Tarski\'s 1944 Polemical Remarks and Naess\' \"Experimental Philosophy\"

Tarski’s 1944 Polemical Remarks and Naess’ “Experimental Philosophy” Robert Barnard and Joseph Ulatowski

Abstract Many of Tarski's better known papers are either about or include lengthy discussions of how to properly define various concepts: truth, logical consequence, semantic concepts, or definability. In general, these papers identify two primary conditions for successful definitions: formal correctness and material (or intuitive) adequacy. Material adequacy requires that the concept expressed by the formal definition capture the intuitive content of truth. Our primary interest in this paper is to better understand Tarski's thinking about material adequacy, and whether components of his view developed over time. More precisely, we are concerned with how Tarski's understanding of the content of the common-sense, every-day usage of truth may have developed over time. We distinguish this concern from the character of the extensional criterion of adequacy Tarski proposes: that a materially adequate definition must entail all instances of Convention T. We will develop our reading of Tarski as follows: first, we will review the "Polemical Remarks," focusing primarily on §§14 and 17, and Tarski's references to Naess' empirical research. Next, we will provide a summary and discussion of Naess' work, especially his findings with respect to Tarski's definition of truth and his research that suggests there is no single common or everyday concept of truth. Third, we will consider several possible objections to our interpretation of the Tarski-Naess dialectic. We will conclude that Tarski's conception of what the material adequacy requirement develop over time, potentially because of what he had learned through his interactions with Naess.

Many of Tarski’s better known papers are either about or include lengthy discussions of how to properly define various concepts: truth (e.g. Tarski 1944, 1983a), logical consequence (1983c), semantic concepts in general (1983b), or definability (1948). In general, these papers identify two primary conditions for successful definitions: formal correctness and material (or intuitive) adequacy. In the case of truth, a definition is formally correct when it is constructed in a manner that allows us to avoid both circular definition and semantic paradoxes (cf. Field 1972, Patterson 2008, Patterson 2012, Soames 1999, Sher 1999). Material adequacy requires that the concept expressed by the formal definition capture the intuitive content of truth. In other words, a stipulative definition of truth is offered, but it must “reflect the notion of truth in its commonsense and every-day usage.”1 Our primary interest in this paper is to better understand Tarski’s thinking about material adequacy, and whether components of his view developed over time. More precisely, we are concerned with how Tarski’s understanding of the content of the common-sense, every-day usage of truth may have developed over time. We distinguish this concern from the character of the extensional criterion of adequacy Tarski proposes: that a materially adequate definition must entail all instances of Convention T. In “The Concept of Truth in Formal Languages” (1983a, 1

Tarski 1944, 360. 1

hereafter referred to as “CTFL”) from 1933, material adequacy requires that the proposed definition be capable of “grasping the intentions which are contained in the so called classical conception of truth (‘true--corresponding with reality’)” (CTFL, 153). Here, the discussion of “everyday” or “colloquial” uses of ‘true’ is one that “every reader possesses in greater or less degree” (Ibid.). Citing the common reader is not to focus upon them and their usage of “true”, etc., but to contrast the classical conception with the pragmatic conception, i.e., true as useful. In “The Semantic Conception of Truth and the Foundations of Semantics” (Tarski 1944; hereafter referred to as “SCT”), material adequacy is discussed at greater length and with an alternative focus. Tarski writes that “the desired definition does not aim to specify the meaning of a familiar word used to denote a novel notion; on the contrary, it aims to catch hold of the actual meaning of an old notion” (SCT, 341). This ‘old notion’ is understood to “do justice to the intuitions which adhere to” the classical Aristotelian notion: “To say of what is that it is not, or of what is not that it is, is false, while to say of what is that it is, or of what is not that it is not, is true” (SCT, 342). But Tarski also considers a variety of other formulations in terms of agreement or correspondence. [T]rue, like other words from our everyday language is certainly not unambiguous. … [A]ll these formulations can lead to various misunderstandings, for none of them is sufficiently precise and clear…; at any rate, none of them can be considered a satisfactory definition of truth. It is up to us, to look for a more precise expression of our intuitions. (SCT, 342-343) This sounds like the same project outlined in CTFL, and Tarski states that part I of SCT is offered as an informal summary of CTFL, but part II of SCT gives us pause. We will suggest that SCT and CTFL can be read as marking two moments in Tarski’s understanding of both the “commonsense” view and more specifically the content of the material adequacy requirement.2 Clearly, the combination of formal correctness and material adequacy is a distinctive feature of Tarski’s efforts to define various concepts.3 While there are many fine expositions of Tarski’s formal work (cf. Field 1972, Horsten 2011, Patterson 2008, Patterson 2012, Soames 1999), our far narrower aim is to sketch an alternative reading of some of the later sections of SCT that highlight the interactions Tarski had with the often overlooked work of Arne Naess (especially

2

Jané (2006) and Betti (2008) have argued that what Tarski meant by the “common concept,” particularly of logical consequence, was common not in ordinary language and its usage by ordinary people but in axiomatics and its usage by specialists. We discuss this approach in §3 below. 3 Künne (2003) also notes the important role of the content of the conception of truth, for Tarski, as opposed to its formal implementation. What makes the formal definition a definition of “truth” (and not a mere stipulation) is that it is “obliged to capture, as far as possible, the meaning of 'true' in ordinary truth-ascriptions...It makes no sense to demand of a stipulative definition that it be materially adequate” (Künne 2003, 213). 2

1938a, 1938b).4 Specifically, we will suggest that the “polemical” sections of SCT, especially those focusing, directly or indirectly, on the issue of material adequacy, are fruitfully informed by a better understanding of Naess’ work; we do not directly call into question the more technical and formal elements of either CTFL or SCT.5 We will develop our reading of SCT as follows: first, we will review the second part of SCT, focusing primarily on §§14 and 17, and Tarski’s references to Naess’ empirical research. Next, we will provide a summary and discussion of Naess’ work, especially his findings with respect to Tarski’s definition of truth and his research that suggests there is no single common or everyday concept of truth. Third, we will consider several possible objections to our interpretation of the Tarski-Naess dialectic. We will conclude that Tarski’s conception of what the material adequacy requirement demands in SCT is different than CTFL, potentially because of what he had learned through his interactions with Naess. 1. Tarski’s Polemical Remarks Recently, Douglas Patterson (2012) has argued that a proper understanding of Tarski’s thought depends upon an appreciation of a complex set of background philosophical commitments.6 Patterson and Künne (2003, 14, 190ff, and passim) both argue that such an understanding of Tarski assumes an appreciation of the Austrian and Polish philosophical and logical context in which works like CTFL were written. We will suggest that Naess’ work might play a similar role when reading SCT. Our proposed reading will suggest that Tarski’s stance toward the ordinary usage of the term “true” developed over time, softening the presumption that the ordinary person’s view of truth is easily known, because, we believe, of what he had learned about Naess’ “experimental philosophy” of truth. While SCT is described by Tarski and by commentators (cf. Sher 1999) as an informal summary of his earlier work, there are some features that set it apart.7 We want to highlight and better understand these differences. CTFL is about how to define truth for specific formal languages. But SCT includes less formal elements; we want to suggest that the non-technical

