Looking across languages: Anglocentrism, cross-linguistic experimental philosophy, and the future of inquiry about truth

Abstract

Analytic debates about truth are wide-ranging, but certain key themes tend to crop up time and again. The three themes that we will examine in this paper are (i) the nature and behaviour of the ordinary concept of truth, (ii) the meaning of discourse about truth, and (iii) the nature of the property truth. We will start by offering a brief overview of the debates centring on these themes. We will then argue that cross-linguistic experimental philosophy has an indispensable yet underappreciated role to play in all of these debates. Recognising the indispensability of cross-linguistic experimental philosophy should compel philosophers to significantly revise the ways in which they inquire about truth. It should also prompt analytic philosophers more generally to consider whether similar revisions might be necessary elsewhere in the field.

1 Introduction

Truth has been a focal topic in analytic philosophy since the tradition’s inception in the late nineteenth century. Analytic theories of truth are conceptually diverse, invoking signature notions ranging from correspondence to fact, verifiability, and superwarrant to transparency, prosentences, pretence, and replacement. However, analytic debates about truth have been largely—and, we think, regrettably—homogeneous along another crucial dimension.

When supporting their own theories of truth or criticising opposing theories, analytic philosophers standardly make heavy use of their own intuitions (and perhaps those of a few colleagues and students). These intuitions are expressed in English and are often about how certain English expressions (‘true’ and ‘truth’ being prime examples) are ordinarily used. Analytic philosophers use such intuitions to assess theories of such things as ‘the concept of truth,’ ‘the meaning of the truth predicate,’ and ‘the nature of truth’ (definite articles abound!). In this way, analytic philosophers exhibit a tendency to debate about theories of truth that are framed in universal terms, even though the intuitions that inform these debates are typically articulated in and solely about English. We can call this overall tendency Anglocentrism.¹

Our aims in this paper will be to argue that the Anglocentric model for inquiring about truth is outmoded and to then take one significant step towards an improved methodology. A key component of the methodology that we favour is cross-linguistic experimental philosophy, i.e. experimental philosophy involving studies which compare the behaviour of people in one linguistic community to that of people in at least one other linguistic community. To start, we will offer a brief overview of three major analytic debates about truth and the ways in which experimental findings have been applied in two of these debates. We will then explain why cross-linguistic experimental philosophy is indispensable to all three of these debates. To close, we will respond to a question and two objections.²

2 Analytic debates about truth and extant experimental research

Needless to say, analytic philosophers have been concerned with numerous issues when pursuing debates about truth.³ To keep things manageable, we will focus in this discussion on debates about three core topics: the concept truth, the meaning of alethic vocabulary such as ‘true’ and ‘truth,’ and the nature of the property/relation truth.⁴

2.1 Debates about truth

Many analytic philosophers have offered speculations on how we think about truth and why we think about truth as we do. These speculations are often framed in terms of the concept truth. In developing theories of truth, analytic philosophers have put forward a number of different views about the sort of entity that truth is.

For instance, in a seminal discussion of the distinction between the concept truth and the property truth, Alston (2002, pp. 12–13) proposes that we identify truth with the meaning of the word ‘true’ when it is used in a particular range of cases (e.g. in an assertive use of the sentence ‘The proposition that Algeria is in North Africa is true’). By contrast, when setting out his functionalist theory of truth, Lynch (2009, p. 7) proposes that our “folk concept” truth is “the way we tacitly think about [truth] in ordinary life.” Additionally, Asay (2021) has recently argued that we should take truth to be the ability to have propositional thoughts such as beliefs, hopes, or desires.⁵

Despite this diversity in theories of truth’s nature, there is a basic conception of truth that is often in play, and often implicitly, in debates about this topic. We might call this the coarse-grained conception of truth, according to which truth is a mental entity of some sort that thinkers deploy whenever they have thoughts involving truth. According to this conception, for instance, if Eric wonders whether the last statement that he read on Wikipedia is true, then he deploys the concept truth. Likewise, if Patrice is convinced that her mechanic’s claims about her carburettor are not true, then she deploys the concept truth. For present purposes, we will adopt this coarse-grained conception of truth.⁶

2.2 Debates about alethic vocabulary

Analytic philosophers have also produced a range of theories concerning the meaning of truth-related, or alethic, vocabulary. In English, alethic vocabulary includes the familiar ‘true’ and ‘truth.’ It also includes ‘correct’ and ‘right,’ when the latter are used in certain sentences, such as ‘Harold believes that the Earth is flat, but his belief is wrong/incorrect, since the Earth isn’t flat.’⁷

There are many theories of alethic vocabulary. For instance, after presenting his famous treadmill argument, Frege (1956, p. 291) draws the primitivist conclusion that “it is probable that the content of the word ‘true’ is unique and indefinable.” By contrast, Horwich (1998, pp. 35–36, 128, 145, 2010, pp. 19, 35, 37, 41, 47, 80, 158–159; ch. 5, nn. 11, 15) maintains that the meaning of ‘true’ is fixed by our explanatorily basic disposition to accept the instances of the Equivalence Schema⁸:

(ES) The proposition that p is true iff p.

