The Prosentential Theory of Truth. Beyond Realism and Anti-realism
María J. Frápolli
Departamento de Filosofía
Universidad de Granada (Spain)
frapolli@ugr.es
1. Conceptual Analysis
The debate realism/antirealism comes in (at least) two flavors: metaphysical and
epistemic; the semantic formulation of the debate due to Dummett that defines realism
as related to classes of statements rather than to classes of entities 1 is reducible to one of
the other two. There are theories of truth that explain truth as a metaphysical notion
(“correspondence” to facts) and others that explain it in epistemic terms (coherence, or
assertibility), and it is not infrequent that the realism/antirealism debate turns into the
correspondence/coherence debate.
Metaphysical realism states the independence of the reality around from our
thought and will. A realist statement about a particular domain (metaphysics, ethics,
esthetics, semantics, logic) is the acknowledgement of the existence of facts of the kind
(i. e metaphysical facts, moral facts, esthetics facts, semantic facts, logical facts). The
standard explanation in this context is that truth is correspondence with facts. Truth is
ascribed to a proposition if there is a fact that makes it true. Epistemic realism, on its
turn, states the objectivity of knowledge. Truth, knowledge, and objectivity lies thus on
the side of realism. And antirealism is left with a picture in which a robust notion of
1
truth is not applicable, in which objectivity is not possible, and knowledge is substituted
by subjective opinions and the relativity of perspectives. This is the standard view.
Nevertheless, truth is neither a metaphysical nor an epistemic notion, and a
complete account of truth that explains the meaning and use of a truth operator is
compatible with any particular position in metaphysics and epistemology. The debate
between realists and antirealists doubtless raises profound philosophical questions, but
none of the parties is justified in claiming exclusive rights on truth, knowledge and
objectivity. A partial reason that explains why truth is generally involved in
metaphysical and epistemic debates lies in the fact that the truth operator is an
indispensable instrument of propositional generalization and metaphysical and
epistemic discourse essentially deals with general thoughts.
I will argue that it is legitimate to ascribe truth to propositions and propositional
contents independently of our realist or antirealist sympathies. The notion of truth
doesn’t belong to this debate, to which it enters, as we have said, as an expressive tool,
but is not essentially involved. Truths ascription play their role once that some
propositional contents have been accepted. The home of the realism/antirealism debate
is the justificatory level, i. e. how and why we assume that some contents are claimable;
only after that truth appears in the picture. This point is particularly relevant for the
question of realism/antirealism for it shows that there can be a neutral definition of truth
that both parties, realists and antirealists, are allowed to use. Besides, removing the
question of truth from the metaphysical and epistemic discussion clarifies both
discussions, and permits individuate the actual difficulties related to the definition of
truth in natural languages as much as those related to the structure of reality and our
access to it.
In these pages I will be concerned with an exercise of conceptual analysis. I will
argue for the thesis that the notion of truth is not essentially involved in the debate
between realists and anti-realists. It is, so to say, put at work in this debate but it does
not belong to one side better than to the other.
The theory of truth I would like to put forward, the prosentential theory of truth,
is a technical proposal about the meaning of the truth operator in natural languages. Or
more precisely, it describes an operator that in many natural languages (certainly in
English, French, and Spanish) corresponds to the meaning of the truth predicate.
Tarski already declared that his semantic theory of truth was neutral between the
general, metaphysic and epistemic, philosophical views that one endorses. Although, on
the other hand, the semantic theory of truth is related to the objectual interpretation of
quantifiers that is one of the realists weapons.
Truth is not an epistemic notion (my Canarias paper).
Truth is not a metaphysical notion either. The basic tenet of a correspondence
theory of truth is that truth is correspondence to facts. Yes, this tenet is correct, but
unfortunately it is correct because it is empty. “is true” is a prosentence builder and the
same is “is a fact”, defining truth as correspondence to facts is moving in circles. This is
one of the reasons of the undeniable appeal that the correspondence theory has enjoyed
throughout the centuries. Another reason is the ubiquitous representation account of
content as provided by the referents of the expressions of language.
C.J.F. Williams, and independently, R Brandom, have explained the actual conceptual
poverty that lies under the correspondence view.
