CHAPTER 4 3 страница

The second kind of substitutional sense-relation to be men­tioned here is incompatibility,which is definable in terms of entailment and negation:

(35) f => ~ g and g => ~ f.

For example, 'red' and 'blue' are defined to be incompatible in this way: if something is (wholly) red it is necessarily not (even partly) blue, and conversely. A special case of incompatibility is complementarity,which holds within two-member lexical sets, where, in addition to (35), the following conditions are also satisfied:

(36) ~ f => g and ~ g => f.

For example, not only does (i) 'married' entail the negation of 'unmarried' and (ii)'unmarried' entail the negation of 'mar­ried', but (iii) the negation of 'married' entails 'unmarried' and (iv) the negation of 'unmarried' entails 'married'. Complemen­tarity is often treated as a kind of antonymy ("oppositeness of meaning").

But antonymy in the narrowest sense - polar antonymy -differs from complementarity in virtue of gradability (in terms of more or less. This means that the conjunction of two negated antonyms is not contradictory. For example, 'good' and 'bad' are polar antonyms and "x is neither good nor bad" is perfectly acceptable, even though " x is not good" might be held to imply " x is bad" (in some looser sense of 'imply') in many contexts. When they are graded in an explicitly compara­tive construction (" x is better than y "), the following holds:

(37) f⁺(x,y) => g⁺(y,x)

4.5 Sense-relations and meaning-postulates 129

where the superscript plus-sign is a non-standard, but conveni­ent, way of symbolizing "more". For example, if/is 'good' and g is 'bad', then f⁺ and g⁺ symbolize the selection of the forms better and worse ("more good" and "more bad"). If we substi­tute expressions referring to particular individuals for x and y, we see that, for example, "John is better than Peter" entails and is entailed by "Peter is worse than John".

In fact, expressions with the meanings "more good" and "more bad" are two-place converses. They are like corre­sponding active and passive verb-expressions ('kill': 'be killed'), and also like such pairs of lexemes as 'husband': 'wife' (due allowance being made in both cases for the associated grammati­cal adjustments). The verbs 'buy' and 'sell' exemplify the class of three-place (lexical) converses:

(38) 'buy'(x,y,z)=>'sell' (z,y,x).

For example, "Mary (x) bought the car (y) from Paul (z)" entails, and is entailed by, "Paul (z) sold the car (y) to Mary (x)". Obviously, what I have here called syntactic adjustments (to avoid the more specific implications of the term 'transforma­tion' in linguistics) need to be precisely specified. Provided that this is done and that we can give a satisfactory account of the relation between sentences, propositions and utterances, we can account formally for sets of entailments such as

"John killed Peter" => "Peter was killed by John",

"Mary is John's wife" => "John is Mary's husband",

"John bought a car from Peter" => "Peter sold a car to
John",

and so on.

This is a big proviso! Before we address ourselves to it in Parts 3 and 4, it is worth emphasizing the fact that in this chapter we have been concerned solely with the descriptive meaning of expressions. Moreover, we have limited ourselves to a brief con­sideration of only the most important of the relations that hold, by virtue of sense, in the vocabularies of natural languages. My main concern has been to give the reader some idea of what is involved in the formalization of lexical structure and to outline

130 The structural approach

two notions that linguists have invoked in this connexion in recent years: sense-relations and meaning-postulates. There is perhaps no reason, in principle, why the non-descriptive meaning of lexemes should not also be formalizable. But so far at least formal semantics has taken the same limited view of lexical structure as we have done here.

PART 3

Sentence-meaning

CHAPTER 5

Meaningful and meaningless sentences

5.0 INTRODUCTION

In the last three chapters we have been concerned with lexical semantics: i.e., with the meaning of lexemes. We now move on, in Part 3, to a consideration of the meaning of sentences.

The distinction between sentences and utterances was introduced in Chapter 1 (see 1.6). The need for drawing this distinction is reinforced by the discussion of grammaticality, acceptability and meaningfulness in the following section (5.1). But our main concern in this short, and relatively non-technical, chapter is the meaningfulness of sentences. Granted that some sentences are meaningful and others meaningless, what grounds do we have for drawing a theoretical distinction between these two classes of sentences? Is it a sharp distinction? Is there only one kind of meaningfulness?

