Comments on David Johnston’s “Identity, Necessity, and Propositions”

 

            Johnston maintains that the notion of a proposition—a language independent (abstract) particular—can be dispensed with in philosophical semantics and replaced with that of a propositional act. A propositional act is a component of a speech act that is responsible for the propositional content of the speech act. Traditionally, it is thought that a propositional act yields the propositional content of a speech act by being an act of expressing a proposition. And it is the expressed proposition that serves as the propositional content of the speech act. Johnston points out, however, that a propositional act is a structured event consisting minimally of a referential act, a predicative act, and a time-designative act. And on Johnston’s view, the propositional content is the structured propositional act itself. (Strictly speaking, Johnston analyzes sameness of propositional content in terms of sameness of propositional act type, from which I, perhaps rashly, inferred that the propositional content of a speech act should be taken to be the propositional act itself). 

            Johnston argues that a semantic analysis in terms of propositional acts enables us to reconcile the necessity of both,

(1)   Hesperus is Phosphorus

(2)   Phosphorus is Phosphorus

with their intuitive difference in meaning, while maintaining the direct reference theory of proper names. Moreover, he argues that invoking propositional acts rather than propositions has the advantage of being able to capture the various senses in which distinct statements might be said to “say the same thing”, as well as that of ontological parsimony. I will address each of these claims in turn, but first, I want to point out that the propositionalist can easily reconcile the necessity of (1) and (2) with their intuitive difference in meaning, without forsaking direct reference. All one needs to do is invoke the Fregean idea that there is a meaning shift in “that”-clauses of (opaque) indirect discourse (and other) ascriptions. In extensional contexts, co-referential names have the same meaning (in the sense of truth-conditional contribution). And, hence, all true identity statements that utilize names are necessarily true. But in opaque ascriptions, co-referential names can differ in meaning. As a result, on this view, there is a difference in meaning between (1) and (2), but one that emerges only in certain non-extensional contexts.

            Johnston argues that in different contexts, and for different conversational purposes, there are distinct criteria for “saying the same thing.” In one context, two utterances say the same thing—that is, have the same propositional content—if the referent, attributed property, and designated time coincide. In another context, the speaker needs to additionally know the referents of the two utterances coincide. In a third, context, both the speaker and the audience need to know this. Since on Johnston’s view, the propositional content of an utterance is the propositional act token that in part constitutes the speech act, he can analyze these distinct criteria for same-saying in terms of distinct ways of typing propositional act tokens. The propositionalist, in contrast, lacks this flexibility. Since the propositional content is taken to be a proposition, only one criterion for same-saying is available: two utterances say the same thing if and only if they express the same proposition. But this is an oversimplification of the propositionalist picture. There is a well motivated distinction between opaque and transparent indirect discourse (and other) ascriptions: roughly, ascriptions are used opaquely when how the ascribee conceives of the objects of thought and talk is at issue, and they are used transparently when the truth-value of thoughts and statements about the world is at stake. And the Fregean can simply take the propositional content of the ascriptive “that”-clause to consist of distinct propositions as between opaque and transparent uses. Moreover, further flexibility can be achieved by appeal to the distinction between the information literally expressed by a statement and the information pragmatically imparted by it. For example, the propositionalist might concede that John’s utterance of “The man in the corner is a thief” suffices for the literal truth of “John said that the Chief Justice of the Supreme Court is a thief”. But it could be maintained that when uttered in the legal context Johnston describes, the ascription pragmatically imparts information to the effect that the audience of John’s original utterance could have reasonably been expected to know that John was referring to the Chief Justice of the Supreme Court—information that may or may not be true.

            According to Johnston, a propositional act type is necessary when (=if and only if?) tokens of that type are true in every possible context in which they occur. And he defines the necessity of sentences in terms of propositional act types as follows: a sentence is necessary when (=if and only if?) it determines a necessary propositional act type, where a sentence S determines a propositional act type T if and only if any utterance of S is an instance of T.[1] The propositional act type determined by (2) is formally necessary: its necessity stems from the fact that any utterance of a sentence of the form “X is X” is true in its context of utterance, regardless of the meaning of “X”. And the propositional act type determined by (1) is semantically necessary[2]: its necessity stems from the fact that “Hesperus” and “Phosphorus” are coreferential rigid designators.

            There are three comments I wish to make here. First, the distinction between the senses in which (1) and (2) are necessary seems to be ill founded. As Russell has taught us, surface grammar is a poor guide to logical form. It is true that a semantically evaluable item whose logical form was “X is X” would be formally necessary in Johnston’s sense. But given the ambiguities of natural language expressions, including proper names, we simply cannot conclude that a sentence whose surface grammatical; form is “X is X” has the corresponding logical form. As a result, we should only speak of certain uses of (1) and (2), as opposed to the sentences themselves, as being in any sense necessary, perhaps uses individuated in terms of the types of propositional act performed. Moreover, the uses of (1) and (2) in question which are candidates for necessity are such on the very same basis: they are uses in which the occurrences of the two name tokens are coreferential. And so if this provides grounds for taking the putative necessity of (certain uses of) (1) to be semantic, it provides equally good grounds for saying the same thing with respect to (2).

