2
The Untranslatatiliby of A-Sentences by Tenseless Date-Sentences
2.1 The Date Theory of A-Sentences
There are two main types of B-sentences, date-sentences and token-reflexive sentences. An example of the former is “John (is) born earlier than May 1, 1960,” and of the latter, “John (is) born earlier than this utterance.” The distinction between these two types provides a means for classifying most of the proponents of the old or new versions of the semantic tenseless theory of time. Some of these proponents hold the date-theory of A-sentences, that tokens of A-sentences are translated (in some sense) by tenseless date-sentences, or at least have truth conditions that are completely statable by these date-sentences. Other proponents hold the token-reflexive theory of A-sentences, that these sentences are translated (in some sense) by token-reflexive tenseless sentences, or at least have truth conditions that are completely statable by these token-reflexive sentences. In this chapter, I shall argue against the date-theory of A-sentences, in chapter 3, against the token-reflexive theory; in chapter 4, 1 shall argue against the other less frequently espoused versions of the semantic tenseless theory and develop a positive theory of A-sentences that conforms to the tensed theory of time. This will prepare the way for my defense of the presentist theory that every sentence, including every tenseless sentence, ascribes a presentness-involving property (see part II).
The following quotations convey the flavor of the date-theory of A-sentences:
“Fido is [was, will be] running” as said at a given time, say t0, expresses the very same proposition expressed by “Fido (is) running at [before, after] t0.” (Milton Fisk)
Take any token of the form
(i) Event E is occurring now.
Translate it, or if you prefer, the proposition which it expresses, by
(ii) Event E occurs (tenseless) at (time) theta.
Here the proper name “theta” is what we might call a “non-descriptively individuating proper name” of the time in question, that is, the time at which the token to be translated was produced. (Paul Fitzgerald)
A certain “ran” is translated by any “runs (tenseless) on Jan. 7, 1948 at noon E.S.T.” (Nelson Goodman)
When I say
(4) 1 was insulted yesterday,
a specific content—what I said—is expressed. Your utterance of the same sentence, or mine on another day, would not express the same content. It is important to note that not just the truth value may change; what is said is itself different. Speaking today, my utterance of (4) will have a content roughly equivalent to that which
(5) David Kaplan is insulted on April 20, 1973
would have when spoken by you or anyone at anytime. (David Kaplan)
Only at twelve noon can someone think the thought consisting of noon and the sense of “The meeting starts at ( )“ by entertaining the sense of “the meeting starts now.” (John Perry)
“Henry Jones of Lee St., Tulsa, is ill” ... uttered as a tensed sentence on July 28, 1940, corresponds to the statement “Henry Jones of Lee St., Tulsa, is [tenseless] ill on July 28, 1940.” (W.V.O. Quine)
When we are told “Mrs. Brown is not at home,” we know the time at which this is said, and therefore we know what is meant. But in order to express explicitly the whole of what is meant, it is necessary to add the date, and then the statement is no longer “variable,” but always true or always false. (Bertrand Russell)
The proposition expressed at any given time, t,,, through the use of a sentence in which “now” is used is the same as the proposition expressed at t through the use of a sentence formed by replacing the “now” in the first sentence with any date- expression that is used to refer to t,,. (Clifford Williams)[1]
Other philosophers who subscribe to some version of the date-sentence theory
include Joseph Almog, Michelle Beer, Tyler Burge, Cohen, Nagel, Lynne Rudder
Baker, Frege, William Lycan, Jerrold Katz, Nelson Pike, Mark Richards, Jan David
Wald.
The date theory of A-sentences is either an indirect reference theory or a direct reference theory, I will explain the versions of these theories that postulate de dicto and de re date-propositions, respectively, as the semantic content of A-sentence- tokens and will at the end of these explanations say something about the nominalist versions of the date theory, such as Quine’s.
The indirect reference theory holds that A-words (the tensed copulae and verbs and the temporal adverbs, adjectives, and pronouns in A-sentences) refer to times indirectly via definite descriptions of them. These definite descriptions are not, of course, “definite descriptions” in the sense of linguistic phrases, since these phrases are not parts of the relevant A-sentences; rather, they are “definite descriptions” in the sense of components of the propositions expressed by the A-sentences. But strictly speaking, the A-words express these propositional definite descriptions not by themselves but only with the help of their contextual features. Frege makes this point about A-sentences and other sentences containing indexicals: “The mere wording, as it is given in writing, is not the complete expression of the thought, but the knowledge of certain accompanying conditions of utterance, which are used as a means of expressing the thought are needed for its correct apprehension.”[2] The “time of utterance,” Frege adds, is the relevant condition that helps to express the descriptive sense of the A-words: “If a time indication is needed by the present tense one must know when the sentence was uttered to apprehend the thought correctly. Therefore the time of utterance is part of the expression of the thought.”[3]
In order to make this theory more precise, it will be helpful to provide definitions of times and dates and delimit the nature of the corresponding propositional descriptions. First of all, let us give the two standard definitions of times, the definitions used in the reductionist and substantival theories. The reductionist theory is often called the “relational” theory, and the substantival theory, the “absolutist” theory; but since “relational versus absolutist theories of time” can mean many different things (John Mackie distinguishes fourteen different senses),[4] it is best to use different terminology to avoid ambiguity. The reductionist theory of time defines a time as a set of simultaneous events. W. H. Newton-Smith defines the “reductionist” theory’s conception of a moment: “We can form the collection of all events simultaneous with any particular event used in identifying the moment. This collection, the reductionist claims, just is the moment. To say that such and such an event occurred at such and such a moment is just to say that the event in question is a member of the set of events constituting the moment.”[5] Russell espouses a reductionist theory of time in his later writings and defines “instant” in a related way: “An ‘instant’, as I propose to define the term, is a class of events having the following two properties: (1) all the events in the class overlap; (2) no event outside the class overlaps with every member of the class.”[6] Examples of reductionist theories of time beside Russell’s are Hans Reichenbach’s, Adolf Grunbaum’s, Mario Bunge’s and Ian Hinckfuss’s.[7]
The substantival theory, on the other hand, defines a time as a particular item that is logically independent of the events (if any) that stand to it in the relation of occupying it. Let us depart from Newton-Smith’s terminological usage and reserve the word moment for a time in the substantival sense. For any given moment M, if M is occupied by some simultaneous events, there is a set S of all these occupying events that is both distinct from M and accidentally related to M. If we suppose that in the actual world, M is occupied by the members of S, there will be some merely possible world in which M is occupied by the members of some other set S* of simultaneous events, and there will be some still other world (containing “empty time”) in which M is not occupied by any events at all (or at least not by any “events” in the narrow sense that involves physical or mental changes or states). Newton-Smith calls these substantival moments “temporal items” and says that (what I am calling) the substantivalist thesis is that “the existence of temporal items is ontologically independent of the existence of things in time. Temporal relations between things in time hold in virtue of temporal relations holding between the times at which the things in time occur.”[8] Examples of substantivalist theories are Isaac Newton’s, Richard Swinburne’s, Sydney Shoemaker’s, and the theory I defended in The Felt Meanings of the World: A Metaphysics of Feeling.[9]
There are some nonstandard definitions of times, such as A. N. Prior’s definition of a time as a conjunction of true propositions[10]; but these nonstandard definitions will make no material difference to my argument, and I shall not discuss them further.
If a time is either a set of simultaneous events or a moment, then a date is definable as a set of simultaneous events (or moment) that stands in a B-relation to the set of simultaneous events (or moment) at which Christ’s birth is (conventionally said to be) located. (Obviously, the zero-point of a dating system can be located at any event; but I am here defining “date” in terms of the system we conventionally use.) On the reductionist theory, an example of a date is the set So of simultaneous events that bears to the set S1 of simultaneous events that includes Christ’s birth the relation of being 1,985 years, 56 days, 10 hours, and 3 minutes E.S.T’. later than it. Some philosophers think it nonsensical (or at least false) to talk about sets as sustaining temporal relations (“sets are abstract objects, and abstract objects by definition are timeless”); but the idea that sets of simultaneous events do not exist simultaneously with their members seems counterintuitive, and I agree with David Lewis that “a set of [spatiotemporally] located things does seem to have a [spatiotemporal] location: it is where [ when] its members are.”[11] But this is not a mere matter of stipulative definition or intuitive seemings; I will argue in chapter 6 that each object, concrete or abstract, possesses temporal relations and properties.
The substantival theory of times represents dates as moments that exemplify date-properties; for example, the moment M is a date, since it has the property of being 1,985 years, 56 days, 10 hours, and 3 minutes E.S.T. later than the moment M which is occupied by Christ’s birth. For brevity’s sake, I will sometimes talk of a date as a set or moment that bears such a B-relation to Christ’s birth.
