STILL UNDER CONSTRUCTION

x re-definition; confusion; invalid argument; zombie TC being kept goingx

 

 

David Lewis's (1970) re-definition of "true", supporting a bluntly fallacious argument

 

The general framework which Lewis adopts and implicitly defends in this paper is the 'truth-conditional approach' to sentence meaning, according to which the meaning of a sentence consists in, or is completely determined by, the conditions of its truth. There is a serious problem with this view, however: Most sentence types actually cannot be true or false; interrogative sentences and directive sentences, for example, cannot. Is "Where did Napoleon die?" true, or false? Is "Go ahead!" true, or is it false? In both cases, the question is misdirected: the question of truth or falsity does not arise at all; to raise it seriously would be to make a categorical mistake. There is only one type of sentence which actually has (sometimes) conditions of truth, the declarative sentence. Since non-declarative sentences have meaning too, the truth-conditional approach to sentence meaning cannot be true. Let us refer to this as the "nondeclarative sentence problem".

 

A meaning for a sentence, we said initially, was at least that which determines the conditions under which the sentence is true or false. But it is only declarative sentences that can be called true or false in any straightforward way. What of non—declarative sentences: commands, questions, and so on? If these do not have truth-values, as they are commonly supposed not to, we cannot very well say that their meanings determine their truth conditions. (Lewis 1970, 54)

 

...

 

...; if we liked, we could also call an imperative or interrogative or other non-declarative sentence true itfits sentence radical has the value truth, but we customarily do not. (Lewis 1970, 56)

 

 

Lewis's attempt to save the truth-conditional approach starts with a putative distinction in the structure of sentence meaning, between a "sentence radical" (or propositional content) on the one hand, and a "performative", or "mood" element. Here are two examples. 

(a) The meaning of "Are you late?" can be paraphrased as 'I ask you whether you are late', where 'I ask you' is a "performative" element, and "you are late" the sentence's "radical" (or "propositional content").

(b) The meaning of "Be late!" can be paraphrased as 'I command you to be late', where 'I command you' is a "performative" element, and "you are late" the sentence's "radical".

According to Lewis, all sentence share this structure, and thus all sentences, including non-declarative sentences, have a paraphrase of the kind under consideration--let us call this a "performative paraphrase".

 

The method Lewis applies to save the truth-conditional approach is this. He assumes the freedom of the scholar to define and re-define her/his terms arbitrarily, as the following passage demonstrates:

We may grant that a declarative sentence [56] is called true iff its sentence radical has the value truth; if we liked, we could also call an imperative or interrogative or other non-declarative sentence true iff its sentence radical has the value truth, [although, as he sees himself] we customarily do not. (Lewis 1970, 55f.)

Using this (putative) freedom, he then introduces a re-definition of the term "true":

I see no problem in letting non-declaratives have the truth-values of the performatives they paraphrase; after all, we need not ever mention their truth-values if we would rather not. (Lewis 1970, 59)

Applying this "construal" to the examples just used, we arrive at the view that ...

(a) The sentence "Are you late?" is true when in uttering it the speaker does ask the audience whether she is late.

(b) The sentence "Be late!" is true when in uttering it the speaker does command the audience to be late.

 

It is apparent that this notion of "truth" is profoundly different from the real notion of truth (what we consistently mean when we use the term in the relevant connections). In order not to confuse this peculiar sense 'of "true"' with the common notion, we better introduce a terminological convention. The notion introduced by Lewis will be marked by a superscript "LEWIS": 

  »trueLEWIS«

 

The aim Lewis pursues with this re-definition of the term "true" is to save the truth-conditional approach to sentence meaning from disproof by means of obvious counter-examples. The argument we are to construe at the back of our minds (Lewis does not state this argument explicitly himself) will proceed along the following lines:

(1) For every non-declarative sentence there is a performative paraphrase, having conditions of truth.

(2) A sentence is »trueLEWIS« when its performative paraphrase is true.