            Second, recently (for better or worse) a fair bit of attention has been paid to a distinction that has been drawn within the framework of two-dimensional modal logic between two types of necessity. A semantically evaluable type is 1-necessary (or “deeply” necessary) just in case its tokens are true in every possible context of occurrence. So, for example, the English sentence “I am here now” is 1-necessary. And a semantically evaluable token is 2-necessary (or “superficially” necessary) just in case its propositional content (in its context of occurrence) is true in all possible circumstances of evaluation (at least those in which the relevant referents exist). For example, the statement “I am Mary”, made in a context in which ‘Mary’ refers to the speaker, is 2-necessary (albeit not 1-necessary) because Mary is self-identical in all possible circumstances (in which she exists). Moreover, one can infer the 2-necesity of a statement from the 1-neccessity of the stated sentence only if the sentence is unambiguous, that is, it has the same (truth conditionally relevant) propositional content in all contexts of utterance. My worry here is that Johnston’s solution to the “necessity puzzle” at best yields a variety of 1-necessity whereas Kripke’s concern regarding (1) and (2) pertained to 2-necessity.

            And third, even given the account of necessity Johnston relies on, as well as restricted attention to only those occurrences of (1) in which the relevant type of propositional act occurs, it is far from obvious that (1) proves to be necessary. Kripke drew an important distinction between a theory of meaning and a theory of referring: the former concerns the analysis of the truth conditions (or, more neutrally, the propositional content) of a dated utterance; the latter concerns the determination of the truth conditions of a sentence type on its various occasions of use. The claim that proper names are rigid designators—that a name has the same referent in all possible worlds—(as well the claim that they are directly referential) is a claim belonging to the theory of meaning: it does not imply that a given name has the same referent in all contexts of utterance. The latter issue is the business of a theory of referring. Kripke’s theory of referring, for example, is the causal-historical theory. Roughly, the referent of a name on an occasion of use, on this view, is the object that played a specified role in an event that initiated a chain of appropriately causally linked events that culminated in the use of the name in question. And given such an account, it is clearly possible for a speaker to refer to distinct objects by means of her use of a single name in different contexts of utterance, even in contexts in which her referential intentions are the same. Consider, for example, possible circumstances—epistemically indistinguishable to our own up to the point at which the identity of Hesperus and Phosphorus was discovered—in which it turned out that ‘Hesperus’ and ‘Phosphorus’ denoted distinct celestial bodies. Now one could simply insist that these would count as contexts in which the speaker was performing distinct types of propositional acts, but this would be a typing scheme that is insensitive to speakers’ intentions and their causal-historical embeddedness in the worlds in which they perform their speech acts. However, unless Johnston simply grants himself some such typing scheme, (1) will simply not prove to be necessary on his account.

                                          Finally, Johnston argues that an analysis of propositional content in terms of propositional acts is to be preferred to the propositionalist account on the grounds of ontological parsimony. After all, propositions are postulated theoretical entities, whereas propositional acts are parts of utterances and hence entities whose existence we need to grant in any event. I worry, however, that given Johnston’s account of propositional attitudes, this is not an advantage he is entitled to claim. In his view, a propositional attitude, such as belief, is a relation between a subject and a propositional act type. And he maintains that he maintains that “someone may have a belief even though an appropriate propositional act type has never been instantiated, either by the believer or by anyone else.”[3] But uninstantiated types are not entities whose existence is uncontroversial: they like propositions are (abstract) theoretical entities. Moreover, the claim that they are in part constituted by elements that exist independently of any particular propositional act token—the denotations of the referential, predicative, and time-designative acts, for example—is of no help here.[4]  This fact no more vindicates uninstantiated types than do (similar) facts about constituents vindicates Russellian propositions.

 

Peter Alward

Department of Philosophy

University of Lethbridge

peter.alward@uleth.ca



[1] The idea that a sentence determines a single propositional act type seems strangely at odds Johnston’s prior discussion of the flexibility of his view. It might be better to suppose that a sentence is necessary when all propositional act types expressible using it are necessary.

[2] It is not formally necessary because sentences of the form “X is Y” can in general be false in many contexts of utterance.

[3] P. 8.

[4] It is far from obvious that the referent of a referential act, e.g., is a constituent of that act. It might be argued that the victim of an act of shooting is not a constituent of that act, but only of an event causally downstream. Similarly, if one endorses a causal theory of reference, one might take the referent of a referential act to be a constituent not of the referential act itself, but of an event causally upstream which culminated in the referential act in question.