Chienkuo Mi（米建國）

Department of Philosophy, Soochow University

cmi@scu.edu.tw

Over last three or four decades, the discussions of what the concept of truth is, and of what a theory of truth can do or ought to be, have always been very important and popular issues in philosophy, but there seems to be no agreement reached among philosophers with respect to how we should understand the concept of truth and why we need a theory of truth.

When we study the concept of truth or try to find a definition for the concept, it does not necessarily lead us to the construction of a theory of truth. Correspondingly, when we endeavor to form a theory of truth, we are not required to focus on the concept of truth itself, nor are we forced to provide any definition for the concept. This situation is very similar to that we find in discussions of the concept of meaning and the theories of meaning. The conceptual analysis of certain concept does not necessarily connect with a theoretical construction bearing on the very concept at issue. A lot of philosophical discussions concerning the nature or the definition of the concept of meaning need not be regarded as constructing a theory of meaning. Some philosophical projects aiming at producing a theory of meaning may not be concerned with the nature of the concept of meaning, and one could do the theory of meaning without even taking up the concept of meaning.¹However, it is worth noting that there is an important difference between the cases of truth and meaning. The point that needs to be made is this: while we might have a theory of meaning without using or involving the concept of meaning, the construction of a theory of truth, if not directed to define the concept of truth, still presupposes or relies on the very concept of truth. Why is the concept of truth so fundamental for constructing a theory of truth, or indeed any theory? Why do we need a theory of truth?

Michael Lynch, in the “Introduction” to his bookThe Nature of Truth, distinguishes all kinds of “Robust theories of truth” from the “Deflationary theories of truth” based on the starting point (or question): “Does truth have a nature?” So, it is immediately obvious for him to sort out various theories just depending on whether the attitudes toward the starting question are positive or negative. However, Lynch

Consider a theory of meaning in the Davidsonian style in which “the one thing meanings do not seem to do is oil the wheels of a theory of meaning——at least as long as we require of such a theory that it non-trivially give the meaning of every sentence in the language.” (Davidson, 1967: 20)

also notices that “there is a growing consensus among some philosophers that neither traditional robust theories nor deflationary theories are right.” If it is so, then he claims that “we must find new ways to think about this old concept.” (Lynch, 2001: 5) It is my belief that we could find a new way to think about the old concept of truth in this paper—that is, viewing the concept from a theory of truth based on the linguistic perspective or language-involving phenomena.

1. Metaphysical Theory of Truth vs. Linguistic Theory of Truth:

Dealing with the concept of truth can be a quite different philosophical concern from constructing a theory of truth, although the two different concerns may be closely related. A distinction Michael Devitt has introduced can help us here. In order to tell what exactly the difference is between the deflationary theory of truth and the correspondence theory of truth, Devitt attempts to make a distinction between the metaphysical and linguistic issues (or foci) with respect to these discussions of truth. He claims that “[w]hereas the focus of the correspondence theory is on the nature and role oftruth, the focus of the deflationary theory is on the nature and role ofthe truth term, for example, of ‘true’. The former focus is metaphysical; the latter, linguistic.” (Devitt, 2001: 580) According to the distinction made by Devitt, it seems that if the focus of our theory isthe concept of truth, then the theory should be regarded as a kind of metaphysical theory; while if the focus of the theory is onthe truth term, then the theory should be linguistic in kind. Devitt’s distinction, it seems to me, is not only misleading (because the distinction might miss or obscure what the truly significant difference is between the correspondence theory and the deflationary theory), but also too trivial (because it is so obvious to treat the nature of truth as an metaphysical issue and the meaning of truth term as a linguistic issue), and too narrow at best (because it cannot capture the whole picture of the debates among so many different theories of truth existing in contemporary philosophy).

Contrast to what Devitt has suggested, I propose to draw a distinction between the metaphysical level and the linguistic level when we discuss the issues concerningthe theory of truth(rather than concerningthe concept of truth). Different from Devitt’s proposal, my distinction of the different levels is based onwhat we try to ask the theory of truth fororwhy we need to construct such a theory(rather than being based onwhat the object the theory focuses). A theory of truth on the metaphysical level (I will call this MT) is a theory concerning with the nature and the role of the concept of truth, aiming at providing a definition or an explication for the concept. A theory of truth on the linguistic (or language-involving) level (LT for short) is a theory purporting to explain how various linguistic phenomena, such as communication,

understanding, and interpretation, are possible by relating the concept of truth with other semantic concepts such as that of meaning and that of belief. We may also say that MT is a theory focusing on the concept of truth, whereas LT is a theory of truth focusing on ordinary linguistic phenomena.