4

See Crockett (1959) and Chapman (2008, Ch. 6) for discussion of Naess and his work. It should be noted that this follows Naess’ own view, that empirical studies of non-philosophers cannot affect the formal elements of Tarski’s definition of truth (TCNP, §97). 6 As Patterson summarizes: “Lurking in the background of Tarski’s project, then, appears to be a rather articulate view of concepts, meanings, definitions, formalism, ordinary language, and related topics--a set of views in philosophy of logic, language, and to some extent mind. In order to understand Tarski’s work we need to know what these views were” (Patterson 2012, 16). One might disagree with Patterson’s interpretation by comparing what he said with what Tarski claims in fn. 4 of SCT. Regardless of which interpretation one supports, the purpose of introducing this summary is to show that a difference of interpretation is possible. 7 For detailed discussions of the specific differences between CTFL and SCT, see Patterson 2012, Introduction and Ch. 8, and David 2008. 5

3

polemical remarks that form the second part of SCT indicate that Tarski’s thinking about how his formal project relates to the common or everyday notion of truth evolved. The second part of SCT is dedicated to presenting “various objections, of unequal value,” (1944, 341) to Tarski’s investigations regarding the definition of truth. Some of these “objections” had appeared in print, e.g., Neurath (1983, 100-114) and Gonseth (cf. Tarski 1944, fn31), and others had emerged from public and private discussions, including those had at the 1935 and 1937 Paris Congresses on the Unity of the Sciences (Mancosu 2010, 418-422). Tarski, however, hoped his remarks would not be taken as purely polemical, “but will be found to contain some constructive contributions to the subject” (SCT, 341). Of primary interest to the question of whether a proposed definition of truth is materially adequate and its connection with the “old notion,” the “everyday,” or “common-sense” conception of truth are §14, “Is the Semantic Conception of Truth the “Right” One?” and §17, “Conformity of the Semantic Conception of Truth with Philosophical and Common-Sense Usage,” of SCT. §14 is a series of remarks about whether the “Semantic Conception” is the single correct conception of truth. Tarski writes that he neither maintains that the semantic conception of truth is “the ‘right’ nor indeed the ‘only possible’” view (SCT, 355); instead, we are directed to attend to the “vagueness and ambiguity” of “common language” and that we should “reconcile ourselves with the fact that we are confronted,” in the case of truth, “not with one concept, but with several different concepts which are denoted by one word” (Ibid.). Tarski suggests that, “we should try to make these concepts as clear as possible (by means of definition, or of an axiomatic procedure, or in some other way); to avoid further confusions, we should agree to use different terms for different concepts” (Ibid.). In turn, these concepts can be systematically studied “which will exhibit their main properties and mutual relations” (Ibid.). §14 is one of the less emphatic polemical sections, for it repeats much of the content of §3, “The Meaning of the Term ‘True’.” For example, in §3, Tarski states: The word “true,” like other words from our everyday language, is certainly not unambiguous. And it does not seem to me that the philosophers who have discussed this concept have helped to diminish its ambiguity. In works and discussions of philosophers we meet many different conceptions of truth and falsity, and we must indicate which conception will be the basis of our discussion. (SCT, 342) In the §14 discussion, Tarski cites the work of Arne Naess to illustrate the “endless, often violent discussions” of the question “What is the right conception of truth?” (SCT, 355). In his 1938 monograph Truth as Conceived by those who are not Philosophers (hereafter referred to as “TCNP”), Naess motivates his project by admitting his dissatisfaction with the diversity of published philosophical theories where each author claims to have uncovered the one, singular, aim of “truth-theories” (TCNP, 12ff; Naess 1938a, 39f). Tarski is here marking an explicit affinity with Naess. They are both frustrated by the manifest indecision between and among

4

philosophers on this question of central philosophical importance, and they both seem to be in agreement that there is no one right conception of truth. What is distinctive about §14 is the final discussion of an intractable conflict between competing precise conceptions of truth. Tarski proposes a “future world congress of the ‘theoreticians of truth’ to decide which conception to baptise as “true”. “But,” he notes, “I cannot imagine that anyone could present cogent arguments to the effect that the semantic conception is ‘wrong’ and should be entirely abandoned” (SCT, 356). As it turns out, Naess’ data suggest that non-philosophers’ responses to questions regarding truth can be understood as compatible with the semantic conception. This does not mean that non-philosophers endorse or fully accept a view as detailed as Tarski’s semantic conception of truth. Rather, Naess finds that a view approximating the less formal elements of Tarski’s semantic conception is one among many notions of truth articulated by at least some of Naess’ titular ‘non-philosophers’. In any event, Tarski is suggesting that the project of trying to isolate a single common concept of truth might be more trouble than some, including his earlier self, may have thought.8 On this point, we think, Tarski and Naess agree. §17 attends to the question of “whether the semantic conception of truth can indeed be regarded as a precise form of the old, classical conception of this notion” (SCT, 359). While Tarski rejected as impractical a process of asking individual truth theorists, some alive, some dead, if they agree with the semantic conception of truth, §17 does contain the following striking passage: I happen to believe that the semantic conception does conform to a very considerable extent with the common-sense usage although I readily admit I may be mistaken. What is more to the point, however, I believe that the issue raised can be settled scientifically, though of course not by a deductive procedure, but with the help of the statistical questionnaire method. As a matter of fact, such research has been carried on, and some of the results have been reported at congresses and in part published. (SCT, 360) Tarski again cites Naess’ TCNP, but this time he appears to endorse the specific method employed by Naess (the statistical questionnaire method). While Tarski does go on to urge caution against the careless use of such a method, he appears to see no real problem with statistical results. Tarski writes:

8

In CTFL, Tarski repeatedly treats the Aristotelian formulation that he calls the classical conception of truth as the target for analysis, and his stated aim in CTFL is to “introduce a consistent and correct use” for the concept by treating it as a primitive notion in an special science, “namely the theory of truth” (see especially CTFL, 152-154). He is clearly interested in isolating a single preferred conception of truth that is intuitively plausible by his lights. Our reading suggests that his appreciation of Naess’ work may have allowed him to recognize the possibility that his classical conception may not be the best way of capturing the ordinary common-sense notion. Indeed, there may not be a single common sense notion. 5

I was by no means surprised to learn (in a discussion devoted to these problems) that in a group of people who were questioned only 15% agreed that “true” means for them “agreeing with reality,” while 90% agreed that a sentence such as “it is snowing” is true, if and only if, it is snowing. Thus, a great majority of these people seemed to reject the classical conception of truth in its “philosophical” formulation, while accepting the same conception when formulated in plain words. (SCT, 360)9 It is still possible to interpret these remarks in incompatible ways. An uncomplicated, perhaps somewhat pollyannaish, reading of this passage has Tarski embracing the philosophical relevance of Naess’ empirical results. The results are informative, untroubling, and he finds exactly what he expected to observe in Naess’ data. On a more pessimistic reading of the same passage, Tarski is presenting the results in order to deflate their relevance. On such a reading, the data show exactly what Tarski thought they would, i.e., that his view and the ordinary person’s view coincide, but the results are treated as uninformative.10 We believe the proper reading will probably fall somewhere in between the pollyanna and the pessimist. That said, we want to try to block the pessimistic reading. If Tarski is saying that these results are irrelevant, then the pessimistic reading invites us to assume that all of the references to Naess in the polemical sections are similarly dismissive. However, such a pessimistic reading is incompatible with Tarski’s citation of Naess in §14, which supports his frustration with an incompatible variety of published theories of truth. Likewise, when in §17 Tarski writes “What is more to the point, however, I believe that the issue raised can be settled scientifically, though of course not by a deductive procedure, but with the help of the statistical questionnaire method,” the pessimistic reading would require that we read Tarski as rejecting a scientific approach to the problem (SCT, 360). Doing so would not make sense on Tarski’s view. Thus, the pessimistic reading does not fit with Tarski’s references to Naess. Material adequacy is, in part, about hitting a mark, or at least conforming “to a very considerable extent with” the common-sense everyday usage of truth.11 Tarski has made clear that he recognizes the wide range of incompatible ways that truth is characterized in the literature and that he might need to employ statistical research in support of his view that the classical conception of truth is the proper subject of his analysis. To better understand Tarski’s claim that a statistical investigation might help to resolve these disagreements we must turn to a more 9

