of the sense of a sentence is on e of the two truth-valu es and since whatever is complete is, by definition, an object, truthvalues are ther efore themselves objects; as such they belong to the realm of reference. The results of apply the method of gene ralization as Frege has done reveals an ambiguity in Frege’s explanation of how valu es, in general, obtain. In “Function and Concept” Frege states that “the value of our function is a truth-value,” where the sense here of “value yielded” (given an argument for the function) is indistinguishable from the sense in which the expression ‘1>0’ names a truth-value (in this case, ‘the True’).


16 Chapter II : The Relation of the Theory of Func tions to the Theory of Assertion There is a philosophically interesting c onsequence of Frege’s strategy of first extending the concept of function to include identity, and then extending (on this basis) his functional analysis, already applied to nom inalized sentences, so as to include the copulative sense of ‘is’. This consequence is uncontroversial and is articulated succinctly by Frege himself in Grundgesetze : ….I do not mean to assert anything if I merely write down an equation, but that I merely designate a truth-value, just as I do not assert anything if I merely write down ‘2’, but merely designate a number. (M. Furth (trans.) 1964) The concept of ‘2’ as a naming expressi on is generalized to include sentences, e.g. ‘2 = 4’. I wish to examine now some further impli cations of Frege’s thesis that sentences are entities of the same logical type as singul ar terms. This identification of sentences with singular terms, which was seen in the la st chapter to be a direct consequence of Frege’s methodology (in particular, MG) would seem to have, in turn, some undesirable theoretical consequences with respect to Frege’s understand ing of the nature of assertion. According to V. H. Dudman, a noted Fr ege scholar, there are two prominent interpretations of Frege’s judgment-stroke – one attributed to Max Black and the other to P. T. Geach. Although upon investigati on it becomes apparent that the Begriffsschrift


17 account accords with neither interpretation, th e extent to which the later Frege (e.g. of Grundgesetze ) was committed to one interpretation or the other is patently unclear. Thus, where Frege’s theory of assertion as it occurs in the context of the later semantic theory of Grundegesetzse (and in “Function and Concept”) is not met with incomprehension, it is recognized by many philosophers of language to be problematic. In what follows I try to give some indi cation of just how problematic that theory is. On Black’s account, the assertion sign, ‘ ’ was introduced to convert a mere (complex) designation to a non-designation.8 Such a conversion then would allow Frege to “restore to the propositional sign its truth-claiming aspect.” V. H. Dudman looks to the following passage in “Function and Concept” in an effort to marshal textual support for this interpretation: …[According] to the Black ve rsion asserted sentences are not names at all: the judgment stroke “does not serve, in conjunction with other signs, to designate an object, ‘ 2 + 3 = 5’ does not designate anything ; it asserts something.”9 Unlike Black, however, Dudman reckons the view expressed here to have already fallen into serious error. Dudman goes on, now ec hoing a point made by Furth in an earlier quoted passage10: In that passage Frege is surely over-reaching. What he ought to say is rather that ‘ 2 + 3 = 5’ does not just express a thought and designate a truth-value, for “over and above this is the acknowledgment that the truth-value is the true. (Dudman 1975) On the proposed amendment, Dudman recommends a view that had in fact been maintained by Alonzo Church. In his Introduction to Mathematical Logic (1951) it is


18 maintained that sentences are still names – wh ether asserted or not – though they are said to differ from other naming expressions in their use. Church gives the following justification for his construi ng sentences in this way: [We] shall require variables for which sentences may be substituted, forms which become sentences upon replacing their free variables by appropriate constants, and associated functions of such forms – th ings which, on the theory of sentences as names, fit natura lly into their proper place in the scheme set forth in §§ 02-03. …[Granted] that sentences are names, we go on, in the light of the discussion in §01, to consider the denotation and the sense of sentences. (Church 1951, 24) I cite this passage, in part because it lays out an (alleged) justification for the ‘theory of sentences as names’. This justification figures into an explanation that I give later as to why Frege’s theory of truth, despite all its metaphysical appurtenances, has nonetheless had a strong influence among subsequent phi losophers of language. The passage is of further interest to us since it immediately fo llows a passage that c ontains both what is later argued to be the lync hpin of Frege’s theory of m eaning, and the source of many metaphysical and linguistic muddles – not th e least important of which I discuss in Chapter VI, “Moore’s Paradox and the Logic of Assertion.” Th e preceding passage is as follows: This [account of sentence meaning] seems unnatural at first sight, because the most conspicuous use of sentences (and indeed the one by which we have just identified or described them) is not barely to name something but to make an assertion. Neverthele ss it is possible to regard sentences as names by distinguishing between the assertive use of a sentence on the one ha nd, and its non-assertive use, on the other hand, as a name and a constituent of a longer sentence (just as other names are used). Even when a sentence is simply asserted, we shall hold that it is still a


19 name, though used in a way not possible for other names. (Church 1951, 24) This assumption leads us to the central not ion underlying Frege’s theory of sense: The sense of a sentence may be described as that which is grasped when one understands the sentence, or as that which two sentences in different languages must have in common in order to be correct translations each of the other. As in the case of names generally, it is possible to grasp the sense of a sentence without therefore necessarily having knowledge of its denota tion (truth-value) otherwise than as determined by this sense. In particular, though the sense is grasped, it may sometimes remain unknown whether the denotation is truth. (Church 1951, 26) The possibility of grasping the sense of a sentence without knowing whether the denotation is truth (or false hood) in the way Church describes implies that one may recognize the thought expressed by the sentence to be correct (w hile at the same time not knowing its denotation. The failure here to represent senses in a grammatically perspicuous fashion leads ultimately to a ki nd of skepticism with respect to truth – makingd the property of truth epistemological ly inaccessible. Truth-skepticism, however, is not so much entailed by Frege’s notion of sense in itself as it is by Frege’s manner of grammatically representing it. The sense-refe rence distinction as typically construed, therefore, might plausibly be seen as spurious. What we should want to preserve in this distinction is the idea that an assertion and a proposition share a conceptual content; however, as I suggest in what follows, we should want to avoi d drawing the distinction in precisely the way Frege and ot hers have done. This commonality is what Frege expressed in his own way by claiming that sense determ ines reference, and the manner in which Frege draws the sense-reference distinction ac hieves this much. But the signinficance of


20 how it is drawn is crucial to understanding it. As we will see, a good deal hangs upon the way it is drawn. The argument that Dudman presents in favor of this interp retation (which he attributes to Geach) is at times difficult, but it mirrors in many respects the confusion into which Russell fell in his early (1903) treatmen t of assertion and so may warrant closer examination. I come to this later. According to Dudman, Black’s view holds that the assertion sign was introduced as a consequence of Frege’s having recogni zed the “namehood” of sentences. It is understandable that this doctrine would then require the use of a sign which would restore to a propositional sign its “truth-claiming aspect ” – particularly in light of the conceptual priority of this doctrine in Frege’s later semantic theory. For Black the necessity of introducing such a device is a consequence of Frege’s constr ual of sentences as complex designations. Let us calls this Frege’s ‘later doctrine’. This is essentially the conjunction of theses (a) an (c) introduced on p. iv, namely, that sentences (normally) have denotations and are of the same logical type as singular terms. On Black’s view, the namehood (its semantic role of naming) of a sentence ipso facto deprives it of its assertoric function. As Dudman observes in th is connection, “to name is not to say.” I wish not to quarrel with this line of interpretation per se but to underscore what it is that Dudman has so far correctly observed – vi z., that this doctrine appears fully in Grundgesetze (Frege 1891). Thus, if the Black interp retation is at all plausible, we must be prepared to admit that Frege had two explanations for the introduction of the judgment-stroke. The sign was introduced in Begriffsschrift in 1879, a considerable time


21 before Frege provided a late r and separate explanation of the judgment-stroke in Grundgesetze .and “Function and Concept” in 1891. It will be useful to inquire as to whet her the theory of functions, from which Frege derives the notion that sentences rather are complex names, is independent of Frege’s early theory of assertion. In an article entitled “Frege’s Judgment-Stroke” V. H. Dudman writes: What are used assertively and non-assertively are alike sentences, and, as sentences, al ready have their own verbs; those verbs are part of what are being us ed assertively or non-assertively, and what is used assertively on one occasion cannot be the same as what is used nonassertively on another if they differ in respect of verb. (Dudman 1975, 155) The crucial – and suppressed – premise in th is passage is this: when the grammatical rendering (the “grammatical expedient”) of ‘ ’ is used to mention or indicate, then it cannot also be used to make an assertion. It cannot also be a “ca ndidate for assertive use”.11 But I think that we are entitled to as k here why not. It is assumed that the declarative form ‘a is F ’ is used both assertively and non-assertively; therefore we should recognize ‘ is ’ as “part of” what is used at one time assertively in ‘a is F ’ another non assertively in ‘a is F ’. On Dudman’s assumption, then, the grammatical difference expressed between the declarative form ‘a is F ’ and the non-declarative ‘the circumstance that a is F ’ is not sufficient to characterize the di stinction marked by Frege’s use of the signs ‘ ’ and ‘ ’. There is very little textual evidence in support of the interpre tation according to which the assertion sign was conceived by Fr ege as merely an index of assertion.


22 Sometimes called the ‘Geach interpretation’, th is reading of the assertion sign as a kind of index maintains that the addition of the ve rtical judgment-stroke to the (horizontal) content-stroke is intended not to affect the semantic role of the expression to which it attaches. Thus, the sign is not a device wh ich serves as a grammatical functor from substantival phrases (or noun phr ases) to declarative sentence s. Nor is it a device that restores to complex designations an assertoric force that is otherwise vitiated simply in virtue of their naming function. It is a device which shows merely when a sentence is being used assertively rather th an non-assertively. On this view there is no corresponding grammatical difference. Frege was simply mi staken to think that the judgment-stroke would alter semantic content in addition to se rving as an index of assertion. The so-called index interpretation ap pears to be what Church has in mind when he distinguishes between “assertive use” and “nonassertive use” of a sentence.12 The initial restriction on the use of the judgment-stroke is that put forth by Russell; and it seems that others, including those we have been discussing, have followed the prescription. On Russell’s view it is only appropriate to pr efix the judgment-stroke to expressions representing con ceptual contents which are “i n principle capable of being held true.” (Russell 1903, Appendix) Both the Geach-Dudman and Russell interpretations appear to be in agreement that falsehoods ar e not capable of being asserted in the same way in which truths are. However, as will be shown in the next chapter, Russell does not fail to see the implication that for this very reason error becomes impossible. In an effort to escape the implication, Russell claims that assertions whose conceptual contents are incapable of being held true are to be regarded as having a primarily psychological status. Russell was not the only one to regard a ssertions in this way. Several prominent


23 philosophers, chief among which are Church, Geach, and Dummett, have made the same move even if they were bli nd to that troublesome implication.13 Dudman has no good reason to claim that2) Dudman is incorrect to say that there is no explanation of why the a ssertion sign is not needed in Begriffsschrift (unlike Grundgesetze and F&B ) or to claim that, whatever Fr ege’s intentions were, §§ 2,3 of Begriffsschrift “embody a readily unders tandable slip on Frege’s part.” There Frege is said to have been unmindful of the fact that assertoric force is susp ended (“cancelled”) in truth-functional contexts. In light of what Freg e actually says in thes e sections (2 and 3), this criticism is unjust. It is not obvious why it is incorrect to say that ‘the circumstance that unlike poles attract ’ cannot “even be a candidate for assertive use,” a nd that because its semantic role is to name it cannot be used to say Further, it does not follow that even if “the circumstance that unlike poles attract ” cannot be a candidate for assertive use, a declarative sentence cannot be used to both name and assert. Dudman’s argument at this point is puzzling: (1) If the vertical stroke is to be an index of assertion (in Geach’s sense), then we are left with no choice bu t to also assign to it the role of what Frege has cal led the ‘common predicate’ (that of a “verb”), “is true,” or what is the same, “is a fact.” This is the role it is said to have in §§ 2 and 3 of Begriffsschrift (2) If the judgment stroke joins to an expression a common predicate upon an expression, then the judgment stroke cannot be merely an index of assertion – i.e. that is, a device to signify when a sentence is being used assertivel y and when it is being used non-assertively. Therefore, the judgment stroke cannot be taken as a kind of functor. Dudman observes that this is a consequence of assuming a de finite description such as “the circumstance that unlike poles attract” to be an adequate renderi ng of an expression


24 ‘A’ in ‘–A’. But Dudman thinks this is just why the Begriffsschrift account is mistaken, for Dudman maintains that the judgment-stroke “cannot be an index of assertion if it is a verb.” The problematic assumption on which this argument rests may be located in the the following passage14: (A2) “What are used assertively and non-assertively are alike sentences, and, as sentences, al ready have their own verbs; those verbs are part of what are being used assertively or non-assertively” Dudman then goes on to claim, “what is us ed assertively on one occasion cannot be the same as what is used non-assertively on another if they differ in respect of a verb. ” Appealing to an argument originatin g in Frege’s later writings, Dudman concludes that any sentence c ontaining the universal predicat e (the verb) is a sentence which itself may also be used non-assertively. The confusion that emerges here – a c onfusion that appears also to underlie Russell’s difficulties concerning the nature of truth in (§§ 51-53) Principles of Mathematics – is one that can perhaps only be cleared up by suggesting yet another interpretation of the assertion sign. Assuming that the verb is part of what is be ing used (as part of a sentence) assertively on one occasion and non-assertively on anothe r, it may be doubted whether we should therefore assume that which is to be used assertively and non-assertively must be a declarative sentence We have as an example of what is to be used assertively and nonassertively alike, “Unlike poles attract.”15 It may be argued that (i) the verb ‘is’, in


25 Russell’s example – ‘Caesar is dead’ – cannot be part of an expression used assertively if that expression (the declarative form) is also to be used non-assertively, for normally we do not say non-assertiv ely ‘Caesar died’. The foregoing points, arguably may be taken to support Dudman’s contention since it is compatible with the later doctri ne that indicative sentences are proper names and as such are debarred from assertoric employment. Of course, Dudman’s contention that Frege’s rendering of ‘— A’ as “the circum stance that unlike poles attract” in order to display grammatically the occurrence of a possible content of judgment was an “understandable slip” on Frege’s part is also consistent with (i) above.16. The compatibility of these claims, it may be specu lated, encouraged Dudman to think that he was on to something and therefore might be forgiven for justifying his interpretation on such slender textual grounds. What support is thereby afforded the Geach inte rpretation seems to collapse given what I take to be another, very plausibl e interpretation of the assertion sign. This interpretation is perh aps closest in letter and spirit to §§ 2,3 of Begrifftsschrift On this interpretation, Frege made no understandable slip and did not misrepresent his own intentions at all. Contrary to Dudman’s claim, Frege’s use of (1) ‘the circumstance th at unlike poles attract’ and (2) ‘Unlike poles attract’ as linguistic expressions for which th e following symbolic expressions stand,


26 (1a) —A (2a) A ,accomplishes just what it was intended to accomplish, namely, to show that (1) is the conceptual content of (2); and to show, therefore, that (2) shows (1) as having been asserted Thus, what has come to be known as the common predicate (of Begriffsschrift ) reading of the assertion sign is correct. But also correct is the contention that (I) the assertion sign is, in a sense, an index of asse rtion. However, it must be added that this contention is not merely incidental to the common predicate reading. That is, we should not therefore read the assertion sign as providing an index of assertion independently of, and in addition to, conferring the predicate upon the participial expres sion. To take it as such is to regard the introduc tion of the assertion sign as an unnecessary accretion to an already established th eory of assertion.17 Rather, it is by virtue of our recognizing the presence of the common predicat e that the assertion sign may then serve as an index of assertion. We are required to take the asse rtion sign as a functor which yields transformations from noun-phrases to comp lete sentences. Though this functor is syntactic, one may argue that th e transformation itself is not merely syntactic: it yields both a logical and semantic transformation fr om definite descriptions to complete sentences – i.e. from propositional contents to truth-values. There is, then, no reason to think that what is viewed asser tively and non-assertively are alike sentences. Therefore, (A2) must be incorrect.18


27 This reading contradicts what Dudman ta kes to be the prevailing interpretation of the assertion sign, and given our earlier glo ss on his argument for this interpretation, we may be inclined to regard (I) as a reductio in favor of the prevai ling view. This would be plausible were there not good reasons, as I suggested there are ear lier, for rejecting Dudman’s key assumption. The assumption in qu estion, it will be recalled, is that the judgment stroke cannot serve as an index of assertion if it is also to suppl y the verb – in these examples, the predicate ‘is true’, or ‘is a fact’. The reason is that sentences are the types of entity capable of being used bot h assertively and non-assertively. But as sentences Dudman claims, they already contain verbs (A2). On a closer examination of Dudman’s discussion we can begin to see where the confusion begins to set in. It would seem to be found the following paragraph, which immediately precedes his argument for his interpretation: Frege comes up with (1) “the circumstance that unlike poles attract” as being an e xpression satisfying the two conditions of (a) having the same conceptual content as (2) “Unlike poles attract,” and (b) lacking assertoric force.” (Dudman 1975) Witness that (1) is an expressi on which satisfies condition (a) – of having the same conceptual content as (2). This implies, of course, that there is some third thing (something like a proposition) which (1) and (2) share But a not unreasonable question might arise as to what this thing could be and in what way it is to be expressed linguistically. In sections 2 and 3 no such thing is implied, since Frege never makes the claims that are attributed to him by Dudman in the passage in question. A car eful examination of


28 these sections will reveal that (1) “the circumstance that unlike poles attract” just is the conceptual content of (2), “Un like poles attract.” It should then be noted that (1) is taken as an unsaturated expression which is conve rted to (2) upon having restored assertoric force. Once the judgment-stroke is seen in this light, we avoid the need to posit the semantic tertium quid to which the Geach-Dudman view must be committed; for, we cannot sensibly hold that (1) has the same conceptual content as (2) if (1) just is that conceptual content. I suggest that this reading is consistent with the spirit of Frege’s early theory of functions, as the horizontal (or content-stroke) would then be seen to, together with the expressions to which it is pr efixed, express a function. Histor ically speaking, this is not without a more general significance since it affords some insight into the causes of Russell’s difficulties in the famous passage of §§51 and 52 of Principles Before turning to this, we shall first have to substant iate the foregoing claims against Dudman’s argument and question the merits of the Geach interpretation, since it would seem that that interpretation is inconsistent with much of what is implied by the proposed interpretation – viz. that assertoric force alters the semantic role of the expression to which it applies. Judgeable Content, Truth, and ‘Judging true’ The Geach interpretation is thought to be supported by several ar guments appearing in discussions after 1891 (in 1896, in two published essays on Peano’s system): the role of the j udgment-stroke in the cited passages is taken by Dudman to be the role accorded to it in Begriffsschrifft that is, as an index of assertion and not as a grammatical functor. I have argued that this is


29 mistaken, but it may explain in part the pre dominance of the Geach interpretation among various interpretations of Frege’s philosophy of language.19 Consistent with the proposed interpretation in §1 are the views expressed in several passages of Begriffsschrifft which may also be taken to support the DudmanGeach interpretation. In their inte rpretation of, for example, §7 of Begriffsschrifft we begin what might appear to be a marked dete rmination to divest Frege’s account of the judgment-stroke of its sy ntactical significance: If the judgment-stroke is absent then, here as elsewhere in the Begriffsschrifft, no judgment is passed. A merely requires the formation of the idea that A does not take place, without expressing whether this idea is true (Geach 1965). Then later on, “Even in section 2 the account of the judgment stroke in terms of “the circumstance that” …is counterbalanced by a passage that supports Geach’s “index of assertion” interpretation: If the small vertical stroke at the left end of the horizontal one is omitted then this is to tran sform the judgment into a mere combination of ideas concerning which the writer does not express whether he acknowledges its truth or not (Geach 1965). Though it is perhaps not so clear how section 7 is supposed to count as further evidence of the point that section 2 is alleged to have already made we may nevertheless see that the oversight of the syntactical distinction here between a me re combination of ideas and judgment is encouraged by a somewhat biased understanding of Freg e’s early theory of


30 judgment. This emerges most notably in the words, “concerning which the writer does not express whether he acknowledge s its truth or not.” The aut hor cites this as a passage (§2) which “counterbalances” the syntactical distinction containe d in the preceding paragraph. But it does more than this: it betrays an insistence upon a wholly psychological conception of judgment accordin g to which judgment amounts to nothing more than ‘veridical commitment.’ On this assumption we are led to the conclusion that the cited passage must countervail Frege’s discussion in §2 ( Begriff. ) in which the syntactic distinction is explicitly stated. However, once we drop this problematic assumption, whatever motivation we may ha ve had for maintaining the Dudman-Geach interpretation is lost. Supporters of the Geach view will perhaps not hesitate to point out that the force of the objection that I have raised depends upon the plausibility of taking the (above) quoted passages as meaning something other than what most Frege scholars have taken them to mean. Moreover, it would seem that pursuing the objection further would lead us into some rather thorny issues surrounding th e nature of judgment more generally. The problem would then become the more gene ral one of understanding what we do in our activity of passing judgment and ho w the relation of our notion of passing judgment – of expressing veridical commitment – is related to what we already take to be the objects that judgments are about. There may be no better entering wedge into this difficult problem than the remarks of several ke y passages in Frege’s later writings.


31 “Two components in that whose external form is a declarative sentence”: (i) The acknowledgement of truth and (ii) The content that is acknowledged to be true: (a) Thought ( Der Gedanke ) (b) Truth-value Frege explains that content is “split” into (a ) thought and (b) truthvalue, and this split emerges “as a consequence of distinguishing between sense and denotation of a sign.”20 Note that Frege construes an ‘acknowle dgement of truth’ in such a way that implies that what was once a component given a semantic role in a theory of judgment is now a component which is left out of account. Se en in this light it then becomes apparent how such a lacuna led various interpreters to mistakenly construe “acknowledgement of truth” as synonymous with veridical co mmitment. The “acknowledgement of truth” ought to have meant the determination of the sign’s referent by the sense of the sign and not, as is implied by Frege, the further determination of a truthvalue by the denotation. If we take the denotation itself to have a denotation, then we are forced to postulate ‘the True’ and ‘the False’ as logical obj ects. With the above passage from Grundgesetze we have the following development of Frege’s views, outlined here diagrammatically:


32 Figure 1. Components of a declarative sentence Frege’s suggestion in “Function and Concept” is that there is a need to maintain a clear “separation of the act from the subject-matte r of judgment,” and to do this we should need to treat indicative sentences ‘5 > 4 ‘, ‘1 + 3 = 5’, e.g., as proper names. Following Dudman’s suggestion, we may take this to be a non-sequitur, and rightly so: the separation of the act from subject-matter of judgment does not en tail the thesis that declarative sentences are proper names. We may, in fact, press the poi nt a bit further than Dudman has done and claim that not only is the namehood thesis of declarative sentences not entailed by a commitment to keep separa te the act and cont ent of judgment, it ought not be entailed by this commitment. More prec isely, if the namehood th esis is entailed in the way Frege claims it is, then his separation of the act from the content of judgment must then be the wrong kind of separation. It would appear that what I take to be Frege’s Acknowledgment of truth Possible content of judgment Thought Truth-value Declarative sentence


33 mistake in this connection was overlooked by Russell who was chiefly instrumental in bringing Frege’s later semantic theory to lig ht in anglo-american world of philosophy and therefore largely responsible for the receive d view of Frege’s theory. Thus, while the practice of treating declarative sentences as proper names has continued in semantic theory well after Grundgesetze the initial reasons for adopti ng that practice was lost sight of. This is an important point bearing on c onceptions of truth which I discuss later in further detail. Figure 2 Semantic Content: The relationships roughly pictured here, put forth explicitly in the previous passage ( Grundgesetze ), show that we are to take the thought of a sentence as having a denotation, in which case we must hold that a (non-defective) senten ce denotes one of the two truth-values. This is a c onsequence of identifying the sense of a sentence with the thought expressed by the sentence. Given the view presented in Grundgesetze according to which the thought as expressed by an indi cative sentence inherits the structure as determined by assumed semantic principles of composition of the sense of the Content Possible content of judgment Sense Reference Thought Truth-value


34 corresponding sentence, a thought stands as a complete entity. The structure and the logical standing of the thought, if we are to analyze its struct ure in the context of Frege’s early view of assertion ( Begriff. ), exhibits this complete ness by showing a ‘possible content of judgment’ as having been asse rted; that is to say, as having been advanced as true To speak of a content as having been adva nced as true is simply to speak of an acknowledgement of truth ( Begriff. ) the grammatical representa tion of which, as we have seen, is the vertical-stroke. Thus, if it is correct to assume that a truth-value obtains with the addition of assertoric force to the incomplete expressi on – which represents a ‘possible content of judgment’ – then we are said to be in possession of a truth value given the denotation of the sense of a sentence. This observation is supported when we attend to Frege’s discussion of the twin notions of desi gntation and denotation and how they are distinguished in section 2 of Grundgesestze : I say: the names “2 = 4” and “3 > 2” denote the same truth-value, which I call for s hort the True. …Likewise, for me “3 = 4” and “1 > 2” denote the same truth-value, which I call for short the False, precisely as the name “2” denotes the number four. Accordingly I call the number four the denotation of “4” and of “2”, and I call the True the denotation of “3 > 2” (Frege 1891, 35) Immediately preceding these remarks, Frege writes: The expressions “0 = 4”, “1 = 4”, “2 = 4”, “3 = 4” are expressions some of true, some of false thoughts. I put this as follows: the value of the function = 4 is either the truth-value of what is true or that of what is false.” (p. 35) (see also S&R)


35 The true thought “2 = 4” is expressed by designating ; that is, the truth-value (here, the True) is designated independently of our asserting that 2 = 4. But here, by the very fact that we are dealing with sentences, simply because we are dealing with expressions involving equalities and inequalitie s, we are liable to be tem porarily unmindful of the fact that such expressions are already complete (i.e. “saturated”) expressions and that inattention to their completeness has certa in implications. One implication is the difficulty that Montgomery Further had drawn a ttention to in remarking that the notions of truth and falsehood "seem to be turning up twice over in the theory, once within the domain of individuals in the guise of the ‘logical objects’ the True and the False, and then again at a different level as success versus failure… of the act of asserting.”21 As complete, these expressions cont ain an occurrence of truth, unless they undergo an operation whereby they are divested of their assertor ic force. Of course, Frege and others (viz. Russell and Geach) explicitly recognize that such an expression must be capable of being expressed without being thereby asserted. One consequence of this, however, is a very one-eye d preoccupation with the psychological aspects of assertion. This, in turn, would seem to be a conse quence of several developments which, as it happens, span the whole of Frege’s work in semantic theory: firstly, extending the function-theoretic approach to include the notions of equality and inequality; secondly, the introduction of the sense/refe rence distinction; and thirdl y, having lost sight of the fact that the internal stru cture of judgment “whose exte rnal form is a declarative sentence” carries with it (cont ains) an occurrence of truth given by the presence of assertoric force. It is the la tter that is at odds with the syntactic interpretation of the judgment stroke.


