Imagine the following type of encounter:
Jones: The Red Sox lost the 1967 World Series.
Smith: That's true.
Jones's remark is the antecedent of Smith's utterance of "that's true." Smith's utterance is not a referring expression, but it is an acknowledgment of the presence of the antecedent. The expression acts semantically like its antecedent. Smith's utterance would fail to make sense if Jones had not said what he said. Smith's truth ascription would hang in the air. This kind of truth has been called the prosentential theory of truth. Such a theory of truth draws truth conditions of an anaphor, "that's true," from the antecedent, in this case "The Red Sox lost the 1967 World Series."
In Chapter 5 of Making It Explicit, Robert Brandom (1994) offers a prosentential theory of truth. The purpose of this paper is to argue that Brandom's prosentential account is unsatisfactory. In particular, I will show that the idea that truth must always be understood as a prosentence that refers anaphorically is incorrect. The similarity of content in anaphoric antecedents and prosentences is a result of the two sentences having the same truth conditions. Since the semantic content of anaphoric antecedents and prosentences presume a relation between a proposition and the world, these truth conditions turn out to be genuine, language-independent properties.
The Prosentential Account of Truth
Brandom offers an expressive role of the concept of truth, and the most successful and most sophisticated candidate is a prosentential theory. A prosentential theory of truth was first devised by Franz Brentano in The True and the Evident (Trans. Oskar Kraus and Roderick Chisholm. New York: Humanities Press, 1966). It was popularized and made more sophisticated by the triumverate of Dorothy Grover, Nuel Belnap, and Joseph Camp in "A Prosentential Theory of Truth," Philosophical Studies 27 (1975): 73-124.
A prosentential theory claims that locutions such as "... is true" and its relatives are proform-forming operators. Despite the subject-predicate structure, these really are \textit{prosentences}. "That is true," a relative of the original locution, is a prosentence, which relates to and inherits content from anaphoric antecedents. Prosentences inherit content in the same way that a pronoun such as 'he' relates to and inherits its content from an anaphoric antecedent. For example, in the sentence "Tiffany lifts weights, so she is strong," the word "she" inherits its content from the anaphoric antecedent, "Tiffany." "That's true" is an anaphor for some antecedent and functions in the same way "she" functions in the example. In simple cases ``that is true'' is a prosentence that inherits its content from an anaphoric antecedent, e.g. "snow is white." In more complex cases, there are quantificational pronouns such as "it" in "Any positive integer is odd if it is not even." We cannot substitute "any positive integer" for "it" without changing the semantic content of the proposition. An example of a quantificational prosentence is "Everything Lester Holt says is true." Brandom's account can handle these troubling cases by simply expanding the claim to "For anything one can say, if Lester Holt says it, then it is true." Once we understand how this works, we need only employ the lazy way to understand how the anaphora inherits content from the antecedent (Brandom 1994, p. 302). The theory holds that the uses of truth talk create sentences that pick up the semantic content of anaphoric antecedents, having whatever content the antecedent has.
Understanding an anaphoric dependent is a two step process. First, we compute the class of antecedent tokenings, and, second, we determine the content of the anaphoric dependent as a function of the contents of its antecedents.
Such an approach to truth has many advantages. First, the anaphoric account handles embedded uses of 'true' very well. When 'true' is used in the antecedent of a conditional statement, no commitment is undertaken, no assertion endorsed, and no pro-attitude expressed. Yet, according to Brandom, it still is propositionally contentful making an essential contribution to the content of the conditional claim since embedded prosentences are as straightforwardly interpretable as embedded sentences. Second, the anaphoric account handles cases in which the anaphoric antecedent contains indexicals, demonstratives, anaphoric dependents, and expressions in foreign languages very well. If the antecedent of my uttering, "that's true" is your saying, "I am hungry," the second stage of the computation of the content of my remark must adjust the indexical from your mouth to mine and yield something equivalent, "you are hungry." Third, the theory that construes 'true' as a prosentence-forming operator generalizes to a treatment of `refers' as a pronoun-forming operator. I
will omit an explanation of reference here since our main concern is with Brandom's notion of truth. Finally, the prosentential theory of truth is metaphysically parsimonious. No longer is there a need to explain the attributions of a special and mysterious property of truth since truth is exhibited as uses of grammatical proforms anaphorically referring only to the sentence tokenings that are their antecedents.