Tarski does not ascribe these statistics to Naess, but the context strongly suggests that these are Naess’ empirical findings since they align well with what he reports in TCNP. 10 Tarski may have actually endorsed Naess’ empirical work in a more concrete way than we let on in our exposition above. According to Naess, Tarski himself acknowledged in a postcard that Naess’ work showing that the semantic conception “applies clearly to a subgroup of occurrences… might be right” (Naess 1993, 18; 2005b, 270; cf. Chapman 2008, 119ff). 11 We should take note that Tarski believes a “fixed and exact meaning” of “true,” like the one he proposes as the formally correct definition, “implies necessarily a certain deviation from the practice of everyday language” but not a complete rejection of ordinary usage (Tarski 1944, 360). 6

detailed examination of Naess. What we hope has become clear is that Tarski is not indifferent to Naess’ results about everyday usage of the term “true.” Indeed, Tarski writes as if these results are salient to the issue of correctly understanding the commonsense concept of truth. 2. Naess and the “Experimental Philosophy” of Truth In 1938, Naess published TCNP (1938b), a monograph presenting and analyzing partial results from a series of empirical studies, and “Common-sense and Truth” (1938a, hereafter “CT”), an article published in Theoria which was largely an overview of the results reported in TCNP. Some of this material originated from Naess’ 1933 masters thesis and his 1936 doctoral dissertation (Chapman 2008, ch. 6; Naess 1936, 20; Naess 1993, 17ff.; Naess 2005b, passim; Naess 2005c, 265ff.). Naess was primarily concerned with the tension between assertions some philosophers had made regarding their ability to intuit what the common person thought about truth, and these same philosophers claiming that their own singular, monolithic conception of truth accurately reflects what the ordinary person’s view of truth is. The project of surveying non-philosophers about the “common characteristic” of truth was conceived as a means to explore this tension empirically and scientifically. How do the philosophers know these things? What is the source of their knowledge? What have they done to arrive at it? Much work with this treatise would have been saved (and - it may be added - many of our theses would in this circumstance turn out to be untenable) if the philosophers had indicated how they investigated the opinions of the non-philosopher (the amateur) and how they arrived at the conclusion that there are thoroughgoing differences between opinions (explicitly or implicitly) of philosophers and non-philosophers. But the fact remains their writings contain almost nothing of this matter. Perhaps some of them have asked their wives or assistants for their opinions on the truth-notion, but there is very little to prove that they actually employed such a method. If they had worked out a method to determine truth-theory of a man inspecting his general behavior, why do they not mention this method? Even very superficial questioning of non-philosophers would make it almost impossible for anyone to believe that the philosophers writing about the opinions of ordinary people actually ask others than themselves. (TCNP, 15) It does not help that philosophers--most of whom claim to be explicating the ordinary person’s account of truth--fail to agree (CT, 39; TCNP, 14f). Indeed, following a survey of papers on truth, Naess describes the confusing state of the professional literature. It is clear that there is little agreement among philosophers about the nature of truth,

7

A bewildering multiplicity of possible interpretations are given and their exponents indulge in acute polemics with each other. Some distinguish sharply between a criterion and a definition of truth, or between a sufficient and necessary criterion of truth and the essence of truth, others think that no such difference exists. Some speak about true beliefs, others about true propositions, and others of true things. Some speak about the truth, others about truths and others about true relations, all criticizing each others’ standpoints. Some find the word true undefinable, others definable. Some find it contrary to scientific tact to use it, others find it impossible to leave it. (TCNP, 12) The discussion has already lasted some 2500 years. The number of participants amounts to a thousand, and the number of articles and books devoted to the discussion is much greater. The number of standpoints felt as different or incompatible may be said to be 2, or 100 or 1000 according to the criteria adopted. (TCNP, 14) The impasse in the professional literature seems easily surmountable to Naess if we are prepared to undertake a scientific investigation of the ordinary person’s conception of truth, collecting the views of ordinary people using a statistical-questionnaire method. To refuse to perform these studies would be tantamount to giving “the impression that some sort of investigation [had] been carried out” (CT, 39), when the philosopher knows full well that they had not. So, Naess decided to ask ordinary people what they think the common characteristic of truth is. Naess’ research program consisted in an analysis of a series of qualitative interviews and survey results for approximately 250 northern Europeans (CT, 45; TCNP, 40).12 Anyone who had compromised their “philosophical virginity,” whether by studying philosophy at university or through a serious study of philosophy on their own, was excluded from the participant pool (CT, 46; TCNP, 46). Interviewers were not philosophically sophisticated and were given specific instructions on how to contend with study participants (cf. TCNP, 20). Naess’ interviewers used open-ended questions allowing respondents’ answers to arise organically in conversation and permitted study participants to give more than one response to any one of the questions. Investigators asked respondents questions like: What is the common characteristic of that which is true? What is the common property of true? What distinguishes that which is true? What are some common characteristics of things that are absolutely true? 12

85.1% of the participants were Norwegian, 8.7% Austrian, 3.4% Swede, and 3.2% either German or English (TCNP, 44). 48% of the respondents had completed grammar school, and 38.6% of the respondents had completed a baccalaureate (TCNP, 45). Participants were aged 12 to 65 years old (CT, 46; TCNP, 44), and of Group-1 (the first 150 participants) 55% were male and 45% were female (TCNP, 45). The duration of the interviews lasted no less than 10 minutes, while 2% carried on more than 10 hours (CT, 44; TCNP, 44). 8

Why do you use the word true? (CT, 46; cf. TCNP, §§7-11) The interview data was coded and analyzed by Naess and his colleagues, based upon a linguistic analysis Naess had prepared prior to the execution of the experiment (TCNP, §§13, 20-23). The data yielded 278 distinct formulations for Group-1, the first 150 participants of Naess’ studies, (TCNP, §21) and some 500 definitions for the total number of study participants (CT, 47). In Group-1, the two formulations most often expressed were correspondence or agreement with reality and correspondence or agreement with facts. He called these two formulations Aand B-formulations, respectively. A-formulations are “short general statements on truth [and similar notions] except “fact”, having the character of definitions” and include participants’ responses like: agreement with reality, a statement (is true) if its contents agree with reality, agreement with real things, it is the case, that it actually is so, it is so, and when I hold that it is like that and it appears to be like that (TCNP, 39). B-formulations are “formulations corresponding to the [A-formulations], but dealing only with the notion of fact” and include participants’ responses like: a fact, it is a fact, agreement with facts, that it is pure fact, what one may ascertain is a fact, that one states lucid facts, and the fact of the case (TCNP, 39f).13 Neither A- or B-formulations, nor any of the other formulations identified by Naess were common enough to be called the one and only common sense definition of truth. Naess writes, “No definition of truth can be said to express the non-philosophers’ view of truth. There is no such view.” (CT, 47) “[My study] reveals how contrary to sound principles it would be to choose or crown any special kind of formulation to express “the non-philosopher’s view of truth” or “the non-philosopher’s theory of truth” when frequency of occurrence is taken as [a] criterion” (TCNP, 69f). The correspondence formulation types were not widely held by participants in his study. Naess wrote, “no group comprises as much as 15% of the [A-formulation]. [Among] the 10 greatest groups according to frequency of occurrence contain, on an average, only 3.4% of the [A- or B-formulations]” (Ibid.). Hence, Naess concluded: It is very difficult to see why one of these (most frequent) groups [so called A- and Bformulations] or both taken together should be called ‘the common-sense theory of truth’. There is no evidence in support of the hypothesis that anyone has made any attempt to investigate “common sense” in the sense required. Philosophic “truth theories” seem wholly to be the fruits of “contemplation” and “intuition.” (Ibid.)14