Fortified with these definitions of times and dates, we can give further precision to the indirect reference date-theory of A-sentences. The theory is that A-expressions express propositional definite description of dates not by themselves but in conjunction with the date at which the A-expressions are tokened. Each A-expres sion-token has the property of the form occurring at D, where D stands for a date. It is this property that is the accompanying condition that (along with the A-expression token) expresses the propositional definite description of D. The relevant A-expression-tokens also introduce additional senses into the proposition expressed; for example, a token of “yesterday” introduces, in addition to the propositional description of the date D at which it occurs, the sense whatever day is one day earlier than (D). Thus, if I utter “Yesterday was cloudy” this utterance has the property, say, of occurring at the daylong moment (or set of events) that is 1,988 years and 90 days later than Christ’s birth; and this property, in conjunction with the utterance of “Yesterday,” expresses the sense whatever daylong moment or (set of events) is one day earlier than the daylong moment (or set of events) that is 1,988 years and 90 days later than Christ’s birth.
Some defenders of the new tenseless theory of time might wish to say that A-sentence-tokens and their date-properties do not express such senses but that B-sentences expressing them suffice to give the truth conditions of the A-sentence- tokens. I shall show that this is not the case in the course of my argument that these senses are not expressed by A-tokens (see section 2.2).
But before I begin these arguments, I want to explain the direct reference date theory of A-sentences, which states that A-expression-tokens, do not express propo sitional definite descriptions of dates but refer directly to the dates. These tokens introduce the dates themselves, rather than a descriptive sense into the proposition expressed by the A-sentence-token. The de re proposition expressed by a noon occurrence of “The meeting starts now” includes the meeting itself (supposing that the incomplete definite description “The meeting” is used referentially, rather than attributively), the property of starting and the set of events or moment that is identical with noon. This proposition contains all and only the members of the set
(S) [The meeting, the property of starting, noon]
where “The meeting” and “noon” are used as directly referring terms.
But as this example arguably suggests, the direct reference date-theory is too rigid as it stands to be even prima facie plausible, since the claim that A-expression- tokens introduce no senses at all into the proposition but merely the date is overtly inconsistent with the syntactic features of certain A-locutions. I am not here making a criticism of the direct reference theory from the point of view of the tensed theory of time (as I shall later) but an internal criticism that presupposes only the tenseless theory. Consider the difference between a pronomial and adverbial use of “now,” the first being illustrated by “Now is when the meeting starts,” the second, by “The meeting starts now.” The direct reference date theory runs into no internal problems with the former sentence, which it translates as “Noon is when the meeting starts” for a noon utterance of this sentence, with “noon” referring directly to noon. But incoherence appears if this interpretation is applied to a noon utterance of “The meeting starts now,” for this would translate as the syntactically incomplete sentence “The meeting starts noon.” A syntactically complete version of this sentence would read “The meeting starts at noon”; but the translation of “now” by “at noon” is forbidden by the direct reference principle that “now” merely refers directly to times and does not have any predicative sense, such as the relational sense at (simultaneously with).
It is instructive to consider how some of the direct reference date-theorists fudge or gloss over this issue. John Perry gives this account of the aforementioned sentence where “now” is used adverbially: “Only at twelve noon can someone think the thought [proposition] consisting of noon and the sense of ‘The meeting starts at ()‘ by entertaining the sense of ‘the meeting starts now’.“[12] Notice that Perry introduces the sense expressed by “at” into the analysans of the proposition expressed at noon by “the meeting starts now,” even though his next premise forbids its introduction: “The demonstrative [“now”] in context gives us the [object, noon itself], the rest of the sentence the [incomplete sense].“[13] If this premise is true, then the demonstrative cannot contribute a part of the incomplete sense (that expressed by “at”) and the analysans should consist only of the rest of the sentence (“The meeting starts”) and a word that merely directly refers to the object (“noon”).
A defender of the direct reference date theory might respond that the sense of “at” is contributed by the verb “starts,” which means “starts at.” But this is plainly false, since if “starts” means “starts at,” then these two expressions are intersubstitutable salvu sensu, which they are not. “Now is when the meeting starts” is so far from being synonymous with “Now is when the meeting starts at” that the former alone is syntactically well formed. The “direct reference” theorist might then respond that verbs like “start” are systematically ambiguous, having different senses when they are used with demonstrative adverbs than when not so used. When they are so used, they acquire an additional sense that relates the rest of their sense to the direct referent of the demonstrative, so that “starts” now expresses what is expressed by “starts at.” But this response is implausible and ad hoc. “Starts” is used in the same way, as a verb, in “The meeting starts now” and “Now is when the meeting starts”; and the relevant difference is in the change from a pronomial to an adverbial use of “now.” Since the syntactic change involves “now,” not “starts,” it is intuitively more credible to suppose that the semantic change involves “now” than to suppose it involves the verb. Indeed, this change is precisely what we would expect in a change from a pronomial to an adverbial use; for a pronoun is merely a referring device, and an adverb is a qualfying expression, something that characterizes what is expressed by a verb. As a qualifying expression, the adverb expresses a property of the semantic content of the verb. It is natural to think that the adverbial use of “now” expresses the relational property being simultaneous with noon, which includes noon itself as a constituent, and that the function of this adverb is to convey that this property is possessed by the meeting’s start. We may still allow, consistently with the direct reference theory, that this adverb refers to noon directly, rather than through a propositional definite description; but we deny that this reference is its sole semantic property, since the adverb, in addition, ascribes to the meeting’s start a property partly composed of its direct referent, being simultaneous with noon.
This internal modification of the direct reference date theory seems to me to be desirable, and I will assume it in what follows.[14] It prejudges nothing against the tenseless theory of time; but it will help prepare the way for my introduction (in chapter 4) of additional senses into A-words like “now,” namely, A-senses.
I have explained the date theory of A-sentences in terms of the propositionalist theories; but the nominalist version deserve a word, as well. According to nominalist date theorists, such as Quine, we do not need to postulate de dicto or de re propositions as semantic contents of the A-sentence-tokens but can explain this content solely in terms of the eternal B-sentences that correspond to these tokens:
But instead of appealing here to propositions, or meanings of eternal sentences, there is no evident reason not to appeal simply to the eternal sentences themselves as truth vehicles. If we undertake to specify the proposition “expressed” by the utterance of some noneternal sentence, e.g .,“The door is open,” on some particular occasion, we do so by bracketting some eternal sentence that means the proposition; thus we have had to compose an appropriate eternal sentence anyway, and we could as well stop there.[15]
The eternal sentences corresponding to A-sentence-tokens include linguistic definite descriptions that are “explicit references to dates or periods.”[16] These references are “left tacit”[17] in the use of A-sentences. This nominalist theory is mistaken if it can be shown that uses of such sentences as “The door is open” do not tacitly refer to dates, which I shall do in section 2.5, using Quine’s own liberal criteria for “translation” (or “paraphrase,” as he prefers to call it).
The rest of the present chapter consists of five sections. In the four sections that immediately follow, I argue that tenseless date-sentences, however conceived, meet none of my four conditions of translating uses of A-sentences. I begin in section 2.2 by arguing that tenseless date-sentences and A-sentence-tokens have different truth conditions and proceed to argue in section 2.3 that they have different confirmation conditions, in section 2.4 that they are not logically equivalent, and in section 3.5 that they are not logically identical. In section 2.5 I shall concede that they are intersubstitutable salva veritate in extensional contexts but note that this is insufficient for them to express the same proposition. I do not believe that any single one of my arguments is sufficient to refute the tenseless date-sentence theory (there is no single knockdown argument). But the cumulative effect of the many arguments 1 shall present suffices, in my judgment, to refute the tenseless theory. This is the reason that I attack the tenseless theory at many different points, rather than concentrate on a single problem. If my tensed theory is to be completely refuted and the tenseless theory to be completely justified, then each of the many arguments offered in the following sections must be shown to be unsound.
2.2 The Truth Conditions of A-Sentence-Tokens and Tenseless Date-Sentences
Previous attempts to refute the date-sentence theory of A-sentences have concen trated on the date theory developed by proponents of the old tenseless theory of time and have typically consisted of arguments to the effect that A-sentence-types have semantic rules different from those of tenseless date-sentence-types and therefore are untranslatable by them. This is the line of argument developed by Richard Taylor and Richard Gale in his tenser period.[18] However, the arguments of Taylor and Gale have been successfully rebuted by Clifford E. Williams,[19] who pointed out that arguments of this sort miss the point of the date-sentence theory, which does not claim that the date-sentence-type translates (in some sense) an A-sentence-type but that it translates some token of the A-sentence-type. The thesis of the date theory is that each successive token of some A-sentence-type corresponds to a distinct date- sentence-type, such that corresponding to the token of “Henry is ill” that occurs on July 28, 1940 is the date-sentence-type “Henry (is) ill on July 28, 1940” and corresponding to the token of this A-sentence that occurs on July 29 there is “Henry (is) ill on July 29, 1940.” Thus, if the date theory of A-sentences is to be refuted, it must be shown that the relevant semantic features of a given token of an A-sentence- type are different from those of the corresponding date-sentence-type. In this section, I shall show that the truth conditions of a given A-sentence-token are different from those of its corresponding date-sentence-type and that this refutes both the old and new versions of the tenseless date-sentence theory.