Additionally, Kölbel (2008, §§ 3–4) has recently proposed that ‘true’ is ambiguous between the meaning that Horwich identifies and a meaning that applies to a proposition p only if p is objective, in the sense that it is a priori that when one thinker believes p and another thinker believes not-p, one of them must be mistaken.⁹

In what follows, we will adopt a conception of alethic vocabulary that uses the notion of a standard translation. Translation is a subtle business, so there is a good deal to say when fleshing out this notion. Here, though, we will just put forward the basic idea. We can say that an expression e of language L is a standard translation of an expression e^* of language L^* iff there is a norm among L speakers that in a certain range of contexts c₁, …, c_n, e^* ought to be translated as e.¹⁰

In some cases, e^* may have exactly one standard translation into L. Examples of this sort may include logical vocabulary such as sentential conjunction or names of numbers such as ‘sixty-three.’ In other cases, e^* may have more than one standard translation into L. An example of this sort is the te reo Māori noun ‘mana,’ which is translated into English in a variety of ways depending on the context. These translations include ‘prestige,’ ‘authority,’ ‘control,’ ‘power,’ ‘influence,’ ‘status,’ ‘spiritual power,’ and ‘charisma.’¹¹ In short, the relation of being a standard translation is variably polyadic, insofar as it can be either one–one or many-one.

It is also worthwhile to note that in the ‘mana’ case, there are at least two L expressions e₁ and e₂ (e.g. ‘prestige’ and ‘charisma’) such that (a) e₁’s meaning ≠ e₂’s meaning, even though (b) e₁ and e₂ are both standard translations of an L^* expression e^*. This suffices to show that the fact that e is a standard translation of e^* does not entail that e’s meaning = e^*’s meaning. With this in mind, we can also say that the relation of being a standard translation is not naively meaning-preserving.

With the basic notion of a standard translation in hand, we can articulate the translational conception of alethic vocabulary that we will adopt here. According to the translational conception, an expression e of language L is a piece of alethic vocabulary iff¹²:

• (i) e is ‘true,’ ‘false,’ ‘right,’ ‘wrong,’ ‘correct,’ or ‘incorrect,’ where these adjectives are used to speak about (a) sentences, mental states, or the contents of either, or (b) entities (such as theories) that are composed of sentences, mental states, or the contents of either;

• (ii) e is a standard translation of one of these adjectives, as used in (i);

• (iii) e is a grammatical variant of one of these adjectives (e.g. ‘truth’ or ‘truly’), where this variant is similarly used; or

• (iv) e is a standard translation of one of these grammatical variants.

2.3 Debates about truth

In addition to reflecting on truth and the meaning of alethic vocabulary, analytic philosophers have long been concerned to identify the nature of the property/relation truth.¹³ Whereas debates about truth concern our thought about truth and debates about alethic vocabulary concern the words that we use to talk about truth, debates about truth’s nature concern the metaphysics of truth.

When it comes to theories of truth’s nature, we are certainly spoiled for choice. Rasmussen (2014), for instance, takes a proposition’s having truth to consist in its corresponding to what he calls an ‘arrangement.’ Edwards (2018) maintains that truth is ‘determined’ by different properties in different domains. The early Russell (1904) suggested that while truth does exist, it is a primitive, indefinable property. By contrast, certain deflationists such as Grover (1992) and Brandom (1994) have argued that truth does not exist at all.

In what follows, we will again use a coarse-grained conception of the property truth. Described in this coarse-grained way, truth is just the property that is possessed by all and only the true truth-bearers. For instance, since the statement that Oslo is the most populous city in Norway is true, it possesses truth. By contrast, since the statement that Bergen is the most populous city in Norway is not true, it fails to possess truth.¹⁴

2.4 Extant experimental research on truth

Thus far, experimental philosophers have shown that experimental research can significantly inform both debates about truth and debates about alethic vocabulary. To illustrate, we will focus on two relevant experimental studies. Thinking through the implications of the studies will set up our discussion of cross-linguistic experimental philosophy in Section 3. The bearing of experimental research on debates about the metaphysics of truth has yet to be explored in detail, so we will devote special attention to this issue in Section 3.3.

Experimental inquiry about truth originates with the pioneering work of Arne Næss (1938a, b, 1953). Næss pointed out that philosophers regularly make claims about the ‘ordinary notion’ or the ‘ordinary concept’ of truth and that these claims are empirical in nature. Accordingly, he suggested, these claims need to be supported with empirical evidence. Næss’ main innovation in this area was to go beyond standard philosophical practice by actually gathering such evidence. He and his research team conducted qualitative, interview-style surveys, with most of the interviewees (71%) having no education in philosophy.¹⁵ During these interviews, they asked their participants questions like the following. Based on the information that Næss provides, it appears that the interviews were conducted entirely in Norwegian or entirely in English¹⁶:

• What is to be understood by the expression “something is true”? Define the expression.