13

Naess reports on other formulations, too. D-formulations “substitute the word “true” for the word “right” or vice versa,” e.g., “what people with authority say is true” or “that it fulfills the requirements we ourselves make of that which we think is right.” M-formulations are those that are “concerned with moral standards,” e.g., “that conscience does not protest against it” or “the opposite of a lie” (TCNP 40, 57-60). 14 This result is of specific interest to some, like Tarski in CTFL, who claim that the common notion of truth is best understood as an expression of something like a correspondence theory. 9

Despite failing to find the common person’s view of truth, Naess felt confident enough to present a few key results: a) a majority of the participants rejected formulations involving correspondence (that do not also involve appeal to facts), b) non-philosophers think professional formulations of the “truth problem” are better than amateur formulations, and c) the positions proposed by professional philosophers and the arguments offered in support of them are “refound among the amateurs” (TCNP, 152). As Tarski indicates in SCT, the results from Naess most relevant to SCT are not published as part of TCNP.15 A non-formal version of Tarski’s Convention T was addressed briefly in a short series of follow-on questions.16 In one question, interviewers were instructed to give a professional philosopher’s definition of truth. “PAf148” was the non-formal version of Tarski’s Convention T.17 Interviewees were asked how willing they were to substitute “p” for “‘p’ is true,” and leaders provided some examples, such as “‘It is raining’ is true if and only if it is raining.”18 Notably, one person (P1) said the definition was “useful in practice,” while another person (P55) claimed, “This seems to turn round in circles” (TCNP, 145). Despite the snarky response of P55, other positive responses were offered, but regrettably these data were not reported in detail in TCNP. Our aim is not to argue that Tarski is either right or wrong. Rather, we note that Naess knew of Tarski’s early work on truth when the relevant survey work was performed, and Tarski was familiar enough with Naess’ empirical work to include two references to unpublished questionnaire results in SCT. Second, in SCT, Tarski says Naess’ results regarding the semantic conception are unsurprising, but these results show that non-philosophers affirm a wide variety of philosophical truth-theories, one of which seems to be compatible with Tarski’s view. Finally, Naess’ results present a picture where there is not one clear candidate for the ordinary person’s view of truth, a conclusion with which Tarski agrees (cf. SCT, §14). We believe that Tarski’s 15

See Tarski 1944, 374n29. The relevant passages are found in TCNP, 144-148. The data most relevant to Tarski’s work was from class “E” questionnaires. This data was omitted from both TCNP and CT (cf. TCNP, 22). Naess does briefly discuss some class “E” questionnaire data in Naess 1937/1938, 383. 16 The follow-on item we mention in this paper asks study participants about whether they agree or disagree with a professional philosopher’s definition of truth. Other follow-on studies asked participants whether another participant’s definition sufficed for truth. (CT, 50f; TCNP, 125153). 17 Naess did not take himself to be evaluating Tarski’s formal work. Rather, Naess was assessing whether something like a T-sentence was an expression of truth in the non-formal sense. Naess specifically writes in a footnote that “The “PAf 148” (Tarski) is not to be identified with the socalled “semantic notion of truth”. To construct this notion the method of formalization is essential. There is, however, a tendency to look at PAf 148 as a definition of non-formal truth. It was therefore included in our lists.” (TCNP, 148, fn1). 18 We should note here that some people might criticize Naess’ approach because Tarski’s view is not interpreted catholically. While we agree with such a criticism, it is what was reported in Naess’ TCNP and it was what Tarski cited in n29, regardless of how we might feel about it as an interpretation of Tarski. 10

polemical remarks in §§14 and 17 of SCT are best understood as reflecting the interactions with Naess and that these two sections were included in order to show that the definition of truth, or components thereof, he had offered might be discovered to be consistent with the ordinary person’s view of truth, though he could not conclude that it was because he had not performed the kind of empirical studies that Naess had completed. There seems to be little evidence supporting the idea that the “commonsense” or “ordinary” person’s view of truth included running an empirical study to discover what it is in the earlier CTFL. His understanding of the role of the ordinary person’s view appears to have been altered by conversation or correspondence with Naess. If the intellectual background against which Tarski was working really matters to understanding Tarski correctly, then the apparent salience of Naess’ work cannot be neglected. If Tarski’s background understanding of the common view changed, then our understanding of what Tarski is up to in 1944 should change too. 3. Discussion Whether Naess’ empirical work informed the development of Tarski’s thinking about truth between 1935 and 1944 seems to be a matter of controversy. The textual evidence in Tarski is scant enough that we cannot definitively say that Naess played an integral role in the development of his thinking. We do have Tarski’s own references to Naess and the apparent endorsement of the statistical questionnaire method. We have argued that Tarski’s thinking about material adequacy, especially the conceptual content of the everyday conception of truth, seems to have changed over time. His undivided concern with the formal definition of truth in the 1930s seems to have given way to a deeper concern with the conceptual basis of Convention T in his later work, particularly 1944, one that involves the results of Naess’ statistical questionnaires. There are three main ways in which a critic might resist our Naess informed reading of SCT. First, we have suggested that Tarski reads Naess sympathetically. This might strike many as contrary to the tradition according to which Popper and Carnap argue that Naess’ empirical work was a response to Tarski in an attempt to undermine the project laid out in CTFL. We will argue that the chronology of Naess’ research and interactions with Tarski do not support this reading. A second criticism is found in the work of those who, like Jané (2006), have argued that Tarski is actually invoking a specialized axiomatic notion of “common” concept of truth. We will argue that although this reading is plausible in light of Tarski’s early work, it will make other remarks of his appear inconsistent or contradictory, especially in the case of SCT. Third, some might argue that SCT should be read as a rejection of the philosophical side of the issue, an objection Douglas Patterson anticipated by claiming that the polemical remarks were a way of getting the rest of the philosophy world to “stop bothering him” about truth. We will suggest that such a reading gives rise to a perplexing instability within SCT. On our reading, taking Tarski at his word seemingly eliminates the putative inconsistencies that crop up for other readings. Thus, “later” Tarski is not necessarily “lighter” Tarski.