36 It would seem that the psychological sense of assertion may be maintained with or without the syntactic dis tinction: given the senten ce ‘Unlike poles attract’ we may say that we assert the sentence wh en we wish to recognize the truth of the sentence and that we hold the sentence in consideration when, e .g., we do not recognize its truth. However, it is assumed in the Begriffsschrift that we are to represent the relationship between (unasserted) judgeable co ntent and truth-value as follows Figure 3 Relation of semantic content to truth: Content Acknowledgement of truth Truth-value It is perhaps well known that this relations hip grows more complicated Frege’s later work. Thus, in Grundgesetze we have the following (Figure4):


37 Figure 4 Two levels of truth Sense Truth or Falsehood The True Problematically, the psychological sense of assertion according to which an assertion is merely an expression of veridi cal commitment (to the sense of a sentence) then prevails over the logical sense in the following passage from “Function and Concept”: If we write down an equation or inequality, e.g. 5 > 4, we ordinarily wish at the same time to express a judgment; in our example, we want to assert that 5 is greater than 4. According to the view I am pres enting ‘5 > 4’ and ‘1 + 3 = 5’ just give us expressions for truth-values, without making Acknowledgement of truth Acknowledgement of truth


38 any assertion… (Frege 1891; compare to passage in ‘On Herr Peano’s Begriffsschrift’, viz. “with ‘(2 > 3) = (7 = 0) a sense of strangeness is at firs t felt…for such a sign serves two distinct purposes…”) The words “we ordinarily wish at the same time to express a judgment” are instructive in that they imply that there are occasions on which we do not wish at the same time pass judgment, e.g., that 5 is greater than 4 when we simply write ‘5 > 4’. In case such as this we should have no need for a grammatic al functor to e xpress the content unasserted It appears that the mind – upon havi ng been given the sentence ‘5 > 4’ covertly supplies the affirmative judgment that ‘5’ is greater than ‘4’. In the case that we judge a content as false, this point is brought out more clearly: given the written senten ce ‘1 + 3 = 5’, we seem to be engaged in what Frege calls expressing a supposition – or rather covertly (and simultaneously) – supplying a negative judgm ent which we may wish to say, following Frege, designates the truth-va lue ‘the False’. On Frege’s vi ew, strictly speaking, this is not a judgment, since in inscri bing the signs ‘1 + 3 = 5’, or in being presented with the sentence ‘1 + 3 = 5’, we may simultaneous ly suspend judgment as to its truth or falsehood. In such cases it is clear that we me rely feign an expression of the (unasserted) judgeable content by assertive means. We are still apt to ask what exactly are we doing in suspending judgment in this way, suggesting the operation is perhaps more complicated. In the next chapter I argue that an understa nding of assertion in terms of a veridical commitment to sentences, where force is now conferred, now suspended only psychologically, leads to serious puzzles. These are puzzles, however, that arise out of the prominence of Frege’s later views. They ma y be solved by appeal ing to the insights of his earlier Begriffschrifft The prominence of his later vi ews owes in large part to Russell’s interpretation of Frege’s work in PofM Russell, and many who later fell under


39 his influence, seized on the psychological approach to a ssertion, thereby inheriting the puzzles and associated tangles concerning th e nature of propositions and facts, and the definition of correspondence. In a certain sens e, what is occurring implicitly in claiming that a sense has a denotation just is the passi ng of judgment – or, to adopt the language of Begriffsschrift an “acknowledgement of truth” – a tr uth claim. It is now a simple matter to see why, given a psychologica l sense of assertion, Frege would vehemently deny that expressions standing for truth-values (i.e. what and how a given expression denotes a truth-value) are expressions of judgments


40 Chapter III: Frege and Russell on Assertion In this chapter I advance a syntactical in terpretation of Frege’s theory of assertion which derives from the views of Kwasi Wi redu (1975) and W. E. Johnson (1921).On this interpretation, the expressions (1) ‘The circumstance that unlike poles attract’ (2) ‘Unlike poles attract’ are to be taken as the grammatical forms of the symbolic expressions, (1a) —A (2a) A It is then argued that Frege (of Begriffsschrift ) accomplished what he intended to do, namely, to show that (1) is the conceptual content of (2) and that (2) shows (1) as having been asserted The semantic role of expressions may be said to be altered by the addition of the (vertical) judgment-stroke in (2a) in such a way th at requires us to either ( i ) convert the sentence to a complex noun phrase, or ( ii ) represent the altera tion by converting the principal verb from its finite to its infinitive form. On the proposed interpretation, we may find a solution of the problem with which Russell struggled in Principles of Mathematics in finding an account of the difference between the finite and infinitive forms of verbs that will accord with an “ult imate notion of assertion.” ( PofM ) The second part of the discussion examines Russell’s problem in li ght of the proposed view. Suggested by this view is a conception of assertion as a primarily logical rather than psychological notion.


41 Russell claims in Principles of Mathematics (1903) that grammar brings us closer to a correct logic than do th e opinions of philosophers. In ch apter IV of that book he sets out to treat the nature of assertion, propositio ns and the terms of gr ammar, prefacing his investigations with the remark, “in what fo llows, grammar, though not our master, will be taken as our guide.” ( PM §46) The remark is somewhat curious, since there is in so much of Russell’s early work (1903) more in common with the vi ews of Frege and of other mathematically-oriented logicians (e.g. Ca ntor and Peano) than there is with the work of logicians whose conception of logi c cleaves closely to the categories of traditional grammatical analysis. Here I have in mind the views of the early 20th century logician, W.E. Johnson, from whom Russell claims in the preface to Principles to have received “many useful hints.” In what follows I advance a thesis seve ral of the arguments pertaining to which are derived from a view of asse rtion advanced by Wiredu (1975).22 On the basis of this view I argue for a syntactic in terpretation of Frege’ s early theory of assertion according to which the two statements: (1) ‘the circumstance that unlike poles attract’ (2) ‘Unlike poles attract’ are formally expressible as (1a) —A (2a) A In this construal of the verti cal and horizontal strokes, as laid out in fi rst section of Begriffsschrifft Frege accomplishes what he intended to accomplish, namely, to show


42 that (1) is the conceptual content of (2) and that (2) shows (1) as having been asserted This way of understanding the judgment str oke – what I will call, after D. A. Bell23, the syntactical interpretation – is in keeping with the common predicate view of the judgment-stroke of Frege’s early Begriffsschrift .24 On this interpre tation, the semantic role of expressions may be said to be altered by the addition of the vertical stroke to the horizontal (content) stroke in (2a) in a way which requires us to either ( i ) convert the sentence to a complex noun phrase, or ( ii ) represent the alteration by converting the principal verb from its finite to its non-finite form (W. E. Johnson).25 In Frege’s own example (above) we have something like a tr ansformation to a noun phrase with the use of indirect speech, but there are cases, as in Russell’s own example – ‘Caesar died’ – that may be treated in accordance with the second option ( ii ). So, in taking the following pair of expressions: (3) ‘Caesar’s being dead’ (or ‘Caesar’s death’) (4) ‘Caesar died’ we have a suitable grammatical expression of the previous distinction ((1a) and (2a)), (3a) ‘—A’ (4a) ‘ A’ Accordingly, the judgment-stroke is construed as an operator yielding transformations from incomplete functiona l expressions (typically descriptions) to complete expressions which standardly take the form of declarative sentences. This conception is founded on the assumption that the construal of the judgment-stroke is syntactical : the interpretation of ‘ ’ as an operator derives ch iefly from the notion of a grammatical functor from noun-phras es to declarative sentences.


43 Recognition of this fact by interpreters of Frege’s Begriffsschrift it seems, has not been forthcoming, and in many cases Fr ege’s own argument for the proposed account is outright dismissed.26 The evidence to support this cl aim abounds. But, leaving that aside, I shall point out that there are many who simply de ny the logical significance of the syntactical interp retation, prominent among them are Michael Dummett, Peter Geach, Max Black and V. H. Dudman. I shall go on to sp ell out some implications of what I take to be a more tenable interpreta tion of the judgment-stroke. A supposed implication of the proposed inte rpretation is the apparent blurring of the programmatic separation of psychology a nd logic to which both Frege and Russell were strongly committed. But it will emerge shortly that if this is an implication of my view, then we ought, for the reasons adduced in favor it, reconsider the whether the logicist’s separation of psychol ogy and logic should have ever been attempted in the first place. Given the plausibility of the syntactical in terpretation, we can dispose of the difficulties that had plagued Russell in the early part of Principles of Mathematics One of these problems was to account for the di fference between the finite and non-finite forms of verbs that accords with what Russell curiously refers to as the “ultimate notion of assertion.”27 It is at this point in his discussion (§§51-53, PofM ) that the guide of grammar affords him a way out of these diffi culties. I will turn to briefly examine Russell’s problem in light of the suggested view of assertion. The main argument in §52 of PofM holds that “every constituent of every proposition must, on pain of self-contradict ion, be capable of being made a logical subject.” Russell goes on to say that “[By] transforming the verb, as it occurs in a


44 proposition, into a verbal noun, th e whole proposition can be tu rned into a single logical subject, no longer asserted, and no longer containing in itself truth or falsehood.”28 What Russell finds puzzling about this possibility is that there is no ostensible difference between the proposition as asserted and the corresponding logical subject; or rather, if there is a difference, we seem unable to say what it could be. Thus, if we go on to ask what is asserted in the proposition ‘Caesar di ed’, we should say that the ‘The death of Caesar is asserted’, or simply, ‘Caesar’s deat h’. In this instance it would seem that it is ‘Caesar’s death’ which is true or false. Bu t it is equally obvious that truth and falsity cannot belong to a logical subject. Several commentators have thus summarized Russell’s argument and have further ,so it would seem, admitted that it is valid. Ho wever, here I think we are entitled to ask how Russell arrives at the conc lusion that ‘Caesar’s death’ is true If we take the logical subject to be an incomplete entity, then while observing the distinction in (3) and (4), we should conclude that ‘Caesar’s death’ is not the sort of thing that could be true or false. We may say that ‘Caesar’s death’, as an e xpression grammatically equivalent to the participial expression, ‘C aesar’s being dead’, transforms to a declarative sentence once pressed into assertoric use. The complex th at is expressed by the verbal noun ‘Caesar’s death’ is not the kind of entity (categorially) of which truth or falsity may be predicated. To think that it is predicable in this way w ould be, to borrow Ryle’s term, a ‘grammatical type mistake’. Sentences, relativ e to contexts of use, however, are the sorts of thing of which we predicate ‘is true’ or ‘is false’.29 Double-aspect problems surface frequently in many of Russell’s writings from his early realist period, specifically in connection with his atte mpt to address the so-called


45 ‘problem of complexes.’ We may put that pr oblem in the form of a dilemma. It is assumed without question that we are capab le of marking any distinction that is thinkable. There is a thinkable distinction between concepts ( “as such”) – i.e. denoted as meanings and concepts denoted as terms We might, following Nicholas Griffin, characterize the distinction in the foll owing way: the con cept ‘one’ denoted as meaning may be symbolized as / a / and the concept ‘one’ denoted as term may be symbolized as / A / .30 The former ‘one’, taken in its adjectival form, is to be understood as the concept qua meaning The latter ‘one’, in its substantival form, is to be understood accordingly as the concept qua term We should then expect the distin ction to be at the very least stateable However, Russell has shown that any atte mpt to state what the difference is, or even to state that there is a difference, lands us imme diately in self-contradiction. On the other hand, if we take the concept denoted as meaning and the concept denoted as term to be one and the same concept – if / a / and / A / are identical – then we are destined to be “enveloped in inextricable diffi culties.” (The difficulty to wh ich Russell is here alluding is Lewis Carroll’s regress objection) The difference between / a / and / A / is therefore said to consist “solely in external relations.” This, according to the Russell of 1903, is a consequence of recognizing that the difference cannot be intrinsic to the nature of the terms, for in merely stating that / a / differs from / A / / a / is eo ipso converted to / A / ; hence the self-contradiction. We must then conclude that any proposit ion about the difference is nece ssarily false – and this is unacceptable for Russell. The problem remains: it is impossible to state the distinction with the needed precision, for we cannot hold that there is a concept denoted as meaning that is not also a concept denoted as term. Yet that there is a distin ction is undeniable.


46 We are told in PofM that the above contradiction is avoidable if we take the difference to be one that is not internal to the terms which constitute the proposition, but rather one that is external to them. The a ppeal to this distinct ion introduces a needless obscurity, but, perhaps more significantl y, it is not patently obvious that we must not speak of the concept denoted as meaning and the concept denoted as term as one and the same concept. If we were prohibited from doing this in the manner in which Russell claims that we are, we should be immediat ely confronted with a double-aspect problem, and perhaps then forced into the position of maintaining that the crucial difference between the concepts consists solely in external relations. But since there is good reason, I think, to suppose that we may speak of the concept ‘as m eaning’ and the concept ‘as term’ as the same concept, the “inextricable difficulties” into which we are led in not following Russell’s prescripti on are not such as to be otherwise inescapable. When a ‘proposition proper’31 – a proposition expressed by a declarative sentence – occurs as the logical subject of another proposition, that proposition is un asserted on Russell’s view. But if Russell is right in saying that asse rted propositions have an internal relation to truth, then we have the following difficulty. For Russell, propositions to which truth is internal may occur as the logical subject of a proposition. If so, we should be prepared to admit that truth is a constituent term of that proposition. Our difficulty here is that because the assertive fo rce of any proposition must be withdrawn once it is made a logical subject, we ca nnot hold that truth is internally related to that proposition. We have now, because of this fact, w itnessed how, in Russell’s vi ew, the “ultimate notion of assertion, given by the verb …is lost as soon as we substitute a verbal noun.” ( PoM ) In respect of the number of terms of the two complexes, ‘C aesar died’ and ‘the death of


47 Caesar’ would appear to be di stinct. This follows given that the complex expressed by the verbal noun, ‘the death of Caesar’, la cks the constituent term possessed by the proposition proper – viz. the te rm that is given by the finite form of the verb. Although the difference in its grammatical rendering is quite plain, Russell, somewhat casually, instructs us to disregard it ( PofM p.48). The reason for this, which remains somewhat inexplicit in these passages, would seem to be that a proposition occurring as a logical subject need not be rendered grammatically as the verbal noun, and so should not be so rendered. Whether the motivation springs from this observation or not, such a move, which I attempt to explain in more detail in what follows, signals a radical departure from the Begriffsschriftt On Russell’s view, we may have an unasse rted proposition of which a truth-value is a constituent – viz. where the declarative form figures in truth-functional contexts and in certain more complex propositions.32 The requirement that such a proposition is unasserted, and so distinct from an asserted proposition of which a truth-value is also a constituent, combines with st ill another requirement that ultimately leads to Russell’s difficulties in providing a satisf actory account of assertion. The latter requirement is that one and the same proposition must occur in th e antecedent of a conditional statement as that which occurs in the second premise of modus ponens The mistake here is to assume that we cannot grammatically represent the proposition qua logical subject of a proposition as a verbal noun But we have found the difference – the alleged unasserted proposition is not truly un asserted, and to explain this we need only re cognize that the difference lies in what is expressed by the grammatical form of a declarative sentence.33


48 Let us consider an earlie r assumption of Russell’s from PofM namely that the distinguishing feature of a comple x is that it is the type of entity that may bear a truthvalue and of which we may predicate ‘is true’ or ‘is false’. Because all propositions are, on Russell’s theory, complexes, propositions are bearers of truth in this sense. When we come to consider ‘the death of Caesar’, how ever, we are reluctant to say that it could have a truth-value, even though the verbal construct denotes a complex on Russell’s view. The proposition ‘Caesar died’ plainly doe s have a truth-value but it is also a complex to which truth or falsity may be predicated As the distinction between internal and external relations here intimates, there is reason prima facie to distinguish between two kinds of truth attribution in addition to distinguishing two types of complex. Russell, however, does not pursue the suggestion further. Aside from the fact that we are reluctant to say that ‘the death of Caes ar’ is the kind of en tity that could have a truth-value, we have reason to believe, as Gr iffin notes, that both ‘Caesar di ed’ and ‘the death of Caesar’ are the same complex on Russell’s view. If we suppose that the proposition containing the finite form of the principal verb, e.g. that expressed by the sent ence ‘Caesar died’, differs fr om the same proposition in which the verbal noun is substituted for the finite verb, then we effectively render modus ponens invalid. We have then at least three inter-related problems that stem from the doubleaspect problem, according to Griffin: ( a ) How can an un asserted proposition contain a truth-value (externally or otherwise), ( b ) How can an unasserted proposition ever be the logical subject of a propo sition, i.e. a term, and ( c ) the problem of pr eserving the validity of the inference schema modus ponens Recurring in Russell’s analysis is what appears to


49 be the often inexplicit claim that we are pr ohibited from represen ting the logical subject of a proposition as a verbal noun or in some comparable grammatical form. For instance, in section 52, it is said that “by transforming the verb, as it occurs in a proposition, into a verbal noun, the whole proposition can be turn ed into a single logical subject, no longer asserted, and no longer contai ning in itself truth of fals ehood. The “death of Caesar” seems to be what is asserted in “Caesar died.” In which case, the reasoning goes, “it is the death of Caesar which is true or false; and yet neither truth nor fals ity belongs to a mere logical subject.” Because truth does not belong to a logical subject, it must be related to it in some way, and this is what Russell means when he claims that truth has an external relation to it. Similarly, since truth (or falsit y) does “belong” to the proposition “Caesar died,” it is said that truth has an inte rnal relation to the proposition. The problem, as Russell sees it, is that “[there] appears to be an ultimate notion of assertion, given by the verb, which is lost as soon as we substitute a verbal noun, and is lost when the proposition in question is made the subject of some other proposition.” This has the consequence of making it impossible to refer to propositional concepts (as opposed to propositions proper ) as unasserted. What motivates Russell to make this claim is the assumption that, insofar as his theo ry of complexes is concerned, ( i ) A (‘Caesar died’) and ( ii ) —A (‘The death of Caesar’) differ in respect of the verb, ‘to die’, and so differ in respect of a term Therefore, ( i ) and ( ii ) denote distinct complexes. This is why we cannot say that they are the same; but then, as Russe ll explains the matter, it is difficult to see precisely how they differ. The problem leads us in turn, into difficulties with respect to justifying modus ponens


50 But certain insights concerning the nature of these difficulties are afforded us if we suppose that ( i ) and ( ii ) do not differ in respect of a term Here, we stand to profit in following W. E. Johnson (1921)34 and K. Wiredu (1975) in construing the verbal noun ( ii ), ‘ The death of Caesar’35, to have the same grammatical standing as th e participial form, ( ii ) —A (‘ Caesar’s being dead ’), wherein lies what Johnson calls a “latent formal element” ( Logic 1921)36. In such a case the verbal element is a constituent in both asserted and unasserted propositions but the verb contained in ( i ) is fully inflected, whereas it is in its non-finite form in (ii) We should therefore expect to find a corresponding difference in the proposition where there is a difference in the form of the verb. What unites the substantive and adjective is, in Johnson’s terminology, the “characterizing tie.” The occurren ce of the principal verb in its finite form (the finite form of ‘to be’) – ‘is’ signifies the presen ce of another relation, which Johnson calls the “ assertive tie.” This marks the addition of assertive force, but it is clear that the force is added to the already existi ng characterizing relation. Herein, then, lies the crucial difference with respect to Russell’s view of assertion: corresponding to this added syntactic element is the determination of a truth-value.37 This determination is primary in the sense that any assertion – any claim of what is so – possesses a truth-value (of either True or False). If we suppos e that the syntactic element possesses no propositional correlate, then not only do we end up with a theory of propositions bereft of an account of that difference (otherwise accounted for by the assertive tie), our foundering again upon the double-aspect problem becomes inevitable. Let us note that because the verbal elem ent is a constituent of both propositions ( i ) and ( ii ), we should not conclude that ‘ A’ and ‘—A’ are therefore logically or semantically


51 equivalent. This was the point of registering the addition of the assertive tie We may say that the propositional content (the ‘judgeable conten t’ in the language of Begriffsschrifft ) is identical in each case, but to glibly re gard the whole expressi ons ‘Caesar died’ and ‘Caesar’s being dead’ themselves as identical is to violate the Fregean tenet to avoid (at all costs) confusing th e act of assertion with what is asserted. The problem seems to be that any attempt to esca pe such a violation comes at the price of compromising the viability of modus ponens : the occurrence of the second premise contains an element (that supplied by the assertive tie) not contained in the occurrence of ‘Caesar died’ in the hypothetical proposition ‘If Caesar died, then he died on the Ides of March’, since the occurrenc e of the proposition ‘Caesar died’ in a hypothetical proposition is unasserted and so a distinct complex from the proposition ‘Caesar died’ as it occurs alone in the second premise. But this is a mistake, as we are misled along the way, out of neglect of the ear lier syntactical dis tinction, into thinking that ‘Caesar died’ as it occurs alone in the context of an argument schema must possess the element (constituent) supp lied by the judgment-stroke. Inasmuch as a proposition may be said to possess unity, a formal relation obtains between its terms. In our example, ‘ Caesar’s being dead ’, the infinitive form of the verb constitutes the latent formal element. It is in virtue of this rela tion that the proposition should be seen to be essentially unasserted rath er than asserted. The infinitive verb that is supposed to express this formal relation, then, ought not to have counted as a constituent term in Russell’s analysis. Only the principal verb in its finite form can count as a term. So we may assume that ‘Caesar died’ as it occurs in the c onditional premise of modus


