CHAPTER 4 4 страница

But to return to the main theme. Philosophers and linguists frequently make the point that sentences containing definite descriptions (for example, 'the wooden door'), or, more obviously, personal pronouns (' I ', 'you', etc.), demonstrative pronouns ('this', 'that') or demonstrative adverbs of place and time ('here', 'there', 'now', 'then') can be used to assert, deny or query indefinitely many true or false propositions. Alltoo often they fail to add that this is also the case for sentences

5.5 Propositions and prepositional content 143

containing proper names and dates. The vast majority of sen­tences in the most familiar natural languages can be used, on par­ticular occasions of utterance, to assert, to query or to deny indefinitely many propositions, each of which has a constant truth-value which is independent of that of each of the others that may be expressed by uttering the same sentence.

But what exactly is the relationship between sentences and propositions? This is a difficult question; and the answer that one will give to it depends in part upon one's theory of meaning. It suffices for present purposes to note that certain assumptions must be made, whether tacitly or explicitly, by anyone who says of sentences that they express propositions. Ayer, it will be noted, is more circumspect, in the quotation given above. He talks of sentences as purporting to express propositions; and it is easy to see why. The purport of a document is the meaning that it conveys by virtue of its appearance, or face-value, and standard assumptions about the interpretation of the author's intentions. Sentences of whatever kind may be uttered, in var­ious circumstances, without there being any question of the assertion or denial of a proposition. For instance, if I am asked to provide someone with an example of an English sentence in the past tense, I might comply with their request by uttering (11). It is quite clear that, in the circumstances envisaged, the sentence that I have uttered is not to be construed as referring to or saying anything about anyone (or anything). Indeed, in one sense of the verb 'say' I have not said anything. For this and other reasons, we cannot say that sentences as such express prop­ositions. What we can do, however, is to interpret the phrase, 'purport to express a proposition' in terms of the notion of char­acteristic use, as explained in Chapter 1. And this is what will be done throughout the next three chapters. We shall assume that all declarative sentences belong to the class of sentences whose members are used, characteristically, to make statements (that is, to assert or deny particular propositions) and that they have this potential for use encoded in their grammatical struc­ture as part of their purport or face-value;that all interrogative sentences have encoded in their grammatical structure their potential for querying particular propositions; and so on.

144 Meaningful and meaningless sentences

Under this interpretation of the notion of purport, or face-value, we can temporarily and provisionally exclude from considera­tion not only a variety of metalinguistic uses of sentences and expressions, but also what will be identified in Part 4 as their performative and indirect uses.

Sentence-meaning is.intrinsically connected with utterance-meaning, but can be distinguished from it by virtue of the dis­tinction between the characteristic use of a sentence (which need not be its most frequent or psychologically most salient use) and its use on particular occasions. I have emphasized the notion of the use of sentences at this point because the so-called use theory of meaning, associated with Wittgenstein, Austin, and others, developed out of and in reaction to verificationism. What I want to do in this book is to throw a bridge between a restricted version of the meaning-as-use theory and the truth-conditional theory of descriptive meaning, which also developed historically out of verificationism. It is essential to the fulfilment of this purpose that what is said here about the purport, or face-value, of a sentence and what is said in Part 4 about the intrinsic connexion between sentence-meaning and utterance-meaning should be properly understood.

It is also important that a distinction should be drawn between the propositions expressed by a sentence on particular occasions of utterance and its prepositional content. I will come to this presently. Strictly speaking, as we shall see, it is not propositions that sentences purport to express, but prepositional content. Provided that this is understood, together with the point made earlier about the purported, or face-value, use of sentences, no confusion will arise if, occasionally and for brevity's sake, we say, as most authors do, that sentences express propositions.

5.6 NON- FACTUAL SIGNIFICANCE AND E M O TI V I S M

There is one final point that may be made in connexion with Ayer's statement: "A sentence is factually significant to a given person if, and only if, he knows how to verify the proposition that it purports to express." This has to do with factual

5.6 Non-factual significance and emotivism 145

significance. It was by means of the verifiability principle that the logical positivists wanted to proscribe as meaningless, or nonsen­sical, sentences which purport to express metaphysical and theo­logical propositions such as, let us say:

(12) 'Everything must have a cause'

or

(13) 'God is good'.

But it was soon realized that the principle of verifiability also ruled out (or, at least, did not obviously allow as meaningful) what many of them held to be the philosophically more respect­able sentences which purport to express propositions of ethics and aesthetics, such as:

' Cannibalism is wrong'
or

'Monet was a better painter than Manet'.,

One way round this problem was to say that, although such sen­tences as (14) and (15) are not factually significant, they have another kind of meaning: an emotive, or expressive, meaning.

Emotivism - the thesis that in making what purport to be factual statements in ethics and aesthetics one is not saying any­thing that is true or false, but giving vent to one's feelings - has now, like logical positivism itself, been abandoned by most of those who once professed it. In its day, it had the beneficial effect of obliging philosophers to look more closely at the logical status of different kinds of both meaningful and meaningless utter­ances. It is this that Ryle had in mind when he said, in the quota­tion given earlier, that the verification principle helped philosophers to see that there are different ways in which an utterance can be significant, or meaningful, and different ways in which it can be nonsensical. One important product of this insight into the diversity of meaning, as we shall see in Part 4, was Austin's theory of speech acts.