11

Karl Popper and Rudolf Carnap have argued that (1) it was Tarski and others attending the 1936 Copenhagen Congress that motivated Naess to do empirical work on the non-philosophers’ views of truth and (2) Naess took up his empirical studies in order to undermine Tarski’s semantic theory of truth. If this is correct, then it appears that it was not Naess who influenced Tarski but the other way around. Several years after the 1936 Copenhagen Congress Popper wrote: At the Copenhagen Congress, in 1936, which Carnap did not attend, I tried to show that the doctrine of the unity of science and of the one universal language was incompatible with Tarski’s theory of truth. Neurath thereupon suggested in the discussion which followed my lecture that Tarski’s theories about the concept of truth must be untenable; and he inspired (if my memory does not deceive me) Arne Naess, who was also present, to undertake an empirical study of the usages of the word ‘truth’, in the hope of thus refuting Tarski. (Popper 2002, 363n44) According to Popper, he and members of the audience at the 1936 Copenhagen Congress, such as Alfred Tarski and Otto von Neurath, inspired Naess to do empirical studies on nonphilosophers’ views of truth in order to refute Tarski’s semantic conception.19 Nowhere in Naess does he argue that his experiments on the ordinary person’s use of ‘true’ (and its cognates) set out to prove Tarski’s semantic conception of truth incorrect (cf. TCNP, 125-130). Nowhere in Naess does he argue that the ordinary person’s responses to the questionnaire ought to be taken as conveying accurately what Tarski’s semantic conception is. In fact, in TCNP §1, Naess tells us how he had grown dissatisfied by the vagaries of philosophical 19

Popper’s interpretation of the events is ambiguous. We do not know whether it was Neurath or Naess that “hoped” for Tarski’s semantic theory to be refuted. Perhaps Neurath despised Tarski’s semantic conception of truth so much so that he recommended Naess perform some experimental studies on the ordinary usage of truth to prove Tarski wrong (cf. Neurath and Cohen 1946; Neurath and Cohen 1973). Naess may have agreed, and if so, Naess aimed to refute Tarski’s theory of truth. On the surface, this is the argument of the passage; however, there is nothing appearing in Naess’ work that supports such an interpretation. An equally good interpretation of Popper’s passage would have it that Neurath hoped Naess would find empirical evidence refuting Tarski’s theory, without Naess taking up a position either for or against Tarski’s semantic conception (cf. TCNP, 10-15). We know that Neurath rejected the formal terminology Tarski’s semantic conception of truth employed because he thought it would lead to “all kinds of metaphysical speculations,” something he and other members of the Vienna Circle despised (Mancosu 2010, 419). In Neurath’s view, Tarski’s moving from formal correctness to the ordinary usage of ‘truth’ was illegitimate (Ibid.). Neurath wanted Tarski to offer an anti-metaphysical and strictly empirically-based theory of truth (Ibid.). Therefore, it is entirely consistent with what Popper has said that it remain an open question whether Naess sought to undermine Tarski or not. There is some evidence that suggests the open question ought to be closed, and, in the beginning of this section, we provide reasons for thinking it ought to be closed. 12

theorizing. “Pragmatic” and “intellectualistic” accounts of truth were both appealing without either one of them being overwhelmingly convincing (TCNP, 11).20 The “truth notion” problem was “to some degree capable of solution” beginning with “the opinion of the ordinary man (“the non-philosopher”)” (CT, 44f). So, Naess, writing in 1938 claims that he set out to explore the interconnections, or lack thereof, between the “intellectualistic” and the “pragmatic” accounts of truth, not to thwart Tarski. The claim that Naess was “inspired to undertake” empirical studies following the Copenhagen Congress seems inaccurate, too. According to Naess, the empirical research on the ordinary conception of truth had been well underway by 1933: “[M]y master’s thesis was about various usages of the term true” (Naess 2005c, 270).21 Naess could not have begun his studies with the intention of undermining Tarski’s semantic theory.22 Naess’ testimony has been supported by correspondence between Neurath and the eminent Polish logician Marja Kokoszyńska. In a letter dated 23 April 1936, two months before the Congress convened, Neurath wrote a letter to Kokoszyńska criticizing her argument opposing his ‘sociological’ account of truth by showing how Naess’ work “undermine[s] the idea that the “semantic” conception had any better right to claim to capture some sort of ordinary notion of truth than the sociological definition” (Mancosu 2010, 431).23 Given that the Congress took place from 21 June 1936 to 26 June 1936, it cannot be that the debate Popper and Neurath had on Tarski’s theory of truth “inspired” Naess to “undertake an empirical study” of truth, because the work was already underway (cf. Holton 1993, 73). Popper was not the only one to criticize Naess’ work. Rudolf Carnap claimed that the data Naess collected were not very useful for problems that arise in semantics.24 Carnap writes: 20

The distinction between “intellectualist” and “pragmatic” accounts of truth is not discussed at length in Naess’ TCNP or CT, but in order to distinguish between the two he cites the debate regarding the notion of truth had between F.C.S. Schiller and Russell in Journal of Philosophy, Psychology, and Scientific Method (Naess CT, 57f). 21 These studies were a part of the research he completed in 1933 while writing his thesis, Diskusjonen omkring sannhetsbegrepet [The Discussion about the Concept of Truth], for the Masters degree at the University of Oslo. The empirical data later reappeared (in part) in his Ph.D. dissertation that was completed in 1936, Erkenntnis und wissenschaftliches Verhalten, and subsequently published in German (Naess 1936). 22 This has been corroborated by conversations Naess had with David Rothenberg (1992). Rothenberg has reported that Naess contended the empirical studies began years before he attended meetings of the Vienna Circle. Because Naess began attending meetings of the Vienna Circle in 1934, Naess started collecting data well before the 1936 Copenhagen meeting (cf. Chapman 2008, 110f). 23 Also, we should note that Neurath’s interpretation of Naess’ data does not actually assert that Tarski’s semantic theory is refuted or disproved by Naess. Rather, the claim seems to be that Naess’ results challenge any claim to the effect that Tarski’s common concept is the only proper way to conceive the ordinary notion of truth. This is consistent with our interpretation. 24 Naess later sought to explain his disagreement with the empiricism of both Carnap and Neurath. His disagreement was directed against the “fundamental theses and trends in the Circle” 13

Tarski says further that the characterization given is also in agreement with the ordinary use of the word ‘true’. [...] [I]t may be remarked that Arne Ness [sic] has expressed some doubts about the assertions, based on systematic questioning of people. At any rate, this question is of a pragmatical (historical, psychological) nature and has not much bearing on the questions of method and results of semantics. (Carnap 1961, 29) For Carnap, there is not much use in calling on empirical data to support or to reject Tarski’s semantic conception of truth because the “pragmatical” features of such a study might not inform semantics. First, Carnap did not believe that the vague notion of truth employed by the untrained novice was identical with the more refined formally correct definition logicians used, yet he seemed to believe that the non-philosopher’s and the logician’s views of truth were compatible in virtue of how we come to appreciate his view of explication. By Carnap’s analysis, a vague term, what he called the explicandum, is transformed into or explicated by a formal term, or the explicatum (Cf. Carnap 1956, 7f; Carnap 1967, 3). Carnap asserts generally, “I do not think that a semantical concept, in order to be fruitful, must necessarily possess a prior pragmatical counterpart” (Carnap 1956, 235). But, in the case of the semantical concept of truth, Carnap admits, “the semantical concept of truth has as its explicandum the concept of truth as used in everyday language” (Carnap 1956, 8).25 Thus, the content of the truth concept being explicated is tied with the content of ordinary usage (cf. Carus 2007a, 273-309; 2007b, 40ff).26 Next, Carnap believed that Naess doubted Tarski’s semantic conception coincided with the ordinary use of true, “based on systematic questioning of people.” Naess’ skepticism may not have been limited to Tarski’s view. Naess may have “expressed some doubts” about Tarski’s that had been mistakenly understood by Neurath merely to be a “proposal for modifications which were already accepted in principle” (Naess 1968, 13f). Naess’ recollections of the events came from a paper he prepared between 1937 and 1939, which was not published until 1956 (Naess 2005a). In a review of the work, Eivind Storheim (1959) argued that Naess found Carnap’s view too myopic to accept citing as his chief complaint against Carnap a reluctance to accept that there ought to be empirical justification of the equivalence between expressions of natural language and of logic. 25 On an interesting note, Carnap seemingly endorses one of Naess’ later studies on the synonymity of expressions. Carnap writes: “Although there may be different opinions concerning some features of the various [empirical] procedures, it seems to me that [Naess’] book marks an important progress in the methodology of empirical meaning analysis for natural languages” (Carnap 1956, 241fn4). To be clear, however, Carnap sees that Naess’ work serves a very specific philosophic niche, one that likely does not affect various semantic concepts upon which he is focused. 26 We should note that a very interesting paper, forthcoming in this journal, has spoken to how Carnapian explication can be used to “clarify” and “extend” experimental philosophy’s appeal. We believe that what we say here regarding Carnap’s methodological considerations for concept formation is compatible with theirs, cf. Shepherd and Justus forthcoming. 14