52 ponens is, formally speaking, more properly re presented by the part icipial ‘Caesar’s being dead’. We might speculate that such a renderi ng of the proposition was seen by Russell, to be unnecessary, or redundant given that the assertive fo rce belonging to the sentence ‘Caesar died’ is suspended in both truth-f unctional contexts and where it occurs in subordinate clauses. This observation led Wittgenstein, in the Tractatus to eliminate the assertion sign altogether. In fact, assertive force is suspended in these contexts, but, crucially, no device was ever empl oyed to mark the transformation. Provided that these remarks are accurate in describing how Russell viewed the matter, he cannot be said to have been c onsistent: any occurren ce of ‘Caesar died’ – where the finite form of the principa l verb signifies the presence of the assertive tie – is an asserted proposition; which is to say that it expresses an a ssertion that occurs, as does any intentional act, at a specific time and place. It occurs as an ep isode in the cognitive history of a unique individual. Presumably, Russell did not wish to suggest that in speaking of modus ponens as an inference schema we ordinarily refer to the sentences which constitute its premises as particular da teable speech-acts, perhaps then relativized to specific speakers and contexts. We may say that ‘Caesar died’, as it occurs in the second premise of modus ponens is more properly construed as a proposition in Johnson’s sense, to be rendered as ‘Caesar’s being dead’, so as to make it a judgeable content. That content plainly corresponds to a proposition whose unity is only formally specified by means of the ‘char acterizing tie’. Symbolically, modus ponens is written as follows38: p q p


53 _____ Therefore, q To bring out the participial character of the propositional va riable, we may symbolize the schema as follows, inserting th e horizontal to show that it goes with the variable: ( — p — q ) ( — p ) ____ Therefore, q In any instance of this schema, the ant ecedent of the conditional premise must be identical to the proposition that occurs al one in the second premise, otherwise the inference does not carry. Russell sees this to be the main obstacle to resolving the dilemma. In the scheme above, p in the first premise is identical to its occurrence in the second premise. Both occurrences may be rendered as ‘ — p’ to which the vertical may then be added in order to s upply assertive force. Thus, when given a statement like ‘If the conversations are monitored, Nixon knows about it’, it should be expressed using the more logically perspicuous phraseology, ‘The conversations being monitored’ implies ‘Nixon’s knowing about it’. Given this interp retation, the propositional variables may be seen to be identical and so we avoid Russell’s difficulties with modus ponens Rather than construe p of the conditional premise as an asserted proposition (in this instance, ‘Conversations were monitored’), it should be construed as a participial form; the same holds for the p of the second premise. Russell’s view renders p as ‘Conversations are monitored’ even where it occurs in a hypothe tical statement. In such contexts and in


54 truth-functional contexts gene rally, assertive force is suspended. Ordinarily we do not take the proposition to have a participial form when it occu rs as the antecedent of a hypothetical proposition, let alone as a premise th at stands alone. This is unfortunate, as it gives the appearance that the proposition po ssesses a constituent term – that which Russell noted is supplied by the finite form of the verb – which it may only be said to possess with the addition of the judgment -stroke. Otherwise the proposition does not possess the term. On account of this appearance, a propositional variable may be seen to represent, at once, both a proposition as asse rted, viz. in the form of the declarative sentence – ‘Conversations are monitored’ – and a proposition whose assertive force has been withdrawn. The suspension of assertive force of a proposition is made possible in virtue of its role in more complex propositions It seems that direct inspection of how we use hypothetical statements confirms this fact But here, the double aspect issue arises. Russell’s remarks may be seen to be espe cially problematic when we consider propositional variables that stand alone. A proposition that stands alone (e.g. in the second premise of modus ponens ) is, on Russell’s view, one that possesses assertive force. Not only do we wish to be able to assert that p, we also wish to say ‘p’ is of the form of an assertion (of a declarative form). But in following Russell even this far we have effectively undermined modus ponens : just as assertive for ce has been supplied to the second premise, the assertive force is withdrawn from the antecedent of the conditional premise, and thus we no longer have identical propositions. The difficulty vanishes, however, if we construe p as it occurs alone as a participial construction. We have so far neglected the other si de of Russell’s dilemma (Griffin’s third problem, (c); Griffin 1993, 51), but I think that it will be s een that this side of the


55 dilemma is less central to the discussion’s concerns. I will nonetheless go on to make a few remarks about it, as it had exercised Russell considerably in the passages of PofM we have been considering. There are difficulties involved in assuming the two occurrences of the proposition p to be identical39: if we take the two occurrences of p in the two-premise account of modus ponens to be identical, we should have to admit that ( i ) p ; ( p q ); therefore, q could be expressed as the conditional, ( ii ) [ p ( p q )] q Russell, as Lewis Carroll did be fore him, recognized that ( i ) is logically basic : any attempt to establish the validity of the inference schema on the basis of appealing to the conditional, p q is vitiated by the infinite regre ss famously described in ‘What the Tortoise said to Achilles’. Russell’s assump tion is that the principle according to which “if the hypothesis in an implication is tr ue, it may be dropped, and the consequent asserted” is indemonstrable; but it is ne vertheless “quite vital to any kind of demonstration.” ( PofM p. 35) This gives rise to the fo llowing problem. Provided that we may be permitted to conditionalize the argument schema in the manner above ( ii ), then from the truth of p and the premise, p q, we should expect to be able to prove q – i.e. assert that q But the threat of Carroll’s regress is generated, togeth er with Russell’s conflation of asserted propositions and “complex concepts,” on the assumption that if we take the conditional ( p q ) as a license to the inference (of q from p ), then, with the truth of p we may detach to get the conclusion. Bu t we have no reason to believe that we have satisfied the student who, although he can come to recognize the truth of the conclusion and p q still cannot manage to draw the inference to q The temptation is


56 then to suggest that the conj unction of these two premises, p ( p q ), licenses the inference from p to q But here we come upon the regress. As Carroll has shown, and as Gilbert Ryle40 later reminded us, we are mistaken to think that a conditional form which justifies the inference is something that we could ever obtain. All that we have done so far is replace one conditional premise with a more complex conditional premise; and all that we could further do is repeat that pro cedure indefinitely in such a way as to get increasingly complex conditional propositions as premises. The suggestion that modus ponens is vitiated by regress arguments of this kind, given the assumption that the antecedent term of the first premise and the term of the second premise are identical, does not amount to the insurmountable difficulty it is thought to have amounted to. The solution is most easily seen given the following considerations, which have been bro ught to my attention by K. Wiredu (1960)41: modus ponens is a valid inference schema because we may justify the inference (when pressed to) by appeal to the conditional [ p ( p q )] q as the principle of the argument This is a sufficient license for inferring q from p for the conditional premise is now a tautology (and p q is not). It would be inconsiste nt to question the validity of [ p ( p q )] q because it is logically true. Thus, what a ppears to us at first to be the start of an infinite regress may be seen to stop at the conditional, [ p ( p q )] q I had earlier remarked that if we must speak in terms of complexes we should have recognized what Russell did not – that an element (or term) is supplied by the finite form of the verb. Though it is somewhat a matte r of historical speculation why Russell did not recognize the significance of this point, I suspect that one reason for this is that it was thought that a proposition o ccurring as a logical subject need not be rendered


57 grammatically as a verbal noun or a participial construction. It would se em that this belief was a consequence of the fact that a proposition which re flects the structure of a declarative sentence can do double duty for a proposition that goes either asserted or unasserted. It may be further supposed that the most plausible explanation for this phenomenon and the double aspect problems to which it has given rise is that the notion of assertion was, for Russell, primarily a psychological one. On the premises of Russell’s early realism, a proposition cannot be true merely by our ta king it to be true. Either it is the case that Caesar died or not. This seems to be the primary impetus for distinguishing between the psychological and logical sense of assertion in the way that Russell (in various periods) and the later Frege did. As it has been suggested in the foregoing reflections, what they perhaps should have said is that the ‘j udgeable content’ cannot be true merely by our thinking that it is.


58 Chapter IV: Theories of Truth Russell on the Nature of Truth and Falsehood The primary aim of this discussion will be to give some indication of the plausibility of an account of tr uth that I attempt to explicate, in its essentials, more fully in §2 of Chapter V. A feature of this account whic h distinguishes it from several others is its similarity to deflationary theories, specifi cally in its repudiation of the notion of fact qua truth-maker. The notion of fact in Russell’ s relational theory of judgment, it should become apparent, introduces a further el ement to Russell’s analysis, thus further complicating the epistemological dimensions of the theory. Rather than attempt to purge these entities from his theory, Russell endeavor s, later in his analysis of judgment, to explain away the need for a single ‘Objective’ in favor of another type of objective. The new theory of judgment (henceforth, the mu ltiple-relation theory) now requires a kind of objective that may be split up into constituent terms, which are then capable of standing in several relations to the mind. The theory is not without its difficulties, several of which are well documented and have even been argued to be irremediable.42 On the account of judgment outlined in the present chapter, whos e fuller explication appe ars later, it seems that we can do away Russellian facts – viz. complexes of terms and their interrelations – altogether and the difficulties that attend them. Escaping the implications of Russellian facts would seem prima facie to be evidence against the very idea of facts construed as entities whose semantic function are to se rve as truth-makers. It is from this consideration, in part, that th e proposed account may be seen to derive some plausibility.


59 Further, given a conception of judgment that duly accounts for the significance of the syntactical function of the “characterizing tie, ” we could avoid the traditional difficulties associated with the problem of propositional unity.43 This is a problem which, as Davidson44, plagues Russell’s multiple-relation theory just as it did the theory of propositions of his ear lier realist period. According to Russell, a relation is ‘mu ltiple’ “if the simplest propositions in which it occurs are propositions involving more than two terms (not counting the relation).” (Russell 1910, 155) Events may be construed to be the correspondents of true judgments.45 However, that there are no events corresponding to fa lse judgments may lead one to the false conclusi on that “when we judge falsely there is nothing that we are judging.” (Russell 1910, 155) It might have been suspected that such a consideration would have likely forced Russell in 1910 to abandon talk of true and fa lse objectives, but this was not in fact the case. The argument that Russell marshals agai nst his earlier theory of propositions does not, strictly speaking, amount to a repudiation of the notion of an objective altogether but rather forces Russell to scout around for a more suitable candidate for the role of objective – something that woul d effectively do the work of true and false objectives while affording a way out of the earlier di fficulties. Russell never abandons the notion. Even with the more epistemologically sophi sticated multiple-relations theory, Russell speaks of there being an “ objective ground” with respect to which judgments are determined to be true or false. It is, rath er, the Meinongian notion of objectives that he abandons. A Meinongian objective is a single structured entity to which a judgment is related. And act of judging then involves the relation of a judgment to the object of the


60 judgment – i.e. what the judgment is about. To take Russell’s example, we may say that ‘Charles I’s death in his bed’ is just another way of saying th at Charles I died in his bed, and therefore the complex ‘Charles I’s deat h in his bed’ cannot be the objective. Curiously, Russell seems not to be tempted to say that ‘Cha rles I’s death on the scaffold’ is just another way of saying that Charles I died in his bed. This is because, so we are left to assume, Charles I’s death on the scaffold was an event that actually occurred, and so accordingly the form of words ‘Charles I’s death on the scaffold’ names that event. A notable feature of this discussion in “T he Nature of Truth and Falsehood” is that the central line of argume nt begins with Russell stressi ng what must have appeared to him to be a truism, namely, that the “first point on which it is impor tant to be clear is the relation of truth and fa lsehood to the mind.” (Russell 1910) When we see the sun shining, Russell writes, “the sun itself is not ‘true’, but the judgment ‘the sun is shining’ is true”. Although we may regard such a remark to be so obvious as to be beyond dispute, in stating this observation Russell obscures a point whic h might otherwise not have escaped his notice had the point been stated di fferently. This is something that is brought to light when he later claims that the truth or falsehood of statements “can be defined in terms of the truth or falsehood of beliefs .” Further, Russell claims, “A statement is true when a person who believes it believes tr uly, and false when he believes falsely.”46 Hence, the truth and falsehood of statements is “derivative” of the truth and falsehood of beliefs. These remarks may lead us to attri bute to Russell the view that belief has some kind of constitutive relation to truth, but this would be mistaken. The realistic notion of correspondence is introduced when Russell late r claims that judging truly or falsely (as the case may be) requires some fact of the matter – some “objective ground” – in addition


61 to the judging mind. If the way in which th is passage is worded is true to Russell’s meaning, it is difficult to see how this could be his actual analysis, ev en in a preliminary sense. The claim in question is th is: when a person believes a true statement he believes it truly; and when a person believes a false statement he believes it falsely. I have reversed the order of the word ing in restating Russell’s clai m (above) because I think that this paraphrase – rather than obscure what is said – has th e advantage of showing that what is stated is, strictly sp eaking, inconsistent with the cl aim that the truth and falsehood of statements is derivative of the truth and falsehood of beliefs In fact, examining this particular passage reveals what we might be inclined to argue is a deep-seated commitment to the opposite view, namely, that the truth and falsehood of beliefs are derivative of the trut h and falsehood of statements. That this is likely to be Russell’s view in “NT&F” is due to the following implication, which the earlier paraphrase of Russell’s claim wa s intended to clarif y. In saying that a statement is true when a person who believes it believes tr uly, there is the sugge stion that what the mind is related to is something that is itself true – and thus someth ing that is apprehended as true Here it is assumed that this something is given as true independently of anyone’s coming to have the belief expressed by the statement – i.e. concerning the content of the belief. Attending even the more sophisticated multiple-relation theory is the idea, which is all-pervasive throughout Russell’s philosophy (postPofM ) is the idea that something makes a statement true. The assumption that something is true (sentence, belief, proposition) is true independently of our beli eving it true combines frequently with the notion of correspondence. In what follows I will suggest th at this notion of correspondence does not in itself lead to a view which would force one to admit truth-


62 makers. In fact, a serviceable notion of co rrespondence may be easily accommodated in an epistemic theory of truth, as may be seen in the logical views of John Dewey. Truth and Judgment: The Frege-Dewey Connection The debate on the nature of tr uth between Russell and Dewey was published in a series of articles between 1939 and 40. It grew out of earlier work by both philosophers. Russell’s view developed early on out of the The Problems of Philosophy (1912) and from even earlier work in logic and the foundations of mathematics. Dewey’s views were, in large part, developed from 1909 – the year in which he wrote A Short Catechism Concerning Truth to 1938, the year in which he wrote his Logic: The Theory of Inquiry My discussion will focus primarily on an article entitled ‘Propositions, WarrantedAssertibility, and Truth’, first published in 1941 in The Journal of Philosophy This article was a response to Russell’ s criticism of 1940, much of which appears in his article “Dewey’s New Logic”.47 Russell speaks of the difference between his and Dewey’s views as consisting of the former’s concern with assertions about particular matters of fact, and the latter as concerned with hypotheses and theories.48 This is, in fact, a distortion of Dewey’s view; it is an idea not an assertion or senten ce, that is the basic unit in terms of which truth is defined. Dewey’s use of the term ‘idea’ is inte nded to draw our attention to an aspect of ideas which the anti-psychologistic bias of early analytic philosophy has at times obscured; that is, the idea construed as a “possible signifi cance.” This is not merely a psychological significance, but an epistemological – and we may assume – even a semantical one. An idea thus conceived is a ba sic element in terms of which we arrive at


63 a warranted assertion, the latt er being applicable to both assertions about particular matters of fact and hypotheses and theories Dewey holds that the “presence of an idea be by way of an existential operation,” and that th is distinct “presence” is what distinguishes his theory from other theories of truth. T hough the claim is most likely correct (leaving aside the precise meaning of “existential operations”), there is a further point that distinguishes Dewey’s theory from the others and this may be called semantical. That difference reveals, in part, the source of both their disagreement about truth and the nature of propositions, and what I will later ar gue is an error on Russell’s part. The error is a consequence of what Russell takes to be the correspondence relation and its terms. His theory introduces a view of corresponde nce according to which when a particular linguistic entity – a sentence – is true, it is in some way related to a particular extralinguistic entity, namely, a fact The correspondence of a ju dgment, or derivatively, a sentence, to the world is more complicated wh en the sentence is false, but we need not enter into these details at this point. The virtue s of this view are furt her discussed later, as well as what I consider to be its shor tcomings. I wish now to give only a rough characterization of Russell’s view for the purposes of contrasting it with Dewey’s conception of truth, a view that might be rega rded as a correspondence theory of truth of a radically different kind in essentials. Dewey’s view is one of correspondence insofar as the relation (of correspondence) obtains, not between sentence (or proposition) and fact, but between idea and fact The relation between the mind an d the world is thus one that relates, on the one hand, the basic conceptual components of judgment and the cognitive operations by which judgments are formed to the ‘existential’ conditions, and which, on the other hand, gives rise to them. I do not thi nk it is necessary to adhere too closely to


64 Dewey’s terminology in order to elucidate the contrast between his view and Russell’s in this specific connection; but it will be necessary to proceed with caution. Dewey was correct to say (Dewey 1941, 168) that he and Russell “cannot understand each other unless im portant differences between [them] are brought out and borne in mind.”49 I believe that these differences ar e most easily brought to light by first comparing aspects of Dewey’s epistemology to those of Frege’s early logical writings. Specifically, I have in mind their rather curi ously similar conceptions of propositions as they appear, respectively, in the above cited work and the Begriffsschrift (1879). Once we have ascertained precisely what Dewey means by the term ‘proposition’, then we are in a position, so I argue, to see wh erein precisely lay the disa greement between Russell and Dewey as concerns the nature of truth.50 Frege first makes the distinction between grasping a thought and judging a thought to be true in his article ‘On Sense and Reference’51, though the source of the distinction may be found in the earlier Begriffsschrift .52 In the later article, Frege describes judging a thought to be true as a transition from thought to judgment whereby one “advances from the thought to the truth-value.” What the truth-value is ontologically speaking, is not a question I wish to pursue here. For the purpose of comparison it will do merely to note that Frege took truth to be a primitive concept. This difference has no bearing on the specific comparison I wish to draw. In fact, it is perhaps fair to say that in this specific connection, Frege and Dewey were separated more by differences of terminology than of substance. Dewey, it may be argued, recognized in his Logic what is tantamount to Frege’s distinction insofar as th ere he stresses the nece ssity of observing a fundamental logical distinction between a proposition and an assertion. This is a


65 distinction that is made somewh at in much of Russell’s work53 which we find either neglected or altogether absent from the writ ings of several student s of Frege’s philosophy of language.54 The point, however, is more fully explicated in Dewey’s “Propositions, Warranted-Assertibility &Truth” of 1941. This distinction is the basis on which Dewey is able to maintain the connection between tr uth and inquiry – twin notions that are divorced in the often platonistically orie nted views of both Russell and Frege. The concept of truth is thus primarily epistemi c on Dewey’s view. Deductive inference is one among several operations of the mind whose essential function is to further rational inquiry. The notion of inquiry alluded to here is considerably broade r than what Russell or Frege may mean by that term. In fact, Dewey often speaks of the logic of inquiry – a locution apt to baffle most analytic philo sophers and logician s (including Frege). However, as concerns the analysis of the structure of judgment, it would seem that Frege and Dewey have joined together in an ‘unholy alliance’ against Russell.55 We may turn now to Frege’s account of supposition in order to make the comparison explicit. Suppositions, as well as t houghts, figure into a stage of inquiry at which ‘propositions’ figure for Dewey, given the sense Dewey accords to the term. Propositions, for Dewey are what I earlier referred to as ‘t he components of judgments’, and, as such, are categorially distinct from judgments in the same way, e.g., an individual is distinct from a sentence in classical trut h-functional logic. Th ey are introduced and combined in accordance with certain operations to construct judgments, and are therefore useful insofar as they figure into judgments w hose aim is to be warrantedly assertible. When we turn to Frege’s analysis of thoughts ( Gedanken ) – specifically, with a view to their role as suppositions – we see the formal resemblance. The transition from thought to


66 judgment wherein, as Frege states, one “advanc es from the thought to the truth-value” (Frege 1892, 31) is the very transition described by Dewey when he claims that the proposition56 occupies a functional role in the construction of a judgment – that is, prior to its being discharged as an assertion If we bear in mind that a truth-value for Frege is possessed, not by a thought in transition but rather by a judgment as completed then the aspect under which the thought may be seen as a component of j udgment should become clear. Dewey sometimes refers to propositions as “instrumentalities,” sometimes as “ideas.” In an idea’s functional role in the process which leads to the end of scientific inquiry, they are said to have the property of efficacy This is liable to be misinterpreted – and has been, along with Dewey’s distinctly pr agmatic remark that “the only alternative to ascribing to some propositions self-sufficien t, self-possessed, and self-evident truth is a theory which finds the test and mark of truth in consequences of some sort…”57 The misinterpretation is evident in almost all of Russell’s discussions of Dewey’s epistemology.58 Having indicated some points of conve rgence between Frege and Dewey, we may now turn to the debate between Russell and Dewey. Dewey attributes a view to Russell that, to quote Dewey, “seems to me to be th e most adequate foundation yet provided for complete skepticism.”59 The view under consideration holds generally that atomic propositions afford a kind of epistemic gr ound for the complex propositions of which they are constituents in a way not unlike the way in which Hume’s ‘impressions of sense’ serve to ground ideas60 of varying complexity. Empirical knowledge for Russell, then consists of justified, true belief as base d on (true) propositions, where the truth of complex propositions is determined on the basi s of their constituent (atomic) propositions


67 in accordance with the truth-functional rules of formation. However, the conditions under which the truth of the basic, atomic propositi ons is determined are mysterious. As Dewey points out, basic (atomic) propositions, taken as the objects of empirical belief, are determined to be true by virtue of their co rrespondence to certain facts. But these basic propositions are also said to be the causal effects of particular “causa l antecedents”, i.e., particular worldly events. Thus, if Dewey is correct, Russell holds th at basic propositions are somehow caused ; yet, these propositions are also capable of being used to refer to the events which cause them. But as it was earli er mentioned, empirical knowledge consists of true, justified belief where truth is defined as corresponde nce with the fact(s). Thus, Russell’s view assumes that, if the cause is a fact, or set of facts, the cause of a proposition, then, is identical to its truth-maker.61 On Dewey’s view, this conception might be amended in a way that gives it at least an appearance of plausibility; he might argue that there may be a way in which we obtain knowledge by experimental means given this conception of propositions. This would proceed by testing the truth of propositions by supplementing a given set of propositions (in Russell’s sense) with ot her propositions. But such an accretion of propositions becomes regular and systematic on Dewey’s view since they are given, or produced, by controlled observation. Whic h propositions are accepted and which are rejected depends on what Dewey calls the “c ooperation of inferent ial and observational subject-matters”62 and, perhaps, on various constraints that determine which propositions cohere with an existing set of propositions. But such a method of ‘testing’ amounts to a criterion by which truth is determined by “conse quences of some sort,” and as such must have appeared to Russell to compound the difficulties associated with traditional


68 coherence theories with those associated with James’ criterion, viz. the thesis according to which beliefs are true insofar as they are useful It may be noted here that Russell simply substitutes proposition for belief in discussing Dewey’s view. Returning to Russell’s assumption that a proposition is identical to its truthmaker63, if we follow Russell in rejecting Dewey’ s pragmatic criterion, we are forced to believe that Russell is proposing what Dewey cal ls a semantic version of the “doctrine of pre-established harmony.” This is “the episte mological miracle … for the doctrine states that a proposition is true when it conforms to that which is not known save through itself.”64 The alternative to this discredited view is, of course, skepticism – a consequence of any theory of truth which defines truth as an epistemically inscrutable relation. This notion of “pre-established ha rmony” which Dewey believes underlies Russell’s view is, in fact, a consequence of Russell’s definition of the correspondence relation – a “Relation at La rge” Such a relation, Dewey cl aims, is a relation “without specification or analysis.”65 Dewey’s remark here, and elsewhere, may be understood in light of remarks made in my earlier disc ussion comparing Dewey’s epistemology with Frege’s early logical views. Dewey’s pragmatic criterion suggests, in part, a way out of traditional forms of skepticism as well as a way of properly identifying the terms of the correspondence relation.66 It is the “idea,” or equivalent ly, the “proposition” in Dewey’s sense, that corresponds and has the essential ca usal relation to events in the world. These events, however, are not facts in the sense that they are en tities or states of affairs pictured by sentences or beliefs, or by, e .g., the structured propositions of Russell’s logical atomism. Rather, they are merely environmental factors which enter into complex causal relations with what Dewey calls “hum an factors.” The proposition, conceived as


69 an instrumental agency, may be said to corr espond with some extra-mental event, and it is by means of this correspondence that a warranted assertion is formed.67 In setting up the correspondence relation between an asser tion and an extra-mental event, Russell thereby (simultaneously) estab lishes an ontology of facts. This is a consequence, in large part, of attempting to fit sentences onto the world. Such an attempt, it should be recognized, makes the logical error of assu ming that the correspondence of propositions (in Dewey’s sense – or ‘thoughts’ in Frege’s sense) to extramental reality is something that again corresponds with reality – with the facts. On this point, I think Dewey’s remark is instructive: “the pragmatist …begins with a theory about judgments and meanings of which the theory of truth is a corollary.”68 This mind-independent entity, as is evid ent from what Russell states elsewhere (Russell 1910, 1918) is sometimes identified with a Meinong’s noti on of objective. The various problems attending the role of objectives in Russell’s earlier theory of judgment deserves fuller treatment than can be given here; but the point that I wish to make presently is that Russell appears in the passage quoted to have construed statements as having the same, or nearly th e same, ontological standing as objectives .69 Witness Russell: “[It] is difficult to abandon the view th at, in some way, the truth or falsehood of a judgment depends upon the presence or absence of a ‘corresponding entity of some sort” (Russell 1910, 152) We might compare this with his later view of 1940: “When we embark upon an inquiry we assume that the propositions about which we are inquiring are either true or false” (Russell 1940, 403) Ontological consider ations notwithstanding, the implication of having identified facts a nd objectives with statements in this way should be recognized to be logica l in nature. That this is in fact a piece of confusion is a


70 point that may be brought to light in considering more clos ely, not so much the question of the mind’s relation to trut h, but the question of what sort of thing, or things, bear the properties of truth and falsehood. In pursuing this question, however, it will be necessary to disambiguate what is commonly take n to fall under the meaning of the term “statement”; and this is perhaps best achieve d by first inquiring into the question of what it is that we believe when we believe a stat ement or proposition to be true. Russell, we shall see, is not always sensitive to this que stion. Let us consider the following claim: “When, for example we see the sun shining, th e sun itself is not ‘true’, but the judgment ‘the sun is shining’ is true”70 (Russell 1910, 148) Russell explai ns in a footnote that the words ‘belief’ and ‘judgment’ are used here synonymously, but the reader is left with little clue as to what preci sely Russell understands the re lation between ‘judgment’ and ‘statement’ to be.