• What are the c.c. (common characteristics) of that which is true?

• Give me an example of something that is true.

• Do you employ the expression “the truth”? (If answered positively:) On which occasions?

Næss reported a number of findings, one of which is particularly relevant to our discussion here¹⁷:

The misconception that non-philosophers adhere—explicitly or implicitly—to a definite type of opinion on the notion of truth is primarily due to an ignorance of the extreme diversity of opinion found among non-philosophers as soon as they are invited to speak about the notion of truth…

All the main standpoints advocated in truth-theories as they are met with in philosophic literature (excluding theories on ‘formal truth’) can be refound among persons without any philosophic education. All the main opinions on the possibility of defining truth, its definition, its verification, its existence, on the existence of absolute truths, on the eventual meaning of this problem, on the law of excluded middle, etc. are refound.

We can summarise this finding—which we will call Næss’ diversity result—as follows:

Naess’ diversity result: the judgments about truth made by Næss’ participants were highly diverse, approximating in diversity the views about truth that were advanced by philosophers up to the late 1930s.

The relevance of this and other results due to Næss was clearly recognised by one of his eminent contemporaries—namely, Alfred Tarski. In a highly influential article defending his semantic conception of truth, Tarski says, citing Næss¹⁸:

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.

As Tarski observes, Næss’ findings bear on debates about alethic vocabulary, as they provide us with information about how alethic vocabulary was ordinarily used at the time of the research. Unfortunately, despite the fact that Næss drew on a linguistically diverse group of participants in the study at issue here (85.33% Norwegian, 8.67% Austrian, 3.33% Swedish, 1.33% German, and 1.33% English), he does not compare the usage of alethic vocabulary across the relevant languages, perhaps because of his own linguistic limitations and/or those of his interviewers.¹⁹ For this reason, the scope of Næss’ diversity result is notably limited, insofar as it pertains only to judgments about truth made in Norwegian or to judgements about truth made in English. We will return to this point in Section 3.2.

Næss’ findings also bear on debates about truth. The ways in which we ordinarily use alethic vocabulary are a source (though certainly not the only source) of evidence regarding the thoughts that we express using this vocabulary (call thought of this sort alethic thought). As a result, data on the usage of alethic vocabulary can in turn improve our understanding of alethic thought. We can then use this empirically-grounded understanding of alethic thought to evaluate philosophical theories of truth.²⁰

Despite Tarski’s highly visible endorsement of Næss’ research, the importance of experimental work on truth was mostly overlooked until the early twenty-first century. While we lack the space to provide a comprehensive overview of current experimental findings, we will briefly describe one study which exemplifies how contemporary experimental work also informs debates about truth and debates about alethic vocabulary.²¹

In a recent study, Kevin Reuter and Georg Brun tested two hypotheses regarding the ordinary meaning of ‘true,’ when this word is applied to empirical statements such as ‘Jill is at the party’ or ‘I have a 1990 Rolex Submariner in my safe²²:’

• Hypothesis 1 (correspondentism): Laypeople consider an empirical statement S of person P to be true or false depending on whether S corresponds with reality, and regardless of S’s coherence with other relevant beliefs of [P].

• Hypothesis 2 (coherentism): Laypeople consider an empirical statement S of person P to be true or false depending on whether S coheres with other relevant beliefs of [P], and regardless of S’s correspondence with reality.

Reuter and Brun evaluated these hypotheses using (i) a vignette in which the protagonist’s belief coheres with other relevant beliefs that they have but fails to correspond with reality and (ii) a vignette in which the protagonist’s belief corresponds to reality but fails to cohere with other relevant beliefs that they have. They report that their participants’ judgments about whether these beliefs are true “were…divided between answers in line with coherentist predictions and answers in line with correspondentist predictions”.²³

Reuter and Brun took this finding to provide evidence that within the empirical domain, ‘true’ is ambiguous between a correspondence-style reading and a coherence-style reading.²⁴ Whether or not this conclusion is in fact warranted, it is clear that Reuter and Brun’s finding bears on debates about alethic vocabulary insofar as it provides significant information about the meaning of ‘true.’

Their finding also bears on debates about truth. Again, our ordinary ways of using alethic vocabulary are a source of evidence regarding our ordinary alethic thought. For this reason, Reuter and Brun’s finding constitutes evidence that English speakers think about truth in at least two different ways when considering statements that pertain to empirical matters. Whether we should infer from this that there are two notably different ways in which English speakers deploy the single concept truth or, on the other hand, that English speakers use at least two truth concepts—one correspondence-theoretic and one coherence-theoretic—is of course a delicate matter.²⁵ Whichever of these inferences proves to be stronger, the point remains that Reuter and Brun’s research is an important contribution to the ongoing debates about truth.

3 The indispensability of cross-linguistic experimental philosophy

Thus far, we have illustrated how experimental research can inform both debates about truth and debates about alethic vocabulary. We now want to go a step further by explaining why cross-linguistic experimental philosophy is indispensable to both these debates and debates about truth.