semantic conception because the data he had collected did not support Tarski’s view alone (CT, 47; TCNP, 96-162). Carnap seems to imply that Naess’ healthy skepticism entails a refutation of Tarski’s semantic conception; however, expressing doubt about a particular view, especially one that depends upon empirical data, does not undermine it. Naess may have known that some data were congruent with Tarski’s semantic conception, while data also might be interpreted as well matched with a host of other philosophical theories of truth (TCNP, 169-171). Therefore, the empirical work neither fully supported nor completely refuted Tarski’s semantic conception.27 In sum, the fact that Naess began his empirical work on truth in relation to his 1933-36 graduate studies and the existence of the Neurath-Kokoszyńska correspondence clearly show that Naess’ research was not undertaken in an effort to refute Tarski and, despite Neurath and Carnap’s insistence, there is no clear evidence that either Naess himself or Tarski saw it as an attempted refutation.28 Whereas defending Naess against Popper and Carnap’s contention that he was responding to Tarski and attempting to undermine his formally correct definition of truth seems rather straightforward, a potentially more troublesome criticism for our interpretation comes from recent work by Ignacio Jané (2006). Jané has argued that Tarski’s “common concept” of consequence is properly read as the concept “common” in axiomatics. Arianna Betti (2008) has supported Jané’s thesis through the historical record, i.e., the work in Polish axiomatics of Kazimierz Ajdukiewicz. The possible objection that comes from Jané and Betti is this: when Tarski developed the material adequacy condition, he meant it to require that a formal definition of truth must conform to the conception of truth common to those working in the relevant domain of axiomatic theory. Although we find the work of Jané and Betti remarkable and doing yeoman service for getting clear on the role of the “common concept” in Tarski’s early work, we remain unconvinced that their analyses apply to the later polemical sections of SCT.29 According to Jané (2006), the “common concept” of consequence that Tarski promoted was the concept common in axiomatics, not the one operative in “everyday language” or “common usage” among ordinary people. Jané has argued that the common concept of consequence 27

The unsurprising statistical results noted by Tarski in SCT §17 are suggestive. Even Tarski himself did not see Naess’ results as a refutation. 28 A few comments are in order here. First, Tarski’s apparent endorsement of the statistical questionnaire method would also belie the claim that Tarski saw Naess’ work as a refutation. Second, we have not found in Naess’ work anything that would suggest he believed the empirical data would render any theory or conception or view of truth unworkable or false; likewise, there is much to suggest that he did not believe the data could tell us which one was acceptable or true. Finally, and perhaps most importantly, Naess appreciated the limitations of empirical data informing (or failing to inform) a philosophical view. There are numerous discussions throughout TCNP where Naess is honest about the advantages and disadvantages of running empirical studies on non-philosophers (cf. TCNP, §§11-12, 25, 56-58, 63, 70-73 (passim), 103105, 107-113. 29 We would like to thank an anonymous referee for pointing out Jané and Betti’s important contribution to the historical debate regarding Tarski’s notion of “common concept.” 15

operative within axiomatics, especially among members of the Peano school, American postulate theorists (like Pieri, Padoa, Huntington, and Veblen), Hilbert, and Fraenkel, is highly compatible with the notion of logical consequence in Tarski’s semantic concept of logical consequence (Jané 2006, 36). Given that Tarski was quite familiar with axiomatics, according to Jané, when Tarski writes that “It seems to me that everyone who understands the content of [the definition of logical consequence] must admit that it agrees quite well with common usage” (Tarski 1983c; our emphasis), this usage is the one employed by logicians who are well-versed in axiomatics (Jané 2006, 31). Notably, Jané’s interpretation seems to be supported by a reading of Ajdukiewicz, which has been provided by Betti: The concept of existence that Ajdukiewicz wants to clarify in his 1921 was the concept in use in axiomatics, and the same holds for the other concepts he discusses, including truth and logical consequence: neither real existence in the sense in which we say that lighthouses, thoughts, or planets exist, nor finding our connections between this notion and that of axiomatics is what interests him. For, Ajdukiewicz claims, it is doubtful whether ‘exists’ as it is used in deductive theories has anything in common with its meaning in everyday language (Ajdukiewicz 1921, 46; trans. 33). The same holds for the other concepts he discusses. (Betti 2008, 62) One might contend that we would have to show why our interpretation rejects the Jané-Betti analysis, but we believe that overestimates the underlying message of our argument. On the contrary, our interpretation of Tarski’s view of the “common concept” suggests that it developed over the course of several years, perhaps beyond the time period in which he broke from his advisor Stanisław Leśniewski (Patterson 2012, 84-107; Betti 2008, 50-56). In Part II of SCT, especially when Tarski calls on the work of Naess, he at least began to consider that the questionnaire method might enlighten us about conceptions of truth the ordinary person employs. An interpretation that sees Tarski’s thinking on the ordinary person’s conception of truth changed over time actually does not take away from either Jané’s or Betti’s remarks that there may be a common concept of truth fully embedded within axiomatics; rather, it just goes to show that Tarski was aware of an alternative view of what the “common concept” might include. Thus, we do not believe that whatever we have said here about SCT is necessarily incompatible with the work of Jané or Betti.30 Similarly, in an effort to allay any disagreement one might have with our view showing an extension of Tarski’s sense of the “common concept,” we would like to cross-examine three 30

Furthermore, we would agree, in principle, with Patterson’s assessment of Jané’s (2006) work when Patterson writes: “The attraction of [Jané’s view] is obvious: since Tarski clearly doesn’t capture essential features of the ordinary notion, one proposes that he wasn’t trying to capture the ordinary notion. However, Tarski is trying to capture the ordinary notion--there are just too many passages in the article that say this, and capturing ordinary notions with expressions constrained by the structure of deductive theories is what he had been trying to do all along” (Patterson 2012, 220). 16

