THE PROBLEM OF SHARED SEMANTIC BEGINNINGS IN THE DEBATE OVER THE PHILOSOPHY OF TIME

Shawn McKinney

The debate in the philosophy of time revolves around whether or not time is tensed or tenseless. For the tenser, events only truly exist in the present, [1] and events possess properties of pastness, presentness, and futureness. The tenseless theorist denies that events possess those transient properties and instead stand in the unchanging relations of earlier than, simultaneous with, and later than, one another. For the detenser the ontological status of an event does not change with regard to its temporal location. For example, the tenser would say that yesterday is less real than today merely by virtue of its pastness, whereas the detenser would claim that yesterday's being earlier than today does not affect yesterday's ontological status (that is, all other things being equal, yesterday is no less real than today merely by being earlier than today).

The focus of this paper attempts to settle this philosophical debate through linguistic analysis. Tensers and detensers both attempt to settle the question of the ontological status of events with respect to time through linguistic analysis, what Quentin Smith calls a "linguistic ascent." [2] It is claimed that A and B theorists (tensers and detensers) can begin the ascent from a common starting point which is that A sentences have tenseless rules of usage and that there are commonly accepted rules governing the truth conditions and entailment relations of sentences. The following lengthy quote from

Smith makes this point more explicit:

[tensers and detensers] agree that there are A-sentences and B-sentences and that there are conventionally accepted rules of usage governing these sentences. For example, both agree that there are commonly accepted rules governing the truth conditions and entailment relations of these sentences. Moreover, both agree that these sentences have rules of usage that are statable in a tenseless language. The tenser would beg the question if he assumed from the outset that A-sentences also have irreducibly tensed truth conditions, confirmation conditions, and so on that imply that events have A-properties; and the detenser would beg the question if he assumed from the outset that A-sentences do not also have such conditions. But since they both agree that A-sentences have tenseless rules of usage, the explication and discussion of these rules forms a common ground from which both the tenser and detenser can build their arguments. [3]

Unfortunately, the detenser and tenser have in actuality no such common ground from which to argue. While it is a given that the two camps should require an agreement on the existence of some data in advance in order to even begin an exchange with the possibility of bearing fruit, it is the nature of the agreement which Smith describes that is problematic and ultimately illusory.

Support for the above claim is found in the ontological positions each side of the debate holds, positions regarding the nature of existence which are not compatible. The tenser believes that only what is present, or now, exists. The detenser does not believe that ‘now’ is an irreducibly meaningful term, nor is presence a property which events may possess, thus neither presence nor nowness can be a relevant factor in determining whether or not something exists. This disagreement about the nature of existence seems to preclude any agreement about the truth conditions or entailment relations of sentences. If these rules of usage differ for the two opposing sets of theorists, then any attempt at building arguments on linguistic analysis is bound to be question begging.

To illustrate the point that sentences have different entailment relations depending upon whether a tensed or tenseless ontology is assumed, consider that the meanings of the terms 'to be,' 'being,' and 'is,' are different for the tenser and detenser. Here I am assuming that the verb 'to be' means to exist. For the tenser, all forms of the verb 'is' will necessarily entail that whatever subject the verb is attached to will be present, or now. On the other hand the verb 'is' will entail no such information for the detenser. For the detenser, the verb will only entail that the subject associated with it occurred at some point. Because the nature of existence is at stake in the outcome of the debate between the tensers and detensers, it can hardly be used as a shared starting point.

Consider a token A of the sentence, "Anna runs." The token cannot entail the same information regarding Anna's ontological status for the detenser and tenser Specifically, the sentence cannot entail Anna's existence to the tenser because Anna only exists if she is doing the running reported by the sentence now, but the sentence can entail Anna's existence to the detenser because it is tenselessly irrelevant whether Anna runs yesterday (prior to the utterance of A), right now (simultaneous with the utterance of A), or 400 years from today (later than the utterance of A). So, if the tenser and detenser cannot agree that Anna's existence is implied by tokens of A, then it is at best unclear what facts they can agree that tokens of A entail.

This problem of entailment relations also complicates any agreement the two sides might wish to have concerning the rules governing the truth conditions of sentences. Anna's existence cannot be a truth condition for tokens of A (at least not one the two sides can agree upon, since any kind of agreement would require them to concede the conclusion each hopes to prove). So, if Anna's existence cannot be a common truth condition of A, then tokens of A would have to be understood as conditionals for them to have agreeable truth conditions. But, to treat tokens of A as conditionals (If Anna exists, then she runs.) prevents those tokens from being able to provide any useful information concerning the nature of time, thus precluding sentences from providing any support for either theory of time. Thus, either the tensers and detensers do not agree on the entailment relations and truth conditions of sentences or they must agree to treat sentences as conditionals that cannot provide any useful information regarding the nature of time. If the later is true, then no linguistic ascent can shed light on the tensed/tenseless debate.