It is legitimate to ascribe truth to propositions and propositional contents
independently of our realist or antirealist sympathies. The notion of truth doesn’t belong
to this debate, to which it enters, as we have said, as an expressive tool, but is not
essentially involved. Truths ascription play their role once that some propositional
contents have been accepted. The home of the realism/antirealism debate is the
justificatory level, i.e how and why we assume that some contents are claimable; only
after that truth appears in the picture. This point is particularly relevant for the question
of realism/antirealism for it shows that there can be a neutral definition of truth that both
parties, realists and antirealists, are allowed to use. Besides, removing the question of
truth from the metaphysical and epistemic discussion clarifies both discussions, and
permits individuate the actual difficulties related to the definition of truth in natural
languages as much as those related to the structure of reality and our access to it.
The truth predicate works, in natural languages, as a builder of prosentences.
Prosentences are the equivalent in natural languages of propositional variables in
artificial languages. A way of offering an exhaustive account of the meaning of truth in
natural languages is following the threefold traditional distinction, due probably to
Peirce and recovered by Morris, and explaining the syntactic, semantic and pragmatic
roles performed by the truth predicate. Roughly stated, its syntactic job is restoring
sentencehood; truth ascriptions are proforms of the propositional kind, and its semantic
job, as it happens with the rest of proforms, is threefold: they work (i) as vehicles of
direct propositional reference, (ii) as vehicles of anaphoric reference, and (iii) as
instruments for propositional generalization. The pragmatic role of truth ascriptions is
the endorsement of propositional contents, i.e. the explicit acceptance of propositional
contents as appropriate items to serve as premises of inferential acts.
Several different authors from different philosophical positions have pointed at
the syntactic, semantic and pragmatic aspects of truth. The syntactic job of the truth
predicate as a way of restoring sentencehood has been acknowledged from diverse
points of view. Prior’s (Prior 1971) and Horwich’s (Horwich 1998) characterization of
the truth operator as a denominalizer and also Quine’s disquotationalism all stress this
syntactic feature.
The view that is nowadays known as the prosentential theory of truth mainly
focuses on the semantic job of truth ascriptions. Prosentences are a kind of proforms.
Proforms are natural languages variables that serve as dummy expressions that
reproduce the role of any instance of the logical category to which they belong.
Pronouns are the best known among them, but they are not the only ones. There also are
pro-adjectives, pro-adverbs and, as we will see, also prosentences. I have said that
proforms reproduce the role of any instance of a logical category. And this needs an
explanation. When linguists qualify an expression as a “pro-noun” they classify it in the
category of singular terms, and in fact a pronoun is a term that can be substituted by any
singular term salva gramatica. Nonetheless, the perspective I will take here is a bit
different. I am classifying expressions according to their logico-semantic behavior
rather than according to their syntactic status. The two ways of classifying pro-forms are
not equivalent. Some syntactic pro-nouns are logically pro-adverbs, pro-adjectives or
even pro-sentences. Terms like “it” and “that” can inherit any content whatsoever, and
are thus all-purpose proforms. This will become clear in what follows. Ramsey
(Ramsey 1927) was the first philosopher to use the term “prosentence” related to the
truth operator. Almost 50 years after Ramsey’s work, Grover, Camp and Belnap
(Grover, Camp and Belnap 1975), on the one hand, and Williams (Williams 1976), on
the other, develop the prosentential account independently.
From a pragmatic perspective, the truth operator makes explicit the speaker’s
endorsement of a content. By qualifying a propositional content as true, the speaker
commits herself to that content as something for which she is ready to give reasons, if
required. By assuming a content as true, one is giving permission to use it as a premise
in an inferential game. This aspect of the truth predicate has been put forward by
Brandom (Brandom 1994). Strawson, when explains the role of truth as a marker of
illocutionary force, hints at this pragmatic feature develop by Brandom. Nevertheless,
Strawson’s treatment of the truth predicate doesn’t stop here and he also accounts for
some of the syntactic and semantic features that I have related here as characteristic of
the prosentential view2.
I endorse the prosentential theory of truth and propose implementing it with the
syntactic insights given by Prior and Horwich, on the one hand, and with the pragmatic
picture developed by Brandom, on the other. All this, I maintain, represents a complete
theory of how the truth operator works, it answers the essential philosophical questions
traditionally related to truth, and also shows where and why most of the traditional
treatments have gone astray.