critical elements Jané and Betti employ. First, Betti cites a critical passage from Ajdukiewicz, which she believes supports the view that the common concept is the one in use in axiomatics. Ajdukiewicz writes: Formal implication is a relation between propositional functions. One may define it in the following way: Φ(x) is formally implied by f(x) if, for every possible substitution of some value for a variable (or variables) x, either f(x) is false or Φ(x) is true. This formal implication is--as it seems--the source of the common notion of consequence. (as quoted in Betti 2008, 62; her emphasis)31 Betti has italicized and emphasized the “common notion of consequence,” as if this supports the view that the common concept is the same as the one in axiomatics, but, if we shift our attention just slightly in reading this passage, it appears to us that it renders a different reading. Notice that the source of the common concept is the one that is operative in axiomatics. We believe that the axiomatic concept, then, can be the source of the common concept without the two having completely overlapping senses. Second, much of Jané’s analysis hinges upon the idea that, in Tarski, when he refers to the common concept, it reflects the perspective of one who is relatively familiar with axiomatics. In other words, we have to already be working within a mathematical paradigm in order for us to comprehend the concept and that is precisely what Tarski means when he discusses the common concept. Jané’s pointing out specific passages in Tarski’s early work on logical consequence, definition, and truth, however, seem to leave open the possibility that the common concept to which Tarski refers in the later works is something above and beyond the common concept he cites in the early work. Assuming that later work by Tarski upholds the view that the “common concept” is the one operative in axiomatics gives rise to a curious problem. If we were to understand the polemical remarks as showing why we ought not take up the “common concept” or why there is confusion about the “common concept,” then, on Jané’s reading, the polemical remarks themselves would seem to be leveling an objection to Tarski’s own view. That is such a radical interpretation and departure from what Tarski must have meant by the polemical remarks that we find it difficult to suggest. Third, it seems to us that by the time of SCT, Tarski’s thinking about truth as a semantic notion was distinct from how he thought about other concepts such as consequence, synonymy, or meaning. When he considers whether the definitional approach developed for truth can be extended to other semantic concepts, Tarski writes in the non-polemic §13 of SCT: “Within theoretical semantics we can define and study some further notions, whose intuitive content is more involved and whose semantic origin is less obvious; we have in mind, for instance, the important notions of consequence, synonymy, and meaning” (SCT, 354 our emphasis). This shows that, by Tarski’s own lights in SCT, applying and generalizing Jané’s reading of the common concept of logical consequence to the semantic conception of truth would require us to 31

Betti draws this quote from Batóg 1995, 56. 17

read the character of what is marked by having a less obvious semantic origin and a more involved intuitive content into what is more obvious and less involved!32 So, even if we allow that the Jané-Betti reading might be appropriate for consequence and that it might apply to CTFL, by the time of SCT Tarski himself indicates that he wants to draw a line between truth and consequence. Now, we would like to turn our attention to a potential third criticism that comes not so much from published work on Tarski but from an implicitly held belief among Tarski scholars. Some have called SCT “Tarski light” because it is less representative of his most serious formal work on truth. SCT is less logically rigorous than Tarski’s earlier more logically rigorous work. If this unstated assumption is true of SCT, then there might not be much hope for figuring out what Tarski really believed about truth. For all the claims we have made to show that one segment of Tarski’s project evolved, one could easily find something that Tarski preaches or practices that denies such a transformation in his thought.33 Throughout this paper, we have called on evidence supporting the interpretation that Tarski’s view of the common conception of truth might have expanded in a way Tarski scholars have omitted from their analyses, even if his formal definition of truth remains firmly embedded within certain mathematical paradigms. Although absent from his early works, Tarski cited Naess’ 1938 empirical work on the non-philosopher’s view of truth twice in SCT to discuss ways in which we might familiarize ourselves with the vagaries operative in the everyday usage of truth (SCT, 360). Tarski and Naess seemingly agree upon the fragmented nature of truth among non-philosophers, which might leave us little hope of coming to address it in formal philosophical works (SCT, 355-370; CT, 58; TCNP, 177). Further, Tarski wrote in a postcard to Naess seemingly endorsing the empirical work he was doing, even if he did not believe it affected his formally defined notion of truth (Naess 1993, 18; 2005c, 270; see also Chapman 2008, 121f). Finally, Popper and Carnap may have been mistaken about the force and intent of Naess’ work, which (if we might engage in some speculation) might have been cleared up if the proposed June 1940 meeting had taken place in Oslo as scheduled. Unfortunately, the Nazis invaded and began their occupation of Norway on 9 April 1940 (Naess 1937/1938, 384). For these reasons, our preference is to see Tarski “light” as Tarski “late.” 4. Conclusion: Tarski “Light” or Tarski “Late?”

32

A defense of the Jané-Betti reading would require that there be some further basis for saying either that the common concept of consequence is not the more involved, less obvious, concept Tarski mentions in SCT or that the differences that Tarski mentions do not matter, i.e. that the more involved, less obvious, parts of consequence are inessential. This would clearly require a more developed account of what the essential vs. inessential parts of the concept are like. 33 We thank an anonymous referee for providing us with this potential criticism of this discussion, and we are grateful for another reviewer pointing out that the criticism is “implausibly strong.” 18

Tarski requires that a definition of truth be formally correct and materially adequate. But, while the former is a matter of how the definition is structured, what to include as a part of material adequacy is less clear and develops over time. In CTFL, the problem of defining truth is counted among the classical questions of philosophy. There, the development of a formal definition of truth for a formal language is recommended as a response to the observation that the project of defining truth for natural languages is logically problematic. This means, that no matter how hard it is to express, there is a formal concept that must align with the materially adequate definition, i.e., a definition that conforms to the everyday conception of truth, and it is a matter of controversy in SCT whether the formal definition actually succeeds in this task. This is the question that Tarski is addressing, primarily in §§14 and 17 of SCT. Regarding §§14 and and 17, we have argued above that Tarski’s references to the empirical work of Naess signal a shift in how Tarski thought about the old-notion of truth, the common concept or the every-day notion of truth. In SCT, the singular classical conception of truth found in CTFL confronts the panoply of competing and incompatible views of various philosophers--each one trying to capture the everyday usage of the term “true.” The advantage of reading Tarski’s remarks, about whether the semantic conception of truth is the right one (§14) or whether the semantic conception conforms with the philosophical and common-sense usage (§17), through the lens of Naess’ empirical work is that the empirical work provides Tarski with a scientific grounding from which to reevaluate his thinking about the content of the common everyday concept of truth. There is no singular and monolithic use of “truth” among non-philosophers. Ordinary people employ the terms “true” and “truth” in many distinct ways. Instead of relying on intuition alone, we can use the statistical questionnaire method to describe the various ways in which “truth” or “true” is used by the folk. (Cf. Linsky 1952, 5).34 With this in mind we can speculate a bit. If this reading is correct, then the everyday notion of truth that must be grasped, is not singular but manifold, not monolithic but diffuse. Perhaps this is why Tarski puzzlingly denies his own original conception of the problem of truth in §18 of SCT: In general, I do not believe that there is such a thing as "the philosophical problem of truth.” I do believe that there are various intelligible and interesting (but not necessarily philosophical) problems concerning the notion of truth, but I also believe that they can be exactly formulated and possibly solved only on the basis of a precise conception of this notion. (SCT, 361)35

34

To be clear, we are not suggesting that there was a wholesale evolution of the formal criterion of material adequacy. Rather, we are claiming that Tarski found in Naess a reason to dwell on the question of the content of the concept that the formal criterion was supposed to track. 35 Patterson (2012) notes this passage and comments on its apparent tension with §14 of SCT and with the opening passages of CTFL. He reads this as evidence of Tarski’s dismissive attitude 19

This shift from the nature of truth being a classical question of philosophy to the claim that there is no such philosophical problem is patent. If Tarski’s thought about how to conceive the common concept of truth is always the intuitive classical concept of CTFL, then Tarski must be read as contradicting himself, or not writing seriously.36 But, if the problems of truth are not necessarily philosophical yet can still be exactly formulated and made precise, then, on a Naess informed reading, these questions might become empirical questions, the sort of questions an analysis of actual everyday usage might answer. Such a radical re-reading of §18 would require more work to sustain than present space allows, but on such a reading the patent inconsistency is resolved. Our main focus in this paper has been on whether Naess’ results can help inform our understanding of Tarski’s notion of material adequacy. As we have shown, of the two critical components of Tarski’s definition of truth, the formally correct definition remained the same throughout the course of his work from the 1930s through at least the 1940s and perhaps to the end of his life, but the character of the material adequacy condition appears to evolve around the time of Tarski’s interactions with Naess and other members of the Vienna Circle. The polemical discussions and both the explicit and the uncited references to Naess in §§14 and 17 reveal an underappreciated degree of affinity between Tarski in SCT and the work Naess reported through personal communication, in TCNP, and CT. We maintain that this affinity warrants a reconsideration of the view of some scholars, like Patterson (2012) and Sher (1999), that SCT is best understood as “Tarski light.” We believe that the Tarski-Naess interaction discussed here should prompt a reconsideration of “Tarski late.”