71 Chapter V: Austin and Strawson, a nd the Vagaries of Correspondence The redundancy view of truth states gene rally that the truth-predicate ‘is true’ does not express an analyzable semantic pr operty; and therefore it is semantically superfluous. To say, then, that a sentence or pr oposition is true is, in a certain sense, to assert nothing other than the se ntence itself. According to Aus tin (1950) this is incorrect. The predicate ‘is true’ is not redundant, if for no other reason than that it is analyzable – not as expressing a property of sentences, but as expressing a property of their use, i.e. insofar as indicative sentences are used to re fer to a “historical si tuation.” We “approach the term,” in Austin’s phrase, “cap and categories in hand.” In what follows I argue that Austin’s vi ew is deficient in a way that Strawson’s ‘performative-redundancy’ view is not; that the semantic categories with which we approach the truth-predicate are inadequate for an analysis of the c oncept of truth. I argue further that the defects in Austin’s view are a consequence of failing to distinguish between two concepts of truth. This distin ction becomes clear once another, closely related distinction is made, namely, that wh ich lies specifically be tween the logical (and syntactic) character of assertion and the semantic characte r of declarative sentences, or sentence-uses. In predicating ‘is true’ of a senten ce or speech-epis ode, we do not quite assert what is already asserted in the way Austin en visages. Austin does, in fact, claim that we “refer” to a given state of affairs in predic ating ‘is true’ of a given sentence, conceding


72 that this referring itself produces another sentence, which then, as asserted involves producing a truth-value on grounds distinct from those on which the initial assertion itself is judged true. While Austin is correct in saying that we make another, numerically distinct, statement (taken as a speech-episode ) in this way, and in doing so assert another truth he does not explain that the process by wh ich this happens is a consequence of the application of the two (aforementioned) con cepts of truth. A cl arification of this distinction, and what it amount s to, is attempted later. Austin’s semantic theory of truth is intended to vindicate the correspondence theory of truth. An outline of the semantic aspect of his definition appears before the definition when he states that there are prim arily two kinds of semantic convention in any form of referential discourse: ( i ) Demonstrative conventions under which certain linguistic expressions are related to “historic situations.” ( ii ) Descriptive conventions under which certain li nguistic expressions are related to types of situations. The definition is as follows: A statement is said to be true when the historic state of affairs to which it is correlated by the demonstrative conventions (the one to which it ‘refers’) is of a type with which the sentence used in making it is correlated by the descriptive conventions. There is, I think, considerable me rit to Austin’s view, a fact th at is not always explicitly recognized by his opponents.71 There is, for example, something obviously correct about


73 Austin’s notion that “a statement is made and its making is an histor ical event,” and that, construed as such, a statement is so mething like a primary truth-bearer.72 Less obvious is the contention that the idea of a primary truth-bearer implies a primary/secondary distinction. Something like this distinction, which often takes sentence-tokens and assertive utterances to be ‘secondary’ trut h-bearers, is a comm onplace in the philosophy of language. Given this, it is comm only taken for granted that the concept of truth appertaining to the distinc tion of truth-bearers is the same concept. It is seldom held that to this primary/secondary distinction th ere is a corresponding distinction between concepts of truth. In the second half of the pape r I indicate why I th ink such a further distinction is needed. In light of this distincti on, which is explained more fully in the second half of the discussion, we shall see that the connect ion between assertion and the concept of truth applicable to assertive utterances goes unappreciated by Austin and his critics. However, the programmatic aim of his analysis – of providing a “purified” correspondence relation – may readily accomm odate this explanatory amendment. There is a special context in which sentences may make use of the truthpredicate.73 Austin gives the following examples as perfectly legitimate uses: (a) “The third sentence on p.5 of his speech is quite false.” (b) “His closing words were very true.” In (a), ‘sentence’, and in (b), ‘words’, refer not to a proposition, fact or truth-value, but to “the sentence as used by a certain person on a certain occasion .”74


74 Austin’s definition of the corresponde nce relation naturally requires a truthmaker, as any correspondence view must. He is somewhat wary of the implications of this demand, the effect of which is the erecting of a realm of facts ( qua truth-makers) – “populating the world with li nguistic Doppleganger.” (Str awson 1950) However, while conceding the need to designate an historic state of affairs as the truth-maker for a statement, Austin is cautious not to lose sight of the important if not truisitic, point that “we can only describe that state of affairs in words .”75 It becomes evident soon that Austin did not exercise sufficient caution in this respect, for he goes on to say: It takes two to make a truth. Hence (obviously) there can be no criterion of truth in the sens e of some feature detectable in the statement itself which will reveal whether it is true or false. Hence, too, a statement cannot without absurdity refer to itself.”76 (my emphasis) As an elementary criticism of traditional cohe rentist theories of truth, this has sometimes been, and perhaps ought to be, taken in ph ilosophical seriousness. However, it would seem that Austin, and several other corresponde nce theorists, have ta ken the criticism so seriously as to let it obscure the significance of the observation that we can only describe a state of affairs in words. The kind of view recommended by this observation, on the one hand, and the sort of view implied by the quoted passage, on the ot her, are ordinarily taken by truth-theorists to be antithetical. Unlike Strawson, Austin sometimes fluctuates between them. Thus, once he articulates an in sight which we may rightly consider to be the touchstone of any deflationary theory of truth he then moves back toward a correspondence view. How, then, Austin asks, is ‘the statement that S is true’ (read: TstST) different from ‘the statement that S’ (read: tstS)? Suppose that,


75 If Mr. Q writes on a notice board ‘Mr. W is a burglar’, then a trial is held to decide whether Mr. Q’s published statement that Mr. W is a burglar is a libel: finding Mr. Q’s published statement was true (in substance and in fact). Thereupon a second trial is hel d, to decide whether Mr. W is a burglar, in which Mr. Q’ s statement is no longer under consideration.77 Austin goes on to claim that a second trial is necessary to decide the separate issue of Mr. W’s guilt or non-guilt. A redundancy theorist such as Strawson would believe that such a trial is in fact unnecessary, for the evidence le ading to the verdict of the one trial is the same evidence as that in the ot her. Strawson, presumably, woul d then wish to argue that the point of the example generalizes so as to apply to the descriptive predicate, ‘is true’: thus the grounds for the semantic verdict for TstST are the same as that for tst More precisely, the truth or falsehood of the one imp lies, and is implied by, that of the other. This piece of reasoning, Austin believes, is the consequence of confusing identity conditions of propositions (as synonymous with sentence meanings) with those sentences themselves. The alleged confusion over sentential a nd propositional identity attends a related misconception concerning assertion. This miscon ception is merely hinted at in Austin’s paper, in which he claims that the meaning of the sentence ‘that Fa is false’ is often confused with that of the negation of ‘ Fa ’. Both negation and assertion refer “directly to the world” not to “statements about the world; ” they are, he avers, “on a level.” This remark is at once odd and instructive, for th e assertion and negati on (denial) are seen not to be “on a level” the moment we consider how the truth or falsity of a statement arises in the first place. In illustrating his point, Austin asks: how are the assertions ‘He is not at


76 home’ and ‘It is false that he is home’ the same in a context in which “no one has said he is at home?”78 Interestingly, it is this consideration that leads Strawson to depart from the earlier proto-redundancy view of Ramsey (1927) and to assign a performatory function to the various predicative uses of ‘is true’ – e.g., ‘it is true that’ (preceding a statement), ‘that is true’ (referring to what is sa id), and so on. Thus, the semantically superfluous predicate ‘is true’ is not completely superfluous, as it does something over and above its semantic role. It may confirm (in certain subs tantival phrases), assent, grant, or concede (in proleptic uses). Austin’s principal objection to all of this is that the performative interpretation of the truth predicate stresses a significant aspect of linguistic meaning to such an extent as to almost ignore the significance of the seman tic content of the predicate. This, Austin claims, amounts to ignoring the fact th at the performative aspect is merely one aspect among others. “To say that you are a cuckold may be to insult you,” Austin remarks, “but it is also and at the same time to make a statement which is true or false.”79 There is, however, one aspect of stat ement making whose importance must have been ignored altogether by Austin and to so me extent Strawson, namely, the assertoric function of statements. It is in respect of this aspect of linguistic meaning that the significance of the logical concept of truth alluded to earlier is brought to light. I suggested that the logical concep t is distinct from the semantic concept; but if Austin has somehow given a vague impression that this is so, it can only be by failing to observe that distinction in the first place. Witness Austin: “There can be no criterion of truth in the sense of some feature detectable in the statem ent itself which will reveal whether it is true or false. Hence, too, a statement cannot without absurdity refer to itself .” (Austin 1950,


77 31) Here, one cannot be blamed for suspecti ng that it is Austin’s undue fixation on a statement’s referring capacity and the associated demonstrative conventions of the act of referring that prevents him fr om recognizing that a sentence may say of itself that it is true. I go on later to argue that it must As it stands, the notion that a statement may be ‘saying of itself’ awaits further clarification.80 It is worth noting that one would be mistaken to claim that Austin never entertai ned the suggestion that a statement can say of itself that it is true. In fact, we see some groping for an explanation of a similar point when Austin asks of sentence-uses “whether there is not some use of ‘is true’ that is primary or some generic name for that which at bottom we are always saying ‘is true’”81. Which, if any, of these expressions is to be taken “ au pied de la lettre ?”82 Austin then quips, “[In] philosophy the foot of the letter is the foot of the ladde r.” His question leads to some more fruitful speculation about the concept of truth, but it is not pursued in the manner in which Strawson – I believe, rightly – pursues it. The result of Austin’s efforts is thus a study of the demons trative conventions by which st atements refer, betraying a somewhat fruitless preoccupation with the stric tly semantic import of the truth predicate. It is in Strawson’s discussion of the na ture of facts where the redundancy theory prevails. The suggestion is that the words “‘fact ’, ‘situation’, and ‘sta te of affairs’ have, like the words ‘statement’ and ‘true’ themselv es, a certain type of word-world-relating discourse (the informative) built in to them.”83 Thus “it would be futile” to elucidate any segment of referential discourse in whic h these several terms naturally occur by analyzing the terms themselves, or in terms of one another, for they “contain the problem, not its solution.”84


78 Implicit in these remarks may be the key to understanding precisely why Austin’s view must be found to be untenable85. The point is reflected in Austin’s telling remark that “it may further be questioned whether ever y ‘statement’ does aim to be true at all.”86 Though it may make no sense to say that statements really aim to be true, statements may be said, in a sense, to be tr ue solely in virtue of their normal assertoric function (what Frege refers to as force )87. The linguistic conventions govern ing assertion are not wholly semantic; there is also a logical and syntactic al dimension to which we have paid little attention in this discussion up to now.88 We may now wish to understand preci sely what Strawson means by the claim “that facts and statements have a type of “word-world-rela ting discourse (the informative) built in to them.”89 Why do Austin and Straws on both make the point that ‘ p ’ and ‘ p is true’ do not mean the same, and that this difference emerges once one considers the conditions under which the questio n of truth and falsity arise? What if no one asserted ‘that p ’? Are we to believe that ‘ p is true’ means ‘that p ’ in a context in which the assertion that p was never made, or is unlikely to ever be made? Both are in agreement that such questions reveal something defective in the redundancy conception, for which they offer their respective remedies We may attempt to elucidate the nature of the problem by appealing to Strawson’s in sights. The terms ‘statement’ and ‘fact’ contain the problem. It is worth noting that Ramsey (1927) came near to making the same point earlier with his suggestion th at ‘truth’ is not a problem that could not be dissolved through a proper analysis of the internal structure of judgment We may understand Ramsey’s remark by attending to the logico-syntactical point concerning standard cases of assertion discussed below.


79 To assert that a is F is, of course, to make a claim. It is to say this is thus so A statement of the form a is F is true in the sense that it contains in its making the truthvalue ‘True’. A truth-value is, in a special sense, thereby conferred upon any judgment or statement that states this is so. This is a consequence of the declarative statement’s logical form. The grammatical rendering of such a judgment typically links subject and predicate terms by means of introducing a fini te verb – in this case, ‘is’. Thus, the function of ‘is’ in ‘is F’ reflects an act of judgment, viz., of judging something to be the case. By ‘subtracting’ from the sentence ‘ a is F’ the finite verb ‘is’, we suspend what W. E. Johnson90 calls the ‘assertive-tie’ of judgment, and so we are left with an expression syntactically unlike the statement ‘ a is F ’ and unlike the corresponding judgment. The result of this suspension is a form of wo rds that may be expressed grammatically in participial form, < a’s being F > Note, then, that when a trut h-value is conferred upon this form (the declarative form of a sentence re presents the formed judgment) the result – a statement – does not name or refer to a ‘state of affairs’ or ‘fact’ in the way the resulting sentence is alleged to name or refer.91 In recognition of this di stinction, we may say that such a statement is an instantiation of a concept of truth92 in virtue of its form – in virtue of what I have called, after Wiredu ( 1975), an occurrence of truth in its primary sense. We might say, to be more precise, that su ch a statement contains an occurrence of a primary concept of truth. None of this is to deny the obvious truth that situations in the world often give rise to judgments, as they obviously do. The mistake, rather, is to think that the truth of a resulting judgment obtains in vi rtue of a relation that obtains ex post facto between it and a given fact, or state of affairs. Frege, nearly sixty years earlier, s uggested that there are


80 problems with the idea of taking facts to be tr uth-makers of judgment s. It is worth noting that the mistake is supported by the widespread belief that (1) there is no primary sense of truth-value for statements to have, and (2) when we inquire as to th e truth of a statement we thereby inquire not into but into the truth expressed by the declarative sentence ‘a is F.’ If, however, we hold an Austin-s tyle correspondence theory, we commit a category mistake in supposing that a sentence corresponds to a state of affairs.93 We have already seen that there are good reasons on intuitive grounds not to suppose that sentences as logically and syntactically complete expressions, name, refer, or correspond, to anything whatsoever. In the primary sense a statement asserted just is true insofar as it pos sesses a truth value (of True or False). I wish now to consider a potential objec tion to the idea that assertions contain a claim to truth, or possess what I have calle d a primary truth-value. The objection might go as follows: “On your view even a sentence like ‘2 + 2 = 5’ is true in virtue of its merely being asserted. Surely, you do not wish to say that .”94 That is, if a statement is true in the primary sense, in virtue of its sy ntactical form as you have explained it, then you must admit that any falsehood is also true in virtue of its being asserted, which is absurd. The force of the objection seems to rest on the mistake of confusing mention with use in the following way. When one presents th e falsehood ‘2 + 2 = 5’ as a clear counterexample to the present view, it is not clear what exactly it is that one is presenting as false. Is one presenting for our consider ation ‘2 + 2 = 5’ as a sentence i.e., as a claim or is one presenting the judgeable content of that se ntence? In any case, it is plain that ‘2 + 2 = 5’, however construed, is being proposed as an obvious falsehood; but presumably the one who proposes it as such knows this in advance. If this is th e case, however, then one


81 is in an important sense mentioning the statement that ‘2 + 2 = 5 is true’ as a falsehood – as a statement (in Austin’s sense, a speech-e pisode) which one has already determined to be false after having evaluated its content from a previous point of view. There is now under consideration a point of view which has al ready been determined to be false. This fact is obscured in part because the intende d counter-example takes the grammatical form of a question – viz. “Is the statement ‘2 + 2 =5’ true on the present view?” The covert judgment may be more easily detected in the question, “Are you prepared to admit that ‘2 + 2 = 5’ is true?” But the claim has bee n, somewhat covertly, evaluated from some previous point of view. Let us suppose that our opponent does not know in advance that ‘2 + 2 = 5’ is false, but rather is asking the question in all sincerity. It would sound odd were she to say, ‘Could you really mean to say that ‘2 + 2 = 5’ is in some sense true ?’, as if to ask, ‘Are you sure that you wish to commit yourself to claiming something so absurd?’ If one did not know in advance that ‘2 + 2 = 5’ is a fa lsehood (and did not mean to suggest that in being offered as a kind of counter-example to the present view we would then have to countenance the absurdity that “‘2 + 2 = 5’ is true”), then, of course one would not have raised the question in the first place. The question arises with something non -declarative in form. It is incorrect to think that when one asks the question ‘Is 2 + 2 = 5 true (or false )?’, one would first present the idea that 2 + 2 = 5 is true as if it were false – let alone as a claim – then proceed to ask whether it is true. Rather, what one actually does first is to consider whether 2 + 2 = 5, and in doing so one considers not ‘2 + 2 = 5’ as a statement but as a question. What is truth-evaluable in the primary sense is a content that has yet to receive a truth-value and theref ore non-declarative in form. But the question


82 thus arises: if this accurately reflects linguistic practice, w hy would the grammatical form of the expression not correspond to the appropr iate mood of entertainment – i.e. of its being a question? Such an act of co nsideration does, in fact, involve a content for whose expression a specific syntactical construction is to be use d. For this consideration the proper formulation of the ini tial question, then, is ‘Does 2 + 2 = 5?’ – or, equivalently, ‘ two and two’s being five ’. In the context of inquiry this is what is in question. Accordingly, it is determined to be the case that two and two is not five, in which case the resulting (primary) truth-value is ‘False’. Note that this w ould expressed by the declarative sentence ‘2 + 2 5’, whereas the sentence “‘2 + 2 = 5’ is false” contains two truth-values, one of which is prim ary the other which is secondary. These reflections may put several of Austin ’s earlier remarks into perspective. It was mentioned earlier that Austin believes that we “refer” to a given state of affairs in predicating ‘is true’ to a give n sentence. It is by now perhap s more clear why this cannot be the case. The difference that ‘is true’ ma kes to a plain assertion is, in part, what Strawson takes it to be – a pe rformative one; but it is not merely performative, a fact to which Austin is sensitive in a way Strawson is not. Austin’s observation is that the act of ‘referring’ to another speech-episode itself entails another statement which, as asserted, receives a truth-value on grounds distinct from those by which we judge true the initial assertion. What is wrong with this is not only the idea that we so mehow refer to other statements (though it is true that we may be said to concur, or agree, with them), but the idea that a bare assertion obtains a secondary truth-value in the construction of a judgment. We do not, in fact, form judgments in the way that Austin suggests. If we did, we should expect that we normally go about f itting judgments onto the world. It is then a


83 small step to the doctrine that the truth of a judgment is an epistemically inscrutable relation that obtains between it and a given state of affairs. But what Austin seems to be right about given all this – a nd what Strawson is perhaps mistaken about – is that we do, in fact, arrive at anothe r judgment in the act of predicating ‘is true’ to a statement.95 That judgment, itself a result of applying the tr uth predicate, is a comment upon a prior judgment wherein occurs the primary concep t of truth. We do not arrive at another judgment, however, by fitting it onto the historic al situation which is supposed to make it true. Rather, in cases of empirical belief, we “fit” a concept onto the world, where this concept in part constitutes our judgment. It may be noted that a judgment, in term s of the concept of truth, because it is the product of a concept discharging its referring function, is never strictly identical with another judgment. This applies also to st atements, speech-episodes, and propositions. Thus, if statements do not refer to states of affairs and cannot refer to themselves (given restrictions on identity), then a statement p as assented to by someone must be accounted for in a way that makes reference to the conditions under which p was initially made.96 The details of such an operation cannot be disc ussed here. The point I wish to make bears on the debate between Austin and Strawson so mewhat obliquely, but it is worth stating. When Austin says that the act of ‘referring’ to another speech-epi sode itself entails another statement, which, as asserted, receives a truth-value on gr ounds separate from those by which we judge true th e initial assertion, he is only partially correct. Were his view amended in such a way as to incor porate the distinction between primary and secondary concepts of truth, it would ha ve perhaps forestalled objections from deflationary quarters while still resp ecting our correspondence intuitions.