What may be described as truth-based theories of the meaning of sentences have been particularly influential in modern times, initially in philosophical semantics, later in linguistic semantics. Two of these were mentioned in Chapter 1: the verifi-cationist theory and the truth-conditional theory (1.7). According, to the former, sentences are meaningful if (and only if) they have a determinate truth-value. In formulating the verification-ist theory of meaning (or meaningfulness) in this way, I am temporarily neglecting to draw a distinction (as many verificationists did) not only between sentences and utterances, but also between propositions and prepositional content, on the one hand, and between truth-values and truth-conditions, on the other. The reasons for drawing these distinctions (which

132 Meaningful and meaningless sentences

were tacitly drawn in the slightly different formulation of the vcrificationist theory that was given in Chapter 1) will be explained below.

As we shall see, the verificationist theory, as such, in the form in which it was originally put forward (in the context of logical positivism), has been abandoned by most, if not all, philosophers oflanguage. I should make it clear, therefore, that my principal aim in this chapter is not to give an account of the verificationist theory of meaning for its own sake, but rather for its historical significance in preparing the, way for the truth-conditional theory of meaning, which was also mentioned in Chapter 1 and which is central in all modern versions of formal seman­tics. In my view, it is much easier to understand the truth-conditional theory of meaning and to see both its strengths and its weaknesses if one knows something about its prede­cessor, the verificationist theory, and the philosophical context in which verificationism arose. That there is a connexion between meaning and truth (as there is a connexion between truth and reality) is almost self-evident and has long been taken for granted by philosophers. In this chapter, we take our first steps towards seeing how this intuitive connexion between meaning and truth has been explicated and exploited in modern linguistic semantics.

5.1 GRAM M AT I C A L I T V, ACCEPTABILITY AND M EAN ING FULNESS

As was noted in an earlier chapter, some utterances, actual or potential, are both grammatical and meaningful; others are ungrammatical and meaningless; and yet others, though fully grammatical and perhaps also meaningful, are, for various reasons, unacceptable (1.6).

To say that an utterance (more precisely, an utterance-type) is unacceptable is to imply that it is unutterable (more precisely, that one of its tokens is unutterable) in all normal contexts other than those involving metalinguistic reference to them.

Many such utterances are unacceptable for socio-cultural reasons. For example, there might be a taboo, in a certain

5.1 Grammatically, acceptability and meaningfulness 133

English-speaking society, upon the use of the verb 'die', rather than some euphemism such as 'pass away', in respect of members of the speaker's or hearer's immediate family. Thus, the fully grammatical and meaningful utterance

(1) His father died last night

might be fully acceptable, but not the equally grammatical and (in one sense of 'meaningful') equally meaningful utterance

(2) My father died last night.

Or again, in some cultures, it might be unacceptable for a social inferior to address a social superior with a second-person pro­noun (meaning "you"), whereas it would be perfectly accept­able for a superior to address an inferior or an equal with the pronoun in question: this is the case (though the sociolinguistic conditions are often more complex than I have indicated here) in many cultures. It follows, that the same utterance with, argu­ably, the same meaning would be acceptable in some contexts but not in others. There are many such culture-dependent dimensions of acceptability. Some of them, as we shall see later, are encoded in the grammar and the vocabulary of particular languages. For this reason and others, one must be sceptical about the validity of the general principle, which is often taken for granted by semanticists, that whatever can be said in one language can be said in another. At the very least one must be sensitive to the different senses in which one can interpret the phrase 'can be said' (or 'can be uttered'). I will come back to this point in Part 4.

Somewhat different are those dimensions of acceptability which have to do with rationality and logical coherence. For example,

(3) / believe that it happened because it is impossible

might be regarded as unacceptable from this point of view. Indeed, if uttered, (3) might well provoke the response:

(4) That doesn't make sense

(though it is paradoxical, rather than being devoid of meaning or contradictory). What makes (3) unacceptable, in most

134 Meaningful and meaningless sentences

contexts, is the fact that the speaker appears to be calling atten­tion to his or her own irrationality; and this is an odd thing to do in most normal circumstances. However, even such utterances may be fully acceptable in certain contexts. In any event, one should not too readily concede, as some semanticists would, either that the sentence in question is uninterpretable or mean­ingless or, alternatively, that the proposition it expresses is neces­sarily false.