Undermining the Anaphoric Chain
We should now consider at least one objection to Brandom's account of a prosentential theory of truth. If we can undermine the relation between prosentences and their anaphoric antecedents, then the advantages that follow from Brandom's account disappear. What we will find is that we should abandon Brandom's theory altogether.
In prosentential theories, 'true' appears to be a sentential operator. Formalizing Brandom's proform forming-operators like "that's true" entails "It-is-true-that..." This would imply that anaphoric phrases function as some kind of logical operator. If a prosentence acts as a logical operator, then it is not easy to see how it is going to be analyzed so that it does not turn out to refer to a property of truth.
On Brandom's account, however, uses of the prosentence and its relatives are not to be understood as logical operators in the sense construed by this objection. On the contrary, we must understand that they are to be used as proform-forming operators - prosentences. To think that these anaphoric phrases are logical operators is to attribute to them a characteristic Brandom explicitly blocks. We cannot construe prosentences as logical operators without being uncharitable to Brandom, so we should consider whether we can undermine the inheritance relation between the anaphora and the antecedent.
A prosentential theory of truth can only express the truth by agreement with an already-made statement, by agreement with a statement you are not in a position to repeat, or by agreement with many statements at once. Sometimes we are not at all clear about what is meant by some sentences. For example, Jones may say
to Smith, "Stand over there," or "That is a pretty red one." In these types of sentences, no clear sense of meaning is being conveyed between Jones and Smith. For Smith to say, "that's true" in response to Jones may be appropriate, but it is unclear to what Jones is referring. The anaphoric antecedent is empty of content. If the anaphoric antecedent is empty, then there is no clear relation between the anaphoric antecedent and the prosentence that depends on it. The prosentence has a direct link to the antecedent, but it is unclear whether the antecedent is actually true. The truth conditions of the antecedent fall into oblivion. Thus, there is no way to tell what is true from what is false in Brandom's account without an account of reference.
The intimate relationship between truth and reference may be extended in the following way. We may challenge the origins of the content in the anaphoric antecedent and the prosentence. Anaphoric antecedents have semantic content. The prosentence inherits semantic content from its anaphoric antecedent. So, the semantic content of anaphoric antecedents and prosentences are identical. Semantic content presumes a relation between a proposition and the world. According to Brandom's theory, we derive truth conditions from the semantic content. If the semantic content presumes a relation between a proposition and the world, the truth conditions must be understood as talk about genuine, language-independent properties of sentences. Thus, the similarity of content is a direct result of the anaphoric antecedent and the prosentence having the same truth conditions - genuine, language-independent properties.
Since the similarity of content is a direct result of the anaphoric antecedent and the prosentence having the same truth conditions, Brandom's account presupposes that truth is a property. The resulting theory is far more robust than Brandom would like.
Brandom emphasizes accounts of the use of the truth claims, and he is not interested in truth conditional theories of the content of sentences. Truth conditional theories of content of sentences consider representations of the semantic content in terms of truth conditions. Brandom wants to avoid this commitment. For Brandom,
"... is true" is just another locution of the language, while what was offered above "... is true" occurs as
a component of the notion of a truth condition.
Conclusion
If Brandom is correct that prosentences are merely locutions in the language, then it is hard to see how we can use these locutions appropriately. When someone says, "that's true," we need to confirm whether what the prosentence inherits is actually true or not. Confirming the content of the antecedent involves a relation between the proposition and the world. On this reading, Brandom's account posits a relation between language and the world to account for truth claims. Just because Brandom's account gives an account of the proper use of certain semantic claims does not mean he can give up the relationship between language and the world.
By wanting to offer an account of truth where the prosentence, "that's true," does not refer to a property, the claim only makes sense on a robust construal of predicate reference. One of the advantages of Brandom's truth theory is that it generalizes to treatment of reference. If the claim that truth is not a property only makes sense on a robust construal of predicate reference, then the truth theory presupposes a treatment of reference. Circularity, therefore, forbids the generalization of truth theory to a theory of reference.