As we indicated in Section 1, one of the reasons that these observations are significant is that by and large, analytic philosophers have evaluated theories of truth, alethic vocabulary, and truth using armchair intuitions that are formulated in and solely about English. In Sects. 3.1–3.3, we will explain why this Anglocentric methodology leaves much to be desired.

An additional reason that our observations here are significant is that experimental philosophers working on truth have also relied in large part (Mizumoto being a notable exception) on studies that are formulated solely in English. In light of our discussion in Sects. 3.1–3.3, it will become clear that Anglocentric experimental research has only a limited impact on debates about truth and that experimentalists would also do well to prioritise cross-linguistic research going forward.

What does it mean for cross-linguistic experimental philosophy to be indispensable to a particular set of debates D? For our purposes, the claim that cross-linguistic experimental philosophy is indispensable to D is the claim that when D is properly pursued, cross-linguistic experimental studies are needed to fully evaluate at least some of the central conjectures within D. We should make two observations about this claim.

The first is that one of the main determinants of what proper pursuit of D would involve is what the theoretical goals of D should be. Identifying these goals is a complex task, as it hinges on a range of factors that must be balanced out in a reasonable way. Among others, these factors include the following: how the participants in D have described D’s subject matter, the sorts of arguments that have been offered by these participants, and the ways in which D is connected to other sets of debates. In what follows, we will focus primarily on the first and second of these factors.

The second observation to make is that the indispensability claim entails that cross-linguistic experimental studies are needed to fully evaluate at least some of the central conjectures within D. This means that the claim is compatible with some, and perhaps all, of the central conjectures within D being partially evaluable using e.g. armchair or monolingual experimental methods. It also compatible with there being a central conjecture within D that is fully evaluable using the latter methods. What the indispensability claim tells us is that we cannot fully evaluate all of the central conjectures within D unless we do some cross-linguistic experimental philosophy.

3.1 Debates about truth

We noted in Section 2.1 that whatever other features it may have, the concept truth is typically taken to be a mental entity of some sort that thinkers deploy whenever they have thoughts involving truth. Moreover, as we illustrated in Section 2.4, extant findings in experimental philosophy can be brought to bear on theories of truth. These findings provide us with information about how we use alethic vocabulary. This information in turn constitutes evidence regarding our ordinary alethic thought, and that evidence can be used to evaluate theories of truth.

Why, then, is cross-linguistic experimental philosophy indispensable to inquiry about truth? A significant detail here is that philosophers standardly take truth to be a single concept that is possessed by a linguistically diverse subset of actual creatures.²⁶ It is not always entirely clear who the members of this subset are meant to be, but for the sake of illustration, we can take the set to contain all and only humans who are at least 2 years of age.²⁷ Call this (or some nearby conjecture) the standard conjecture about truth.

To fully evaluate the standard conjecture, it is necessary to conduct a range of studies on how people in various linguistic communities deploy alethic vocabulary. If we fail to do such studies, then we will not know whether ordinary alethic thought is notably uniform across all linguistic communities or whether there are significant differences in such thought across certain linguistic communities. Absent this information, we will not be able to confidently determine whether we should indeed posit a single concept truth that is possessed by all of the relevant members of every linguistic community. That is, in the absence of these cross-linguistic studies, we will not be able to fully evaluate the standard conjecture about truth. For this reason, cross-linguistic experimental philosophy is indispensable to inquiry about truth.²⁸

3.2 Debates about alethic vocabulary

As we indicated in Section 2.2, the main aim of a theory of alethic vocabulary is to specify what alethic vocabulary means. A primary source of evidence regarding the meaning of alethic vocabulary is how that vocabulary is used. Notably, alethic vocabulary is used not only by working academics but by competent speakers generally. For this reason, philosophers interested in the meaning of alethic vocabulary should investigate its usage by competent speakers as such. Empirical methods, e.g. those used by Reuter & Brun (2022), provide the most reliable way of doing this, and for this reason, these methods are vital to debates about alethic vocabulary.

A further point that we want to emphasise here is that alethic vocabulary is present in many, and perhaps all, existing natural languages.²⁹ Philosophers inquiring into the meaning of alethic vocabulary should thus not restrict their attention to alethic vocabulary in, say, NZ English. Instead, their aim should be to investigate alethic vocabulary as such, i.e. the behaviour of alethic vocabulary across the world’s many natural languages. As mentioned above, this cross-linguistic inquiry should be driven by empirical methods, which shows that when debates about alethic vocabulary are properly pursued, cross-linguistic experimental philosophy is a necessary component of those debates. In other words, cross-linguistic experimental philosophy is indispensable to debates about alethic vocabulary.