146 Meaningful and meaningless sentences

5.7 TRUTH-CONDITIONS

The truth-conditional theory of meaning, like verificationismi one of its historical antecedents, comes in several slightly differ­ent versions. What they have in common is their acceptance of the following thesis: to give an account of the meaning of a sen­tence is to specify the conditions under which it would be true or false of the situation, or state of the world, that it purports to describe. Alternatively, it is said that to know the meaning of a sentence is to know the conditions under which it (or the state­ment made by uttering it) would be true or false. Neither of these formulations is very precise as it stands, and they are not necessarily equivalent. For example, neither of them actually identifies the meaning of a sentence with its truth-conditions; and the second leaves open the question of what precisely is meant by knowing the truth-conditions of a sentence. We shall return to such questions in the following chapter.

For the present it suffices to draw readers' attention to the dif­ference between the truth-value of a proposition and the truth-conditions of a sentence. To take a simple example:

(16) 'John Smith is unmarried'

purports to express a set of propositions, each of which has a par­ticular truth-value according to whether whoever is being referred to by 'John Smith', on particular occasions of utterance, is unmarried (at the time of the utterance). We do not need to know who (or what) is being referred to on all or any of the occa­sions of the utterance of the sentence 'John Smith is unmarried' or whether the person being referred to (on the assumption that it is a person) is unmarried in order to know what conditions the world must satisfy for the proposition "John Smith is unmar­ried" to be true. In cases like this at least, we know how we might verify (or falsify) empirically any one of the propositions that a sentence purports to express.

Also, independently of any empirical investigation relating to a given John Smith's marital status, we can argue, on the basis of our knowledge of English, whether

5.7 Truth-conditions 147

'John Smith is not married'
or even

'John Smith is a bachelor'

has the same truth-conditions as (16). If (and only if) they have the same truth-conditions, we will say that they have the same propositional content. And a moment's reflection will tell us that (18) differs truth-conditionally from both of the others. Not every unmarried individual is a bachelor. For example, unmarried women are not bachelors (and, to reiterate a point made earlier, there is nothing in the structure of English that prevents a woman from bearing the name 'John Smith': we have only to think of the well-known women novelists George Eliot and George Sand). Or again, a child with the name 'John Smith' - or a racehorse, or a yacht, or indeed any entity whatso­ever that is not only not married, but also not marriageable, and can be appropriately referred to with the name 'John Smith' - will fulfil the truth-conditions of (17), but not of (18). The situation with respect to (16) and (17) is less clear-cut. It is arguable (though not all native speakers will take this view) that an individual cannot be unmarried unless he or she (or it) could in principle have been married: i.e., is (or has been) mar­riageable. Those Who take this view might say that sentences such as

'That racehorse is unmarried'
and

'That square-rigged schooner is unmarried'

are meaningless: that they do not make sense. Others might say that (19) and (20), though odd, are tautologous (and therefore meaningful) because each of the p'ropqsitions that they could be used to express is analytic (and therefore true: see 5.8). Qthers, again, might wish to draw a potentially relevant distinction between (19) and (20); they might argue that the former is less obviously, or less definitely, categorially incongruous (and therefore less obviously meaningless) than the latter, in that it is

148 Meaningful and meaningless sentences

quite easy to conceive of a culture in which racehorses (but not ships, on the assumption that they are indeed, by natural neces­sity, inanimate and incapable of mating and reproduction) are brought within the scope of the same laws as human beings with respect to cohabitation, the legitimacy of their offspring, etc.

As we saw earlier (and it is a point that will be emphasized throughout this book), if we are seriously concerned about both the theoretical and the empirical foundations of linguistic semantics, we must not dismiss as facetious or irrelevant the deliberate manipulation of a particular society's normal onto-logical assumptions when it comes to the testing of native speakers' (including one's own) intuitive judgements of meaningfulness or semantic equivalence. In this section we are concerned with truth-conditional equivalence as an important, if not the sole, component of the semantic equivalence of sen­tences. The principle of truth-conditional equivalence holds independently of the facts of the matter in particular instances:

Sentences have the same prepositional content if and only if they have the same truth-conditions.

Readers are now invited to put to the test their understanding of the principle of truth-conditional equivalence, as formulated in (21), by trying to falsify the statement that (16a) and (17a) have the same prepositional content:

(16a)'That man is unmarried' and

(17a) 'That man is not married'.

(These two sentences differ from (16) and (17), it will be noted, in that I have substituted the phrase 'that man' for the proper name 'John Smith'.) Are there any circumstances - in the actual world as we know it - under which it can be said truly (and properly) of the same fully adult (and therefore, let us assume, marriageable) male person, x, that x is both not married and not unmarried? Are there circumstances in which x could be truly and properly said to be both married and unmarried?