2
In many of the cases in which we are doing something besides merely stating that X is
Y, we are available, for use in suitable contexts, certain abbreviatory devices which
enable us to state that X is Y […] without using the sentence-pattern “X is Y”. Thus, if
someone asks us “Is X Y?”, we may state (in the way of denial) that X is not Y, by
saying “It is not” or by saying “That’s not true”; […]. It seems to me plain that in these
cases “true” and “not true” (we rarely use “false”) are functioning as abbreviatory
statement-devices of the same general kind as the other quoted (1950, pp. 174-175).
According to the prosentential theory, “is true” is a prosentence builder, and so
is “is a fact”. Correspondentist theories that define truth by the equivalence “something
is true iff it is a fact” or any variant of it don’t affirm anything wrong at the prize of not
affirming anything at all. The claim that something is true iff it is a fact is shown by the
prosentential account to be a generalization, whose appropriateness to represent a
definition of truth cannot be assessed until its instances have been displayed.
The insight behind redundancy proposals is explained relying on the syntactic character
of the truth predicate as a denominalizer. In natural languages there also are
nominalizers, i.e., operators whose jobs is converting non-nominal phrases into singular
terms. Inverted commas and the particle “that” at the beginning of a sentence have this
effect. If one applies a nominalizer to a sentence, inverted commas for instance, and
converts it in a singular term, and afterwards applies a denominalizer, it is no wonder
that one arrives at the same result with which one had begun. This is exactly what
happens with Tarskian T-schema.
Truth is not an epistemic notion. Even so, the truth predicate is omnipresent in
epistemological discourse, and that not even the most basic theses in epistemology can
be stated without essentially using the truth predicate. Besides, the endorsement role
that the truth predicate performs in natural languages is applied in many cases to the
items that have passed the kind of justificatory filters sanctioned by epistemology. The
prosentential account can explain the insight that traces a connection between truth and
justification; as the truth operator is a means of forming prosentences, i.e., propositional
variables, it (or any equivalent operator) has to be around always that propositional
generalizations are needed. The truth operator, in its generalization use, is the natural
language counterpart of propositional quantifiers and proposicional variables of
artificial languages. Epistemology and philosophy of science are paradigm contexts in
which we deal with packs of propositions, and the only way natural languages have to
talk about propositions in general is by means of propositional variables, i.e.
prosentences.
Truth is not a metaphysical notion either. Metaphysics is another context in which the
use of prosentences is essential. The predicates “is true” and “is a fact” help to express
metaphysical concerns. But here, as in the case of epistemology, the truth predicate is
put at work; it is doing its job as a means of propositional generalization.
If the prosentential theory is correct, then the analysis of truth is neutral with respect to
the realism/anti-realism debate. This debate has to do with the structure of reality and
our access to it. But the truth operator operates, so to say, at a second stage, i. e. it
operates on the outputs of the justification processes. These processes can be positioned
on any zone of the justificatory spectrum, they can be scientific procedures or
assumptions of common sense, and they can be empirical or not, formal or not. All this
belongs to epistemology and pragmatics, and would constitute the first step on top of
which an explicit ascription of truth would be the second one.
Mixing up the realism/anti-realism debate with the definition of truth is the effect of a
poor understanding of the way in which the truth operator works. The debate
realism/anti-realism surely tough upon fundamental philosophical questions, but none
on which the truth predicate is essentially involved.
References
Brandom, R. (1994), Making it Explicit. Reasoning, Representing, and Discursive
Commitment, Harvard University Press
Grover, D., Camp, J., and Belnap, N. (1975), ‘A prosentential theory of truth’.
Philosophical Studies, vol. 27, pp. 73-125. Also in Grover (1992).
Horwich, P. (1998), Truth. Oxford Clarendon Press
Kneale, W. and Kneale, M. (1962), The Development of Logic. Oxford Clarendon Press
Prior, A. (1971), Objects of Thought. Oxford: Clarendon Press
Ramsey, F.: (1929), “The nature of Truth”. N. Rescher y U. Majer (eds.), On Truth.
Original Manuscript Materials (1927-1929) from the Ramsey Collection at the
University of Pittsburgh, Kluwer Academic Publishers, 1991, pp. 6-24.
Strawson, P. F.: (1950), ‘Truth’. In Simon Blackburn and Keith Simmons (eds.): Truth.
Oxford Readings in Philosophy. Oxford University Press, 1999, pp. 162-182.
Williams, C.J.F.: (1976), What is Truth? Cambridge University Press