Let me begin by pointing out that a token of an A-sentence has token-reflexive truth conditions, whereas no tenseless date-sentence-type and no token of this type has such conditions (except in exceptional cases to be noted). The truth value of an A-sentence-token is dependent upon the B-relation of the token to the event the token is about. A token of a future tensed A-sentence is true if and only if it occurs earlier than the event it is about; a token of a present tensed A-sentence is true if and only if it occurs simultaneously with the event it is about; and a token of a past tensed A-sentence is true if and only if it occurs later than the event it is about. For instance, a token of “Henry was ill” is true if and only if it occurs later than Henry’s illness. By contrast, when a tenseless date-sentence is used is irrelevant to its truth value. For example, it is false that the date-sentence “Henry (is) ill before April 5, 1989,” or any given token of this sentence, is true if and only if it is used later than the event it is about. It is rather the case that the tenseless date-sentence or any one of its tokens is true if and only if the event it is about has the appropriate B-relation to the date the sentence refers to. “Henry (is) ill before April 5, 1989” is true if and only if Henry’s illness is earlier than April 5, 1989 and if this condition obtains, the sentence is true regardless of whether it is used before, after, or simultaneously with Henry’s illness. Of course, some tenseless date-sentences have token-reflexive truth conditions (“This sentence-token occurs later than Henry’s illness and on April 5, 1989” is true if and only if the token occurs later than Henry’s illness and on April 5, 1989); but these are abnormal cases and are not the sorts of translating or truth- condition-stating date-sentences the detensers have in mind.
It is important to make explicitly clear that in this contrast I am specifying only tenseless truth conditions of the tenseless truth values of A-sentence-tokens. It is my contention, for example, that
(1) A token of “Henry was ill” (is) true if and only if this token (is) later than Henry’s illness.
By describing the truth conditions of A-sentence-tokens in a tenseless language and by defining them in terms of the obtaining of certain B-relations between the tokens and the events they are about, I am not begging the question against the date-theorist by assuming from the outset that A-sentence-tokens have truth conditions and truth values that involve A-properties and are statable only in a tensed language. I believe that A-sentence-tokens do have such truth conditions and values, for example, that a token of “Henry was ill” is presently true if and only if Henry’s illness is now past; but my strategy is not to assume this but to prove it by arguing from the differences in tenseless truth conditions and values (and other tenseless semantic properties, discussed in later sections), between A-sentence-tokens and date-sentences. These tenseless differences are my concern in this chapter.
At first glance, this account of A-sentence-tokens and tenseless date-sentences might seem to show that these locutions have different truth conditions and therefore that date-sentences cannot translate A-sentence-tokens. It might also seem to show that the date theory espoused by proponents of the new tenseless theory of time is mistaken, since if A-sentence-tokens have token-reflexive truth conditions and not date-involving truth conditions, then date-sentences cannot give the truth conditions of A-sentence-tokens. But a refutation of the old and new tenseless date theories is not as easy as this, since some of the proponents of these theories have a seemingly powerful argument that A-sentence-tokens do have date-involving truth conditions—in fact, the very same truth conditions as their corresponding tenseless date-sentences. J.J.C. Smart, holds, for example, that “when a person P utters at a time t the sentence ‘Event E is present’ his assertion [utterance] is true if and only if E is at t. More trivially, when P says at t ‘time t is now’ his assertion is true if and only if t is at t so that if P says at t ‘t is now’ his assertion is thereby true.”[20]
If the expression “t” is replaced by a particular date-description and “P” by a person’s name, we can obtain truth conditions of particular utterances of A-sen tences. Using the example of “Henry is ill,” we may say
(2) “Henry is ill” is true as spoken by John on July 28, 1940 if and only if Henry (is) ill on July 28, 1940.
The truth condition clause following the biconditional is exactly the same as the clause following the biconditional in the statement of the truth conditions of the corresponding tenseless date-sentence:
(3) “Henry (is) ill on July 28, 1940” is true if and only if Henry (is) ill on July 28, 1940.
Proponents of the old tenseless theory of time will take these claims as evidence that tenseless date-sentences translate the A-sentence-tokens, and defenders of the new tenseless theory (such as Smart) will take them as showing that the date-sentences suffice to give the truth conditions and thereby the meaning of the A-sentence- tokens, from which it follows that these tokens convey no information about time not conveyed by the date-sentences.
I believe, contrary to these allegations, that A-sentence-tokens do not have these date-involving truth conditions. If we call the utterance mentioned in (2) the utterance U, then we may say that (2) states that a necessary condition of U’s truth is that Henry be ill on July 28, 1940. In possible-worlds terminology, this means that U is not true in any world in which Henry is not ill on July 28, 1940. But this is not the case. In the actual world, U occurs at a time that possesses the date-property of being 1,939 years, 6 months, and 27 days later than Christ’s birth. Let us first suppose that the reductionist theory of times is true and that this time is a set of simultaneous events, two of which are U and Henry’s illness. This set of events does not possess the aforementioned date-property in each possible world in which it exists. In one of these worlds, W1 Christ was not born at all; and in another world, W2 Christ was born 1,938 years earlier than the set of events that contains Henry’s illness. In both of these worlds, U is true, since U is simultaneous with Henry’s illness. But “Henry (is) ill on July 28, 1940” is false in W1 and W2 since, in these worlds, Henry is not ill at whatever set of events has the property of being 1,939 years, 6 months, and 27 days later than Christ’s birth.
The same result follows if we assume the substantival theory of times. In the actual world, U and Henry’s illness occupy the moment that has the date-property of being 1,939 years, 6 months, and 27 days later than Christ’s birth. But in W1 and W2 the moment occupied by U and Henry’s illness does not possess this date property; consequently, the date-sentence—but not U—is false in these worlds.
It is conceivable that a defender of the direct reference date theory would object that my account presupposes an indirect reference theory of A-sentence-tokens and tenseless date-sentences. My remarks presuppose the view that the reference to the date is mediated by a propositional definite description such as that expressed by an attributive use of “the time that is 1,939 years, 6 months, and 27 days later than Christ’s birth.” Since this definite description refers to whatever time possesses this property, it refers to different times in different worlds, since in different worlds different times possess this property. In the actual world, set S or the moment M possess it; but in some other world, the set S* or moment M* possesses this property and thereby the reference of the description shifts to S* or M*. But suppose the direct reference theory is true, such that the date-sentence stating the truth conditions of U, “Henry (is) ill on July 28, 1940,” involves a referential, rather than attributive, use of “July 28, 1940.” In this case, this description directly refers to the set or moment that is, in fact, 1,939 years, 6 months, and 27 days later than Christ’s birth and refers to this same set or moment in each possible world in which it exists, regardless of whether or not it possesses this date-property there. If we use “S” (“M”) as a directly referential name of this set (moment), we may say that the truth conditions of U are that it is true if and only if Henry (is) ill at S (M).
This direct reference theory provides an adequate response to my scenario involving W1 and W2 that assumed the reductionist theory of time, for in each of the worlds in which the set S that actually contains U exists, U is true regardless of whether or not S possesses its actual date-property; for the set S that actually contains U also actually contains Henry’s illness and therefore, since sets contain their members essentially, essentially contains both events, rendering U true in each world in which this set exists. Consequently, “Henry (is) ill at S,” like U, is true in W1 and W2.
The direct reference theory’s success is limited, however, since there are worlds in which U is true but the referentially used date-description is not. There is a world W3 in which S does not exist but in which U occupies a set S1 that contains all and only the events that S contains except that S1 contains a certain dustlike particle on the planet Venus hitting the ground, whereas S does not. In this world, U is true, since it is simultaneous with Henry’s illness, but “Henry (is) ill at 5” is false, since Henry is ill at S1 instead.
If we adopt the substantival theory, the direct reference theory will also fail to give necessary conditions for the truth of U. There is a world W4 in which U and Henry’s illness and all the other events in S (or S1) occupy M*, rather than M; consequently, U is true, but “Henry (is) ill at M” is false.
I conclude that it is false that U is true if and only if Henry (is) ill on July 28, 1940, regardless of whether this date-description be read attributively or referentially. Tenseless date-sentences do not state the truth conditions of U and other A-sentence-tokens; consequently, the old and new versions of the tenseless date theory are false.
But the tenseless date-theorists are not without response. Smart has responded to an earlier and abbreviated version of my criticism by changing his date theory so that it gives world-indexed truth conditions, rather than truth conditions simpliciter.[21] Smart now holds that it is not the case that “B is present” as spoken by P at t is true if and only if E is at t but that
(4) “E is present” as spoken by P at tin world W is true in W if and only if E is at t in W.
I agree with Smart that (4) is true, but this fails to save the date-sentence theory, since the fact that the non-world-indexed version of (4) is false suffices to refute the theory that date-sentences give the meaning of A-sentence-tokens. World-indexed truth condition sentences do not give the semantic content of utterances, since the criterion of truth for these sentences is that the clause on the right side of the biconditional that has the world-indexed phrase appended to it has the same truth value in the mentioned world as the utterance denoted by the clause on the left side of the biconditional. This allows any clause with the same truth value in W as “E is at t” to be substituted for “E is at t” and this prevents (4) from giving the meaning of the utterance. Consider a concrete example,
(5) “Henry is ill” as spoken by John on July 28, 1940 in W is true in W if and only if Henry (is) ill on July 28, 1940 in W.