More generally (if I may now invoke the distinction between sentences and utterances), one should not take too restrictive a view of the meaningfulness of uncontextualized (or decontextua-lized) sentences: the semantic acceptability, or interpretability, of sentences is not something that can be decided independently of the context in which they might or might not be uttered.

5.2 THE MEANINGFULNESS OF SENTENCES

Sentences are, by definition, grammatically well-formed. There is no such thing, therefore, as an ungrammatical sentence. Sen­tences however may be either meaningful (semantically well-formed) or meaningless (semantically ill-formed). Utterances, in contrast with sentences, may be either grammatical or ungrammatical. Many pf the utterances which are produced in normal everyday circumstances are ungrammatical in various respects. Some of these are ihterpretable without difficulty in the context in which they occur. Indeed, they might well be regarded by most of those who are competent in the language in question as fully acceptable. As we saw in Chapter 1, gram-maticality must not be identified with acceptability; and, as we saw in the preceding section of this chapter, acceptability must not be identified with meaningfulness. But what do we mean by 'meaningfulness'?

In the preceding section we were careful to relate the notion of acceptability to utterances. At this point we will restrict our attention to what would generally be regarded as sentences and we will continue to operate with the assumption that the sen­tences of a language are readily identifiable as such by those who are competent in it, and more especially by its native

5.2 The meaningfulness of sentences 135

speakers. As we shall see in due course, this assumption must be qualified. The distinction between grammatical and semantic well-formedness is not as sharp as, for the moment, we are tak­ing it to be. Nevertheless, to say that the distinction between grammatical and semantic well-formedness — and conse­quently between grammar and semantics - is not clear-cut in all instances is not to say that it is never clear-cut at all.

There are many utterances whose unacceptability is quite definitely a matter of grammar, rather than of semantics. For example,

(5) I want that he come

is definitely ungrammatical in Standard English in contrast with

(6) / want him to come.

If (5) were produced by a foreigner, it would probably be con­strued, and therefore understood, as an incorrect version of (6). There is nothing in what appears to be the intended meaning of (5) which makes it ungrammatical. And many languages, including French, would translate (6) into something which is grammatically comparable with (5).

If someone, having uttered (5), not only refused the proffered correction, but insisted that it meant something different from the corrected version, we should simply have to tell them that, as far as Standard English is concerned, they are wrong. We can classify their utterance, unhesitatingly, as ungrammatical.

There are other, actual or potential, utterances which we can classify, no less readily, as grammatical, but meaningless. Among them, we can list, with their authors, such famous ex­amples as

Colourless green ideas sleep furiously (Noam Chomsky)

Quadruplicity drinks procrastination (Bertrand Russell)

Thursday is in bed with Friday (Gilbert Ryle).

Of course, none of these is uninterpretable, if it is appropriately contextualized and the meaning of one or more of its component expressions is extended beyond its normal, or literal, lexical meaning by means of such traditionally recognized rhetorical

136 Meaningful and meaningless sentences

principles as metaphor, metonymy or synecdoche. The fact that this can be done - and indeed has been done on several occasions to considerable effect - merely proves the point that is being made here. As far as (9) is concerned, it is of course readily and immediately interpreted, both literally and meta­phorically, if 'Thursday' and 'Friday' are construed as refer­ring to persons (as in G. K. Chesterton's The Man Who Was Thursday and Daniel Defoe's Robinson Crusoe). Indeed, a moment's reflection will show that there is a euphemistic inter­pretation which is half-way between the fully literal and the definitely metaphorical. In order to assign an interpretation to (7)-(9), one does not identify, and tacitly correct, some general rule or principle which governs the grammatical structure of English, as we did in the case of (5); one tries to make sense of what, at first sight, does not of itself make sense on a literal, face-value, interpretation of the expressions which it contains. We shall need to look later at the question of literal inter­pretation (see Chapter 91. All that needs to be said here is that (7)-(9) are grammatically well-formed and that, despite their grammaticality, they are literally meaningless. Any generative grammar of English will therefore generate, or admit as gram­matically well-formed, not of course the utterances (7)-(9), but the sentences which correspond to them and from which (as will be explained in Chapter 8) they can be derived:

(7a) 'Colourless green ideas sleep furiously'

(8a) 'Quadruplicity drinks procrastination' (9a) 'Thursday is in bed-with Friday'.