The illusory agreement between the A&B theorists concerning the rules of usage governing sentences leads to the very question begging Smith warns against. For instance, a token R of the sentence, "Anna is reading now," will necessarily have irreducibly tensed truth conditions for the tensed theorist but will not necessarily have irreducibly tensed truth conditions for the tenseless theorist. For R to be true, then both Anna and whatever Anna is reading must exist. The two camps would certainly like to agree on that much, but to do so would be to equivocate the meaning of exist. The tenser clearly means to 'to be present' by exist, "if time also consists of A-properties, then some events are ontologically distinguished by virtue of their temporal determinations; the events that are or exist in the tensed sense, the events that possess the A-property of presentness, have a reality not possessed by other events." [4] On the other hand, the presence of Anna and her reading material are not necessarily truth conditions of R according to the B-theorist. For the detenser presence is not a property, merely a mental construct which assists humans in the navigation of our experience; "It [presentness] is either the simultaneity of the events with our experiences of them or the times themselves at which events occur – times that are simultaneous with our experiences of events." [5] Both theorists would like to agree that tokens of R entail that Anna and her reading material exist, but the two cannot agree at the outset on such existence because they each assume that reality works according to their theory's ontological claims. The problem is that the nature of reality is what provides the backdrop against which the truth conditions of a sentence token can be deductively discerned. Neither set of truth conditions entails a logical contradiction because the very meanings of the terms in the sentence token are contingent upon reality. That is, what it means to be Anna is to be a person, and a person is a real thing. The token of R is not about a fictional Anna; the referent of "Anna" is a real person. To some extent I am only assuming that this is the case (that the referent of the term "Anna" is a real person possessing all the properties which are commonly entailed by the term). Such an assumption should not incite disagreement among any party because if tokens of R were not understood as such then we would only be analyzing theoretical, or fictional sentences which would not be intended, nor would directly relate to the reality which is the subject of the debate at hand.

To show that this confusion is present (tenselessly) in the tensed/tenseless debate and that the confusion traps both sides in question begging arguments, contemplate the linguistic arguments of Smith and Oaklander. [6]

The question both theorists attempt to answer is, "do tokens of tensed sentences have only those tenseless truth conditions stated by tokens of tenseless sentences?" [7] For Smith the answer is no, "there is room for truth conditions of S other than the tenseless ones, namely, the tensed facts about the presentness of S and 1980 that are necessary conditions for the tensed truth of S." [8] (S is any token of "It is not 1980.") Predictably, Oaklander will give the opposite answer to the question in his response to Smith, "Smith argues that they do not, since tensed facts must also be introduced, but I shall argue that his arguments fail." [9]

Smith argues that tokens of "It is now 1980" (S) logically entail the sentence "1980 is present" (V). [10] How can Smith support his claim that tokens of S entail tokens of V? The claim seems to be solely supported by an assumption that reality is tensed; there is no other reason to assume this entailment. [11] Once Smith makes this assumption about the entailment relation between S and V he goes on to disprove Mellor's previous claim that a token U of "S occurs in 1980" has the same truth conditions as tokens of S;

a fact statable by S implies a fact statable by V, and consequently a fact statable by V is among the truth conditions of S. But these considerations are incompatible with the claim that S has the same truth conditions as any token U of "S occurs in 1980," for a fact statable by V is not among the truth conditions of U. The tenseless fact stated by V, that V occurs in 1980, is not implied by the fact that S occurs in 1980, for "It is now 1980" could be tokened in 1980 even if "1980 is present is not.... Therefore, since a fact statable by V is a truth condition of S but not of U, it follows that S and U have different truth conditions and fail to translate each other. [12]

The point of Smith's criticism is to show tenseless sentences not only cannot translate tensed ones, but also that tenseless translations of tensed sentences have different truth conditions than the tensed sentences they attempt to translate. But, notice how Smith goes about the refutation; he uncovers a contradiction in Mellor's argument "by adopting a tensed theory of time." [13] One should hope that assuming a tensed theory of time would reveal problems with a tenseless description of time. But isn’t such an assumption blatantly dishonoring the agreement that there are commonly accepted rules governing the truth conditions and entailment relations of sentences? Recall that Smith says in his earlier work that, "The tenser would beg the question if he assumed from the outset that A-sentences also have irreducibly tensed truth conditions," [14] and that is exactly what Smith has done in his attempt to refute Mellor. All that can be said in Smith's defense is that his "goal here is not so much to develop a full-blown tenser theory as it is to present an internal critique of the new tenseless theory of time." [15] Even though Smith is not attempting to provide an argument explicitly supporting tensed time, his counter argument still relies on an assumption that is clearly unacceptable.