This point becomes apparent, for instance, when we consider the prominent accounts of alethic vocabulary that have been defended by deflationists about truth. There are different varieties of deflationism on offer, but a core idea that animates these views is that once we clearly understand the meaning and function of alethic vocabulary in natural language, we will see that we have no strong reasons for positing a substantial property truth.³⁰

In the first few pages of his groundbreaking book Truth, the leading deflationist Paul Horwich nicely articulates this line of argument. He suggests that because traditional debates about truth have failed to generate significant consensus regarding truth’s nature, they have left us with the unsettling result that “a conception of [truth’s] underlying nature appears to be at once necessary and impossible” (1998. p. 2). Horwich aims to dispel this appearance by drawing our attention inter alia to the meaning and function of our talk about truth.

Put very briefly (recall Section 2.2), Horwich’s view is that to understand the meaning of ‘true’ is to be disposed to accept every instance of the following schema in the absence of supporting evidence:

(ES) The proposition that p is true iff p.

He also holds that the primary function of ‘true’ is to enable us to make utterances (e.g. ‘What Oscar said is true,’ when one is not sure which proposition Oscar asserted) that we could not otherwise make, due to our linguistic limitations. Drawing on this account of ‘true,’ Horwich (ibid., p. 5) advances the bold conclusion that “the traditional attempt to discern the essence of truth—to analyse that special quality which all truths supposedly have in common—is just a pseudo-problem based on syntactic overgeneralization. Unlike most other properties, being true is unsusceptible to conceptual or scientific analysis. No wonder that its ‘underlying nature’ has so stubbornly resisted philosophical elaboration; for there is simply no such thing.”

This is, to be sure, an intriguing line of argument. A notable limitation of the argument, though, is its Anglocentrism. The argument focuses exclusively on English alethic vocabulary, and for that reason, it leaves open the possibility that there is an expression e in a natural language L other than English such that (i) ‘true’ is a standard translation of e into English, even though (ii) the meaning of ‘true’ ≠ the meaning of e (recall from Section 2.2 that the relation of being a standard translation is not naively meaning-preserving). In this way, Horwich’s argument leaves open the possibility that there are natural languages containing alethic vocabulary that cannot be accounted for along his preferred deflationary lines. If there are such languages, then it may turn out that it is entirely worthwhile to investigate the underlying natures of properties that are expressed or denoted by certain pieces of alethic vocabulary. As a result, to fully evaluate Horwich’s wholesale dismissal of traditional debates about the metaphysics of truth, it is imperative for us to determine whether there are languages of this sort. That is, it is imperative for us to do cross-linguistic experimental philosophy.³¹

3.3 Debates about truth

3.3.1 The role of experimental philosophy

Lastly, we turn to debates about the nature of truth. As we mentioned in Section 2.4, the relevance of experimental studies to debates about truth has yet to be explored at length. This means that in thinking about why cross-linguistic experimental philosophy is indispensable to these debates, we first need to consider a prior question: in what ways should experimental philosophy inform these debates?

At first blush, it might seem that experimental philosophy simply has no relevance to debates about truth. Næss’ studies or that of Reuter & Brun (2022), for instance, seem to only provide information about our ordinary use of alethic vocabulary and by extension, our ordinary alethic thought. It is reasonable to suspect that this sort of information cannot improve our understanding of truth’s nature, but only, in effect, our understanding of how we ordinarily think and speak about truth. However, we would suggest that there are at least two ways in which experimental findings impact debates about the nature of truth.³²

Experimental findings can provide insight into our ordinary beliefs about truth. It seems likely that these beliefs are largely implicit, insofar as we do not ordinarily ask metaphysical questions like ‘What is the nature of truth?’ in everyday contexts. Whether they are implicit or explicit, our ordinary beliefs about truth are relevant to debates about truth’s nature insofar as they constitute preliminary evidence in these debates.

To see this, we should first note that we use our ordinary beliefs about truth to navigate the world.³³ Examples include (i) evaluating other speakers’ assertions as true or false (‘What John said is true/false’), (ii) drawing conclusions based on what we already believe (‘Given that p, it must be true that q’), and (iii) selecting trustworthy information sources (‘Website A reports the truth much more often than website B, so I trust A more than B’).

Moreover, this navigation is often successful. The relevant assertion may actually be true/false, the truth of the relevant conclusion may indeed follow logically from the truth of the relevant belief, or the more trustworthy information source may in fact be favoured over the less trustworthy. Absent evidence that it is caused by factors that are not conducive to the truth of the relevant beliefs about truth, this success constitutes evidence that truth is as it is represented in these beliefs, i.e. that these beliefs are true. It is for this reason that our ordinary beliefs about truth serve as preliminary evidence in debates about truth’s nature.