84 I wish now to consider a potential objecti on to the view that truth is an epistemic property, an objection that might be made to the idea, in partic ular, that assertions contain a claim to truth (in some non-psychological sense) or a primary truth-attribution. The objection might go as follows: “On your view a sentence like ‘2 + 2 = 5’ is true in virtue of its merely being asserted. Surely, you do not wish to say that .” That is, if a statement is true in the primary sense, in virtue of its syntactical form, then we must admit that any falsehood is true in virtue of its being asserted, which is absurd. Or, the objection may be put as follows: if your view is correct, then an ything may be said to be true “in virtue of using a declarative sentence (e.g. ‘2 + 2 = 5’). ” On the present view, we have deviated from orthodoxy in maintaining that a declarative sentence cannot be used non-assertively in any legitimate sense. The force of the objection seems to depend on the mistake of reading into the epistemic view a kind of realist notion of correspondence of truth. This reading is accomplished by confusing mention with use When one presents to us the falsehood, say, ‘2 + 2 = 5’, as a clear counter-exa mple to the present view, I think that we are entitled to ask what is it exactly that one is proposing, thereby elucidating what is intended by the critic to be taken as the primar y truth-bearer. That is we are entitled to ask whether one is presenting for our considerati on ‘2 + 2 = 5’ as a sentence, that is, as a claim which is expressed in the indicative mood. It is plain that ‘2 + 2 = 5’ is being proposed as an obvious falsehood, and presumably the one who proposes it as such knows in advance that it is a falsehood; but if this is the case, then the one who is presenting the objection is, in a sense, mentioning ‘2 + 2 = 5 is true’ as a falsehood. Suppose that the critic who is offering ‘2 + 2 = 5’ for our consideration does not know in advance that it is false, but is aski ng the question in all si ncerity. It would sound


85 odd were one to say, ‘Could you really mean to say that ‘2 + 2 = 5’ is in some sense true ?’ (as if to ask, ‘Are you sure that you wish to commit yourself to claiming something so absurd?’ while asking the question sin cerely). Of course, if one did not know in advance that ‘2 + 2 = 5’ is a falsehood, then one would not have known to raise the question in the first place. This is because one would not know the sentence ‘2 + 2 = 5’ is a counterexample. It is difficult to imagine a situation in which someone who asks the question sincerely would first present the idea that 2 + 2 = 5 is true as if it were false – that is declaratively – then proceed to ask the question ‘Is it true that 2 + 2 = 5’? This is, of course, very unlikely, for we normally ask whether something is true or false by presenting questions where the propositional contents of questions occur in the interrogative mood rather than the indicat ive. That is, what one actually does first in contexts of inquiry is to consider whether 2 + 2 = 5, and in doing so one considers not ‘2 + 2 = 5’ as a statement but as a question, or perhaps even as a kind of hypothesis. The use of ‘whether’, and generally expressions of i ndirect discourse, signi fies the asking of a question. The question thus aris es: if this is what one nor mally does in the context of inquiry, then why would one not us e an expression like “whether p ”97, or even express p in the interrogative mood? We have already seen that there is an a ppropriate grammatical expression corresponding to the act of consider ing. Such an act of consideration involves a judgeable content for whose expression we have employed a specific syntactical construction. If this is th e appropriate consideration, the proper formulation of the proponent’s initial question, then, is ‘Does 2 + 2 = 5?’ – or, when considered as a truthapt propositional content – ‘ two and two’s being five ’. This is what is in question in normal contexts of inquiry. Thus it is dete rmined to be the cas e that two and two is not


86 five, and thus determined to be the case that ‘2 + 2 = 5’ is false. We may again look to Frege’s early conceptual notation to express all this in a more perspicuous manner: ( i ) ‘two and two’s being five’ [ 2 +2 = 5 ] ( ii ) ‘two and two is five’ [2 + 2 = 5 ]; or, in the case of a negative judgment: ‘ is false’ [ 2 + 2 = 5 ] The objection, therefore, depends upon omitting the transition from (i) to (ii) and upon the assumption that in considering whether ‘2 + 2 = 5’, one starts with (ii), One may reply to the ini tial objection by pointing out th at, on the present view, one does not omit the transition from (i) to (ii) – an omission of which the critic herself is guilty. But if one does argue against the present view in the manner we are now considering, then one is in effect eliding over this transition. The critic’s reasoning may be represented similarly as follows: ( 2 ) ‘two and two is five’ 2 + 2 = 5 ( 3 ) ‘ is true’ ‘2 + 2 = 5 ’ is true In this instance what one starts with is th e claim ‘2 + 2 = 5’, which implies that one ipso facto admits to having asserted that 2 + 2 = 5. But if this is the case, then we have already committed to the absurdity that we were alleged to have committed at the outset when presented with the init ial objection. There is yet a nother error committed by the here by the critic: one has not only omitted the tr ansition from (i) to (ii), but in so doing erroneously takes the epistemic view in quest ion to be doing the same. But if we hold the critic to observe the necessary di stinction and the transition from ( i ) to ( ii ) which depends


87 on it, then she will see that we are no more committed than she is to holding true the sentence ‘2 + 2 = 5’. We may appeal to the dis tinction between ‘use’ and ‘m ention’ thus construed in order to show more precisely at what point the confusion se ts in with respect to the potential objection raised a few pages back (p.43). The point of the objection is to force the epistemic theorist of truth to admit th at 2 + 2 = 5 is true, thereby showing the epistemic view to be in error. The ar gument implicit in the objection is a reductio But, as is obvious given the needed distinctions, the admitted absurdity is of the form [ T2( T1p)] which must be taken as a mention of the previous asserti on, an instantiation of ( T1p). The present view does not co mmit one to having to admit as true ‘2 + 2 = 5’ unless one is committed asserting ‘2 + 2 = 5’ as the former depends essentially on the first. But what reason do we have to expect another to bed committed to asserting ‘2 + 2 = 5’? If one asserts that 2 + 2 = 5, then, of course, one is committed to countenancing such an absurdity; but in such a case one already has evaluated ‘2 and 2’s being 5’ as true insofar as there is a primary occurrence of truth. But in this case, even an opponent to the present view has, by the same token, committed to the same absurdity – for we should wish to say simply that the sum of 2 and 2 is not 5. Presumably this is not how one the objection is really intended to be taken. Rather what one means to say is something like this: accord ing to the epistemic view under consideration, are you claiming that 2 +2 = 5? – to which the epistemic theorist may respond with propriety, ‘no’. Our adversary then concludes that I am not claiming that ‘2 + 2 = 5’ is true. But as was seen, talking in this way implies that one is using ‘is true’ as predicative of a


88 sentence and this involves, on the present view, a secondary occurrence of truth. Although this is what the standard use of the truth predicate implies, presumably this is not what one really means in asking whether 2 + 2 = 5 is true. What is now becoming clear is that one is here confusing primary and secondary occurrences, as if they are one and the same or as if they may occur in any context (saving meaning). The primary occurrence of truth consists the making of an assertion, which consists, as has already been shown, in the instituting of the assertive tie given the judgeable content as the primary truth bearer This is, given a simple syntactical consideration, not identical to the secondary occurrence of truth so-called because its corresponding truth-bearer is not a judgeable content, but rather a complete thought as expressed by a sentence – i.e. as a (compl ete) thought. Thus, to the question: ‘Are you claiming that the sentence ‘2 + 2 = 5’ is true?’ the answer is ‘no’, fo r I have not asserted ‘2 + 2 = 5’ as a sentence. Because I have not asserted ‘2 + 2 = 5’, and thus do not have before me a complete thought to which the trut h predicate can be applied, the question of whether the sentence ‘2 + 2 = 5’ is true or false simply does not arise. We might imagine the initial objection to be reformulated as follows. Suppose, then, that you have asserted that 2 + 2 = 5; on your own view, you have no choice but to admit that this (the hypothesized assertion) is true given that you have asserted the sentence, and here the kind of epistemic view that you are advocating mu st admit that the question as to its truth or falsehood does arise. But the response would be that the ques tion as to its truth or falsehood arises given that I have been forced to assert that the sentence ‘2 + 2 = 5’ is now false. But here, as the critic will be quick to point out, in declaring the sentence to be false I have


89 contradicted my own view. In th is instance, what would have escaped him is that there is no reason to make such an assertion, insofar as it is admitted as something for which the question of truth and falsity aris es at all. Of course, in nor mal contexts, there is no reason for anyone to make such an asser tion. One may proceed to say of it (i.e. the sentence itself) that it is false which would enjoin applying th e truth predicate to a sentence, where the sentence is now the truth-bearer – not the propositional content carried by it. Thus the predicating of ‘is true’ to the sent ence as a truth-bearer necessarily involves a secondary occurrence of truth; and it may be se en from this instance that every secondary occurrence depends, in a logical sense, upon a primary occurrence of truth. That this relation is logical may be seen by considering examples in which the use of the truth predicate in its primary role i.e. where ‘is true’ is predicated of something syntactically in complete, e.g. That p is true, where the occurrence of ‘that’ transforms ‘p’ to a judgeable content, or, in cases relevant to se mantic ascent, ‘what she said is true’. But this is not how one need answer the objections Rather, at this point it is to be observed that that the critic of the epistemic view has committed the error of identifying the primary with the secondary occurrence of tr uth. Given the distinc tions set out, this identification necessarily commits her to furt her conflating the trut h-bearers appropriate to each occurrence. Once this has been brought to the attention of the critic, she may again try to reformulate her question in order that this du al conflation may be avoided. In doing this, however, the initial objection is rendered powerless. She would have to resort to asking: ‘Given the distinction we have raised, I am asking: does not your view of assertion commit you to admitting that 2 + 2 = 5 is true by virtue of its simply being asserted, as if I am asking you ‘Is it the case that 2 + 2 = 5? In other words, does this


90 view commit one to asserting ‘2 + 2 = 5’ (w hich we have expressed more formally as ( T1p))? The response would be, again, ‘no’; for although I may consider the idea that 2 + 2 = 5 – of two and two’s being five – I have still not asserted ‘2 + 2 = 5’. In so considering a judgeable conten t I am no more required to assert 2 + 2 = 5 – where ‘=’ carries with it the assertoric force of the finite form, ‘is’ – than is a rational person who denies all that we have said so far in support of the view under consideration or subscribes to a radically different theory of truth. Of course, the sentence has been presented to me for my consideration, but this does not thereby make the claim ‘2 + 2 = 5’ mine in the sense that I am rationally committed to what is asserted by having asserted it. At this point it would be helpful to in troduce a distinction cl osely allied to, and entailed by, the one discussed thus fa r. This concerns differences in point of view What is typically said to be either true or fa lse is a sentence or statement. However, it is a truth-evaluable content that recei ves a truth-value, and this co ntent is not itself a sentence or a statement, since a sentence or statemen t is what one ends up with once the content receives a truth-value. The point of this remark, and much of the previous discussion, is to show that truth is applicable to two diffe rent things. It remains to be shown that to these two kinds of expression there correspond two concepts of trut h-value. It should become clear that my employment of the use/ mention distinction, in the way that I wish to construe it, derives from the distincti on between primary and secondary concepts of truth-value. In the present chapter I attempt an elucidation of this connection. First, it would be useful to see exactly how the primar y/secondary distinction applies to different kinds of linguistic expression and how it may be appealed to in order to address the


91 potential objection raised a few pages back. I wish to consider how in particular, the distinction applies to certain forms of indirect discour se. The sentence “Snow is white” contains an occurrence of the primary concep t of truth (what I have been calling, simply, a primary occurrence of truth). The expression “Snow’s being white” cannot be said to be true (or false) since it is not a claim. It may be made into a claim, however. For it to be made into a claim, it must be supplied w ith assertoric force (see figure on p.45). The addition of assertoric force here leads to th e syntactical (and logica l) transformation of a judgeable content (which is inco mplete) to an assertion, or ju dgment (which is complete). Thus, not only do we have complete and incomplete expressions, the kind of truth-value assignable to incomplete expressions is distin ct from the kind of truth-value assignable to complete expressions. What is incomplete is appropriately expressed by the participial form of a sentence, whereas what is comple te is expressed by a d eclarative sentence. We may get at what is peculiarly epistemic about this transformation by considering what assertoric fo rce adds to an unsaturated j udgeable content in the context of rational inquiry. It is cl ear that supplying assertoric force amounts to nothing other than the supplying of a claim from the point of view of the assertor. This point was made by Frege when he defined assertion as an “a dvancing toward the True .” What appears not to have overly concerned Frege in this defini tion of assertion is the implication that a claim is something that results from what someone does Claims are, of course, not advanced ex nihilo It would be even more to the point to say that advancing claims is something that people do. Moreover, to th e extent that people act on, and otherwise depend in various ways on the claims of ot hers, we typically understand people to be involved in undertaking various commitments when making claims. In making any claim,


92 one normally undertakes a commitment, e.g. to withdraw a claim when it is shown to be untrue, to justify a claim when the claim is justly questioned, or to be held responsible when a claim is not withdrawn and it is show n to be untrue. We may take a claim to involve any of these normative commitments to constitute, in large part, a commitment to truth. We may now return to the question of wh at assertoric force adds to judgeable contents of propositions. We may think of an act of assertion under normal circumstances as involving commitment to these judgeable co ntents. Assertoric for ce, in particular, is what contributes this commitment. The sense of “pr oposition” here may be clarified in the following way: a proposition is a structured entity that does not differ essentially from a sentence. We may think of it some sense as what is expressed by a sentence so long as what is expressed is not regarded as that which carries meaning independently of the sentence which expresses it. It is assumed th at sentences are not me re concatenations of signs to which propositions (as expressed by those sentences) are then thought somehow to supply meaning. Propositions, in this sense, may be taken as being synonymous with statements – as essentially complete expressions. It is perhaps difficult to understand commitment at th e level of ab straction at which we have been discussing the issue so fa r. Commitment to something being true is a normative phenomenon that must be understood re lative to an individual or a group of individuals whose linguistic beha vior is rule-governed in a variety of complex ways. To maintain that commitment to the truth of a pa rticular judgeable cont ent of a proposition is what assertoric force adds to that content is to maintain that someone has committed to it and that someone has supplied th e requisite assertoric force.


93 A further implication that I wish to draw out is that, in terms of what has been explained so far (see discussion on Austin and Strawson) corresponding to the commitment to the judgeable contents of a proposition – to, let us say, the idea of something – is a primary occurrence of truth; in fact, there may be reason to think that truth is partially definable in terms of commitment.98 Since we typically speak of someone being committed to something, we s hould therefore take into account that the act of commitment is what someone within some particular epistemic (and linguistic context) does; that is, an act of commitment is relative in a particular way to a person in an epistemic context. A person may, of course, be committed to holding a proposition (thereby being committed to its truth) over her lifetime.99 A person may be committed to a proposition for a relatively short period of time. Because of this – i.e. because commitments are allowed to vary during the cour se of a person’s lifetime – it is necessary not to identify a point of view with a pe rson. It would be more accurate to say that although commitment the truth may vary over a period of time, it cannot vary over a single point of view, as a point of view just is an act of committing to the truth of some proposition.100 A commitment to truth may then be taken to be in a certain narrow sense relative to a point of view in a way to be distinguished from a commitment’s being relative to a person. A point of view is something that a person may have at time t0 but does not have at t1 ; or may have at time t0, not have at t1 and have again at t2. This variability of points of view (relative to a person) is a basic feature of discursive reasoning, in general, and is reflected in our pr actice of assertion at every level of social discourse. Nonetheless, commitment to tr uth amidst this potential plurality (and variablility) of points of view is possible.


94 The concept of point of view arises out of the primary/secondary distinction in the following way. Given that commitment is rela tive to point of view in the way just explained, it is crucial to recognize that, ex cept in cases in which one mentions a claim made by another, the making of a claim necessarily implies cotemporaneous commitment to it, and this cannot be completely divorced from its truth. Therefore when we say that truth is a warranted assertion we mean that it is a claim from (or relative to) a point of view that carries a commitment cotemporaneous to the context of the sincere utterance of the assertion.101 We may get in another way at what is peculiarly epistemic about the transformation by returning to some points previously made about other philosophers’ views on assertion. In an important respect, philosophers’ understandi ng of the nature of assertion depends on certain conceptions of obj ectivity, or certain c onceptions that have been formed on the basis of common-sensi cal notions of objectivity. These notions manifest themselves in more sophisticated phi losophical analysis as one form or another of realism. We may bring such manifestations into focus in considering some traditional ways assertion has been thought to be re lated to the notion of the proposition. The connection between these notions and asser tion comes into focus in considering a commonly made distinction be tween proposition and asserti on: a proposition is said to have the structure of the sentence which e xpresses it, and an asse rtion is the act of sincerely uttering the sentence. It has been suggested that th is sort of view has certain difficulties, and that a view that construe s a proposition to have the structure of something incomplete has certain advantages.102


95 On such a view, given that the truthvalue of a proposition remains what it is whether or not anyone happens to be asserting it, it is either true or untrue. We may be said to apprehend its truth or fa lsity, or said to be acquainted with it, in which case, if we sincerely utter the sentence that expresses it, we may be said to have asserted a true sentence. But, as is well known, even if one were to consciously believe and have justification for believing the proposition expr essed by any given sentence, one may still not have its truth (where the tr uth of a proposition is justifie d inferentially). The problems with this are well documented, especially as c oncerns traditional defi nitions of truth in terms of warranted assertibilit y. On a realist theory of propos itions, the tendency in some quarters has been to simply resign to skepticism and to maintain that the traditional model of knowledge does not model what we actually do as epistemic agents. There just is no principled means by which it can be determined that one can say she is in possession of a true proposition, whether or not the justific ation condition is satisfied. Because certain strong forms of skepticism imply the impossib ility of knowledge, at least as it has been traditionally defined, for this reason skepti cism is thought to amount to a kind of refutation of the traditional model. Of c ourse, others have concluded from this implication that skepticism (in the strong, global form) is thereby refuted, not the traditional model. But it would seem that th e philosophers who do come to this (latter) conclusion, do not fully recognize what it mean s: the implications of recognizing that a skeptical strategy is itself self-refuting are such that no theory of knowledge according to which truth is defined as a mind-independe nt semantic property that outruns our cognitive capacities is really possible. The idea is this: so long as truth is taken to be an evidence-transcendent property of propositions, ye t it is taken to be a necessary condition

PAGE 100

96 of any claim to knowledge, there will always be cases of justified be lief that are possibly false. The innumerable elabor ations on, and modifications of the traditional Justified, True Belief, analysis of knowledge that fo llowed after Edmund Gettier’s article, “Is Knowledge Justified True Belief?” attests to th is fact and to the vague sense one gets of their futility. But this would seem avoidable if the distinction between kinds of truthbearer were consistently maintained in current theorizing about truth. However, a consistent application of the distinction is necessary, though not sufficient for an account of truth that provides a fuller analysis of the conceptual relati ons between truth and judgment. This brings us to the epistemologi cal point: the advancing of a claim (what is the same as advancing toward the True for Frege) is necessarily an occurrence of a primary truth-value. As it was mentioned a fe w pages back, such a claim is a point of view within some context. Therefore, an occurrence of a primary truth-value is a point of view.103 The suggestion is this: because we have made the logical (syntactical) distinction between primary and secondary concepts of tr uth-value, and because an occurrence of a secondary truth-value is necessa rily distinct from an occurrence of a primary truth-value (assuming the judgeable content is identical in both occurrences), there is necessarily a corresponding difference in point of view – the idea being that ev ery unique assertion entails a unique point of view. A vertical stroke is added to p so that we have p, an occurrence of a primary concept of truth. A point of vi ew attaches to the primary o ccurrence, and we may indicate this occurrence by writing ‘T’ for the value ‘true’ and by us ing a subscript to indicate identity of point of view – ‘T1’.

PAGE 101

97 Note that, in accordance with Frege’s c oncept script, we may symbolize the truthfunction of negation as ‘’ in such a way that does not involve asserting anything. In this way it is clear that the negation stroke is a functional element in Frege’s system, whereas the assertion sign is not. The addition of the vertical -stroke to the content-stroke supplies the assertoric force (which was earlier characterized as the ‘assertive tie’) to the judgeable content, p, converting the content to an assertion. It is useful to observe that, in Frege’s notation, it is not clea r how one could add the vertical -stroke to an assertion once it has been formed out of both the content a nd vertical stroke, alt hough there is a certain sense in which the supplying of assertoric force to an assertion is a commonplace in the context of normal linguistic practice. Cases in which we supply assertoric force to an assertion just are statements like “’Snow is wh ite’ is true,” or “It is true that snow is white.” Here, the assertoric force occurs twice over, as is indicated by the fact that “is” occurs twice: we are saying that the asserti on or judgment, “Snow is white” is, in fact, true. In recognizing, then, that assertoric force occurs twic e over in this way, we are simply admitting that the concept of truth is being predicated of a truth-bearer that is not incomplete. This concept is what we have been calling the secondary concept. Though Frege’s symbolism does not display how it is pos sible to supply asse rtoric force to an assertion. Also, in terms of Frege’s notation, the secondary concept cannot be expressed: p becomes p with the addition of assertoric for ce, but it is difficult to see how we may symbolize adding assertoric force to the expression, p. This would be redundant in Frege’s system and it is not surprising that we would run into di fficulties were we to

PAGE 102

98 articifially produce a device within that system that would express the idea of a secondary concept of truth. Just as there are two distinct expressions to which truth is applicable, there are two concepts of truth-value. And the rele vance of the distinction to grammatical constructions in general is that an occurrence of the secondary concep t is predicative in a way that an occurrence of a primary truth-value is not. It is apparent that if we admit that every truth comes in the in the form of a truthclaim, and that every truth-claim entails someone’s making a claim, then we are in the position of also having to admit that there ar e necessarily two different points of view involved. For example, there are two points of view in “’Snow is white’ is true”, or “It is true that Snow is white,” as there are two diffe rent occurrences of truth in “It is true that Snow is white” (as there are in “What he said is true.” The on e that occurs in the sentence “Snow is white,” which is primary, and the one that occurs in “It is true that snow is white,” which is secondary. Having observed th is, to deny that there are necessarily two different points of view involved in “It is true that snow is white” would be to deny that any truth must come in the form of a truth-claim. Even realists do not deny this What they do deny is the idea that truth could ever be an episte mic property, a denial which, in turn, implies the denial that any claim to tr uth could ever actually be true. For certain realists this idea must be reconciled ultimately with the idea, which is usually accepted, that it is people who make judgments and w ho therefore construct and use sentences for the task of making judgments. A concept of truth that does not contain the idea of its desirability in these terms would seem to be empty.

PAGE 103

99 Chapter VI: Judgment and Propositions: some traditional problems revisited Having indicated that Russell’s theo ry remains unsatisfactory on account of having left the problem of predication unres olved, I endeavor now to spell out more clearly the reasons why I think this is so. The following is a more detailed statement of Russell’s multiple relations theory: The theory of judgment which I am advocating is, that judgment is not a dual relation of the mind to a single objective, but a multiple relation of the mind to the various other terms with which the judgment is concerned. Thus if I judge that A loves B, that is not a relation of me to ‘A’s love for B’ but a relation between me and A and love and B. If it were a relation of me to ‘A’s love for B’, it would be impossible unless there were such a thing as ‘A’s love for B’, i.e. unless A loved B, i.e. unless the judgment were true; but in fact false judgments are possible. When the judgment is taken as a relation between A and love and B, the mere fact that the judgme nt occurs does not involve any relation between its objects A and love and B; thus the possibility of false judgment is fully allowed for. When the judgment is true, A loves B; thus in this case there is a relation between the objects of the judgment. (Russell 1910, 155) One is tempted perhaps to say that there just is no thing called Desdemona’s infidelity. The implication seems to be th at the statement ‘Othello believes that Desdemona is unfaithful’ has been made but has been later determined to be false. Given this determination, it must therefore be a fact that Desdemona is not unfaithful. This seems to describe one tendency in our ordina ry ways of thinking about facts and factstating discourse. If is an accurate descrip tion, then it is natural to think that it is

PAGE 104

100 ultimately responsible for the hypostasization propositions. This seems to be, in large part, a consequence of eliding the role contributed by the prim ary truth-attribution in the construction of a proposition. On Russell’s multiple-relation theory, the issue that concerns us in the case of ‘Othello believe s in the unfaithfulness of Desdemona’ is that there can be no relation of the mind to the proposition as designated by the noun phrase ‘that Desdemona is unfaithful’. This follows because there is no such thing as Desdemona’s infidelity. But the analysis may be seen to be incorrect insofar as Russell is taking the judgment ‘Othello believes that De sdemona is unfaithful’ to be in possession of a secondary truth-attribution.104 We should now examine the reasons in s upport of the traditional conception of propositions – a conception that we may fa irly assume includes Russell’s multiplerelation theory.105 To escape the difficulty earlier mentioned, Russell proposes that in the case of false beliefs, the mind is not related to a fact ( qua truth-maker) but rather to the constituents of what woul d have been the proposition were it true. In Russell’s example, Othello’s belief relates Othello to Desdemona and infidelity. On Russell’s analysis, it becomes possible to distinguish between the truth and falsehood of judgments, which is an improvement over the views contained in PofM As A. N. Prior notes, however, the move to paraphrase away ‘Othello believes Desdemona is unfaithful’ to get instead ‘Othello ascribes infidelity to Desdemona ’ amounts to a false economy. (Prior 1971) Only a “too one-eyed concentration” could ma ke plausible an analysis which explains away the abstract object expressed by the noun phrase ‘that Desdemona is unfaithful’ in order that we are left instead with the uninstantiated abstract object ‘Infidelity’.106

PAGE 105

101 Nicholas Griffin, in a recent study, write s that Russell “sought several different solutions to what is called “the problem of complexes” but the solutions had the unwelcome consequence of producing problems elsewhere. There is a slight irony in Griffin’s discussion of the di alectical background out of whic h the problem of complexes emerged. The revolt against monism, which sought to place in its stead a pluralist metaphysics, involved its own difficulties Yet Russell could not whole-heartedly embrace monism for the reason that it would inter alia “preclude the possibility of judgment.” (Griffin 2001, 161) Griffin notes that the multiple-relation theory of judgment purports to explain away proposit ions, just as the theory of descriptions purports to explain away non-existent ent ities referred to by certain de noting phrases referents of denoting phrases. The virtue of the multiple-rel ation theory consists in its elimination of the chief difficulty for a theory of propositions conceived from the realist’s point of view, namely, the possibility of false propositions. But it would seem that whatever virtues we find in the technique leading to these solutions is mitigated by that technique’s failure to provide a satisfactory solution to the double-aspect problem in its various and sometimes subtle manifestations.107 It is clear that the double-asp ect problem still arises for the multiple-relation theory as it had for Russe ll’s earlier theory of propositions of 1903. Following an objection raised in Geach ( 1957) and more recently Davidson (2001), Griffin observes that the di stinction between relations qua meanings and relations qua terms becomes all the more necessary on the new theory. The distinction between relations parallels that co ncerning concepts, generally: ‘one’ .e.g. is a concept as meaning when it occurs as an adjective; ‘one’ is a concept as term when occurring as a substantive. Russell seems to believe that the distinction is applicab le on the assumption