Oaklander's response to Smith in the following essay does not fair much better. He acknowledges that Smith has revealed a difficulty with Mellor's argument, but only an apparent one and not a genuine challenge to Mellor. Oaklander's response is to cite David Kaplan's theory of demonstratives and indexicals, "According to Kaplan (and Mellor), the meaning of an indexical sentence-type (and all its tokens) is a semantic function (rule) from facts about tokens of that type (their context of utterance) to their tenseless truth conditions." [16] Without delving too deeply into Kaplan's theory of demonstratives and indexicals (which would be satisfying and no doubt useful in clarifying the tenser/detenser debate, but is unfortunately beyond the scope of this paper) it can still be demonstrated that Oaklander (and Mellor) are assuming a tenseless ontology in their use of Kaplan's theory to defend Mellor from Smith's criticisms.

Kaplan's theory of indexicals (nondemonstrative indexicals; "I," "here," ”now," "today," "yesterday," etc.) essentially states that, "the linguistic conventions which constitute meaning [for these words/indexicals] consist of rules specifying the referent of a given occurrence of the word (we might say, a given token, or even utterance, of the word, if we are willing to be somewhat less abstract) in terms of various features of the context of the occurrence." [17] Oaklander's scholarly sleight of hand is to impregnate Kaplan's theory with tenseless ontological assumptions. Kaplan's claim is that the rules governing the meaning of indexicals indicate that the referent of the term is determined by the details of the context in which the word occurs. Oaklander seems to be assuming that the type of context described is a tenseless one, but the details of the context in which

“now” occurs can just as well include presence as they can simultaneity with the occurrence of now. Kaplan's theory does not rule out the possibility of tensed features of the context in which the indexical occurs. What Oaklander should have claimed [18] is that "According to Kaplan, the meaning of an indexical sentence-type (and all its tokens) is a semantic function (rule) from facts about tokens of that type (their context of utterance) to their truth conditions." If Oaklander had avoided making the assumption that the truth conditions were solely tenseless, then he would not have been able to support his conclusion that, "any token of [It is now 1980] is true with respect to the context in which it is produced (namely, the time at which it is uttered), if and only if the year of that context is 1980, and the same may be said of any token of [1980 is present]." [19] Not assuming a tenseless ontology would have prevented him from justifiably claiming that the necessary and sufficient truth conditions merely include the tenseless fact that 1980 was occurring simultaneously with the occurrence of the token. It would be entirely consistent with Kaplan's theory to include the tensed fact that 1980's presence is a feature of the context in which the token is uttered. Simply put, the features of the context in which the indexical is uttered remain to be determined by the outcome of the tensed/tenseless debate. Kaplan himself wisely refrains from bogging down his "findings regarding the semantics of demonstratives and other indexicals" [20] with unnecessary claims about the ontological features of the contexts in which said indexicals are uttered. Oaklander's assumption that the referent of the indexical is merely tenseless presupposes the claim he attempts to prove, that the entailment of tokens of V by tokens of S can be explained with only tenseless facts.

In the above two examples it was shown that both the tensed and tenseless theorist each assumes a different ontological backdrop to explain the entailment relations of sentences (A-sentence types in these examples). These assumptions run counter to any attempt to work from an agreement about the entailment relations and truth conditions of sentences. Unfortunately it seems that any attempt to work from a shared linguistic principle is bound to entail little more than circular reasoning so long as the two sides wish to use 'existence' as a starting point because they simply hold irreconcilable accounts of what it means to exist; and if they did wish to continue using this sort of starting point then they would have to agree that 'existence' has no appreciable or useful meaning and that it can neither be inferred that something which exists is either present or not present. Once that point is agreed upon I fail to see how any information useful to the debate can be inferred from sentences at all.

While the tenser and detenser cannot seem to have any shared semantic starting point, and the belief that they did led both camps to engage in question begging arguments, both sides should still be able to agree on some data as their starting point. Oaklander even suggests that the two camps are not working with the same language at all and that Smith's linguistic argument is only applicable to the old token-reflexive version of the tenseless theory. He claims that,

the new tenseless theory is abandoning the analytic ideal of arriving at a single language that is adequate for both ontological and logical investigations. Once these two functions of language are separated and kept distinct, it is open to the defender of the tenseless view to maintain that logical connections among sentences in ordinary language do not represent ontological connections between

facts in the world. [21]

If it is the case that there are two different languages being discussed, and it is the goal of

the detensers to show how one of those languages does not represent reality, then the use

of linguistic analysis in the debate becomes even less clear. So, this lack of agreement

should lead both tensers and detensers to either radically rework their approach to

linguistic arguments or motivate them to abandon linguistic analysis altogether in favor of some genuinely more agreeable starting data.