 

The reader is reminded at this point that here, as throughout this book, utterances (in the sense of utterance-inscriptions or stretches of text) are represented in italics, whereas sentences, like other expressions, are represented by means of their citation-form enclosed in single quotation-marks. To be contrasted with (7a)-(9a) are

(7b) * Green ideas sleeps furiously,

(8b) * Drinks quadruplicity procrastination,

5.2 The meaningfulness of sentences 137

(9b) * Thursday am on bed when Friday.

 

In (7b)-(9b) the asterisk indicates grammatical ill-formedness. (7b) breaks the grammatical rule of agreement between the sub­ject and the verb in English; (8b) is ungrammatical (in present-day English), not only as a declarative sentence, but also as an interrogative sentence, because it breaks the rules of word-order; and (9b), like (7b), breaks the rule of subject-predicate agreement and, additionally, uses a count noun without a deter­miner (*on bed, which, in contrast with in bed, is not a grammati­cal idiom) and uses a conjunction in a position which syntactically requires a preposition (when, unlike since, cannot fulfil both functions).

It might seem pointless, at this stage, to distinguish notation-ally, as I have done, between sentences and utterances, but the reasons for doing so will be made clear in Part 3. As we shall see, sentences are expressions which may have several forms, including context-dependent elliptical forms.

It is also worth emphasizing that a distinction is being drawn here, implicitly, between ungrammatical strings of forms, such as (7b)-(9b), on the one hand, and non-grammatical gibberish, on the other, such as

(10) On when am Thursday furiously bed,

which cannot be said to violate any specific grammatical rules of English. This distinction is not generally drawn in generative grammar, because generative grammars, as formalized origin­ally by Chomsky, partition strings of forms into two complemen­tary subsets: A, the set of all grammatically well-formed strings (which are then identified with the sentences of the language in question), and B, its complement, the set of strings which by vir­tue of not being grammatical are defined to be ungrammatical. Strings of recognizably English word-forms, such as (10), which are neither grammatical nor ungrammatical, are not only not grammatical: they are, as it were, not even trying to be gram­matical, and the question whether they are grammatically well-formed or ill-formed does not arise. More to the point, in the present connexion, they do not make sense and cannot be made

 

 

138 Meaningful and meaningless sentences

to make sense by any kind of adjustment or correction. They are perhaps meaningless or nonsensical in the everyday use of the words 'meaningless' and 'nonsensical', but they are perhaps not rightly described as semantically ill-formed. The expressions 'well-formed' and 'ill-formed' first came into linguistics as part of the terminology of generative grammar: as they are commonly employed, they imply cpnformity to a set, or system, of precisely formulated rules or principles. As we shall see later, so-called formal semantics takes the view that, as there are rules (or principles) of grammatical (well-formedness, so also there are rules (or principles) of semantic well-formedness. Whether this is or is not the case is, a question that we can postpone until later. Here I am concerned to emphasize, first, that meaning-fulness, or semantic well-formedness (if we use that term and, for the present at least, accept what it implies), is readily distinguishable, in clear cases, from grammaticality, and, second, that not every utterance which is judged to be unacceptable on the grounds that it does not make sense is properly regarded as semantically ill-formed.

But if the intuitive notion of making sense is not a reliable guide, what are the criteria which lead us to decide that an utter­ance, actual or potential, is semantically well-formed or ill-formed? We shall address this question in the following section.

5.3 CORRIGIBILITY AND TRANSLATABILITY

As we have seen, semantic well-formedness must be distin­guished from grammatical well-formedness (grammaticality): both of them are included within, or overlap with, acceptability, as semantic ill-formedness and grammatical ill-formedness are included within, or overlap with, unacceptability. But - to repeat the question that was posed at the end of the preceding section - what are the criteria other than the intuitive notion of making sense which lead us to decide that an utterance is or is not semantically well-formed?

One of the criteria that was invoked earlier in connexion with grammaticality is what we may now label the criterion of corrigibility (5.2). Whereas

5.3 Corrigibility and translatability 139

(5) I want that he come

can be corrected - by some speakers to

(6) I want him to come
and by others perhaps to
(6a) I want for him to come

- without any change in what is assumed to. be the intended meaning, Chomsky's classic example,

(7) Colourless green ideas sleep furiously

cannot. In those instances in which the distinction between grammatical and semantic unacceptability can be most clearly drawn, the former are corrigible and the latter are not.