Suppose that W is the actual world and that the utterance mentioned is true. Then, we may substitute any true clause for “Henry (is) ill on July 28, 1940” and (5) will remain true. Thus, it is true that
(6) “Henry is ill” as spoken by John on July 28, 1940 in W is true in W if and only if the sun is ninety-three million miles from the earth in W.
If (5) gives the meaning of the utterance of “Henry is ill” by stating with truth world-indexed truth conditions of this utterance, then (6) would give the meaning of this utterance for the same reason. But, obviously, “Henry is ill” as used by John does not mean that the sun is ninety-three million miles from the earth. A truth condition sentence that gives this semantic content (or at least gives it up to logical equivalence) must state conditions that obtain in all and only the worlds in which this utterance is true; and this can only be done in terms of truth conditions that are not world-indexed.
The world-indexing of the truth conditions runs into the same problem as does an extensional reading of the biconditional in truth condition sentences that are not world-indexed. It is worthwhile to point this out, since Smart writes that his preferred response to my criticism is to make the biconditional extensional, rather than to read it as intensional and world-index the truth conditions.[22] But if we make extensional the biconditional in
(2) “Henry is ill” is true as spoken by John on July 28, 1940 if and only if Henry (is) ill on July 28, 1940,
then (2) no longer gives the meaning of the utterance of “Henry is ill.” If (2), as read extensionally, did give the meaning of the utterance of “Henry is ill,” this would be due to the fact that this utterance has the same actual truth value as “Henry (is) ill on July 28, 1940,” since the criterion of truth for such extensional truth condition sentences is that the clause after the biconditional has the same actual truth value as the utterance mentioned before the biconditional. But since the utterance of “Henry is ill” also has the same actual truth value as “The sun is ninety-three million miles from the earth,” the conclusion that this utterance meant that the sun was ninety- three million miles from the earth would be equally warranted. Only a stricter criterion of truth can make truth condition sentences meaning-giving, namely, the criterion that the clause mentioned after the biconditional is true in all and only the possible worlds in which the mentioned utterance is true. D. H. Mellor makes this point in an intuitively clear manner in Real Time:
For a sentence’s truth conditions to give its meaning, its being true in them must be more than a coincidence. Otherwise, so far as truth conditions go, the English sentence “Snow is white” could just as well mean that grass is green, since “snow is white” is true and grass is green. “Snow is white” is indeed true if and only if grass is green [if the biconditional is read extensionally] But that of course, is just a coincidence. Even if grass were not green, “Snow is white” would still be true— provided snow was still white. “Snow is white” is not only true in the real world, it would also be true in any other world in which snow was white, and false in any world in which it wasn’t. That is really why the sentence means in English what it does, rather than meaning that grass is green. To give meanings, therefore, truth conditions generally have to include imaginary conditions as well as real ones.[23]
Of course some alternate versions of extensional semantics have been developed (e.g., by H. Field, M. Platts, J. McDowell, and the progressively more complicated versions of D. Davidson);[24] but in a well-argued article,[25] Nicholas Asher has shown these versions succumb either to the problem I ascribed to the extensional reading of (2) or to some equally fatal problem. (Further criticisms of extensional semantics can also be found in chapter 5 of Stephen Schiffer’s Remnants of Meaning.)[26]
Oaklander’s objection to an argument I presented in an earlier article[27] may be reduced to Smart’s response and thereby rejected for the same reason. Referring to my example of a true 1814 utterance U of “The Battle of Waterloo is present,” Oaklander concedes that my example of a possible world in which U is simultaneous with this battle but does not occur in 1814 (where “1814” is read referentially) shows that the truth condition sentence “U is true if and only if the Battle of Waterloo occurs in 1814” is false. In this world (call it W*), U is a member of a different set of simultaneous events, the set directly referred to by “1814*” and in this world, the date-involving truth conditions of U are given (according to Oaklander) by a sentence stating that the token U “of ‘The Battle of Waterloo is present’ uttered in 1814* will be true (if’ the Battle of Waterloo occurs in 1814*,”[28] But this statement is either false or must be read as an extensional or world-indexed truth condition sentence, in which case, it is semantically inadequate for the same reason as Smart’s response. This statement is false if “if” is read as an intensional, rather than extensional, biconditional and the statement is not a world-indexing one, since there is some other world W’ in which U is simultaneous with the Battle of Waterloo (and thus is true) but is not a member of the set 1814*. But if Oaklander’s truth condition statement is true, then the “iff” is extensional or the statement is implicitly world-indexing (so that “in W*” belongs after each occurrence of “in 1814*”); and in either case, the statement runs into the same problems as Smart’s modified theory.
Oaklander offers a second argument, but it runs into the same problems. If we call the set of events directly designated by “1814” the set B, such that U is actually a member of B but is not a member of B in some other possible worlds, then it follows, according to Oaklander, that “if U does not occur in B, then U cannot be a true 1814 token of ‘The Battle of Waterloo is present’. Thus, . . . Smith cannot claim that (an 1814 token) U is true even if the Battle of Waterloo does not occur in 1814.”[29] Oaklander’s remarks may be understood to be implicitly world-indexed, as meaning, “If U does not occur in B in then U does not possess in W* both the properties of being true and occurring in 1814. Thus, Smith cannot claim that in W*, U is both true and occurs in 1814 if the Battle of Waterloo does not occur in 1814 in W’ Understood this way, Oaklander’s remarks are true but irrelevant, since the claims that I in fact made were not about the world-indexed truth conditions of U but about its non-world-indexed and nonextensional truth conditions; and these claims are true, since U is true in worlds in which the Battle of Waterloo does not occur in 1814. Oaklander’s remarks are also true if they are interpreted as claims about the extensional truth conditions of U; but this is again irrelevant, since these claims are consistent with my claims about the intensional and non-world-indexed truth conditions of U. Consequently, Oaklander, like Smart, fails to resolve the vital problem with the date-sentence theory of A-sentences, that date-sentences fail to express (contra the old tenseless theory) or give (contra the new tenseless theory) the semantic content (meaning) of A-sentence-tokens, since they do not state the intensional and non-world-indexed truth conditions of A-sentence-tokens. The intensional and non-world-indexed truth conditions of these tokens, stated in tenseless terms, are instead token-reflexive truth conditions of the form “T (is) related by B to E,” where T stands for the A-sentence-token; B, for a B-relation of simultaneity, earlier, or later; and E for the event that the sentence-token is about. In chapter 4 I shall show that the fact that A-sentence-tokens have tenseless token-reflexive truth conditions, as tenseless date-sentences do not, and the fact that tenseless date- sentences have date-involving truth conditions, as A-sentence-tokens do not, is ultimately explained only by the hypothesis that A-sentence-tokens, but not tense- less date-sentences, ascribe A-properties.
But we are still a long way from being able to argue for the soundness of this explanation, since there are other detenser arguments pertaining to other semantic properties of A-sentence-tokens that might seem to show that these sentence-tokens, despite the aforementioned differences from tenseless date-sentences, have the same semantic contents as the latter. These arguments will be considered in the following three sections.
2.3 The Confirmation Conditions of A-Sentence-Tokens and Tenseless Date-Sentences
Two sentences or sentence-tokens have the same confirmation conditions if and only if they are confirmed or disconfirmed to the same degree by observations of the same sort. That A-sentence-tokens have confirmation conditions different from their corresponding date-sentences follows immediately from their difference in truth conditions. Consider “Henry Jones of Lee St., Tulsa is ill” as tokened on some occasion. An observation that confirms this token as true is an observation that Henry’s illness occurs simultaneously with the token. However, no observation of a similar sort could confirm any token of “Henry Jones of Lee St., Tulsa (is) ill on July 28, 1940.” To observe that Henry’s illness is simultaneous with some token of this sentence neither confirms nor disconfirms this token; for suppose that some token of this sentence is observed to be simultaneous with Henry’s illness but that the date of Henry’s illness is unknown; we would then not be in possession of enough information to determine the truth value of the token. Generalizing, we can say that the first difference in confirmation conditions between A-sentence-tokens and date-sentences and their tokens is that observations of B-relations between the event reported and the reporting sentence-token are sufficient to confirm the A-sentence-tokens but insufficient to confirm the date-sentence-tokens.
The second difference is that observations of the B-relation between the event reported and the date are sufficient to confirm the date-sentence but are insufficient to confirm the A-sentence-token. Each token of “Hemy Jones of Lee St., Tulsa (is) ill on July 28, 1940” is confirmed by the observation that Henry’s illness occurs on July 28, 1940; but no such observation confirms the A-sentence-token. For this observation to count toward the confirmation of the A-token, I must possess, in addition, observational evidence that the A-token occurs on July 28, 1940, that is, that the A-token is simultaneous with Henry’s illness. And this is tantamount to saying that the observational evidence that Henry’s illness occurs on July 28, 1940 is not by itself sufficient to confirm the A-sentence-token.