I think that most contemporary discussions regarding the concept of truth have been loaded with traditional metaphysical burdens, and that many varieties of theories of truth are metaphysical in nature. Most contemporary theories of truth are concerned with the nature of truth, and hope to find the definition and philosophical significances of the concept.²This situation should remind us of the controversy over the concept of “universal” begun in medieval times. Just like the three different kinds of metaphysical theories of universals (i.e., realism, conceptualism, and nominalism respectively), I suggest that there are three matching metaphysical theories with respect to the concept of truth too. Comparable to realism about “universals”, we have the correspondence theory of truth; to conceptualism about “universals”, we have the coherence theory of truth and the pragmatic theory of truth; and to nominalism about “universals”, the deflationary theory of truth. The last two groups of theories of truth, just like their counter parts in the theories of universals, are anti-realist in nature.

The debates between realism and anti-realism can be characterized as the controversies as to whether there is at least something real that is independent of the representations of minds for its being. However, the controversies are not as simple as they look, because the domains or the objects in question vary with different contexts and discussions. For example, common-sense realists would assert that there is at least something existing in the physical world (the world in which we live) that is independent of the human minds, while common-sense anti-realists deny this. Scientific realists assert that there are at least some theoretical entities dealt in scientific theories that are independent of scientific representations of scientist minds for their beings, while scientific anti-realists deny that there are any of these independent entities. Realism about “universals” is a theory asserting that universals exist independent of human minds, while anti-realism about “universals” denies any such realistic universal. Realism about “truth” or alethic realism³is a theory asserting that what makes a sentence or belief true is not merely dependent on the human minds, while anti-realism about truth denies that there is any mind-independent truth-maker. It is therefore important to notice the multi-domains (or different objects) involved in the realism-antirealism debates, if we want to make the best of this controversy.

Scott Soames has nicely generalized three main things that we may generally

expect a theory of truth to do (Soames, 1984: 411). According to Soames’s proposal, a theory of truth is generally expected:

(i) to give the meaning of natural-language truth predicates;

(ii) to replace such predicates with substitutes, often formally defined, designed to further some reductionist program; or

(iii) to use some antecedently understood notion of truth for broader philosophical purposes, such as explicating the notion of meaning or defending one or another metaphysical view.

With respect to Soames’s classification, we may expect that there are at least three different kinds of theories of truth. Let’s call them T1 (corresponding to (i)), T2 (to (ii)) and T3 (to (iii)) respectively.

The reason why Soames classifies these three different expectations for a theory of truth is to clarify what Tarski’s theory of truth is meant to do. He claims that Tarski’s definition of truth is neither an attempt to analyze the meaning of natural-language truth predicates (not a theory in the sense of T1), nor an attempt to use the notion of truth for broad philosophical purposes (not a theory in the sense of T3 either). Tarski’s theory of truth is not a theory in the sense of T1 (not an attempt to give the meaning of natural-language truth predicates) because he restricts his definition to cases in which truth is predicated of sentences of certain formalized languages only. Nor is his theory of truth a theory in the sense of T3 (not an attempt to use the notion of truth for broad philosophical purposes) because for Tarski the concept of truth itself is just what has to be explicated and legitimated. Tarski’s goal, according to Soames’s understanding, is to replace truth predicates used in our natural languages with certain restricted but formally defined substitutes. Tarski’s reductionist program of replacing natural-language truth predicates by the formally designed substitutes has two motivations: first, to remove the doubts of certain scientifically minded truth skeptics; second, to eliminate what he takes to be the inconsistency in our ordinary notion of truth brought out by the liar paradox. Therefore, Tarski’s theory of truth should be understood as a theory in the sense of T2.