toward philosophy. We think it is further evidence of Tarski re-thinking the nature of the everyday usage of truth (see Patterson 2012, Ch. 8, especially 230f). 36 This is the substance of Patterson’s (2012) reading. 20

References Batóg, Tadeusz. 1995. Ajdukiewicz and the development of formal logic. In The Heritage of Kazimierz Ajdukiewicz, edited by V. Sinisi and Jan Wolenski, 53-67. Amsterdam: Rodopi. Betti, Arianna. 2008. Polish Axiomatics and its Truth: On Tarski’s Lesniewskian Background and the Ajdukiewicz Connection. In New Essays on Tarski and Philosophy, edited by Douglas Patterson, 44-72. Oxford, UK: Oxford University Press. Carnap, Rudolf. 1956. Meaning and synonymy in natural languages. In Meaning and Necessity. 2nd edition. Chicago: The University of Chicago Press. Carnap, Rudolf. 1961. Introduction to Semantics and Formalization of Logic. Cambridge, MA: Harvard University Press. Carus, A.W. 2007a. Carnap and Twentieth-Century Thought: Explication as Enlightenment. Cambridge, UK: Cambridge University Press. Carus, A.W. 2007b. Carnap’s intellectual development. In The Cambridge Companion to Carnap, ed. Michael Friedman and Richard Creath. Cambridge, UK: Cambridge University Press. 19-42. Chapman, Siobhan. 2008. Language and Empiricism: After the Vienna Circle. Basingstoke, UK: Palgrave Macmillan. Crockett, Campbell. 1959. “An Attack Upon Revelation in Semantics.” Journal of Philosophy 56:103–111. David, Marian. 2008. Tarski’s convention T and the concept of truth. In New Essays on Tarski and Philosophy, ed. Douglas Patterson. Oxford: Oxford University Press. 133-156. Field, Hartry. 1972. Tarski’s theory of truth. Journal of Philosophy 69: 347-375. Holton, Gerald. 1993. “From the Vienna Circle to Harvard Square.” In Scientific Philosophy: Origins and Developments, ed. Friedrich Stadler. Dordrecht, The Netherlands: Kluwer Academic Publishers. 47-73. Horsten, Leon. 2011. The Tarskian Turn: Deflationism and Axiomatic Truth. Cambridge, MA: MIT Press.

21

Jané, Ignacio. 2006. “What is Tarski’s Common Concept of Consequence?” The Bulletin of Symbolic Logic 12: 1-41. Künne, W. 2003. Conceptions of truth. Oxford: Oxford University Press. Linsky, L. (ed.) 1952. Semantics and the Philosophy of Language. Urbana-Champaign, IL: The University of Illinois Press. Mancosu, Paolo. 2010. The Adventure of Reason. Oxford: Oxford University Press. Naess, Arne. 1936. Erkenntnis und wissenschaftliches Verhalten (Skrifter / Norske VidenskapAkademi i Oslo Il. Hist.-Filos. Klass 1936 No. 2). Oslo: Komisjon Hos Jacob Dybwad. Naess, Arne. 1937/1938. “Zum Vortrag von Kokoszynska uber Einheitswissenschaft.” Erkenntnis 7: 382-384. Naess, Arne. 1938a. “Common Sense and Truth.” Theoria: A Swedish Journal of Philosophy 4:39–58. Naess, Arne. 1938b. “Truth” as Conceived by Those Who Are Not Professional Philosophers (Skrifter Utgitt av Det Norske Videnskaps-Akademi I Oslo Il. Hist.- Filos. Klass 1938 No. 4). Oslo: I Komisjon Hos Jacob Dybwad. Naess, Arne. 1953. An Empirical Study of the Expressions “True,” “Perfectly Certain” and “Extremely Probable”. Oslo: I Kommisjon Hos Jacob Dybwad. Naess, Arne. 1968. Scepticism. London: Routledge and Kegan Paul. Naess, Arne. 1993. Logical empiricism and the uniqueness of the Schlick seminar: A personal experience with consequences. In Scientific Philosophy: Origins and Developments, ed. Friedrich Stadler. Dordrecht, The Netherlands: Springer, pp. 11-25.

Naess, Arne. 2005a. How can the empirical movement be promoted today? A discussion of empiricism of Otto Neurath and Rudolf Carnap. In The Selected Works of Arne Naess, Volume VIII: Common Sense, Knowledge, and Truth, ed. Harold Glasser & Alan Drengson. Dordrecht, The Netherlands: Springer, pp. 163-215.

22

Naess, Arne. 2005b. Logical empiricism and the uniqueness of the Schlick seminar: A personal experience with consequences. In The Selected Works of Arne Naess, Volume VIII: Common Sense, Knowledge, and Truth, ed. Harold Glasser & Alan Drengson. Dordrecht, The Netherlands: Springer, pp. 261-278. Neurath, Marie and Robert Cohen (eds). 1946. Philosophical Papers of Otto Neurath. Neurath, Marie and Robert Cohen (eds.). 1973. Empiricism and Sociology. Patterson, Douglas (ed.). 2008. New Essays on Tarski and Philosophy. Oxford: Oxford University Press. Patterson, Douglas. 2012. Alfred Tarski: Philosophy of Language and Logic. Basingstoke, UK: Palgrave Macmillan. Popper, Karl. 2002. Conjectures and Refutations. Revised ed. London: Routledge. Rothenberg, David. 1992. Is It Painful to Think?: Conversations with Arne Naess. Minneapolis: University of Minnesota Press. Shepherd, Joshua and James Justus. forthcoming. X-Phi and Carnapian Explication. Erkenntnis: 1-22. Sher, Gila. 1999. What is Tarski’s theory of truth? Topoi 18: 149-166. Soames, Scott. 1999. Understanding Truth. Oxford: Oxford University Press. Storheim, Eivnid. 1959. “Review of Arne Naess Wie fordert man heute die empirische Bewegung? Eine Auseinandersetzung mit dem Empiricismus von Otto Neurath und Rudolph Carnap.” Theoria 25: 187-191. Tarski, Alfred. 1944. “The Semantic Conception of Truth and the Foundations of Semantics.” Philosophy and Phenomenological Research 4:341–375. Tarski, Alfred. 1948. “A Problem Concerning the Notion of Definability.” Journal of Symbolic Logic 13: 107–111. Tarski, Alfred. 1969. “Truth and Proof.” Scientific American 220 (6): 63-77.

23

Tarski, Alfred. 1983a. The Concept of Truth in Formalized Languages. In Logic, Semantics, Metamathematics, ed. Alfred Tarski. Indianapolis, IN: Hackett Publishing pp. 152–278. Tarski, Alfred. 1983b. The Establishment of Scientific Semantics. In Logic, Semantics, Metamathematics, ed. Alfred Tarski. Indianapolis, IN: Hackett Publishing pp. 401–408. Tarski, Alfred. 1983c. On the Concept of Logical Consequence. In Logic, Semantics, Metamathematics, ed. Alfred Tarski. Indianapolis, IN: Hackett Publishing pp. 409–420.

24