The difference in confirmation conditions between an A-sentence-token and its corresponding tenseless date-sentence become especially acute in the case of A-date-sentence-tokens and their corresponding tenseless date-sentences. An exam ple of an A-date-sentence is “It is noon” or “Today is February 5.” The confirmation conditions of these A-date-sentences and their alleged tenseless translations will be my particular focus in this section. I shall first critically examine GrUnbaum’s theory of A-date-sentence-tokens, then William Lycan’s theory, David Kaplan’s and Nathan Salmon’s theories, and finally Michelle Beer’s theory.
A. Adolf Grünbaum’s Theory of A-Date-Sentence-Tokens
Consider a token T1 of “10:30 A.M. E.S.T., February 26, 1986 is present” that occurs on this date. T1 is a logically contingent sentence-token that is confirmed by observations that it is simultaneous with 10:30 (etc.). This point may be obvious, but it is worth driving home. It is manifest that in order to determine whether T1 is true or false it is not sufficient that I simply understand T1. I must also be in possession of observational evidence about T1‘s B-relation to 10:30. For instance, if somebody utters T1, I cannot simply nod my head and say, “Of course that’s true.” I must also look at my watch to see if T1 is borne out by the empirical evidence. But consider now the tenseless date-sentence that corresponds to T1: “10:30 A.M., E.S.T., February 26, 1986 (is) at 10:30 A.M., E.S.T., February 26, 1986.” This sentence is logically necessary; the date described by the first date-expression is necessarily simultaneous with the date described by the second date-expression, since they are one and the same date. Consequently, I know this sentence to be true independently of any observational evidence and simply through understanding it. This is prima facie evidence that the indirect reference date theory is wrong in stating that T1 and the date-sentence express the same proposition.
It is worth making clear that according to the indirect reference date theory, the aforementioned tenseless date-sentence is the one that corresponds to T1. It is a tenet of this theory that the A-predicate “present” conjoined with the A-copula “is” refer indirectly to the date on which they are tokened. If “x is present” is uttered at 10:30, then “is present” expresses the propositional definite description that refers to this date. This entails that in the case of “10:30 (etc.) is present,” the item referred to by the rest of the sentence is a date, which entails that the date is referred to twice, once by the date-expression and once by “is present.” What the token T1 of this sentence asserts, according to the date theory, is that the date referred to by the date- expression is simultaneous with the date referred to by “is present,” that is, that 10:30 (etc.) is simultaneous with 10:30 (etc.). And thus the logic of the indirect reference date theory drives us to the seeming paradox that a logically contingent sentence-token expresses the same proposition as a logically necessary sentence.
A defender of the indirect reference date theory might try to solve this problem by arguing that the A-date-sentences or their tokens are no less logically necessary than their corresponding tenseless date-sentences. Adolf Grunbaum, taking this approach, writes that the tenser
must construe the assertion “It is 3 P.M., E.S.T., now” as claiming non-trivially that when the clock strikes 3 P.M. on the day in question, this clock event and all the events simultaneous with it intrinsically have the unanalyzable property of nowness or presentness. But I am totally at a loss to see that anything non-trivial can possibly be asserted by the claim that at 3 P.M. nowness (presentness) inheres in the events of 3 P.M. For all I am able to discern here is that the events of 3 P.M. are indeed those of 3 P.M. on the day in question.[30]
Grunbaum’s argument is based on a confusion between
(1) It is 3:00 P.M. now
and
(2) At 3:00 P.M. nowness inheres in the events of 3:00 P.M.
Grünbaum believes that (1) and (2) make the same assertion and that since (2) is trivial, (1) is also trivial. But manifestly, “It is 3:00 P.M. now” and “At 3:00 P.M. nowness inheres in the events of 3 P.M.” make different assertions. Tokens of the first sentence are in some cases false, namely, when they occur at times earlier or later than 3:00 P.M. Surely it is an obvious fact about English that if I utter “It is 3:00 P.M. now” at 5:00 P.M., I have uttered something false. If so, “It is 3:00 P.M. flow” cannot be a trivial truth; and surely, it is equally obvious that if I utter “At 3:00 P.M. nowness inheres in the events of 3:00 P.M. ,“ I say something true no matter when I say it, be it 3:00 P.M. or 5:00 P.M. This sentence is no less trivially true than “At 3:00 P.M. pastness inheres in the events of 3:00 P.M.” is trivially false. Consequently, one cannot infer from the triviality of sentences like (2) to the triviality of sentences like (1); and the attempt made by Grunbaum to show that A-date sentences like (1) are trivially true must be counted a failure.
B. William Lycan ‘s Theory
The particular problem we found with the indirect reference theory of A-date- sentences is avoided by the direct reference theory of these sentences, but a new and related problem arises to take its place. The direct reference theory holds that expressions such as “now” refer directly to the date on which they are tokened, without the intermediary of a propositional definite description of this date. This seems to demand that the corresponding tenseless date-sentence also possess a phrase that directly refers to a date and that thereby captures the A-words’ direct reference to the date. This demand was first met in William Lycan’s account of these tenseless date-sentences. Lycan recognizes that what is expressed by an utterance of “It is 4:30 P.M.” at 4:30 P.M., March 1, 1967, in Lincoln, Nebraska is logically contingent; but he argues that what is expressed by this utterance is also expressible by “It is 4:30 P.M., March 1, 1967, in Lincoln, Nebraska at 4:30 P.M., March 1, 1967, in Lincoln, Nebraska,” a sentence that Lycan abbreviates as “It is 4:30 at 4:30.” Lycan continues:
There is a natural way of construing “It is 4:30 at 4:30” according to which the sentence is contingent and non-trivial. . . . Let us construe the second occurrence of “4:30” as a fused referring expression, serving to pick out a particular moment (the moment at which the speaker happened to speak, under whatever designation); the whole sentence may then be read as a singular predication having the second occurrence of “4:30” as its subject. Thus, one who utters the sentence would be predicating something of the moment 4:30 (however referred to), viz., the property of its-having-been-4:30 that moment.[31]
Lycan later indicates that the property of its-having-been-4:30-at that moment is more fully specifiable as some such property as being four-and-one-half hours later than the moment at which the sun reaches its highest point in the sky. He also corrects the impression that might be conveyed by this passage to the effect that the second occurrence of “4:30” refers to a different moment at each different moment it is used, an impression conveyable by his assertion that this expression serves to pick out “the moment at which the speaker happened to speak.” Lycan believes the contrary is true; since “4:30” “in fact denotes 4:30 in our language, . . . in saying ‘It is 4:30 at 4:30,’ . . . we know that we are talking about the moment 4:30, and the statement therein made will not fluctuate from speaker to speaker.”[32] Thus, “4:30” refers to the same time, 4:30, at each different time that “It is 4:30 at 4:30” is uttered. (Recall that “4:30” is an abbreviation for “4:30 P.M., March 1, 1967, in Lincoln, Nebraska.”) Lycan’s remark that “4:30” serves to pick out “the moment at which the speaker happened to speak” must accordingly be interpreted as offering an example of a speaker who happened to speak at 4:30 and, for this reason, referred to the moment of his speaking.
The core of Lycan’s interpretation of “It is 4:30 at 4:30” as logically contingent lies in his remark that the second “4:30” is a “fused referring expression.” Lycan hints that he means by this that this “4:30” is a referentially, rather than attributively, used definite description, in Donellan’s sense.[33] Lycan does not elaborate upon this hint; but his meaning seems plain, namely, that if “4:30” is used referentially, it directly refers to the date, 4:30, and if attributively, it indirectly refers to this date by expressing the propositional definite description, whatever set of events (or moment) is four-and-one-half hours later than the sun’s zenith. If “4:30” is used referentially, then the sentence “It is 4:30 at 4:30” will express the de re proposition consisting of the ordered pair of the date 4:30 and the property being four-and-one-half hours later than the sun’s zenith:
(3) (4:30, the property of being four-and-one-half hours later than the sun’s zenith).
If (3) is the proposition expressed by “It is 4:30 at 4:30,” then this sentence is logically contingent. This is because the date 4:30 (4:30 P.M., March 1, 1967, in Lincoln, Nebraska) is a certain set S1 of events or moment M1 that in fact has the property of being four and one-half hours later than the sun’s zenith (on this day) and does not have this property in some other possible worlds.
This de re translation of “It is 4:30” seems to solve the problem confronting the de dicto translation of A-date-sentence-tokens. But in fact, two new problems emerge. One is that the confirmation conditions of the de re reading of “It is 4:30 at 4:30” are different from those of the token T1 of “It is 4:30.” Observational evidence that confirms the de re reading of “It is 4:30 at 4:30” is evidence that the set of events S1 or moment M1 possesses the property of being four-and-one-half hours later than the sun’s zenith. But this evidence is insufficient to confirm the token T1 of “It is 4:30” that occurs at 4:30. If I did not possess observational evidence that T is simultaneous with 4:30 and only possessed evidence that S1 or M1 is four-and- one-half hours later than the sun’s zenith, I would not be in a position to determine whether T1 was true or false. In order to confirm T1, I must possess, in addition to the observations that confirm “It is 4:30 at 4:30,” observations of a different sort, observations of the B-relation between the sentence-token and the time S1 or M1. (S1 or M1 are picked out or identified observationally in terms of events that belong to them.)