Hence,...

(3) For every sentence there are conditions of its »truthLEWIS«.

I shall refer to this as "the implicit argument".

The objection against the truth-conditional approach to sentence meaning just mentioned, according to which some sentences do have meaning, but have no truth-conditions, is supposed to be refuted by (3).

 

The implicit argument in Lewis (1970) had massive impact on the meaning-theoretical debate by stabilizing the truth-conditional approach, which became very popular through Davidson's "Meaning and Truth" (1967). Even today, when you confront adherents of this paradigm with the non-declarative sentence problem, you almost regularly receive replies like "This problem was solved long ago: don't you know Lewis' 'General Semantics'?", or, less bold, "There actually are several ways of ascribing truth-conditions to sentences: consider the performative analysis!". Additionally, in the late 60ies and 70ies a number of suggestions were put forward as to how the meaning of non-declarative meaning could be represented in the form of a proposition; but the implicit argument proposed in Lewis (1970) was probably the most influential factor.

 

The argument, however, fails clearly, due to a logical mistake in Lewis' argument, which in turn goes back to the mistaken view that the definition of a scholarly term is largely a matter of arbitrary decision.

 

First of all, we must recognise that Lewis' definition of "truth", as a definition of "truth", is just inadequate. As a representation of the meaning of "true", it misses its target clearly. This is why we had to introduce the terminological variant »trueLEWIS«. It can be seen by examining its consequences.

 

As we saw, Lewis's suggestion implies, and Lewis has no problem accepting that it implies, that ...

(a) The sentence "Are you late?" is true when in uttering it the speaker does ask the audience whether she is late.

(b) The sentence "Be late!" is true when in uttering it the speaker does command the audience to be late.

Although Lewis endorses this result, it is clearly odd. We may, for the sake of the argument, even assume that the performative paraphrases of the sentences under consideration are true or false, that means, that "I ask you whether you are late" and "I command you to be late" are either true or false (which actually is highly doubtful, but let's set this aside). Even then, it is clear that the non-declarative sentences under consideration, "Are you late?" is not either true or false, and the same holds for "Be late!". There simply are no conditions under which we would call either of these sentences "true"--to ascribe truth to them would be a categorical mistake.

 

What is contended in the implicit argument above (1-3) is that since every non-declarative sentence has a performative paraphrase, having conditions of truth, and since the truth of the performative paraphrase makes the non-declarative sentence »trueLEWIS«, every non-declarative sentence has condition of its »truthLEWIS«. Now what exactly is the non-declarative sentence problem? Consider the following variants: 

(P1) Not all sentences have conditions of »truthLEWIS«.

(P2) Not all sentences have conditions of truth.

Clearly, the non-declarative sentence problem is not P1, but rather P2. The appearance of success of the implicit argument is there only when and as long as we confuse »truthLEWIS« with real truth. As soon as we stop confusing »truthLEWIS« with truth, that means, as soon as we keep different things separate, we see that the implicit argument fails miserably. Even if we granted that interrogative and directive sentences are subject to »truthLEWIS« (which we may do for the sake of the argument), it is clear that this does not make them either true or false in the sense of the non-declarative sentence problem.

 

But how can it be that the two things, »truthLEWIS« and truth, are so widely confused with each other? The reason is obvious: In contrast to us, Lewis does not mark the difference between his notion of »truthLEWIS« and the common notion of truth. Instead, Lews re-defines the word "true", thus causing its ambiguity. This ambiguity, in turn, is the reason why his readers (along with Lewis himself) overlooked the difference between what the truth of a sentence actually is (say, the correspondence of that sentence's meaning with a state of affairs realised in the world), and what Lewis defined (the truth, not of the sentence itself, but of another sentence which, allegedly, serves as a paraphrase of that sentence). The mistaken view that Lewis's argument saves the truth-conditional approach from refutation is triggered by an unperceived ambiguity of the word "true", caused by Lewis's arbitrary re-definition of that word.