That said, since this evidence is preliminary, it can of course be overridden by other evidence. For instance, suppose that experimental inquiry indicates that in ordinary contexts, we tend to (implicitly) take truth to consist in some kind of correspondence to fact. The preliminary evidence that is provided by this ordinary belief can be overridden by philosophical arguments which demonstrate that correspondence theories of truth are incoherent, or scientific findings that our minds or language do not ‘represent’ the world in a sense that can underpin a correspondence theory of truth. Even so, our main point is that experimental findings regarding our ordinary beliefs about truth provide an evidential point of departure in debates about truth’s nature. They provide an indication as to where we should start our metaphysical investigations, even though they cannot tell us how these investigations will conclude.³⁴

A second way in which experimental philosophy can impact debates about truth is by providing evidence for or against objections against theories of truth which turn on (tacit) empirical conjectures. When it comes to debates about objections of this sort, the standard practice is for each party to trade competing hunches about the relevant conjectures, insisting on the plausibility of their favoured hunch. After a few salvoes with no clear winner, the debate usually stalls and the parties move on to some other topic.

Experimental philosophy offers a significant improvement to this standard practice. Thinking experimentally, the basic procedure in cases of this sort will be to identify the relevant conjectures and then determine how to evaluate them experimentally. Once they have been so evaluated, it will be much easier to diagnose the objections that they underwrite as being strong or weak.

We can illustrate these points by considering an objection that G.E. Moore offers against what he takes to be William James’ theory of truth³⁵:

It seems to me…that we very often have true ideas which we cannot verify; true ideas, which, in all probability, no man ever will be able to verify…[I]t is plain that…historians are very frequently occupied with true ideas, which it is doubtful whether they can verify. One historian thinks that a certain event took place, and another that it did not; and both may admit that they cannot verify their idea. Subsequent historians may, no doubt, sometimes be able to verify one or the other. New evidence may be discovered or men may learn to make a better use of evidence already in existence. But is it certain that this will always happen? Is it certain that every question, about which historians have doubted, will some day be able to be settled by verification of one or the other hypothesis? Surely the probability is that in the case of an immense number of events, with regard to which we should like to know whether they happened or not, it never will be possible for any man to verify either the one hypothesis or the other…And the theory against which I am protesting is the positive assertion that we can verify all our true ideas—that some one some day certainly will be able to verify every one of them. This theory, I urge, has all probability against it.

Moore’s objection starts with the suggestion that according to James, a belief b’s having truth consists in b being verifiable by either a present or a future human. However, Moore claims that there are true beliefs (e.g. beliefs about whether a particular historical event did or didn’t occur) which are not verifiable by any present human and will (probably) not be verifiable by any future human. Accordingly, Moore concludes that a belief b’s having truth (probably) does not consist in b being verifiable by either a present or a future human.

We can grant the first step of the objection for the sake of argument. The central issue for present purposes is what the support for the second step is meant to be. Moore’s suggestion seems to be that it is obvious that one of the historians he describes must have a true belief, even though that belief (probably) cannot be verified by any present or future human.

However, it goes without saying that philosophers can be mistaken about what is or is not obvious. To determine whether Moore’s suggestion should be accepted, we should proceed experimentally. In particular, we should determine whether Moore’s claim about the historian will be widely accepted by speakers who can competently use ‘true’ (or a standard translation thereof). If it is, then provided that Moore has accurately described James’ theory of truth, his objection against James’ theory will be strong. By contrast, if Moore’s claim is not widely accepted by such speakers, then we will be warranted in concluding that his objection is weak.

3.3.2 The role of cross-linguistic experimental philosophy

Now that we have outlined a couple of ways in which experimental philosophy should inform debates about truth, it is a relatively short step to explain why cross-linguistic experimental philosophy is indispensable to these debates.

As we indicated in Section 2.3, a theory of truth aims to identify the nature of the property/relation truth. It will be granted by all, or nearly all, parties to debates about truth that if truth has a nature, then its nature does not vary across linguistic communities. Rather, truth is standardly taken to have the nature that it has throughout the actual world (and perhaps in all possible worlds). Moreover, we lack reasons at present to believe that there are linguistic communities L₁, …, L_n whose ordinary beliefs about truth provide stronger preliminary evidence regarding truth’s nature than the ordinary beliefs about truth that are held by members of other linguistic communities. These two considerations indicate that when we aim to determine whether a theory T of truth aligns with our ordinary beliefs about truth, it would be unwarranted to study only the ordinary beliefs about truth that are held by members of a particular range of linguistic communities.

Instead, the course of action that is warranted by these considerations is to study humans’ ordinary beliefs about truth as such. Doing so will involve comparing the beliefs that are held by members of different demographic groups, including linguistic communities, to determine (i) how similar these beliefs are to one another and (ii) whether these beliefs are evidentially on a par or whether certain sets of these beliefs are evidentially superior to other sets. This shows that when debates about truth are properly pursued, cross-linguistic experimental philosophy is a necessary component of these debates.³⁶

Similar considerations apply in connection with objections to theories of truth such as Moore’s objection against James. The metaphysical upshot of this objection is meant to be that having truth does not consist in being verifiable by either a present or a future human. Again, this upshot is not meant to hold only within certain linguistic communities. Rather, it is meant to hold throughout the actual world (and perhaps in all possible worlds). Moreover, as mentioned above, we lack reasons at present to believe that there are linguistic communities L₁, …, L_n whose ordinary beliefs about truth provide stronger preliminary evidence regarding truth’s nature than the ordinary beliefs about truth that are held by members of other linguistic communities. Accordingly, when assessing the tacit empirical conjecture that is built into Moore’s objection, we should determine whether this conjecture is borne out among humans as such. Following this course of action will again require us to do cross-linguistic experimental philosophy.