A second problem is that A-date-sentence-tokens whose date-expressions are directly referential are confirmed by certain observations and disconfirmed by others, but their tenseless de re translations are trivially true or false. In other words, a problem similar to the one confronting the indirect reference date theory arises for the direct reference date theory, albeit for a different reason. The problem for the indirect reference date theory concerns A-date-sentence-tokens wherein the date description is used attributively. The direct reference date theory is able to provide logically contingent translations of these A-date-sentence-tokens by arguing that the A-expressions in the A-date-sentence-tokens are translated by tenseless date descriptions used referentially, rather than attributively. But there are other A-date sentence-tokens in which the date-expressions are used referentially. Suppose I utter “It is 4:30” at 4:30 and use “4:30” referentially to pick out the set of events S1, the set that is, in fact, four-and-one-half hours later than the sun’s zenith (on March 1, 1967 in Lincoln, Nebraska). My utterance is not trivially true but needs to be confirmed by the observation that it is in fact a member of S1 Somebody who is not in possession of this information would not be able to evaluate the truth of my utterance. But if the direct reference date theory is correct and my 4:30 utterance of the present tense “It is,” is to be translated by the de re date-description “It is 4:30 at . . .“ (where the copula is, of course, tenseless), then the result is the tautology “It is 4:30 at 4:30.” This is a tautology because both occurrences of “4:30” are referentially used definite descriptions that refer directly to S1 such that this date- sentence is synonymous with “It is S1 at S1.”
Let us highlight this difference in confirmation conditions between these refer entially used A-date-sentence-tokens and their alleged de re translations. Suppose this conversation occurs at S1 at 4:30, and “4:30” is used referentially by both conversationalists:
A: Is it 4:30 yet?
B: Yes, it is 4:30.
A: I see. Thanks for the information.
Clearly, B’s answer does not state what “S1 occurs at S1” states; for A already knew that trivial fact and would hardly thank B for conveying it to her. Rather, she learned some observationally derived information that she was not in possession of before hand, information that (in tenseless terms) S1 is occurring simultaneously with the conversation.
C. David Kaplan’s and Nathan Salmon’s Implicit Theories
But these arguments would not trouble all direct reference date-theorists, since some of them believe they have a ready explanation of these differences in confirma tion conditions. David Kaplan, for example, would acknowledge that a use of a sentence containing both a temporal indexical and a de re date-description imparts a posteriori and observationally verifiable information but would insist that this is consistent with this sentence use expressing the same trivially true proposition as the corresponding tenseless date-sentence. Kaplan would point out that the empirical information supplied by the sentence-token that contains the token of the temporal indexical is not supplied by the semantic content of this token—the proposition it expresses—but by the pertinent semantic character, specifically, by the rule of use that determines the reference of the temporal indexical on each occasion of its use. Using the phrase “cognitive significance” to refer to the informativeness of sentence-tokens, Kaplan writes, “We identify object of thought [the proposition or propositional constituent expressed] with content and the cognitive significance with character.”[34] Consider John’s utterance at noon of “It is now noon.” Kaplan would say that John’s utterance, call it U, expresses the same tautologous de re proposition that is expressed by “Noon (is) at noon” (where “noon” is an abbreviation of some complete de re date-description) but that U has a different “cognitive significance” or information value than the tenseless utterance by virtue of being a token of a sentence that has a different “character” or rule of use. The sentence “It is now noon” is governed by the rule that “It is now noon” refers directly to the time of its use and asserts of it that it (is) simultaneous with noon, and the tenseless sentence is not governed by this rule. By virtue of John and his addressee’s grasping this rule of use, they find U to have a “cognitive significance” and to convey empirically confirmable information that the use of the tenseless sentence does not.
But I believe this theory does not succeed in explaining the cognitive signifi cance of John’s utterance of U. If I grasp the character
(4) The sentence “It is now noon” refers directly to the time of its use and asserts of it that (is) simultaneous with noon,
that does not provide me with the empirical information that John’s utterance U of “It is now noon” gives me, for (4) does not entail that U occurs at noon. Indeed, (4) entails no information about U or about noon, since (4) is a general statement that abstracts from particular events and times. In fact, (4) does not even have a truth value, since it is a conventional stipulation and stipulations are neither true nor false. This remark ties in with a criticism that Hector-Neri Castafleda and Palle Yourgrau have made of Kaplan’s theory of cognitive significance, that his characters are general formulae for determining the semantic contents of expressions and that the cognitive significance of uses of indexical-containing sentences are truth-valued pieces of information about individual times, places, or persons.[35]
But it seems Kaplan’s theory provides the key to constructing a theory of cognitive significance as something other than the semantic content of sentence- tokens, for characters may be taken as proper parts of this significance. There is a distinction made by Nathan Salmon, Howard Wettstein, Michael Tye, Raymond Bradley, Norman Swartz, Tom McKay, and R. M. Sainsbury, between the semantic information of a sentence-token, (corresponding to Kaplan’s semantic content) and the pragmatically imparted information of a sentence-token (a wider notion than Kaplan’s character but which may include his characters as proper parts).[36] Al though these authors have not applied this distinction to A-sentence-tokens that include de re date-descriptions, this application seems relatively straightforward. Let us suppose the cognitive significance of some sentence-tokens may include a character and the subsumption of the sentence-token under the character; that is, the cognitive significance is the information that the utterance is an individual instance of the expression-type mentioned in the statement of the character. If we adopt this theory, we will say that (4), in conjunction with the subsumption
(5) John’s utterance U (is) a correctly used token of “It is now noon,”
provides the cognitive significance of U; for (4) and (5) jointly entail
(6) U pragmatically imparts the information that it (is) simultaneous with
noon.
With this conception of cognitive significance we have, finally, some success for the tenseless date-theory; for (4)-(6) meet a necessary condition of stating the cognitive significance of U, namely, they entail the verifiable information about U’s token- reflexive B-property, being simultaneous with noon. This theory of cognitive significance thus constitutes an advance over the theory of Kaplan and the previously discussed theories of Lycan and Grünbaum. But I would now like to introduce a new consideration that shows that (4)-(6) fail to meet another necessary condition of stating the cognitive significance of U. This consideration can be presented most exactly in terms of the most completely developed of the aforementioned post- Kaplan theories of cognitive significance, the version developed by Nathan Salmon in Frege’s Puzzle.
Salmon holds that A-sentence-tokens directly refer to dates and do not ascribe A-properties,[37] which implies that the token U of “It is now noon” expresses the logically necessary proposition, “Noon (is) at noon.” This is the semantically imparted information of U; but its pragmatically imparted information involves a propositional guise (not to be confused with Castañeda’s propositional guises), which is a “manner of presentation” or “mode of appearance” of the proposition. Salmon writes that this distinction allows that “the semantically encoded information may be knowable a priori even when the sentence’s pragmatic impartations are knowable only a posteriori.”[38] If we apply this theory to our example, we may say that among the pragmatic impartations of “It is now noon” is that this sentence is used simultaneously with noon, an impartation that can be confirmed to correspond to the facts only through empirical observations. If we take this imparted information to be the guise x by means of which the singular proposition noon is at noon is grasped, we may apply Salmon’s schema for belief to our example. The schema[39] is
(7) “A believes p” may be analyzed as (3x) [A grasps p by means of x & BEL (A,p,x)],
where p is the proposition, x the propositional guise, and EEL a triadic belief relation. Applied to our example, (7) may be given the interpretation:
(8) “Alex believes the proposition expressed by John’s utterance U of ‘It is now noon” may be analyzed as follows: Alex grasps the proposition that noon is at noon by means of the propositional guise being the proposition that ascribes self-simultaneity to the time with which U is simultaneous; and Alex stands in the triadic belief relation BEL to that noon is at noon [ that this is the proposition that ascribes self-simultaneity to the time with which U is simultaneous].
Every verb and copula in (8) is tenseless. My argument is that (8) is true only if
(9) Alex believes that noon is at noon [such that this is the proposition that ascribes self-simultaneity to the time with which U is simultaneous] is now true
entails
(10) Alex believes that what John semantically and pragmatically conveyed to him by his noon utterance of “It is now noon” is now true.
But (9) does not entail (10), since it is now 1:30 and Alex now believes (correctly) that noon is at noon [such that this is the proposition that ascribes self-simultaneity to the time with which U is simultaneous] is now true, but he nonetheless now believes (also correctly) that what John’s noon utterance of “It is now noon” con veyed is now false. Of course, Alex now believes that what U conveyed was true when U was uttered; but this is consistent with his believing that what U conveyed, unlike the tenseless-proposition-cum-tenseless-guise is no longer true. If one prefers these facts stated in terms of beliefs in or about U, it may be said that Alex can consistently believe that it is now true that John’s utterance U of “It is now noon” is simultaneous with noon and that noon is at noon without believing that U is now true, which entails that the belief in the simultaneity of noon with itself and with U is not logically equivalent to the belief in U.