PAGE 106

102 that relations are a kind of c oncept. Whether the distinction is any more needed on the new theory, we may note that it is not suffi cient (as will become more clear in due course) to explain how Russell could have pr ovided a theory of propositions that would not have led to equally deep and vexing pr oblems in the articulation of a satisfactory theory of judgment. In the judgment, e.g., ‘Charles I died in his bed’, a pers on stands in the relation of belief to Charles I, his bed, and dying. Wher e the constituent terms of the judgment are related in the right way, inter se they form a complex in virtue of which the judgment is made true. But since there is no complex in this instance – fo r the terms are not so related – there is nothing to make the judgment true. Bu t in the case of the belief that ‘Charles I died on the scaffold’ the constituent terms ar e related to one anothe r as expressed by the judgment that Charles I died on the scaffold. There is the complex that we would name ‘The death of Charles I on the scaffold’, and so the judgment is true. The mind stands, as it does in the case of false judgments, in rela tion to each of the constituent terms, but unlike the false judgment that Charles I died in his bed, Charles I and the scaffold are, in fact, related to the term ‘dying’ in such a way as to form a complex. A point raised by several of Russell’s critics is that on this account the mind must stand in relation to both the concept as term and the concept as meaning (i.e. the relation as relating ); and although the need for this distinction does not arise with respect to propositions, since they are gone, it nevertheless arises now with respect to complexes. In his critical discussion of Russell, Davidson (2001) admits that there may be some deep truth in the ‘Wittgensteinian thought’ that is motivating Russell’s new theory.108 Since there just are no propositions, the problem of propositional unity – viz. of explaining what gives a

PAGE 107

103 proposition its unity – disappears. But the eff ect of this move is that now the same problem, or something very much like it, reappears under a new guise, under the domain of the mind: there is a shift in emphasis from an analysis of propositions conceived as mind-independent entities to an analysis of judgment as a mental act109. But such a shift signals a significant step toward a solu tion to the problem of propositional unity. Whatever difficulties we may have multiplied in the process, we may credit Russell with having relocated the problem to its proper domain. Davidson takes the new theory to have the further implication that if what unifies a sentence is “no part or aspect” of the sentence itself ( satz an sich ), then we must be committed to a conception of propositions acco rding to which sentences amount merely to “strings of names,” and therefore must deny the predicative role of verbs. However, as it is suggested in Chapter 1, there is no reas on to think that rele gating the problem of propositional unity,110 as it has been traditionally con ceived, to a theory of judgment entails this denial. However, the multiple-rela tion theory, which had been seen as a solution to the problem of propositional unit y, fails insofar as it encounters the doubleaspect problem in a new form. The repeat ed confrontation with the double-aspect problem, from about 1903 to 1912, and up to th e final version of the multiple-relation theory as it appears in Theory of Knowledge (1914), should give some indication that only a conception of assertion as primarily logical can afford an escape from potential double-aspect problems. This is a claim that both Davidson and Griffin seem to come close to making, but neither goes so far as to attribute the source of double-aspect problems to Russell’s not possessing such a conception of assertion.111

PAGE 108

104 Griffin provides an account of one of Ru ssell’s earlier attempts to come to terms with the double-aspect problem, locating the solution in an unpublished paper entitled “On Functions” (1904). 112 On the treatment given there, complexes were definable in terms of constituents – the concept of which was taken as primitive – and a mode of combination by which constituents form complexes. But the mode of combination is “not itself one of the constituents of the complex” (Russell 1904); otherwise it would count as a constituent to be combined with other cons tituents of the complex, and we would then have to specify a mode of combination which would combine the first mode of combination with the constituents of the complex, and so on ad infinitum The mode of combination, then, while an entity, is not an entity belonging to the complex which it forms. Thus, Russell appears to have avoided both the threat of a Bradley-type regress argument on one hand (as concerns modes of combination), and on the other, the selfcontradiction that would have otherwise come about in not taking a mode of combination to be an entity. It is not clear exactly what ontolog ical significance a mode of combination is accorded on this view. But it is clear that introducing the idea was important to avoid both difficulties. A mode of combination must be capable of being made the logical subject of some other pr oposition, for in denying that it could be a logical subject, one thereby implies that it is. This difficulty is thus parallel to those Russell encountered in 1903.113 In the case of functions, Russell saw the need to account for the contradiction generally as follows: given a function /ƒ’ / it must be allowed that /ƒ / may be referred to as a constituent of a proposition, /ƒ’ /. Here we are to take / satisfies the function /ƒ’ // as the mode of combinati on. Russell’s difficulties now consist in accounting for how it is a mode of combination can sometimes be a constituent

PAGE 109

105 of a complex and sometimes an entity not occurring in the complex of which it is the mode.114 With respect to the manner in which the double-aspect problem pertains to verbs, Russell is led to ente rtain a curious solution. In PofM Russell considers the predicate ‘differs from’ and its corresponding singular term ‘di fference’ and arrives at the following conclusion: Yet these constituents, thus placed side by side, do not reconstitute the proposition. Th e difference which occurs in the proposition actually relates A and B, whereas the difference after analysis is a notion which has no connection with A and B …The verb, when used as a verb, embodies the unity of the proposition, and is thus distinguishable from the verb considered as a term, though I do not know how to give a precise account of the nature of the distincti on (Russell 1903, 49) Griffin notes that Russell’s solution to this problem i nvolves construing the satisfaction relation as a concept that “cha nges with the function satisfied”. (Russell 1904) We have briefly discussed several as pects of Russell’s phi losophy that tend in one way or another to run afoul of double-as pect problems, but it is his later multiplerelations theory of judgment that I wish to discuss in more detail, as the double-aspect problem reappears there in a subtler form. It would also appear that an understanding of how that problem is manifested in the theory would give us sufficient reason to question certain of Russell’s assumptions as to th e nature of the rela tion between truth and judgment. We must therefore return to and further elaborate on the central point of the previous chapter. Let us recall that, on Russell’s earlier theo ry of propositions, if we are to ask what is asserted in the proposition ‘Caesar died’, we would have to say that it is the ‘The death

PAGE 110

106 of Caesar is asserted’, or simply, ‘Caesar ’s death’. Thus it would seem obvious that ‘Caesar’s death’ is here what is capable of being true or fa lse; but it is equally obvious that truth and falsity cannot belong to a logical subject. As it was earlier suggested, we are entitled to ask how Russell arrives at the conclusion that ‘Caesar’s death’ is true Let us recall the distinction raised earlie r in “Frege and Ru ssell on Assertion”: (3) ‘ Caesar’s being dead ’ (or, what is the same, ‘ Caesar’s death ’) and (4) ‘ Caesar died ’ If we take the logical subject to be an incomplete entity, duly registering the difference corresponding to the grammatical distinction, then we ought to conclude that ‘Caesar’s death’ is not the sort of thing that could be true or false. I suggest that what warrants the claim that ‘Caesar’s death’ is true is also what supports the claim on the later multiplerelation theory that the mind cannot ex hypothesi bear any relation to ‘false complexes’, and is therefore what ultimately motivated Russell to introduce the multiple-relations theory. The element that I refer to is obs cured, in part, by the ease with which we sometimes use participial forms and noun phr ases interchangeably. There is nothing problematic in doing this from a strictly grammatical poin t of view. But the transition from, e.g. ‘Desdemona’s being unfaithful’ to ‘D esdemona’s infidelity’ is perhaps liable to confer upon the latter expression a semblance of its having an assertive element that it does not in fact possess. For Russell, the senten ce ‘Desdemona is unfaithful’ is given as an example of a judgment expressing a fals ehood; but there is no such thing as the

PAGE 111

107 complex to which the judgment refers. Russell’s solution is, as we have seen, to explain away the fictitious complex by maintaini ng that the mind is related not to any single ‘objective’ as had been previous ly supposed, but to the several constituents of the judgment. Of course, there is no such thing as ‘Desdemona’s infidelity’, but there is a strong temptation to forget that this is itself a claim that is either being advanced presently or (presumably) had been advanced at an earlier time. I think an interesting question suggests itself, viz. whence does such a temptation originate? Precisely what about Russell’s later theory is responsible for it? The answer to this question seems to be that the temptation arises out of assuming that the judgment in question is in possession of a truthvalue as given by a secondary truth-attribution (or truthvaluation). This assumption is made, however, without Russell’s first having acknowledged that such a judgment as ‘Desdemona is unfaithful’ must have first received a primary truth-attribution in its construction as a judgment and this truth-attribution must have involved a transformation from a j udgeable content to a judgment. Assuming the judgment to be in possession of a secondary truth-attribution without acknowledgement of the role contribut ed by the primary truth-attribution in the formation of the judgment, moreover, presents the act of judgment as destitute of any element of subjectivity.115 The presentation of such a judgm ent (as expressing a fact) thus reinforces the idea that the truth of the judgment obtains independently of the person making the judgment; but more problematically, it reinforces at the same time the idea that the person judging has no c ognitive access at all to the truth of the judgment. The latter is a consequence of the former, and though this may be seen as undesirable, even for the semantic realist – for it introduces truth-skepticism – denying the consequence

PAGE 112

108 would come at too high cost – viz. of de nying that the truth of a judgment obtains independently of our thinking it true. This would be, therefore, to trade off skepticism for a deep form of subjectivism. We are led to the conclusion by way of a faulty inference, however. Because Russell has elid ed the role contributed by th e primary truth-attribution in the judgment ‘Desdemona is unfaithful’, th e judgment is accorded whatever semantic significance the secondary truth-at tribution might surreptitiously afford it. And therefore, Russell wishes to say now, as this move has now justified him, that it must simply be a fact ( qua truth-maker) that Desdemona is not unfaithful. The appeal to the notion of fact here arises out of the assumption that it is the case that ‘Desdemona is unfaithful’ independent of anyone having judged it to be so where ‘independent of’ has the sense of, conceptually, an perhaps temporally, ‘prior to’. From the standpoint of an epistemic conception of truth (e.g. a theory that define s truth as warranted a ssertibility), the notion of fact arises out of a real ist tendency in our thinking; but for such a conception the assumption is unwarranted, as we have no r ecourse to facts as Russell conceives them. On a view of truth as warranted assertibilit y, we may explain the realist tendency in our thinking as follows: by the very use of the word s ‘there is no such thing as ‘Desdemona’s infidelity’ we assume the truth of the judgment ‘Desdemona is not unfaithful’ is as if the judgment has already been made, and this is understandable given that the role that is otherwise contributed by the primary truth-attribution is not discharged. This role is never discharged is simply because there is no such role to be played in the case of a true judgment; and the reason why there is no role for true judgments is that, for Russell, to admit otherwise would be to admit that the tr uth of a judgment, and derivatively that of a sentence, does not obtain given the way the world is:

PAGE 113

109 ….[It] is plain, also, that the truth or falsehood of a given judgment depends in no way upon the person judging, but solely upon the facts about which he judges (Russell 1910, 149) Of course, we may insist that epistemic accounts along these lines are incorrect and so, given the extent to which it is entrenched in our intuitions, ther e is no good reason to forfeit the realist notion of a fact. But, I will suggest, that the insistence on the idea that we must preserve the independence of mi nd from world seems to be the mechanism which operates very strongly in Russell’s correspondence theory and the mechanism by which the theory of meaning is insulated from epistemological analysis. As Michael Dummett has said in this connection, “Although we no longer accept the correspondence theory, we remain realist au fond ; we retain in our thinking a fundamentally realist conception of truth.” (Dummett 1959, 14) Altho ugh this only applies partly to Russell (Russell had remained a correspondence theori st throughout his life too), it accurately characterizes the source of many difficulties that arise in association with realism and its relation to judgment. We ought to recognize that this is a mech anism that is at work in the earliest phase of Russell’s development and is one which is in large part responsible for the double-aspect problems which we’ve discusse d. As Griffin notes, Russell was led, in response to these problems, to distinguish between a proposition and a propositional concept, where the former designates the complex /Caesar died/ and the latter designates the complex /the death of Caesar/. But the di stinction cannot get us out of double-aspect problems, since, as Griffin notes, a propositi onal concept may be the logical subject of another proposition; but it remains that a proposition proper may also be the logical subject of another propositi on, and this has been shown to follow on pain of self-

PAGE 114

110 contradiction. But as Griffin further points out propositions are genera lly of interest to logicians; and we should be permitted to say, borrowing Griffin’s example, / Caesar died / and / Caesar was assassinated / are both subjects of the proposition // Caesar was assassinated / implies / Caesar died // .116 As observed in the previous ch apter, there is a difficulty with such a formulation as the one above. It was remarked that no grammatical transformation occurs precisely where it is needed. It should be stressed that the absence of the needed grammatical transformation conceals a point of logical significance, signali ng the presence of a certain confusion. We may also note that Russell’s introduction of the notion of a propositional concept in the attempt to explain the diffe rence between what is expressed by /Caesar died/ and /the death of Caesar / runs into Bradley-type re gress argument. No matter how many new terms are introduced, the question re mains: in virtue of what does a concept possess its dual aspect? If there is nothing about the nature of the c oncept itself that can reconcile the idea that a complex may differ with respect to having radically different properties with the idea that they express the same thing, then no device will be able to do so. Mark Sainsbury’s view117 of Russell’s difficulties is less charitable than Griffin’s., the difference with respect to verbs led Russe ll “to characterize the predicative role as one which suffices for truth”, and “plainly this is absurd.” (Sainsbury 1979, 21) Sainsbury goes on to claim: “the temp tation is responsible for his saying, in PofM that in a logical sense ‘only true propositions are asserted’ and for his holding that what is puzzling about belief is how there can be false belief.” At this point we may credit

PAGE 115

111 Russell with at least having been tempted in th e right direction. It is understandable to think absurd Russell’s claim that the predicativ e role of the verb suffices for truth, but the claim is perhaps not as absurd as Sainsbury believes it to be. But instead of pursue the question of what an account of assertion in its logical sense shoul d be, Russell struggles in various ways (from 1903-1914) to reconcile a conception of asserti on with a theory of meaning, subordinating the development of a full account of asser tion to his realist commitments. When asked what is asserted in /Caesar died/ it is /Caesar’s death/, but from the standpoint of judgment, the former is asserted in virtue of its declarative form; it is the judging mind itself that supplies the logical s ubject /Caesar’s death/ with another term. There should therefore be no puzzle as to fi nding the point of difference between these two forms if Russell had recognized the logi cal significance of the transformation, since the “predicative role” of the finite verb in /Caesar died/ contributes the added constituent by which the two forms may be distinguished. But in considering, as Russell does, the complex /Caesar’s death/ – “ where it is true ” – it would seem to be implied that there is per impossible a kind of coalescence of language with the world. Yet in quite a different sense of the phrase “/Caesar’s death/, where it is true seems to imply that the form of words expressing the singular term /Caesar’s death/ also expresses a proposition in the realist’s sense – what may be entertained as something that is true without anyone taking it to be true. This is a notion that is cons istent with, if not a requirement of, Russell’s early realism.

PAGE 116

112 Chapter VII: A Semantic Treatment of Relative Truth Epistemological questions have traditiona lly been segregated from those which have been declared to be the province of na tural language semantics. It would seem from a survey of the philosophical literature th at some of the most extensive work being written in what is called the ‘the ory of truth’ is confined to solving the problem of truth in those terms in which it has been, and conti nues to be, formulated by philosophers of language, viz. as a problem of semantics. Th is should be seen as a phenomenon that has fostered certain insular prejudices among thos e presently working within the discipline. Among many of these theorists, it seems not to have been recognized that without a sufficiently developed theoretical framework within which the problem of truth may be usefully analyzed, the manner in which the two dimensions intersect remains somewhat obscure. A suitably developed framework w ould, ideally, comprehend not only the study of indexical languages – what is now called ‘epistemology of language’ – but also a wider range of pragmatic phenomena than has been investigated. W ith due recognition of this fact, the frustrations engendered by th e traditional semantic approaches would emerge in a clearer light. Absent this recogni tion, it seems, such frustrations will continue to multiply, as may be seen, e.g., in recen t writing on the interfaces among various subdisciplines – between formal and natu ral language semantics, semantics and pragmatics, and epistemology and semantics. The problem is that, since we expect a

PAGE 117

113 solution to be given in the terms in which the problem has been formulated, we are bound to be dissatisfied even once a solution is proposed. The following is an attempt to explain cer tain features of the proposed epistemic view of truth (as discussed in the first two chapters) by means of contrasting this view with what has been called a relativistic view of truth by John MacFarlane (2005) and which is given, borrowing some ideas from Kaplan (1983), a semantic treatment. The details of the contrast between the two vi ews are discussed below. I will preface the discussion by first summarizing several thes es concerning the concept of truth and assertion so far presented in earlier chapters I then attempt to e xplain the key theses within the semantic framew ork suggested by MacFarlane118: the strategy is to explicate the concept of truth (rather than define it ) by examining its role within an account of assertion. Then it will be seen whether we stand to gain by further elucidating this account assertion by appeal to the notion of context. It was seen in earlier chapters that the use of “is true” in indicative sentences in ordinary English, carries with it certain logi cal implications. The structure which results from applying the truth-predicat e to a sentence consists of a juxtaposition of two verbal expressions wherein two judgmen ts are expressed: a judgment, on one hand, and what we might think of as a performative judgment on the antecedent judgment. There is, strictly speaking, a performative elemen t in the act of any judgmen t or predication. But the construction that results from a (performativ e) judgment on the antecedent judgment is, logically, more complex than it is apt to se em in calling the act, by which the construction is produced, performative119 The logical complexity of a judgment on an antecedent judgment owes to the fact that the expression involves an occurrence of both primary and

PAGE 118

114 secondary truth-values.120 For this reason, it is unders tandable that another point concerning the concept of truth is liable to go unnoticed: that point, which is crucial to grasping the epistemological significance underlying the ap plication of the truthpredicate, is that, in terms of our ordinary practice of a ssertion, the cons truction arises from the juxtaposition of two judgments which are made within a primary context121. A primary context may be taken as a function fr om a judgeable content to a primary truthvalue122 and consists of a situation in which one is evaluating either (1) A statement, P, (expressed by, and possessing the structure of, a declarative sentence), or (2) A judgeable content (propos itional content), p, which is expressed in natural discourse, as a participial construction or that-clause.123 Let us call (1) and (2) situations in which the truth-predic ate may be applied to a statement or judgeable content124, respectively. Philosophers who talk of the truthpredicate in connection with the theory of truth usually have in mind (1), wherein P is taken to be the object to which the predic ate is applied. In va rious (epistemological) contexts in which “is tr ue” is predicated of P P normally occurs antecedently to a truthevaluation of p although there are certa in exceptions. As Strawson notes in his (1950) article, “Truth”, one may “antici patorily” concede the truth of P “in order to neutralize a possible objection.” The use of “true”, Stra wson remarks, “always glances backwards or forwards to the actual or envisaged making of a statement by someone.” While I think Strawson’s remark brings to light somethi ng important concerning the epistemological dimension of predication, the point on which I wish to lay stress is a logical one: in situation (1) the outcome of the use of the truth-predic ate – that is, the resulting construction, is ‘ P is true’, whereas in situa tion (2) the outcome is simply‘ P ’. Here, it is

PAGE 119

115 not enough to notice simply that the outcomes differ, although this difference is important. It should be noted that the juxtaposition of P and “is true” involves one kind of predication.125 We might say, to be more precise, that this juxtaposition involves two instances of one kind of predication. In order to bring out its epistemo logical significance, Wiredu (1981) has expressed the point as follows: “Initiating an inquiry into P automatically…converts it into a question ‘ P ?’ So to answer the question of truth we have to answer this question.”126 As concerns the concept of truth, th e broader epistemological dimension – in terms of which we understa nd how the question of truth arises – intersects with the semantical dimension, in terms of which the problem of truth is customarily articulated. I maintain that the problem is soluble by appeal to this epistemological dimension in some way or another. This is an argument both th e plausibility and implications of which I have been so far exploring. We may distinguish between two concep ts of truth value by approaching the analysis of assertion from another direction: rather than focus on the aim of assertion, we might focus on the sorts of commitment typi cally undertaken in th e assertion of some (non-evaluative) declarative sentence. We often think of a si ncere assertion as committing one to the truth (relative to some context) to what is asserted, but commitment to truth may in turn be exp licated, in part, in terms of the normative consequences of a speech-act of sincere as sertion. This is an approach that John MacFarlane (2003, 2005) has taken in an atte mpt to exhibit various grades of truthrelativity that could obtain – at least in principle – in a na tural language such as English. For the purposes of understanding how MacFar lane’s own brand of truth-relativism may

PAGE 120

116 be related to the notion of point of view, it will be useful to incorporate the concept of point of view into his semantic framework, which I will explain in what follows. We might think of the point-of-view asso ciated with an occurrence of the primary concept of truth-value as constituting part of the context of use of a sentence, S.127 Similarly, we may think of the point-of-view associ ated with the occurrence of the secondary concept of truth-value as constituting the context of assessment of S According to MacFarlane’s view, a context of assessment is “ a concrete situation in which the use of a sentence is being assessed ” – which is to say, a truth-evaluation of S relative to context CA. A context of use, CU, is a situation in which the “circum stances of the context” are fixed.128 The context of use plays a dual role in determini ng the truth-value of a sentence: at the initial stage the context of use de termines which proposition S expresses, whereas at the second stage it determines the truth valu e of the proposition expressed by S once propositional content is fixed at the initi al stage. As MacFarlane explains, in recognizing the significance of this dual role it becomes clear why we should distinguish between context-sensitivity and context-indexi cality. But it is on the question of how this is done, that I differ from MacFarlane’s approach. C ontext-sensitivity is nonetheless the salient notion in terms of which the idea of truth-as-rel ative to a point of view is to be explained. Incorporating this notion of context into an account of assertion which stresses the “normative consequences” of a speech-act of assertion, MacFarlane suggests the following preliminary characterization: To assert a sentence S (at a c ontext U) is (inter alia) to commit oneself to providing adequate grounds for the truth of S (relative to U and one’s current context of assessment) in response to any appropriate challenge, or (when appropriate) to defer this res ponsibility to another asserter

PAGE 121

117 on whose testimony one is relying by withdrawing the assertion.” (MacFarlane 2003, p. 20) Contexts of use and assessment play a ro le in a language in which utterance truth is to be relative in an “unt ame” way. A typical manifestation of “untame” truth-relativism is the ( prima facie ) incoherent idea that what is true for one person may not be true for another in some absolute sense, i.e. wher e reference to a feature of an inividual’s situation is essential to the concept of truth and where that feature cannot be said to have any normative or law-like relation to truth. This is not the sense of relativity that I hold to be relevant to the concept of truth. The kind of relativity that is re levant to the proposed view is benign (or tame) and requires eluc idation with respect to the notion of justification; what is necessary to point out is that it is not what is meant by “untamed.” Given a role for contexts of both use and assessment the semantics for truth in MacFarlane’s framework allows us to provide d us with a recursive definition of truth. The semantics for which truth is defined recurs ively is to be taken as an “input” for what is called “postsemantics”129 The truth value of contingent sentences may vary over worlds and counterfactual contex ts; therefore, in addition to use-sensitivity, the notion of assessment-sensitivity is needed to account fo r contingently true (o r false) sentences in such a way that we may track variation of truth-value in sentences across counterfactual contexts. The kind of language that MacF arlane describes is one in which the “absoluteness of utterance truth” may be re ndered doubtful and the idea that an assertion being “true for Joe but not for Sally” is inte lligible. This language which would exhibit what MacFarlane refers to as a third -grade of truth relativity – a form of relativity that has not been explored in previ ous work on relativism with re spect to truth – is brought to light by employing the doubly-cont extualized truth-predicate: “true relative to context of