4 A question, objections, and responses

To close this discussion, we will respond to a question and two objections that we take to be especially pressing.

4.1 Cross-linguistic diversity and theories of truth

Question: Consider two theories T₁ and T₂ of the nature of truth, and suppose for the sake of illustration that the world contains only two languages L₁ and L₂. Suppose further that ordinary beliefs about truth vary significantly across the world’s linguistic communities, so that T₁, but not T₂, aligns with the beliefs of L₁ speakers, and vice versa for L₂ speakers. For instance, T₁ might be a correspondence theory of truth and T₂ a coherence theory of truth, and L₁ speakers might (implicitly) take having truth to consist in correspondence to reality, whereas L₂ speakers (implicitly) take having truth to consist in coherence with some idealised set of beliefs. In this situation, which of T₁ and T₂ is better supported by the relevant preliminary evidence?

Response: We take this sort of case to be relatively straightforward. Given the description of this case, when it comes to evaluating T₁ and T₂, the ordinary beliefs about truth held by speakers of L₁ or L₂ are on an evidential par. That is, the quality of the preliminary evidence concerning T₁ and T₂ that is provided by beliefs in the former set is on a par with the quality of the preliminary evidence concerning T₁ and T₂ that is provided by beliefs in the latter set.

As a result, the key question is which of T₁ and T₂ aligns with the majority of the beliefs about truth that are held by speakers of either L₁ or L₂. If T₁ does so, then it is better supported by the preliminary evidence that these beliefs provide, whereas if T₂ does so, then it is better supported by this evidence. If the alignment is split (approximately) evenly, then as far as this preliminary evidence goes, we have a wash.

That said, further investigation might reveal that, say, L₁ speakers’ ordinary beliefs about truth are evidentially superior to those of L₂ speakers. This could happen, for instance, if L₂ speakers’ ordinary beliefs about truth were partially caused by a non-truth-conducive cognitive bias, whereas L₁ speakers’ ordinary beliefs about truth were not affected by such biases. In this case, the preliminary evidence that is provided by these beliefs would on balance support T₁ over T₂.³⁷

4.2 Experimental philosophy and debates about truth

Objection: Contrary to the arguments in Sects. 2.4 and 3.1, Jamin Asay (2022, forthcoming) has recently shown that truth-related findings in experimental philosophy, whether cross-linguistic or not, do not have clear consequences for debates about truth.

Asay draws an important distinction between conceptions and concepts. He takes a person A’s conception of an entity x to be A’s “set of beliefs or theories (whether implicitly or explicitly held) about [x].”³⁸ By contrast, Asay characterises the concept truth along the lines mentioned in Section 2.1, remarking that “[t]hose who possess truth can think and talk about truth as such.”³⁹

With this distinction in hand, Asay points out that while extant experimental studies on truth, e.g. those by Næss and Reuter and Brun, do look to provide us with valuable information about ordinary conceptions of truth, it is not clear that they provide us with information about the ordinary concept truth. As Asay puts the point, ordinary conceptions of truth have clear “empirical traces.”⁴⁰ To investigate ordinary conceptions of truth, we need to investigate the judgments that people ordinarily make about truth, and Asay concedes that experimental studies are an integral part of this project. However, he also observes that it is not clear when we are entitled to draw conclusions about a person’s concept of x from premises about their conception of x. Absent well-defined criteria for evaluating such inferences, we should be hesitant to draw any conclusions about truth from extant experimental results.

Response: This objection points in the right direction, but we should be clear about what it shows and what it does not show. It does not show that extant experimental studies fail to provide important information about truth. For this reason, it does not undermine the arguments that we offer in Sects. 2.4 and 3.1. Rather, the crucial point that the objection brings home is that when advancing hypotheses about truth, we need to be upfront about our background theory of concepts and in particular, what sort of entity truth is meant to be. If we fail to offer this information, then it will not be especially clear how the relevant hypotheses should be evaluated.

To illustrate, one proposal regarding truth’s nature that is admirably clear is that of Asay (2021). According to Asay, we should take truth to be the ability to have propositional thoughts. If this is indeed the best way to think about truth’s nature, then it seems that admittedly, extant truth-related findings in experimental philosophy don’t tell us much about truth. Most directly, these findings provide us with information about ordinary conceptions of truth, as well as information about how alethic vocabulary is ordinarily used. That sort of information would not seem to shed any light on issues such as why humans have the ability to have propositional thoughts, at what stage in development we acquire this ability, or what other cognitive abilities we have in virtue of having this ability. The best way to learn about all of those things would be to do research in e.g. developmental and comparative psychology.