If it is objected that truth values of utterances cannot change over time and that the assumption that they do presupposes the tensed theory of time, I would respond by denying both claims. Regarding the latter, change of truth value over time does not presuppose the tensed theory, since these changes can be described in purely tenseless terms and in terms of B-relations. For example, it can be said (with every verb and copula being tenseless),
(11) At 1:30, Alex believes that what U conveys is then (at 1:30) false but is true earlier, at noon
and
(12) At noon, what U conveys is true and at 1:30 what U conveys is false.
“What U conveys” in (11) and (12) means “what U conveys at noon.” Now (11) and (12) mention no A-properties and contain no tenses, so they can be consistently represented as descriptions of some of the tenseless semantic properties of A- sentence-tokens. Furthermore, the fact that the thesis some truth vehicles change their truth value over time does not presuppose the tensed theory of time is further evinced (or suggested) by its endorsement by some proponents of the tenseless theory of time. Both Kaplan[40] and Ernest Sosa[41] hold that propositions do not ascribe A-properties but nonetheless maintain that some propositions change their truth value over time. If it is insisted, nonetheless, that truth-bearers cannot change their truth value, I would point out that this claim is a theoretical postulate and cannot legitimately be said to be a description of the linguistic data that the tenser and detenser are both trying to explain. To use Castañeda’s terminology, (11) and (12) belong to protophilosophy (data description), rather than diaphilosophy (data explanation and theory building), whereas “Truth-bearers cannot change their truth value” is a diaphilosophical statement.[42] If it is denied that (11) and (12) are statements in protophilosophy, I respond that they and their tensed analogues reflect “how we ordinarily talk” or “what we would ordinarily say” and that this can be borne out by observation. For example, we ordinarily say such things as “What John stated when he said ‘It is now noon’ was true when he said it but is now false.” I believe, accordingly, that it is fair to conclude that the Salmon-based theory of the cognitive significance of A-date-sentences, although it meets one necessary condition of explaining their significance (i.e., it explains the information they convey about their B-relatedness to dates), fails to meet the second necessary condition, namely, that is explain their cognitive significance’s change in truth value over time.
However, I do not want to leave the impression that my argument for the tensed theory of time crucially requires the premise that truth vehicles can change their truth value over time, since my argument could then be too easily dismissed by the die-hard supporters of unchangeable truth values. Accordingly, I shall provide, against the detenser theory I constructed in terms of Salmon’s belief schema, an argument that uses premises logically independent of the premise that U changes its truth value. I shall show that this theory fails to meet a third necessary condition of stating the cognitive significance of U, namely, that it explain the logical equiva lence of the cognitive significance of U with the cognitive significance of a simul taneous token of a synonymous sentence. If a person A believes an utterance U at noon of “It is now noon” and also believes an utterance V of the synonymous sentence “It is noon now” that occurs simultaneously, then she has the third (perhaps implicit) belief that the objects of her first two beliefs are logically equivalent. Relative to the theory we are considering, “beliefs” designates the three-termed relation BEL that Salmon introduces. But if the Salmon-based theory is true, the object of A’s belief regarding the utterance U is the proposition-cum-propositional guise
(13) that noon is at noon [such that this is the proposition that ascribes self- simultaneity to the time with which U is simultaneous],
and the object of A’s belief regarding V is
(14) that noon is at noon [such that this is the proposition that ascribes self- simultaneity to the time with which V is simultaneous].
But (13) and (14) are not logically equivalent, since the fact that the proposition that noon is at noon has the property of being about the time with which U is simul taneous does not entail that it has the property of being about the time with which V is simultaneous (somebody could have uttered U without anybody uttering V). Therefore, this Salmon-based theory fails to meet this third necessary condition of adequacy that a theory of cognitive significance must meet, namely, that the cognitive significances of simultaneous utterances of synonymous A-sentences in exactly the same circumstances be logically equivalent. Further criticisms of this Salmon- token-reflexive based theory of the nature of the cognitive significance of A- sentence-tokens are implicit in the criticisms I shall put forth in chapter 3, where I discuss the token-reflexive theory of A-sentences.
D. Michelle Beer’s Theory
I shall critically examine other detenser theories of cognitive significance and of the semantic/pragmatic distinction in the following two sections, but I should like to close this section by saying a few words about a detenser theory of A-date-sentences that differs from all of those considered so far. I have in mind the theory of A-date sentences that belongs to Michelle Beer’s new tenseless theory of time.
Beer argues that A-sentence-tokens do not ascribe A-properties (conceived as irreducible to B-relations), even though these tokens express propositions different from those of B-sentences. Thus, unlike all the date-theorists considered so far in this section, she admits the thesis for which I have been arguing, that an utterance of an A-date-sentence expresses a proposition different from that of the corresponding tenseless date-sentence. For example, Beer avers that an utterance at t7 of “It is t7 now” expresses a proposition different from an utterance of “It is t7 at t7 such that the two utterances “differ in sense.”[43] Beer then claims that this fact is consistent with the A-utterance and the B-utterance’s being such that “they refer to the same subjects or relata [viz., t7 and t7 and ascribe the same relation between them [simultaneity]” and concludes from this that the A-utterance does not ascribe an A-property (conceived as irreducible to a B-relation).[44] But this argument, although it has convinced Richard Gale[45] and L. Nathan Oaklander,[46] seems to me unsound. Note, to begin with, that Beer’s premise that the A and B utterances differ in sense but refer to the same relata and ascribe the same relation between them is ambigu ous, yielding two interpretations. On one reading, it means that the two utterances differ in sense but refer in the same way(s) to all and only the same items and ascribe to these items the same n-adic properties. On this reading, the premise is self- contradictory, since differences in sense entail differences in referents, or ways of referring, or n-adic properties ascribed to the referents. On a second reading, the premise is self-consistent and means that two utterances both refer to t7 and ascribe to it being simultaneous with itself, even though (by virtue of their “difference in sense”) one of them refers to something the other does not or in a way the other does not or ascribes some n-adic property the other does not. But on this reading, the conclusion that the A-utterance does not ascribe an irreducible A-property does not follow, since the premise is consistent with the A-utterance differing in sense from the B-utterance by virtue of ascribing an irreducible A-property to t7 or referring to t7 in a way that involves the propositional definite description the time that exemplifies the irreducible A-property of presentness.
It is possible that a detenser might try to defend Beer’s argument along the following lines. The detenser will note that at one point Beer mentions the relevant difference in sense between the A-utterance and the B-utterance. She says they differ in sense in that “the proposition expressed by now [at t7] using ‘t7 is now’ informs us that t7 is present or now, whereas that expressed by ‘t7 is (tenselessly) simultaneous with t7 does not.”[47] This statement is not consistent with the tensed theory of time, it will be said, since Beer holds that “an event’s having an A-determination—its being past, present, or future—is identical with that event’s bearing a temporal relation [a B-relation] to some moment of time.”[48]
But this line of defense is plainly untenable. For one thing, it does not avoid the first horn of the aforementioned dilemma, the contradiction that the A and B utterances differ in sense but refer in the same ways to the same items and ascribe to them the same n-adic properties. If t7‘s being present is identical with t7‘s being simultaneous with t7 then the A-utterance, by virtue of ascribing presentness tot or referring to t as present, does not ascribe to t7 any n-adic property or refer to t7 in any way that the B utterance does not, which is inconsistent with their differing in sense. How could the information that t7 is present be different from the information that t is simultaneous with t if t7 being present is identical with its being simultaneous with t7? Second, the proffered line of defense avoids the second horn of the dilemma, that the difference in sense is consistent with the A-utterance’s ascribing an irreducible A-property, only at the price of being overtly question- begging, since the thesis that “t7 being present is identical with t7‘s being simultaneous with t7 now becomes a premise, rather than the conclusion of the argument against the tensed theory of time. I conclude that Beer’s new tenseless theory of time—Gale’s and Oaklander’s endorsement notwithstanding—fails to rescue the tenseless date theory from the quandary posed by A-date-sentences. (Further criticisms of Beer’s theory are presented elsewhere.)[49]
2.4 The Entailment Relations of A-Sentence-Tokens and Tenseless Date-Sentences
I shall discuss in this section two sets of entailment relations that are possessed by A-sentence-tokens but not by their corresponding tenseless date-sentences. The first set concerns entailments possessed by some declarative A-sentence-tokens, the second, entailments possessed by certain modal A-sentence-tokens.
The account of the truth conditions and confirmation conditions of A-sentence- tokens in the last two sections provides the key for uncovering the first set of entailment relations to be discussed. In terms of our “Henry is ill” example, we may say that each token of this sentence entails
(1) The illness of Henry (is) simultaneous with some sentence-token about this illness;
for each token of this A-sentence is true only if it is simultaneous with Henry’s illness. However, (1) is not entailed by “Henry (is) ill on July 28, 1940” or any one of its tokens; for each token of this sentence could be true even if no tensed or tenseless token about Henry’s illness occurred simultaneously with his illness.