It is not my concern here to evaluate whether or not Soames’s interpretation of Tarski’s theory of truth is correct. I have to admit that Soames’s classification of different kinds of theories of truth with respect to what those theories try to do may help us clarify some potential confusion and misunderstandings. However, I also have to point out that Soames’s classification isad hocat best. As I mentioned above, his distinction of what one can expect from a theory of truth is meant to locate Tarski’s theory in the right place in order to clarify and justify Tarski’s reductionist program for the concept of truth. Besides, the distinction between T1 and T2 is not needed. The only reason Soames separates the expectation for a theory of truth involved in T1

from that in T2 seems to be that he wants to emphasize what Tarski has done for the concept of truth must be viewed in the context of a formalized language rather than any natural language. And the reason why Tarski restricts his explications of truth within the context of a formalized language is to avoid some truth skepticism and the well-known liar paradox. However, if our concern here is with what we may expect a theory of truth to do or what a theory of truth ought to be, T1 (giving the meaning of natural-language truth predicates) and T2 (replacing natural-language truth predicates with formally defined substitutes) are both fundamentally metaphysical in nature. The differences between T1 and T2 will depend on what the theory takes to be its metaphysical standpoint (i.e., whether we could reductively define the concept of truth or find the nature for it, or we could only minimally interpret or understand the truth predicate (or the truth term) in natural languages). The concerns about the truth skeptics or the liar paradox, and about its possible scope applied to formalized or natural languages are beside the point, or at least are not directly related to the issue. T1 and T2 are, according to my distinction drawn above, both theories of truth based on the sense of MT. T3 can be regarded as a kind of LT, because the purpose of that theory is not to define the concept of truth or to find the nature of it, but rather to use some antecedently understood notion of truth for broader philosophical purposes (for shedding light on various linguistic phenomena as I see it). Therefore, I generalize two main things that we may generally expect a theory of truth to do (corresponding to the distinction between MT and LT): either (i) to find the nature of the truth and to define the concept, or (ii) to explain how various kinds of language-based phenomena, such as communication, understanding, and interpretation, are possible.

I want to claim that (1) we don’t need a theory of truth in the sense of MT: any attempt to find the nature of truth or to define the concept would either be circular or fail to have the explanatory power; (2) we do need a theory of truth in the sense of LT: the general concept of truth is relatively clear and fundamental on which a satisfactory theory of meaning and predication can be built. In order to argue for these two claims, I will focus mainly on the correspondence theory of truth, an oldest perspective and the most commonsensical view about truth, and show why this theory is troublesome; I will also rely on a Davidsonian point of view and explain why truth is not definable while at same time is essential for constructing a theory of truth that aims to explain various linguistic phenomena.

2. Why We Don’t Need a Metaphysical Theory of Truth

As we shall see, following Davidson’s claims regarding MT we will get all the negative points of views, while following his claims concerning LT we will

see more positive perspectives and gain more interesting insights about the concept of truth and approach toward the theory of truth.

[Metaphysical Claims about the Theories of Truth]

(M1) Correspondence theories of truth are empty as definitions, and the concept of truth is not “radically non-epistemic”; but these theories do capture the general intuition that truth depends on how the world is.

(M2) Coherence and pragmatic theories of truth are mistaken in concluding that reality and truth are merely constructs of thought, and the concepts of truth is not “radically epistemic”; but these theories have the merits of relating the concept of truth to human concerns, like language, belief, thought and intentional action, and it is these connections which make truth the key to how mind apprehends the world.

(M3) A deflationary attitude to, or a disquotational view of, the concept of truth is not encouraged by reflection on Tarski’s work, and the prospects for a deflationary or a minimalist theory of truth are dim. But Davidson is sympathetic with the deflationists because the attempts to pump more content into the concept of truth are not appealing for the most part their views can only serve negatively as avoiding well-marked dead ends and recognizable pitfalls.

(M4) The concept of truth is not definable, for it cannot be reduced to other concepts that are simpler, clearer, and more basic. The concept of truth may not be a goal of inquiry if it is to find substantiating evidence for our beliefs or to identify some entities for the concept to represent for or correspond to.

Claims (M1), (M2) and (M3) have appeared in several of Davidson’s articles and strongly suggest that Davidson does not side with any contemporary philosophical attempts to define the concept of truth. Those attempts include correspondence theories of truth (or the “objective” theories so called by Davidson⁴), coherence theories and pragmatic theories of truth that in one way or another make truth an epistemic concept (or the “subjective” theories), and minimalist or deflationary theories of truth—or as I prefer to call them, realist theories (the objective views), conceptualist theories (the subjective views) and nominalist theories (the deflationary views) about truth.

Realism about truth has its focus on ontology, because it tries to define truth

For the discussions of, and the distinction between, objective and subjective views on truth, see Davidson’s “Epistemology and Truth”.

in terms of some objective entities to which the truth bearers can correspond. However the focus turns out to be the fatal wound for the theory itself. It has always been challenged that the realists fail to individuate, identify, or locate the fact of the world or the part of reality to which a true sentence cor