However, there are other accounts of truth’s nature that are worth considering. One such account is that truth is a mental representation which purports to represent the property truth.⁴¹ Given this sort of background theory of the kind of entity that truth is, extant experimental findings do look to have significant implications for theories of truth.

Consider, for instance, the view that truth is a mental representation which purports to represent the property being a proposition that would be believed at the end of inquiry.⁴² How should we evaluate this view? One promising strategy would be to investigate ordinary conceptions of truth with the aim of determining whether humans ordinarily conceive of a proposition’s having truth as consisting in its being a proposition that would be believed at the end of inquiry. This strategy could be pursued, for instance, using methods much like the ones that were employed by Reuter and Brun (Section 2.4).

If this line of investigation turns out negative results, then those will constitute evidence against the target theory of truth. Of course, these results would not prove that this theory of truth is false. However, this state of affairs is entirely familiar, given that when we evaluate theories of truth against our best evidence, induction and abduction are the standard tools of the trade.

What Asay’s objection brings out, then, is the fact that when we assess theories of truth in light of truth-related findings in experimental philosophy, a crucial background consideration is how we should think about the nature of truth. As Asay indicates, given that this issue has received insufficient attention, a great deal of work remains to be done here.

4.3 The normativity of inquiry about truth

Objection: Inquiry about truth—whether it is focused on truth, alethic vocabulary, or truth—is inherently normative.⁴³ When we investigate truth and alethic vocabulary, what we ultimately want to know is how we ought to use this concept and these expressions. This is clearly illustrated, for instance, by debates about semantic paradoxes such as the Liar paradox. Put very briefly, the Liar paradox is generated by three factors: (i) the logic that we use (e.g. classical logic), (ii) the truth predicate that we use (e.g. one that obeys Horwich’s Equivalence Schema), and (iii) our capacity to refer to expressions of e.g. English in English itself.⁴⁴ Accordingly, when we determine how to respond to this paradox, one of the main questions that we have to consider is whether we should continue to speak (and think) about truth as we do, or whether it would be better to use a different truth predicate(s) (and a different truth concept(s)).⁴⁵

Similarly, when we investigate truth, what we ultimately want to know is how we ought to think about truth’s nature, in light of our best evidence. This evidence comes from philosophical arguments as well as scientific research on e.g. mental and linguistic representation.

That inquiry about truth is in these ways inherently normative shows that the methodology proposed here is inadequate. At most, cross-linguistic experimental philosophy can provide us with descriptive information about how individuals in various linguistic communities speak and think about truth. This information does not further the normative aims of philosophical inquiry about truth.

Response: We agree that inquiry about truth is partially normative. We also agree that the information that is provided by cross-linguistic experimental studies is descriptive. What we would emphasise, however, is that when investigating how we ought to think and speak about truth, we should be mindful of how we do in fact think and speak about truth. It is here that cross-linguistic experimental philosophy proves to be vital.

The main considerations that should be mentioned here are nicely articulated by Frank Jackson in a discussion of the role that ‘folk morality’ should play in moral theorising. Jackson (1998, p. 135) explains that “we must start from somewhere in current folk morality, otherwise we start from somewhere unintuitive, and that can hardly be a good place to start from. And we must seek a theory that stands up to critical reflection: it can hardly be desirable to end up with a theory that fails to stand up to critical reflection.”

Applied to debates about truth, the initial point is that in developing a theory of how we ought to think and speak about truth, we as theorists must have a starting point. The question we confront is: should we start by identifying the ways in which we already think and speak about truth, or should we simply try to envisage optimal ways of thinking and speaking about truth?

Taking our cue from Jackson, we would expand upon a point that we made in Section 3.3.1. Despite their potential flaws (e.g. vagueness or subtle inconsistency), our ordinary ways of thinking and speaking about truth regularly enable us to successfully navigate the world. They are not just everyday ways of thinking and speaking—they are ways of thinking and speaking that tend to work. When developing normative theories of truth, alethic vocabulary, or truth, we would thus be well-advised to start with these ways of thinking and speaking, revising them only when we have strong reasons to do so. If you already have an instrument that works well, then you should not try to invent a new instrument from scratch. Rather, you should use the current instrument, noting its advantages and limitations and making adjustments to it as you go.⁴⁶

The additional point to underscore here is that to determine how we—that is, we humans—tend to think and speak about truth, it is not enough to offer speculations from the armchair. Instead, we need to roll up our sleeves and do some empirical work, including cross-linguistic experimental philosophy.

5 Concluding remarks

In this paper, we have argued that cross-linguistic experimental philosophy is indispensable to three overarching debates about truth in analytic philosophy: debates about the concept truth, debates about alethic vocabulary, and debates about truth’s nature. In light of its indispensability, we propose that the dominant Anglocentric model for inquiring about truth should be abandoned.

At present, there are very few cross-linguistic studies of truth. A priority going forward, then, should be to conduct more studies of this kind and to expand the range of languages covered by such studies. In doing so, we will be able to generate significant progress in debates about truth and promote cross-cultural understanding within and beyond philosophy.⁴⁷