Nelson Goodman believes that an argument to the effect that A-tokens do, and tenseless date-tokens do not, convey information about the simultaneity of the event with the tokens is not sufficient to show that A-tokens cannot be translated by the date-tokens. Goodman calls tokens like A-tokens “indicators”; an indicator is any token whose replica has a different denotation. He writes:
We may seek a translation that contains no name of the indicator itself, but rather another name for what the indicator names. Thus a certain “here” is translated by any “Philadelphia” and a certain “ran” is translated by any
“runs [tenseless] on Jan. 7, 1948 at noon E.S.T.”
Against such translations, it is sometimes urged that they do not really convey the content of the originals. A spoken
“Randy is running now”
tells us that the action takes place at the very moment of speaking, while a
“Randy runs [ on October 17, 1948 at 10 P.M., E.S.T.”
does not tell us that the action takes place simultaneously with either utterance unless we know in addition that the time of the utterance is October 17, 1948 at 10 P.M. E.S.T. Since—the argument runs—we recognize the tenseless token[50] as a translation of the tensed one only in the light of outside knowledge, we have here no genuine translation at all. But this seems to be no more cogent than would the parallel argument that “L’Angleterre” is not a genuine translation of “England” because we recognize it as a translation only if we know that L’Angleterre is England.[51]
Goodman’s point is that if a token of “Randy is running now” occurs on October 17, 1948 at 10:00 P.M., E.S.T., then it is translated by a token of “Randy (runs) on October 17, 1948 at 10:00P.M., E.S.T.,” which we will suppose occurs on the same date. If we know that the latter token occurs on the same date as the action, we know that it is simultaneous with the action. Consequently, we will be in possession of the same information in regard to the tenseless token as is conveyed by the tensed token, namely, that the action takes place at the very moment the token occurs.
Contrary to Goodman’s contention, the knowledge that the tenseless token is simultaneous with the action is an illegimate “outside knowledge” and thus renders the translation spurious. A necessary condition of some token T1‘s genuinely translating another token T2 is that T1 can be known to translate T2 solely on the basis of a knowledge of the rules of usage of the tokens in the class of tokens that includes T1 and all tokens similar to T1 and of the tokens in the class of tokens that includes T2 and its replicas. The knowledge involved here is not “outside knowledge” but is just that knowledge needed for any translation. Now the translation of some token of “England” by some token of “L’Angleterre” meets this condition, for I recognize the token of “L’Angleterre” to be a translation of the token of “England” solely on the basis of my knowledge that all tokens in the class of tokens that includes the English token and all tokens similar to the English token are used to refer to the same nation as are all tokens in the class of tokens that includes the French token and its replicas. The translation of the token of “Randy is running now” that occurs on October 17, 1948 at 10:00 P.M., E.S.T. by the token of “Randy (runs) on October 17, 1948 at 10:00 p.m., E. S. T.” that occurs on the same date does not meet this condition; for this translation requires a knowledge of a different sort. This is a consequence of the fact that all tokens in the class of tokens including only the tensed token and its replicas have different rules of usage from the tokens in the class of tokens including only the tenseless token and its replicas. The tensed token and its replicas are used to refer to events that are simultaneous with the tokens, but the tenseless token and its replicas are not so used. The latter tokens are, instead, used to refer to events that occur on October 17, 1948 at 10:00 P.M., E.S.T. Consequently, I know by virtue of understanding the rules of usage of the tensed token and its replicas that the tensed token occurs simultaneously with the action (supposing the tensed token is true); but I cannot know that the tenseless token (supposing it also is true) is simultaneous with the action simply by virtue of understanding the rules of usage of this token and its replicas. Instead, I must resort to outside knowledge, that is, observational evidence that somebody uttered the tenseless token simultaneously with Randy’s running on October 17, 1948 at 10:00P.M., E.S.T. In a word, the tensed token, by virtue of its rules of usage, entails that the action is simultaneous with the token; and the tenseless token does not. The two tokens are logically nonequivalent and cannot genuinely translate each other.
Some date-theorists, such as Steven Boer and William Lycan, might respond to my argument by asserting that the difference between the A-token and the B-token is not semantic but pragmatic. In their article “Who, Me?” Boer and Lycan claim:
“Now” is simply a purely referential or rigid designator whose referent is a moment in time, and there is a pragmatic rule to the effect that a token of “now” always refers to the moment of its utterance. . . . There is nothing special about the semantic content of the propositions expressed [by the A-sentence-tokens that distinguishes them from the propositions expressed by the corresponding tenseless date-sentences]; all that is distinctive are the pragmatic rules that compute the indexical terms’ denotata.[52]
Although the main argument Boer and Lycan present for this semantic/pragmatic distinction concerns first-person indexicals, a similar argument can be constructed for temporal indexicals. The argument would be that a token of “Randy is running now” does not entail that there is some token simultaneous with Randy’s running but, instead, that by virtue of its pragmatic rules, it implicates this (in Grice’s sense). It follows from this that the aforementioned difference between this A-token and the corresponding B-token is not a difference in the entailment relations of the two tokens and, therefore, that this difference is compatible with the two tokens’ being logically equivalent.
But this argument is unsound, since a necessary condition of a sentence being a Gricean implicature, rather than an entailment, is that its falsity is consistent with the truth of the sentence or sentence-token that implicates it. But there is no possible world in which the token of “Randy is running now” is true and in which
(2) There is some sentence-token about Randy’s running that (is) simultaneous with Randy’s running
is false.
The date-theorist might respond by adopting a more liberal conception of the pragmatic/semantic distinction than Boer and Lycan’s and assert that some entailments are pragmatic and that difference in pragmatic entailments is consistent with having the same semantic content (expressing the same proposition). Semantic entailments, it will be said, are due to the proposition expressed by the sentence- token, and pragmatic entailments are not. If a token T1 expresses a proposition P1 that entails, and is entailed by, the proposition P2 expressed by the token T2 but T1 entails something T2 does not, then T1 and T2 are semantically logically equivalent but pragmatically logically nonequivalent. The date-theorist will say that semantic, rather than pragmatic, logical equivalence is a necessary condition for translation in the relevant sense (where “translates” means “expresses the same proposition as”). The date-theorist will insist that the A-token’s entailment of its simultaneity with the event it is about is pragmatic and therefore consistent with the A-token’s semantic logical equivalence with the B-token.
There are two ways I could respond to this objection. First, I could argue that the A-token’s entailment of its simultaneity with the event it is about demands that it be interpreted as expressing a different proposition from that of the date-sentence- token. Second, I could assume for the sake of argument the date-theorist’s claim that the entailment in question is merely pragmatic and consistent with the semantic logical equivalence of A- and B-tokens but go on to show that there are additional logical nonequivalences that are clearly semantic, even by the date-theorist’s own standards. This second line of defense will be easier to deploy, since it appeals to principles the date-theorist accepts; and since it is sufficient to refute the date- theorist, I shall use only it in what follows.
I will concentrate on modal sentences and use them to show that many A-sentence-tokens are so far from being semantically logically equivalent to the corresponding tenseless date-sentences or tokens that they are not even materially equivalent to them
Consider a token T3 of
(3) Christ might not have been born during the present year
that occurs during the year of Christ’s birth, 1 A.D. T3 is true when it occurs, which implies that the proposition it expresses is true at least at this time. But T3 does not entail its corresponding de dicto date-sentence,
(4) Christ might not have been born during whatever year includes Christ’s
birth,
and the proposition expressed by T3 does not entail the proposition expressed by (4); for (4) and the proposition it expresses are permanently false. To see that (4) really is the corresponding de dicto date-sentence, note that the translation of a 1 A.D. utterance of “the present year” is “1 A.D.,” used attributively to express the sense of “whatever year includes Christ’s birth.” “1 A. D.” does not refer directly to the year that in fact includes Christ’s birth; rather, it refers indirectly to whatever year is the one that has the property of including Christ’s birth. Item (4) implies that there is some possible world in which whatever year has the property of including Christ’s birth also has the property of not including his birth. This of course is self contradictory and prevents the proposition expressed by (4) from being equivalent, even materially, to the one expressed by T3.
This nonequivalence extends to the counterfactual propositions about Christ’s birth that are expressible at any time by A-sentence-tokens. A 1986 utterance of
(5) Christ might not have been born earlier than now
expresses a true proposition; but its corresponding de dicto date-sentence,
(6) Christ might not have been born earlier than whatever year is 1,985 years later than Christ’s birth
is self-contradictory. Of course, if we insert “actually” between “is” and “1,985,” the proposition becomes true; but the standard de dicto date-theory does not regard the propositional definite descriptions of dates to be world-indexed. Even if we do world-index the descriptions, the de dicto date-theory will encounter the same problem as does the de re date-theory. This problem arises if we assume the redu tionist theory of time. Given this theory, the de re translation of the token T3 of (3) is
(7) Christ might not have been born during 1 A.D.,
where “1 A. D.” directly refers to the year-long set of events that includes Christ’s birth. (7) does not make the self-contradictory assertion that Christ might not have been born during whatever year-long set of events has the property of including his birth, but it makes a different self-contradictory assertion. It implies
(8) There is some year-long set of events that actually includes Christ’s birth but in some merely possible world does not include his birth.