5.8 Tautologies and contradictions 149

In this chapter, I have deliberately emphasized the historical connexion between verificationism and truth-conditional semantics. Most authors nowadays would not have done this on the grounds that verificationism as a philosophical doctrine is all but obsolete. But all the points made above about verifica­tionism are relevant, in my view, to a proper understanding of truth-conditional semantics; and we shall draw upon them later. They could have been made in respect of truth-conditional semantics without mentioning logical positivism and verifiabil-ity. But there is much in present-day formal semantics which derives from its positivist origins.

In any case, it is important to realize that when it comes to the construction of a truth-conditional theory of meaning for nat­ural languages, verifiability (or falsifiability) continues to pre­sent problems, not just of practice, but also of principle. It will not do to dismiss these problems on the grounds that verifica­tionism itself has failed. As we have seen several times already, it is unreasonable to expect that competent speakers of a language should always be able to decide whether two express­ions are necessarily true of the same classes of entities or not. If the truth-conditional theory of semantics is so formulated that it rules out what seems to be a genuine indeterminacy in the semantic structure of natural languages, it may be rejected without more ado. But, as we shall see in due course, it need not be formulated in this way.

5.8 TAUTOLOGIES AND CONTRADICTIONS

Two kinds of propositions that are of particular concern to logi­cians and semanticists are tautologies (in a technical sense of 'tautology') and contradictions. The former, as traditionally defined, are propositions which are necessarily true by virtue of their logical form. An example would be

(22) "Either it is raining or it is not raining".

Contradictions, on the other hand, are propositions that are necessarily false by virtue of their logical form. For example:

150 Meaningful and meaningless sentences

(23) "It is raining and it is not raining".

What is meant by 'logical form' in this context varies somewhat according to which system of logic we are operating with. But the above propositions would be shown to be tautologous and contradictory, respectively, in standard prepositional logic by the definition of negation ("not"), conjunction ("both... and"), and disjunction ("either... or...").

It will be noted that I am using double quotation-marks at this point, because we are not concerned with English sentences as such, but rather with their prepositional content or with the propositions which they purport to express. (This use of double quotation-marks has been established in earlier chapters and is consistent with the general convention whereby expressions are distinguished notationally from their meanings.) It is important to emphasize once again that propositions, not sentences, are the bearers of truth and falsity.

Obviously, in construing "It is raining and it is not raining" as contradictory we have to make certain assumptions about the time and place being referred to: in particular, we must assume that we are not referring to different times and/or different places in the two constituent simpler propositions. "It is raining in Manchester and it is not raining in Timbuktu" is not contra­dictory. One might thipk that nothing but pedantry is involved in making points like this explicit. But, as we shall see later, there are important theoretical reasons for keeping such seemingly trivial points in mind.

Provided that we do keep the point that has just been made in mind and draw the distinction between sentences and proposi­tions when it needs to be drawn, we can extend the application of the terms 'tautology' and 'contradiction' to sentences in a nat­ural way. We can say of the sentences

(24) 'Either it is raining or it is not (raining)'

and

(25) 'It is raining and it is not (raining)'

5.8 Tautologies and contradictions 151

that, taken at face-value, they are tautologous and contradic­tory, respectively. (By taking them at face-value, I mean inter­preting them in terms of their purported propositional content and on the assumption that they are being used characteristi-cally: see 5.5.) One of the principal tasks of semantic theory is to show how and why competent speakers of a language will recognize that some sentences are tautologous and others con­tradictory (unless there are good reasons in context for constru­ing them otherwise than at their face value).

Logical truths, or tautologies, are a subclass of analytic truths: that is, propositions whose truth is determined wholly by their meaning (cf. Chapter 4). However, linguists commonly extend the terms 'tautology' and 'contradiction' to cover, not only those propositions (and sentences) whose truth or falsity is determined by logical form as this is traditionally conceived, but all kinds of analytically true or false propositions (and sen­tences). Thus, they would say that

(26) 'This bachelor is unmarried' is a tautologous sentence, and

(27) 'This bachelor is married'

is a contradictory sentence, in that the first purports to express a tautology and the second a contradiction (on the assumption that 'bachelor' is taken in the relevant sense). We shall follow this practice.

Tautologies and, especially, contradictions are sometimes classified as being semantically anomalous. Taken at face-value, they are uninformative: they cannot be used to tell some­one facts which they did not previously know or could not deduce themselves on the basis of their knowledge of the language and the ability to draw valid inferences from what they already know. And yet, whatever 'semantically anomalous' or 'meaningless' means in relation to tautologies and contradictions, it cannot mean "devoid of sense" (if 'sense' means "propositional content"). For tautologies and contradic­tions, as we have just seen, are by definition necessarily true and necessarily false respectively; and this implies that contradictory

152 Meaningful and meaningless sentences

sentences, no less than tautologous sentences, must have deter-minable truth-conditions. The former are false and the latter true, as Leibniz put it, in all possible worlds (4.4). We can argue on both theoretical and empirical grounds about the range of data that is, or should be, covered by the terms 'tautol­ogy' and 'contradiction' (that is to say, about the coverage of the term 'analytic'). But we canot without inconsistency abandon the principle that analytically true and analytically false sen­tences are meaningful in the sense of having a truth-conditionally explicable propositional content.