20

Meaning and the Future

Ken Dickey

1. Introducing Vocabulary

First, I want to distinguish between what I’ll call ordinary facts and semantic facts. At first glance, semantic facts include such things as i) that the name ‘George W. Bush’ refers to a certain man, now our President, ii) that the predicate ‘is a chair’ picks out the property or concept of being a chair, or the set containing all chairs, including this one and the one you are sitting on, or some such thing, and iii) that the sentence ‘snow is white’ expresses the proposition or state of affairs to that effect. Ordinary facts are the other things: i) that George W. Bush is our current President, ii) that this is a chair, and iii) that snow is indeed white. One might want to say that semantic facts are essentially linguistic, whereas ordinary facts are not. But this is not quite right, since I would want to include among ordinary facts that the name ‘George W. Bush’ exists, that it was uttered a moment ago, etc. You see, words are as much an ordinary part of our world as are bus schedules and soccer tournaments. So semantic facts are more specific; they concern relations between words (one part of the ordinary world) and other parts of the world. Typically, these relations are described as meaning or reference, so semantic facts are facts about meaning or reference.

Second, I want to point out that many ordinary facts can be pegged to certain times, however vague the time specification might be. For example, the fact that the Titanic sank is tied to that fateful day in 1912. The fact that World War I resulted in the loss of over 7 million lives is tied to several years in the early 1900s. And the ongoing fact that Arethra Franklin was and is a singer is tied to many years, ranging from the fifties to the present and beyond. But I want to emphasize for each of these, however vague the range of times the facts are about, there are certain times they are not about. These include times well before the events in question ever began (say, the year 1500), and most importantly, times well afterward. Clearly, the fact that the Titanic sank is not about the world in 2007 (even though it is a fact in 2007 that the Titanic sank in 1912). Instead, on April 14th, 1912, certain events occurred which gave rise to the fact that the Titanic sank. As I intend it generally, an ordinary fact is pegged to that time or range(s) of times in which the events giving rise to it occurred.[1] (As we shall see in what follows, some semantic facts are different, and can be said to be temporally unusual.)

Next, I want to get all parties to this discussion on the same page by insisting on Strawson’s distinction between a term or sentence on the one hand, and an utterance of a term or sentence on the other.[2] Many logicians and philosophers of language write as if a term refers to an object and a sentence expresses a proposition. As a serious proposal, this could at best work within an artificial language, where terms and sentences were stipulated to always have the same semantic value on every occasion. But as a proposal for any natural language, it can only be seen as a convenient idealization. Surely, the term ‘this cup’ would have to pick out different cups on different occasions of use, and the sentence ‘This cup is full’ would therefore have to express a variety of propositions. Strawson clarifies the matter greatly by pointing out that while the term in isolation has a meaning (roughly, the linguistic conventions governing its use – in the case of ‘this cup’, it would be something like “the demonstrated drinking vessel”), it is only a use or utterance of the term that refers to an object.[3] Similarly, while a sentence alone can have a meaning, it is only a use or utterance of a sentence that expresses a proposition.

As a final preliminary note, I want to bring in the notion of supervenience. Despite a history of some confusion, there is no reason to be suspicious of this innocuous concept. Simply put, to say that the xs supervene on the ys is to say there can be no difference in the xs without there also being a difference in the ys. Intuitively, one would assert a supervenience thesis when one believes the xs depend on the ys in some way, but doesn’t fully understand the nature of that dependence. It might be because the xs are identical with the ys, or because the xs are caused by the ys, or because the xs emerge from the ys in some other way. Supervenience alone doesn’t tell us which, if any, of these claims is correct, and thereby makes a less controversial claim than any one of them alone. For example, I might maintain that mental states supervene on brain states. Then my view is compatible with that of the identity theorist, the eliminative materialist, emergent theories of mind, and even Occasionalism! Still, my view would not be completely vacuous, since it rules out the idea that I could change my mind about something without there being any change whatsoever in my brain.

It is time now to put all of these preliminary ideas together to formulate what I shall call the General Semantic Supervenience Thesis (GSST).

(GSST): For all times t, semantic facts about any utterance at t supervene on ordinary facts about the world up to and including t.

Note, this is called the general semantic supervenience thesis, for we shall want to distinguish specific variations concerning semantic facts about singular terms, predicates, and sentences.

(TSST): For all times t, terms n, and objects o, the fact that an utterance of n at t refers to o (as opposed to some other object or none at all) supervenes on ordinary facts about the world up to and including t.[4]

(PSST): For all times t, predicates p, and properties, concepts, etc. φ, the fact that an utterance of p at t picks out φ (as opposed to some other property or none at all) supervenes on ordinary facts about the world up to and including t.

(SSST): For all times t, sentences s, and propositions Ψ, the fact that an utterance of s at t expresses Ψ (as opposed to some other proposition or none at all) supervenes on ordinary facts about the world up to and including t.

Also note that all versions of the semantic supervenience thesis make essential use of the time of utterance. A simpler, less controversial semantic supervenience thesis could be formulated by omitting concern about the time of utterance. But it is among my goals in this paper to focus attention on this very issue.

At first glance at least, all versions of the semantic supervenience thesis would seem to be correct; semantic facts do seem to depend only on the past and present, not the future.[5] I shall show in the next section that some version of the semantic supervenience thesis is presupposed by most if not all traditional semantic theories. Then I shall present two different types of arguments (in sections 3 and 4) for believing the semantic supervenience thesis is in fact false. I conclude by relating these arguments to a more general movement toward externalism in the philosophy of language.