PAGE 122

118 use and to context of assessment.” What th e point of such a language would be is arguable, although it is argued in “Future Cont ingents and Relative Truth” that such a language contains the semantic resources to account for our evaluati ng future contingent propositions without assuming absolute trut h-values for utterances. It seems that MacFarlane succeeds in providing a kind of description of such a language. We may question whether we have a use for, or even know what it means to use, such a language. MacFarlane’s answer lay in his appeal to a speech-act analysis of assertion which would accommodate the semantic characterization of tr uth. On this analysis, the concept of truth is to be explicated in terms of a ssertoric commitment: to assert that p is to be committed to the truth of p To be committed to the truth of some proposition is to be inter alia be obligated to withdraw the commitment to, or provide justification for, the assertion that p under the appropriate circumstances relative to the context of us e. Justifying one’s assertion relative to Cu, for example, means providing grounds for the truth of the sentence asserted. It is at th is point, namely that of spelling out exactly what justification is to consist in that MacFarlane seems to smuggle into his th eory the notion of justification the “absoluteness” that was to be avoided with respect to truth. The idea of assessmentsensitivity MacFarlane explains, should not be confused with assessmentindexicality The truth-value of an assessment-sensitive proposition varies with context, whereas in speaking of an indexical, what is meant is that propositional content itself (as opposed to sentence-content) varies with context. A context may contain several “parameters.”130 These may include world, time, and place of utterance; but it may also include speaker. Propositions exhibit contextsensitivity with respect to a single speaker, S (perduring over time) when, e.g., S

PAGE 123

119 evaluates p as true at t0 and p as false at a later time, t1. A simple example of this is the proposition ‘I believed that p but p is false’, where the part containing the belief clause marks one truth-evaluation and the clau se containing “is false” marks another.131 Here, context sensitivity occurs over time but may be invariant with respect to any class of speakers, S1, S 2, S 3 Similarly, a proposition’s truth may vary with respect to a class of several speakers, S1, S 2 S 3 … etc., simultaneously, at time t As far as the proposed view is concerned, however, it is important to not e that the “variance” of a proposition’s truth with respect to a class of speakers is a slightly misleading way of saying that S1 and S2 disagree as to what the truth value of p is. To say that the truth-value “varies” in this case is to say, rather, that one ag rees with S1. For example, S1 may evaluate p at t0 as true, while S 2 may evaluate p as false at t0. The latter kind of sensitivity is liable to strike some as more controversial than the former kind, but I will argue in what follows that, logically speaking, context sensitivity in the speaker’s sense is really no more problematic than context sensitivity in the temporal sense just described. In fact, context-sensitivity of this kind is essential to the view I am advocating. The context of use, CU may be allowed to diverge from the context of assessment, CA for any truth-evaluable sentence. This is so because the doubly-contextualized predicate, “true at CU and CA”, may involve a primary as we ll as a secondary truth-value. That is, CU may contain the primary truth-value and CA the secondary truth-value. As was noted earlier, these two truth-values can di ffer: for example, “The Dodo is extinct in 2008,” and “It is false that the Dodo is exti nct in 2008” – differen t truth-values for different contexts. Perhaps it is prudent to insert a word of caution here. A truthevaluation is not being claimed to be relative in the sense that the individual truth-value

PAGE 124

120 determined at CU or CA will depend on either CU or CA. That a truth-evaluable content, p or truth-evaluable proposition, P may vary in truth-value over contexts depends not on context but on whether such a content or pr oposition is warrantedly-assertible at a context. It may be objected that in my havi ng spoken of warranted-assertibility at this level of generality, I have neglec ted to specify any criterion or standard of warrant, or justification. I have de liberately left this question in ab eyance for several reasons. One is that, for the purposes of disti nguishing the present view of truth from relativism, it is more important to see how such a standard c ould be included in a context with the other parameters ( w s t p ), and how such a standard would be treated in the context as a constant parameter. Were propositions and propositional contents not truth-evaluable relative to a fixed standard of justification, a more contro versial and untamed relativism would likely result132. It is just such a relativism that is thought to be implied whenever an agent (or speaker) is taken to be a c ontext parameter along with world, time, and place. But this, I suggest, is not an implication what it means to relativize the concept of truth-value to points of view, as I shall argue shortly. In order to dist inguish the senses of relativity involved, we may say that a pr opositional content is truth-evaluable in reference to CU whenever “is true” (or fals e) is predicated of a sentence To make the distinction more precise, however, it will be ne cessary to identify certain aspects of MacFarlane’s semantic fr amework with which the concept of point of view is compatible, as I have explained it, and those with which it is not. The sense of ‘context’ which I would like to adopt allows for the expr ession of the idea that usesensitivity may diverge from assessment-sens itivity in a significant way. There are several ways in which this can happen: the truth-evaluation of a proposition at CU may

PAGE 125

121 occasion the use of ‘is true’, as a speaker at CA is not in such a case merely determining an issue – i.e., either affirmativel y of negatively – but is “commenting133” on an existing judgment made at CU. A speaker, then, thereby recognizes how the secondary concept of truth-value is instantiated. Fo llowing terminology introduced in the last chapter, let us call such an instantiation an “occurrence”. It should be clear that an occurrence of the secondary truth-concept en tails the divergence of CU and CA since the use of the truthpredicate, as I have just characterized it, entails that CU CA Critics of the epistemic conception sometimes point out that if truth is to be taken to mean nothing more than warranted-assertibilit y, then one must admit that a belief which was formerly held to be false may no w be true. What had once been warrantedlyassertible may no longer be so in light of newly available evidence. Such an epistemic theory of truth of the kind just describe d, the objection goes, must somehow violate the law of non-contradiction. As it concerns the role that point of view play s in a context, this objection confuses CU with CA. The confusion is brought to li ght in first recognizing that a truth-evaluation of a “belief” is tantamount to ( inter alia ) a truth-evaluation of a propositional content (as the object of a belief). It should be clear that since any determination of truth must be in the form of a claim to truth. It is also clear that since a truth-claim is itself a judgment, one’s commitme nt to another point of view – i.e. to holding another claim as true – can only issue in the form of another commitment that is cotemporaneous with the present Thus, any agreement with a former point of view implies a judgment to that effect from a new and distinct point of view. This alone does not preclude one from agreeing or disagree ing with a previous point of view at CU. The two things that are cote mporaneous are (1) the reference – in this case, reference to an

PAGE 126

122 earlier point of view, P at world W and (2) the commitment to the truth of the judgeable content contained in the assertion of the earlier point of view, at world, W What is precluded is the possibility of being in possession of a truth that is point of view transcendent. The outcome of a truth-evaluation may be either affirmative or negative (assuming the principle of bivalence). Thus with respect to c ontexts of use and assessment, a proposition’s truth is invariant when the truth-value of p at CU and CA is the same. For example, it is the same whenever at CU the determination is affirmative and at CA it is also affirmative. It varies if, for example, at CU the determination is affirmative and at CA it is negative. With respect to the former case we may say that there is a kind of congruence of points of view. In the latter (where truth-value varies) there is an incongruence. But it is, nonethel ess, crucial to recognize that just as the points of view are numerically distinct in the latter case they are also di stinct in the former and that, in the case of incongruence of points of view, th e points of view themselves vary – not strictly speaking the truth-valu e. I argue in what follows that this numerical distinctness of points of view implies a kind of context-sensitivity without implying relativism Before commenting on how the notion of poi nt of view may be related to the kind of truth-relativism that MacFarlane advocat es, it may be useful to summarize the main objectives of his exposition. MacFarlane’s aim is to challenge the or thodox view of truthrelativism as a view that is plainly absurd by presenting a “clear counterexample to the absoluteness of utterance truth.”134 According to the orthodox view, utterance truth is absolute, whereas propositional truth is relati ve to a context of use. According to MacFarlane ordinary talk of the term ‘utte rance’ is often ambiguous. An utterance may

PAGE 127

123 be taken in one sense as a speech-act (as an ut tering) or in another sense as the object of an act of uttering – what, e.g., MacFarlane ta kes as a particular in scription or “acoustic blast.” Given MacFarlane’s semantic char acterization, there is a way to understand propositional relativism with respect to trut h. For MacFarlance, this involves recognizing simply that at least some propositions are assessment-sensitive. According to MacFarlane, [Sentence] S is true at a context of use CU and context of assessment CA iff there is a proposition p such that (a) S expresses p at CU, and (b) p is true at the world of CU and CA.135 It is not entirely clear whether MacF arlane has articulated a form of truthrelativism that is really intel ligible, as he claims to have done. In what follows, I contend is that that conception fails to be applicable to the formulation of point of view on the proposed account of the concept of truth. Cruc ial to maintaining this position are several ideas. One is that the idea of a point of vi ew should not be taken as something on which truth-values depend in any “controversial” sense. Such a controversial sense of ‘relative’ is what MacFarlane – following Max Kolbel – attempts to articulate. The semantic framework that he provides fails to apply to the following point: a proposition may be context-sensitive in a given language without implying the “untame” truth-relativism that MacFarlane wishes to vindicate. This is the case where p is assessment-sensitive at a world, W of CA, while the standard of justification, j remains fixed at both CU and CA Just as the truth-value of the proposition “The glass is full” at a given context depends on the world in which the senten ce is used, the truth-value of the proposition depends on the world in which it is being assessed.136 A proposition is said to be assessment-sensitive if its truth-value varies with CA while CU remains fixed.

PAGE 128

124 Here, it is useful to point out, amending the previous definiti on, that a sentence S is true at CU and CA iff (1) S expresses p at CU (2) p is true at the world of CU and the standard of justification j of the assessor at CA. A proposition is true at CU and CA iff p is true at a circumstance of evaluation determined by CU and CA where the circumstance of evaluation is composed of the world of CU and standard of justification of assessor at CA I would suggest the following point about the Kaplanian notion of context as it concerns a view according to which truth is ex plicated in terms of warranted-assertibility. If a standard of justif ication is included as one of the parameters in a context, then a circumstance of evaluation would then be composed of CU and CA ( w,t,s,j ). But in this case the justification parameter would have to be construed as the ‘O ne correct standard of justification’. In stressing the specific logical role played here by the standard of justification we are allowed to distinguish the epistemic conception of truth with other untame forms of truth-relativism. The role of justification figures crucially in this epistemic conception. It is necessary also to stress the following point concerning judgment as it relates to any st andard of justification: in the context of inquiry the judgments which constitute points of view are, as Wiredu has claimed, “in accordance with the canons of rational justification.” (Wiredu, 1980) Wi redu further lays stress on this point by clarifying the in tended sense of the term ‘relative’, the use of which in philosophical discourse is ofte n shrouded in obscurity. “Truth is not relative to point of view,” Wiredu claims, “[It] is in one sense, a point of view. But it is a point of view born

PAGE 129

125 out of rational inquiry, and th e canons of rational inquiry ha ve a human application.” In light of this remark, let us rewrite (2) as follows: (3) p is true at the world of CU and the ‘ One correct standard of justification ’ at CA.137 While this does not allow for one feature of c ontext, namely standard of justification, to vary, it does allow for the other features of context to vary. Parameters, as they are defined in MacFarlane and Kapl an’s framework, may be varied independently Therefore, a proposition’s truth-value may, in a sense, vary accordingly.138 Thus, the truth-value of p at a world, W of CU may differ from the truth-value of p at W of CA. Since there may be indefinitely many contexts of assessment, a distinct context of assessment for, as MacFarlane points out, for Joe, for Jim and for Sally. The truth-value may ‘vary’ among those contexts in the way just described. This is a kind of assessment-sensitivity. It amounts to the recognition that there is a pluralit y of truth-values that come in the form of points of view. We might agre e with MacFarlane in wanting to preserve this notion of sensitivity: CU ‘The glass is full’ CA ‘The glass is full’ While incorporating the concept of point of view and standard of justification as parameters into the context, we should allow for the possibility that the world of CA will not be the same as that of CU. But, while we would like to suppose that a refutation of p given by a challenger at CA, where the asserter (speaker) may evaluate the putative

PAGE 130

126 refutation, we would also need to allow for a world of CU in which the glass is full and another world, of CA, in which the glass is empty. Sim ilarly, it should be possible that there is a world of CA(1) in which the glass is full and a world of CA(2) in which the glass is empty. Therefore, there is nothing incohe rent in allowing the truth-value of p to vary with the context of assessment (while the contex t of use remains fixed). However, it is important to recognize that a proposition may be assessment-sensitiv e in at least two ways, one of which leads to untame relativism and one of which that does not. That is, assessment-sensitivity may be due either to not fixing the standard of justification at the relevant CA or not fixing the world of the speaker.139 Only the first kind amounts to untame relativism with respect to truth.

PAGE 131

127 Chapter VIII: Moore’s Paradox and the Logic of Assertion ‘Moorean sentences’, i.e. sentences of the form ‘ p but I do not believe that p ’ or ‘I believe that p but it is not the case that p ’, appear to be paradoxical because the propositions they express are inconsistent with what they pragmatically convey. Many philosophers readily acknowledge the oddity of these sentences, attempting to explain away the oddity on the basis of recasting the distinction between ‘standard meaning’ (descriptive content) and pragmatic effect. I argue that this approach is mistaken. I suggest that the pragmatic/sema ntic distinction, despite recent attempts to ‘stabilize’ it, obscures a deeper problem which, once suffici ently analyzed, reveals that the various appeals to the distinction cannot afford a solution to Moore’s paradox. One reason for this is that the problem is both a logical and syntactical one. Because of this, I suggest that a solution to Moore’s paradox should not exploit the pragmatic/s emantic distinction in the manner that has now become customar y. Rather, I show that the paradox arises from the failure of standard semantic analys es to account for the role of the concept of truth in the practice of assertion. The concept of point of view which is central to understanding this role of the concept of truth, affords us with an explanation as to why we are tempted to think that Moorean sentences are (possibly) true. Once the role of the concept of truth is proper ly understood, it becomes clea r how Moorean sentences are analyzable simply as logical contradictions.

PAGE 132

128 Moorean sentences are senten ces such as the following: (1) ‘I don’t believe it’s raining, but as a matter of fact it is’. (2) ‘Oysters are edible bu t I don’t think they are’.140 Such sentences would seem not to be amenable to the kind of treatment which the garden-variety semantic paradox receives, the reason being that neithe r appears to be in any way self-contradictory, as neit her could be said to be used to assert two contradictory propositions. In asserting a sentence of the form ‘ p but I do not believe that p ’ we at least standardly assert that p is the case and imply its negati on. There has been, however, little agreement among philosophers as to precisely ho w this sense of ‘imply’ is to be properly analyzed. The component sentences possess the dua l function of asse rtion and belief conveyance (Figure 5): What is said ; what it implies Oysters are edible, but I don’t believe it What is said; implying the negation of what is said.

PAGE 133

129 It may be said that an assertion implies that the corresponding belief is true, but an analysis that rests on this assumption would appear to fail to meet what Roy Sorensen and Jane Heal identify as adequacy conditions for any satisfactory solution of Moore’s paradox. Following Heal and Sorensen, I assu me the burden of having to meet these conditions: (C1 ) Moorean sentences must be shown to be reducible to contradictions (C2 ) An explanation must be given as to why we are tempted to say that Moorean sentences can be true ev en if they cannot be rationally believed.141 We may see the difference between a fairly standard view (E. H. Wolgast) and the one that I propose, using (S1) to represent th e former, (S2) to represent the latter, and brackets to highlight th e different in emphasis142 (Figure 6): (S1) The meaning implies (in some unspecified sense) that this is the speaker’s belief; paradox results from conflicting beliefs It is raining, but I do not believe it’ (S2) The expression of this belief commits the speaker to the truth of what is believed – either It is not raining” or “I have no attitude toward the proposition ‘that it is raining’” One consequence of (S2) is that what is seen to be paradoxical a bout the sentence results from a conflict between believing ‘It is raining’ and ‘It is not raini ng’. I suggest that the

PAGE 134

130 failure to realize the possibility that each c onjunct contains an asse rtion from a different point of view obscures an otherwise a flagrant inconsistency.143 Supposing that, in some sense, (a) ‘I believe it is raining, but I do not believe it’ is a defensible interpretation of Moor ean sentences, on S1 the second conjunct would remain ambiguous, and would thereby present se veral difficulties into which we need not enter here. But if we suppose, on the other hand that ‘It is raining but I do not believe it’ is analyzable straightforw ardly as a contradiction, (b) ‘It is raining, but it is not raining’ then (b) would appear to satisfy our first condition. S1 and S2 may be further explicated in accordance with the following principles: (P1) An assertion (standardly) expresses a belief (P2) Articulating a belief standardly expresses an assertion; and stating, or asserting a belief in the first-person, t hus from the first-pe rson point of view, commits one to the trut h of what is believed On the Wolgast and Black view, the expression of a given belief, according to (P1) is central to the meaning of the sentence which is used to express the corresponding belief. I do not wish to dispute this claim, but it would seem that a lingering difficulty with such a

PAGE 135

131 view (represented somewhat crudely in S1) is that it leaves the source of Moore’s paradox unexplained. Specifically, th e demands on such a view would be to show that the ‘paradox’ amounts to a contradi ction, and S1 does not allow th is to be done absent a detailed analysis as to how declarative se ntences are to be “doxastically contexted.” These details are spelled out carefully in vari ous treatments of Moor ean sentences (e.g. in recent work by Uriah Kriegel), but much of th at work would seem unnecessary if we can meet those demands yet forgo the complications attending views along the lines of S1. The question then remains as to how Moorean sentences are analyzable as contradictions in accordance with S2 – the answer to which I suggest points in the direction of a simpler analysis of Moorean sentences. I suggest that the reducibility of Moorean sentences to assertions may be permitted, assuming such sentences are seen to issue from the first-person point of view Further, I suggest that this strategy may be extended so as to include belief reports generally. The suggestion plainly fails for the sentence ‘ p but she does not believe it’, since what is asserted in the first conjunct is from a point of view that is distinct from the point of view involved in what is believed in the second. In the se cond conjunct something is believed from another point of view, such that what is asserted is that someone other than the speaker does not believe what is asserted in the first. We may see that differences in point of view also arise in connection w ith tense. Thus, not surprisingly, we find no Moorean element in sentences su ch as ‘It is raining, but I did not believe it’. The reason is that, in saying ‘I did not believe it’, I am speaking about a point of view that I once held but no longer consciously hold. Thus, the suggestion is that in cases of believing that p from two numerically dis tinct points of view, one may not derive a logical

PAGE 136

132 contradiction; rather, to do this one must assert that p and ~ p from one and the same point of view. We may observe that in Moorean contexts the semantic content of ‘I’ is reflexive, and so is determined at any given time by the point of view associated with the speaker at the time of utterance. It is in virtue of the fact that I am sincerely uttering that p that I am thereby expressing the conscious belief that p But I should like to suggest that a sincere utterance of ‘ p and I do not believe p ’ is not a pragmatic inconsistency, for the inconsistency lies between what is asserted in both conjuncts. Appeals to conversational implicature seem misguided here. On the view that is indicated here, the first conjunct contains an implicit free variable whose value is determined by the point of view associated with the occurrence of the first-person pronoun. A point of view is re presented by the token-reflexive ‘I’ and the surface grammar of a sentence in which it occurs would appear to allow for its elimination. It would seem that most, if not all, declarative sentences possess this grammar. But here, as elsewhere in philo sophy, surface grammar is misleading as to logical grammar, and a sentence in which the token-reflexive character of a given point of view is implicit awaits fu rther analysis. Such an anal ysis would supply the sentence with an index to which the truth of a give n token of the sentence may be relativized. From the standpoint of the pragmatics of a point of view, a speaker’s present point of view and understood by reference to several factors which constitute the narrow context of an utterance. It may not be immediately clear how, as regards th e particulars, a more thorough account along these lines may be give n, but the following considerations would appear to lead to a presumption in its favor.

PAGE 137

133 In taking Moore’s original example, ‘I we nt to the pictures last Tuesday, but I don’t believe that I did’ we are told that “ what is asserted is something which is perfectly possible logically” (Moore, 1944). The sentence, ‘I went to the pictures last Tuesday but do not believe that I did’, is absurd when asserted but this is because an assertion from a given point of view entails a commitment to its truth (P2) ; and, it may be further added that an assertion is necessarily from a point of view. What makes it ‘p erfectly possible’ is that I am able to imagine someone else presently entertaining the same proposition (thus suspending commitment to truth) with re spect to myself. Similarly, another person saying of me that J. Kelly went to the pictures but that J. Kelly does not believe it, may be correct, but this is because th e entire sentence issues from another point of view The notion of ‘ what is asserted ’ is indispensable to rational discourse, since without it we should not be able to expl ain how it is possible to consider what it is that a sentence expresses; we should not be able to explai n how the accusatives of thought are possible. Such considerations, however, issue from di fferent points of view; and while they do not enjoin commitment to truth – for such a commitment is suspended in our merely considering some proposition – we should avoid the temptation to believe then that we may maintain the suspension of commitment when we sincerely assert a sentence. Given this, to have before us a Moorean sentence, a declarative sentence whose assertive force is reflected in its syntactical structure, and then go on to as k ‘How is it possible that this is true?’ is to yield to this very temptation. The sentence ‘I went to the pictures but do not believe it’ is possibly true only insofar as I am able to refer to myself having gone to the pictures at an earlier time, i.e. from another point of view. This could be any given hypothetical point of view; and it would seem that the possibility – of entertaining a

PAGE 138

134 hypothetical point of view – explains why it is we want to say that Moorean sentences, despite their absurdity, may nonetheless be true. The temptation to regard Moorean sentences as possibly true would thus a ppear to be a consequence of confusing two distinct points of view – viz. the point of view associated with the (absurd) 1stperson assertion of a Moorean sentence and the point of view associated with the 3rd-person assertion of the Moorean sentence, which, of c ourse, may be true. In both cases, we have the same proposition, taking a broadly Fregean view of propositions. It may then be seen to follow that if any interpretation of Moorean sentences as being possibly true or false is born of such confusion, so would c onsequent intimations of paradox. As asserted from a first-person point of view, a Moorean assertion is absurd, whereas from a third-person point of view such an assertion is either true or false. Thus, someone other than myself may say truly of me that I went to the pictures but that I do not believe it, or, on a possible-worlds model, it may be true in some possible world other than the one that I inhabit. I may come to the same conclusion as that of the person in that world, but if I do so it is only in virtue of inhabiting some other possible world. The plausibility of the possible-worlds model, in connection with the concept of point of view, depends on clarifying a syntactic dis tinction between propositions in the sense given to them here and d eclarative sentences. Insofar as I may be tempte d to regard both parts of (i) ‘I went to the pictures la st Tuesday, but I do not believe it’

PAGE 139

135 as possibly true, I must effectively dissociat e the assertion of the sentence from my present point of view so as to imagine the en tire sentence being true of me at an earlier time. As I already mentioned, what gives one the sense that the sentence could nonetheless be true must be that, given any asse rtive utterance of (i), (i) is surreptitiously converted to an assertion from another distinct point of view. As is the case with speech acts in general, a point of view is episodic in nature. This fact, however, is not reflected in the logical grammar of (i). Similarly, I may imagine what is said being uttered by someone else, or someone else may say of me (ii) ‘He went to the pictures last Tuesday but he does not believe it.’ In either case I divorce the propositional conten t of the assertion from the assertion itself and go on to make an asserti on; but in doing so I adopt someone else’s point of view, in which case, the sentence should read: (iii) ‘I went to the pict ures last Tuesday but did not believe it.’ Therefore, if it is to be true of anyone, there is necessarily a commitment to its truth by its speaker, and to this commitment there necessarily corresponds, by (P2) a unique point of view. In believing (or asserti ng) that ‘what is a sserted’ is ‘possible’ I thereby no longer consider the assertion from the same point of view, as I am not truly considering an assertion – unless commenting upon it. Entertaining some proposition that p is thus possible without being committed to the truth of that proposition, and in this way, p

PAGE 140

136 appears to be point-of-view-less But here we are misguided into believing that from a proposition presented under one mood (of ente rtainment) as point-of-viewless, we may infer that a proposition under any mood (e.g. the mood of assertion) is also point-ofviewless: being wedded to a point of view doe s not necessarily commit one to the truth of any proposition, but in asserting any proposition to be true, one thereby commits oneself to a point of view (More on what this commitment amounts to (e.g. normative consequences)) Moorean sentences are therefore unique inso far as they are signal manifestations of a failure to realize the implications of co mmitment to truth that is involved in any act of assertion. We might susp ect, given the foregoing reflections, that it is to this uniqueness that Wittgenstein refers when he famously said in the Philosophical Investigations that Moorean sentences reveal some thing important about “the logic of assertion.” We may think of this commitment as what an assertion adds to a bare proposition. If I am inclined to think that I might consider the assertion of a Moorean sentence from one and the same point of vi ew in such a way that it appears to be simultaneously absurd and possibly true, I need only attend to the tense of the principal verb of the second conjunct to see that an absurdity could result – insofar as I am considering the sentence as asserted. ‘I went to the pict ures last Tuesday but do not believe it’ is absurd, yet the proposition as a judgeable content is possibly true when I conceive of a state of affairs in which I went to the movies and later did not remember it. But to do this is not to conceive of the same state of affairs referred to by the proposition expressed by that sentence. It is quite a different thing to say to oneself ‘I went to the movies and later did not remember going’. The further rider may then be added: ‘But I

PAGE 141

137 remember now’. It should be clear that this is another assertion and th at the assertion is a conversion of the judgeable contents, (expr essed as two separate noun-phrases) ‘my having gone to the pictures last Tuesday’ and ‘m y not believing that I went to the pictures last Tuesday’, to a declarative sentence which is both assertible and possibly true. Thus, the sentence ‘I went to the pi ctures last Tuesday but I don’ t believe it.’ asserted from another point of view may be regarded as another sentence altogether.