2. Survey of Traditional Theories

Frege

Gottlob Frege famously held the theory of sense and reference whereby singular terms do not directly refer to objects, but are instead always mediated by sense. Thus, ‘the morning star’ and ‘the evening star’ both refer to the planet Venus, but they do so via different senses. And while this example is phrased in terms of definite descriptions, the same point could be made using the names ‘Hesperus’ and ‘Phosphorus’.

Frege’s concern with how singular terms refer might suggest that (TSST) would be the proper semantic supervenience thesis to measure his theory against. However, his appeal to the notion of sense complicates matters, in two different ways. First, we are never told exactly what senses are, but it is reasonable to conclude from Frege’s hints that they are entities that not only have a long past, but will endure into the future. Perhaps they, like natural laws or Platonic Forms, are timeless entities. In that case, since senses mediate reference, it would be hard to know on Frege’s view whether the reference of an utterance of, say, ‘the morning star’ is determined solely by facts prior to and contemporaneous with the utterance.

Second, because Frege’s account of singular terms involves conditions of satisfaction, there is a straightforward example he could offer for denying (TSST). Consider the singular term ‘the winning lottery ticket’ in my declaring “I home mine will turn out to be the winning lottery ticket.” In this situation, it is clear that what ‘the winning lottery ticket’ refers to is still up for grabs at the time of utterance, even though the conditions of satisfaction are perfectly well known. (Otherwise, no one would play the lottery.)

We can avoid both complications by instead considering (SSST) in connection with Frege’s theory. When I utter the sentence “The morning star is very bright”, on Frege’s view, I express a proposition which contains the senses of that sentence’s components.[6] And Frege does say enough about senses to suggest that he does buy into (SSST), if not (TSST), viz. “The sense of a [term] is grasped by everybody who is sufficiently familiar with the language”, and that senses are “a common store of thoughts which are transmitted from one generation to another.” So even though senses might be timeless or eternal, that an utterance expresses a particular sense is well within the grasp of the speaker’s community at the time of utterance.

Russell

Bertrand Russell eschews Frege’s notion of sense and consequently makes a sharp distinction in his treatment of definite descriptions on the one hand, and names on the other. Well actually, Russell has to further clarify a distinction between ordinary names such as ‘Aristotle’, which he assimilates as disguised definite descriptions, and what he calls logically proper names. Russell’s driving concern is epistemological; we must be directly acquainted with the constituents of the propositions we express, otherwise in some sense we would not know what we are talking about. So, for example, since I am not in direct mental contact with Aristotle, when I say “Aristotle lived a long life”, I do not refer directly to Aristotle, nor do I express a proposition with him as a constituent. It is only in the case of a logically proper name (Russell suggests ‘this’ or ‘that’ as candidates) that I refer directly to an object of immediate sensory experience. Instead, in uttering “Aristotle lived a long life”, I express a complex proposition, such as “The student of Plato who taught Alexander and lectured for years at the Lyceum lived a long life”,[7] each constituent of which I am directly acquainted with.

Clearly, in either case, Russell is a believer par excellence, in the semantic supervenience thesis. In the case of any utterance of a logically proper name, we can attribute to Russell a belief in (TSST). Because immediate acquaintance with the referent is built into the very notion of a logically proper name, clearly future facts can’t be relevant to determining the reference in such a case. And in all other cases, we can attribute (PSST) and (SSST) to Russell. This is again because Russell insists on immediate acquaintance with the constituents of any proposition expressed when one makes a statement.

Kripke

Saul Kripke maintains that names, including ordinary proper names, are rigid designators. This means that a name such as ‘Aristotle’ picks out the same object in every possible world in which that object exists. Kripke means to contrast his view with a descriptivist view such as Russell’s or even a sense view such as Frege’s. For suppose one associates with the name ‘Aristotle’ either the description or the sense expressed by ‘the student of Plato who taught Alexander and lectured for years at the Lyceum’. Then if someone else had done those things, the name ‘Aristotle’ would have referred to him or her. That is, consider the world in which Plato had a son who he took on as a student, and who subsequently taught Alexander the Great and lectured at the Lyceum. On the view we are considering, the name ‘Aristotle’ refers to this other person in that world. But this is counter-intuitive, Kripke would insist: “We are speaking of [Aristotle] and asking what, in certain counterfactual circumstances, would have been true of him.”[8]

If not a description, what, then, does determine the referent when I use the name ‘Aristotle’ in a sentence? Kripke suggests it is a causal chain leading back from my utterance of the name, through prior utterances by other language users from whom I acquired the use of the name, ultimately eventuating in some baptismal event in which the person Aristotle was originally dubbed with that name.

Kripke’s view of ordinary names does have considerable advantages over those of Frege or Russell. When I utter “Hillary Clinton might be our next President”, I certainly do seem to be speaking essentially of Hillary, not just someone who happens to satisfy some description or other. And I can do this even though I have never met H.C., nor had any other direct contact with her.

But plausible or not, it is important for present purposes to note that Kripke’s causal chain account commits him to (TSST), at least where names are concerned. The causal chain runs back in time, through other utterances and ultimately to a prior dubbing event. So clearly it is not future facts, but only past and present facts that determine what I refer to when I utter an ordinary name.[9]

Kaplan

We conclude this survey by considering the theory of indexicals offered by David Kaplan. Any correct treatment of indexicals requires a distinction between the term (such as ‘I’, ‘here’, and ‘now’) and an utterance of the term. For obviously, different utterances of the word ‘I’ (or ‘here’) refer to different people (or places). And it is not that the word ‘I’ (or ‘here’) is just highly ambiguous. Rather, there is a systematic connection between any utterance of ‘I’ and its referent. The rule is that ‘I’ refers to the speaker of that utterance (and ‘here’ refers to the place where the utterance occurs).