PAGE 142

138 Notes 1 R. S. Wells, “Frege’s Ontology” p. 16 2 Ibid. 3 See Grundgesetze (ed. M. Furth, 1964), section 2, for a nearly parallel discussion. 4 Though we find, for example, a discussi on of a function being doubly incomplete ( Grundgesetze ), Frege 5 The “syntactical” distinctions appealed to are in keeping mo re with e.g. W.E. Johnson’s employment of the term “synt actical” than, e.g., Carnap’s ( The Logical Syntax of Language (London; 1937)), the latter of which pe rtains exclusively to the study of signs 6 Sometimes called ‘propositional sign’, e.g. in ‘Function and Concept’ (1891) 7 “Frege’s Ontology,” The Philosophical Review ; 1960 8 Dudman 9 Here Dudman also cites Russell’s co mplaint in a similar connection in Principles of Mathematics : “Asserted propositions have no indication.” ( PoM ) 2nd ed. 1937, p. 504 10 See p. iii of the introduction. 11 Dudman (1975) 12 See p. 24. 13 See, e.g., Geach, 1975. 14 Here we find echoes of Church’s view. 15 Or, to use Russell’s example, ‘Caesar is dead’ 16 Let us recall that Frege is alleged to have overlooked the fact that assertoric force is withdrawn in truth-functional contexts and so the use of a definite description to exemplify ‘—A’ was unnecessary. 17 In fact, this is Dummett’s understanding of the matter. Frege: The Philosophy of Language (Harvard; 1980) 18 See p. 31 19 See Dudman, p. 157. 20 See Grundgesetze pp.6f 21 I repeat and earlier quotation from Ch II, p. 22. (M. Furth (trans.), Berkeley, 1964) 22 See “Truth as a Logical C onstant; with an application to the principle of excluded middle” The Philosophical Quarterly Vol. 25, No. 101 (Oct., 1975), 23 Frege's Theory of Judgement David Andrew Bell, Oxford: Clarendon Press: Oxford University Press, I979 24 Russell often speaks of this construction as the verbal noun 25 Principles of Mathematics New York: Norton (1931) p.48 29 George Pitcher has expressed some apprehen sion that a kind of schizophrenia here begins to set in to the theory of truth in th is connection. Given that what I’ve said so far

PAGE 143

139 implies the possibility of two kinds of truth-bearers, Pitche r is perhaps right. However, what I shall argue is that this “schizophrenia” is virtuous fr om the point of view of an analysis of the role that truth plays in the practice of assertio n. In the pages that follow I distinguish between primary and secondary form s of truth-attribution and argue that it is sentences relative to contexts, not utterances, that are bearers of secondary truthattribution. However, for the purpose of distinguishing betw een primary and secondary it is not necessary to distinguish here between sentences and their use. See Pitcher (ed.), Truth Englewood Cliffs: Prentice-Hall, 1964. Se e also Bar-Hillel, ‘Primary TruthBearers’, Dialectica (1973) 30 Griffin, Nicholas. 1993. “Terms Relations, Complexes’ in Eds. Irvine and Wedeking Russell and Analytic Philosophy Toronto: University of Toronto Press. 31 This is distinguished from what Russe ll calls a ‘propositiona l concept’, which is expressed typically by a noun phrase – or verbal noun. 32 This is Griffin’s observation. That Russell did not wish to assert this is perhaps arguable, but it seems to be an unlikely possibility. 34 W. E. Johnson. 1921. Logic 35 This phrase would appear to de note a datable event, whereas ii seems less apt to do denote the same datable event. This is mere ly an appearance, however. Strictly speaking, noun phrases do not denote anything. Strawson’ s remark regarding this matter in “On Referring” seems to show us why this is so: a phrase in itself doe s not denote anything. A phrase is something a person may use to denote something. The added qualification is then necessary: in the referri ng act, a noun phrase may denote so mething (in this case, an event). This is only possible, however, given the addition of the ‘a ssertive tie’, whereby the unsaturated expression (e.g. a noun phr ase) becomes saturated. This is a transformation of the jugeable content to a judgement. By vi rtue of this transformation, an act of denoting just is a claim – i.e., a j udgment. For this reason, it would seem that we must use a declarative sentence, not a noun phras e, to express the idea of a phrase having a denotation. 36 By grammatical sameness of the definite de scription and the participial phrase I mean that both are unsat urated expressions. 37 Wiredu (1975) distinguishes between this kind of determination (of truth-value) and that involved in the sort of attribution of truth by which the truth predicate is employed. 39 This has also been discussed and recognized to be a real difficulty by Russell (1903), Geach (1975) and Dudman (1975) 40 Proceedings of the Aristotelian Society, vol. XLVI, 1946 41 ‘Knowledge, Truth and Reason’ Dissertation, Oxford (1960) 42 See Davidson (2001) 43 See Ch. II for a discussion of the “characterizing tie.” 44 Geach (1957), Griffin (2001) 45 Meinong’s “Objectives”

PAGE 144

140 46 “On the Nature of Truth,” in Philosophical Essays 1966, p.149 47 This was published later in The Philosophy of John Dewey; The Library of Living Philosophers vol. 1, ed. Paul A. Sch ilpp (Open Court, 1990) 48 This is stated in Inquiry into Meaning and Truth, (Allen and Unwin Ltd.; London) 1948, p.148; hereafter IMT 49 This comment was made in response to Russell’s remark ( IM&T) that “there is an important difference between his (Dewey’s) views and mine, which will not be elicited unless we can understand each other.” (p.401) 50 The connection between Russell’ s conception of propositions and the nature of truth is well documented in a discussi on dating back to Russell’s Principles of Mathematics (1903) in which Russell struggles, famously, to distinguish between lo gical subjects with respect to which truth is said to be external, and assertions with respect to which truth is said to be internal. 51 “On Sense and Meaning”, trans. M. Black, (Basil Blackwell, 1952) 52 Begriffsschrift trans. P. T. Geac h, (Basil Blackwell, 1952) 53 See, e.g., §§ 51-53, Principles of Mathematics (W. W. Norton, 1996) 54 I have in mind, e.g., Alonzo Church (1956), Peter Geach (1965), Max Black (1964), Michael Dummett (1993), and V T. Dudman (1970). I do not mention Wittgenstein because he did, of course, treat th e assertion sign of Frege in the Tractatus Though the details of this treatment are quite signifi cant with respect to the development of the philosophy of language in general, they fall outside the scope of our discussion. 55 C. I. Lewis ( An Analysis of Knowledge and Valuation ; 1946) joins their company in this respect. 56 For Frege, these propositions are the bearers of thoughts. 57 PWAT p.172 58 Both my criticism of Russe ll’s correspondence theory and my interpretation of Dewey here and throughout derive from the views of K. Wiredu’s, to be found in several discussions on these subjects. See, for exam ple, “Truth: The Correspondence Theory of Judgment," ( TCTJ ) African Philosophical Inquiry (January 1987). 59 PWA&T p.171. 60 The use of idea here is not to be under stood in Dewey’s sense, but rather in a (classical) empiricist – i.e. as implying some primitive datum of experience whose function is to represent the object it is (in some sense) about. 61 See p. 178 of PWA&T 62 Ibid. 63 This is what Dewey calls a pr oposition’s “sufficient verifier”63 64 Ibid., p.179 65 SCCT p. 158 66 I refer here to a form of truth-skepticis m arising out of certain correspondence views that carry the implication that the truth-ma ker relation is asymme trical and therefore cognititvely inaccessible. 67 My attention was drawn to th ese observations by K. Wiredu, TCTJ (1987). 68 SCCT p. 165

PAGE 145

141 69 See p. 65 70 The point here is made later in Russ ell’s lectures on the Philosophy of Logical Atomism. 71 I include, in addition to Strawson (1950), tw o notable truth-theorists, J. L. Mackie (1973) and Kwasi Wiredu (1973) 72 Ibid. 73 This a context to which Frege was sens itive early on despite exhibiting a strongly deflationary bias. See Frege (1879) 74 Ibid. p.20; Strawson, more frequently than Au stin himself, uses the synonymous term, ‘speech-episode’. 75 Ibid., p.23. 76 Ibid. 77 Ibid., p. 26 78 Ibid., p.27 79 Ibid., p. 31 11 It is worth noting that Moore seems to have detected the se nse of a sentence ‘saying of itself that it is true that I wish to explain. On this score, however, it is evident that Austin did not follow Moore’s lead. 81 My emphasis. 82 Ibid., p. 19 83 Strawson, ‘Truth,’ (1950). Subsequent ci tations refer to the G. Pitcher, ed. Truth (Prentice Hall, NJ, 1964) pp. 32-53. 84 Ibid., p. 42 85 This may apply also to Russell’ s correspondence view of 1912. See his Problems of Philosophy 86 Ibid., p. 29 87 Caution must be taken here to di stinguish Frege’s early concept of force of his Begriffschrifft (1879) from the later concep t as it appears, e.g., in Grundgesetze (1891), as the differences between them, though subtle, l ead to radically differe nt interpretations. I refer here to the former. The critical litera ture on Frege from Russell to D. M. Armstrong would suggest that these differences in in terpretations lead, in turn, to different ontologies. 88 It is not possible to discuss this aspect of assertion satisfactorily here; it is an aspect which, as I mentioned earlier, makes essential re ference to ‘point of view,’ a significant although somewhat technical term of art. The idea of point of view, insofar as it bears on the logic of assertion, is discu ssed fully in Wiredu’s “Truth as a Logical Constant with an application to the principle of excluded middle” (1975). Many of the views contained in the present discussion are appli cations of the main ideas ad vanced in that paper. 89 Ibid., p. 41. 90 Logic (Cambridge: Univer sity Press, 1921) 91 I was first made aware of the primary and s econdary distinction in reading K. Wiredu’s “Truth as a Logical Constant, with an App lication to the Principl e of Excluded Middle,” (1975); see also his “Deduc ibility and Inferability” Mind (1973). 92 I.e., has a primary value of ‘True’

PAGE 146

142 93 This is what Strawson means, as I unders tand him, when he remarks that it is a “logically fundamental type-mistake ” to suppose that there is something in the world to which a statement can be related (or be “about ”) other than that to which the referring part of the statement refers and that to which the describing part “fits or fails to fit.” To understand how these functions are discharged in normal cases of a ssertion is to see precisely what it is a statement is about Cf. Strawson, §2 of ‘Truth’ (1950). 94 Or we might say ‘in virtue of using a declar ative sentence ‘2 + 2 = 5’. I have deviated from orthodoxy in maintaining that a declarative sentence cannot be used non-assertively (e.g. as a complex name). 95 I should point out here not just the differe nce between primary and secondary concepts but that the secondary concept is derivative of the primary. This point is developed clearly in both K. Wiredu (1975) and K. Wiredu’s “Truth: The Correspondence Theory of Judgment,” (1987) 96 The idea presented here, which I do not develop, may be found in Kwasi Wiredu, op. cit., (1973), and Philosophy and an African Culture (1980). 97 Where ‘ p ’ stands for a declarative sentence. 98 I pursue this idea in a later discussi on on the idea of relative truth. 99 We may suppose that a commitment rema ins in force unless or until a person withdraws the assertion that carries it. 100 This point will have implications for a later discussion of relativism. 101 See Wiredu, 1975 102 For clarification of what th ese advantages are, see my pr evious discussion of assertion in Ch. II, p. 50 103 Precisely in what way a truth-value is iden tified with a point of view is explained on the next page. 104 See Chapter V, p.114 105 A. N. Prior Objects of Thought 106 Ibid. 107 See Chapter II. 108 Truth and Predication (2001) 109 In the context of Russell’s later ph ilosophy of mind, judgment so conceived was referred to as a ‘propositional action’. 110 Griffin suggests that such a relegation w ould demand an account of the “propositional action” of the mind ( Terms, Relations, Complexes ). 111 There is good evidence that suggests Da vidson’s one views fail to recognize this point. Davidson’s theory of truth does not, e.g., account for the logical aspect of assertion for reasons having to do with the fact that Davidson intends his theory of truth to serve also as a theory of meaning in a unique way. In light of the role that interpretation plays in this theory of meaning, the theory of truth then is suppos ed to describe the “critical core of speakers’ actual and potential lingu istic behavior – in e ffect how the speaker intends his utterances to be interpreted.” (Davidson, 2001). Davidson further claims: “There is one intention not touched on by a theory of truth which a speaker must intend an interpreter to perceive: the force of an utterance.” (emphasis added). Here, Davidson understands, as Russell did before him, assertoric force to be primarily psychological I

PAGE 147

143 have argued in “Frege an Russell on Assertion” how this construal is liable to error. Davidson later claims, “An interpreter must, if he is to understand a speaker, be able to tell whether an utterance is intended as a j oke, an assertion, an or der, a question, and so forth. I do not believe there are rules or conven tions that govern this essential aspect of language. It is something that language users ca n convey to hearers and hearers can often enough detect…” He therefore concludes: “I believe there are sound reasons for thinking that nothing like a serious theo ry is possible concerning this dimension of language.” 112 I refer to Griffin’s article, ‘Terms, Rela tions, Complexes’, in which he discusses the views of Russell’s unpu blished manuscript. 113The act of passing judgment in the case desc ribed here would yield a judgment that fails to fall, in C. I. Lewis’s phrase, under a “mood of entertainment.” ( Analysis of Knowledge and Valuation ) 114 Griffin, 2001. 117Russell (London: Routledge, 1979) 118 MacFarlane, “Making Sense of Relative Truth” (2005) 119 It is performative in something like the sense in which Straws on understood the uses of the words “is true,” but it is not redundant. 120 I follow the analysis given in Wiredu’s “The Correspondence Theory of Judgment” (1981). The structure of the expression may be represented as follows: T, in the case that “is true” is predicated of a (complete) proposition; F in the case that “is false” is predicated of a (complete) proposition. 121 I borrow the term from Wiredu. See “T he Correspondence Theory of Judgment” 122 The latter being an affirmative or negative determination with respect to a judgeable content. See, “The Correspondence Theory of Judgment” 123 According to Wiredu, a judgeab le content (what he calls an “ideational content) may be represented more formally as a truth-func tional variable in a classical logic. While I cannot here discuss the specifics of this s uggested interpretation of propositional content, I use the upper and lower-case letters as Wi redu does in order to distinguish between statement and judgeable content 124 I will use ‘judgeable’ conten t and ‘propositional’ content synonymously throughout. 125 I say the same “kind” in order to dis tinguish what Russell calls ‘positive’ and ‘negative’ forms of predication. See Russell (1905); see also Bernard Linsky, Russell’s Early Metaphysical Logic 126 Ibid 127 We may think of context, e.g., as involv ing a speaker, world, time and place, which we shall refer to as parameters of the cont ext. Further parameters may in principle be added to this list, such as a standard of justification) 128 MacFarlane, “Making Sense of Relative Tr uth,” p. 326. There are exceptions, as noted in F. Recanati’s Perspectival Though: A Plea for (Moderate) Relativism (2007) 129 Ibid. 130 “Parameter” is MacFarlane’s term and is equivalent to what Kaplan calls the “coordinate” of an index. An index may contain a world, time, place, and agent coordinate, ( i = w, t, p, a… ). According to Kaplan, “All th ese coordinates can be varied,

PAGE 148

144 possibly independently, and thus affect the truth-values of statements which have indirect references to these coordinates.” Da vid Kaplan, “Demonstratives” (p. 508) Themes from Kaplan Almog, Wettstein, Perry (eds.) 1989 131 See Chapter VI, “Moore’s Paradox and the Logic of Assertion” 132 I borrow this term from Max Kolbel. 133 The idea of an act of asserting as co mmenting upon an existing assertion may be found in K. Wiredu’s “The Correspondence Theory of Judgment” (1980) 134 See p. 14, “Three Grades of Truth-Relativity” 135 I paraphrase this, leaving out “and the aesthetic standards of the assessor at” in (b). 136 The same would apply context-indexicality at CU and CA A proposition is said to be assessment-sensitive if its truth-value varies with CU while CA remains fixed. 137 I borrow the phrase, “The one correc t standard” from MacFarlane (2001) 138 See footnote 9. 139 On MacFarlane’s view the context of assessment at which the putative refutation is evaluated by the asserter is privileged, where the asserter is the same speaker at CU and CA 139 140 This example is Max Black’s “Saying and Disbelieving,” Analysis (1952) 141 Sorensen, R. A., Blindspots (Oxford University Pre ss/Clarendon Press: 1988). Although I find a good deal of Sorensen’s discus sion to be instructiv e, my treatment of the paradox diverges significan tly at points from his own. 142 Wolgast, E. H., Paradoxes of Knowledge (Ithaca, NY Cornell University Press, 1977)

PAGE 149

145 Bibliography Armstrong, D. M. 1993. A World of States of Affairs Cambridge. Austin, J. L. 1950. “Truth.” Proceedings of the Aristotelian Society, Vol. XXIV Reprinted in Austin, J. L. 1961. Philosophical Papers Ed. J. O. Urmson and G. J. Warnock, Oxford: Clarendon Press. Bell, David Andrew.1979. Frege's Theory of Judgement Oxford: Clarendon Press: Oxford University Press. Burge, Tyler. 2005. Truth, Thought, Reason: Essays on Frege Oxford: Clarendon Press. Church, Alonzo. 1951. Introduction to Mathematical Logic (Princeton: Princeton University Press. Davidson, Donald. 2001. Truth and Predication Oxford: Oxford University Press) Dewey, John. 1990. The Philosophy of John Dewey; The Library of Living Philosophers vol. 1, ed. Paul A. Schilpp. Open Court. Dudman, V. H. 1975. “Frege’s Judgment-stroke”. The Philosophical Quarterly 20: 150161. Dummett, Michael.1968. “Truth” Truth and Other Enigmas Cambridge: Harvard University Press. Dummett, Michael.1973. Frege: Philosophy of Language Cambridge: Harvard University Press. Frege, Gottlob.1879. Begriffschrifft eine der arithmetischen nachgebildete Formelsprache des reinen Denkens Halle: L. Nebert Translated as Begriffsschrift, a Formula Language, Modeled upon that of Arithmetic, for Pure Thought (1879) Also in Bynum, Terrell (ed.) 1972. Conceptual Notation and Related Articles London: Oxford University Press. Frege, Gottlob.1891. Grundgesetze der Arithmetik Jena: Verlag Hermann Pohle, Band I. Translated as The Basic Laws of Arithmetic. 1964. (ed.) Furth, Montgomery. Berkeley: University of California Press. “Function and Concept” 1891. Basic Philosophical Writings ed. Geach and Black. “Sense and Reference” 1892. Basic Philosophical Writings ed. Geach and Black. Grundlagen 1884 Griffin, Nicholas. 1993. “Terms, Relations, Complexes’ Irvine and Wedeking (eds.) Russell and Analytic Philosophy Toronto: University of Toronto Press. Johnson, W. E. 1921. Logic Cambridge: University Press. Kaplan, David. 1989. “Demonstratives” in Themes from Kaplan Almog, Wettstein, Perry (eds.) Oxford University Press. Kolbel, Max. 2002. Truth without Objectivity New York: Routledge. Lewis, C. I. 1946. An Analysis of Knowledge and Valuation. Open Court. Linsky, Bernard 1983. Russell’s Early Me taphysical Logic. Stanford; CSLI publications. MacFarlane, John. 2003. “Three Grades of Tr uth Relativity,” Unpub lished manuscript. MacFarlane, John. 2005. “Making Sens e of Relative Truth.” in Proceedings of the Aristotelian Society vol. 105: 321–39. Mackie, J. L. 1973. Truth, Probability, and Paradox Oxford: Oxford University Press.

PAGE 150

146 Pitcher, George. 1964. ed. Truth Prentice Hall, NJ, 1964 Prior, A. N. 1971. Objects of Thought Ed. Geach and Kenny. Oxford: OUP. Ramsey, F. P. 1927. “Facts and Propositions”, Proceedings of the Aristotelian Society Supp. Vol. VII : 153-170 Russell, B. 1912. The Problems of Philosophy Oxford: Oxford University Press. Russell, B. 1903. Principles of Mathematics New York: Norton. Russell, B. 1948. Inquiry into Meaning and Truth. London: Allen and Unwin Ltd. Russell, B. 1910. “On the Nature of Truth and Falsehood” Philosophical Essays (1966) Russell, B. Philosophy of Logical Atomism (1918), David Pears (e d.) (Open Court, 1999) Russell, B. 1999 Theory of Knowledge: The 1913 Manuscript London: Routledge. Russell, B. 1904. “On Functi ons” Unpublished manuscript Ryle, Gilbert. 1946. “”If”,”so” and “Because”” Proceedings of the Aristotelian Society vol. XLVI. Sainsbury, M. Russell. 1979. London: Routledge. Sorensen, R. A. 1988. Blindspots Oxford: Clarendon Press. Strawson, P. F. 1950. ‘Truth’, Proceedings of the Aristotelian Society Supp. Vol. XXIV Wells, R. S. 1968. Ed. E. D. Klemke. Urbana: Univ. of Ill. Press. Wells, R. S., 1968. Ed. E. D. Klemke. “Frege’s Ontology” Urbana: Univ. of Ill. Press. Wiredu, Kwasi 1960. ‘Knowledge, Truth and Reason’, Dissertation, Oxford. Wiredu, Kwasi. 1973. “Deducibility and Inferability” Mind Wiredu, Kwasi. 1980 Philosophy and an African Culture (Cambridge and New York: Cambridge University Press. Wiredu, Kwasi. 1987. “Truth: The Correspondence Theory of Judgment” African Philosophical Inquiry Wiredu, Kwasi. 1975. “Truth as a Logical Constant with an Applicati on to the Principle of Excluded Middle” The Philosophical Quarterly October. Wittgenstein, Ludwig. 1918. Philosophical Investigations Wolgast, Elizabeth H. 1977. Paradoxes of Knowledge Ithaca, NY Cornell University Press.

xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 2200409Ka 4500
controlfield tag 001 002029677
005 20090918155240.0
007 cr bnu|||uuuuu
008 090918s2009 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0002956
B72 (Online)
1 100
Kelly, Jeremy J.
0 245
Truth and judgment
h [electronic resource] /
by Jeremy J. Kelly.
[Tampa, Fla] :
b University of South Florida,
Title from PDF of title page.
Document formatted into pages; contains 146 pages.
Includes vita.
Dissertation (Ph.D.)--University of South Florida, 2009.
Includes bibliographical references.
Text (Electronic dissertation) in PDF format.
3 520
ABSTRACT: I examine the difficulties that several philosophers of language are liable to encounter in their attempts to provide an account of the connection between truth and assertion. I then attempt to provide an account of this connection. The analysis is concerned chiefly with difficulties which consist in elucidating the conceptual connection between truth and assertion in a way that respects certain linguistic intuitions while at the same time rendering the concept of truth amenable to a semantic interpretation. The proposed view suggests one way in which we might go about meeting the theoretical demands implicit in addressing this concern, among others, demonstrating the extent to which a theory of truth should be regarded as belonging to the province of epistemology. Insofar as semantical considerations figure into such a theory, a more systematic investigation of the interface between epistemology and natural language semantics is recommended. The solution to many problems at this interface, I argue, lay in an analysis of judgment.
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
Co-advisor: Kwasi Wiredu, B.Phil.
Co-advisor: Stephen Turner, Ph.D.
Natural language semantics
Philosophy of language
Dissertations, Academic
x Philosophy
t USF Electronic Theses and Dissertations.
4 856