Other kinds of unacceptability, some of which at first sight seem to be a matter of meaning, also fall within the scope of the notion of corrigibility. For example,

(2) My father died last night might be corrected to, say,

(2a) My father passed away last night

in a language-community (of the kind referred to in section 5.1) in which the use of 'die' is prohibited with expressions referring to members of one's own family. But the unacceptability of (2) in such circumstances, is not such that we would say that it does not make sense. Its unacceptability is a matter of social, rather than descriptive, meaning. (And there are independent reasons for saying that, though corrigible, it is fully grammatical.)

In other instances, as we shall see later, the situation is less clear-cut. But, interestingly enough, the criterion of corrigibility and incorrigibility is still relevant in that it shows the pre-theoretically indeterminate cases to be genuinely indeterminate.

Another criterion that is sometimes mentioned by linguists is translatability. This rests on the view that semantic, but not grammatical, distinctions can be matched across languages. However, as we shall see later, it is not clear that what is

 

 

140 Meaningful and meaningless sentences

semantically unacceptable in some languages is semantically unacceptable in all languages. The criterion of translatability can supplement, but it does not supplant, our main criterion, that of corrigibility.

We turn now to a discussion of a famous and influential philo­sophical criterion of meaningfulness: verifiability.

5.4 V E R I F I A B I L I T Y A N D V E R I F I C AT IO N I S M

The verificationist theory of meaning - verificationism, for short - was mentioned in Chapter 1. As its name suggests, it has to do with truth. It was originally associated with the philo­sophical movement known as logical positivism, initiated by members of the Vienna Circle in the period immediately preced­ing the Second World War., Although logical positivism, and with it verificationism, is all but dead, it has been of enormous importance in the development of modern philosophical seman­tics. On the one hand, many of its proponents - notably Rudolf Carnap and Hans Reichenbach - were active in the construc­tion of systems for the analysis of language which have led, more or less directly, to the methods of modern formal seman­tics. On the other, the very excesses and defects of logical positi­vism forced its opponents, including Wittgenstein in his later work and the so-called ordinary-language philosophers, to make explicit some of their own assumptions about meaning. As Ryle (1951: 250) has said of vcrificationism: "It has helped to reveal the important fact that we talk sense in lots of different ways".

We shall not pursue Ryle's point at this stage. Instead, I will take one version of the famous verifiability principle and, in the next few sections, use this to introduce the notion of truth-conditions and other notions that will be of use to us later. The principle may be stated, initially and for our purposes, as follows: "A sentence is factually significant to a given person if, and only if. he knows how to verify the proposition which it purports to express" (Ayer, 1946: 35). This formulation by A. J. Ayer, it will be noted, does not say that the meaning of sentences (or alternatively of propositions) is their method of verification. It

5.5 Propositions and prepositional content 141

simply provides a criterion of one particular kind of meaning -factual significance; it does not define meaning as such.

Even so, it raises a number of problems. The logical positivists wanted to say that all verification is ultimately a matter of obser­vation. Yet, as Karl Popper has pointed out, universal state­ments of the kind that scientists tend to make cannot, in principle, be verified, though they may be falsified, by means of observation. For example, the statement that all swans are white can be falsified, by observing just a single instance of a black swan, but it can never be proved to be true on the basis of empirical investigation. Popper's point that falsifiability, rather than verifiability, is the hallmark of scientific hypotheses is now widely accepted (though it has its critics and requires to be for­mulated more carefully than it has been here).

 

 

5.5 PROPOSITIONS AND P ROPOS I T I O N A L CONTENT

Ayer's formulation of the verifiability criterion draws upon (though it does not explain) the distinction between sentences and propositions. The nature of propositions is philosophically controversial. But those philosophers who accept that proposi­tions differ, on the one hand, from sentences and, on the other, from statements, questions, commands, etc., will usually say that propositions

(i) are either true or false;

(ii) may be known, believed or doubted;

(iii) may be asserted, denied or queried;

(iv) are held constant under translation from one language to

another.

There are difficulties, as we shall see later, about reconciling all four of these different criteria: (ii) and (iii) seem to be in conflict as far as some natural languages are concerned; and (iv) makes dubious assumptions about intertranslatability.

However, granted that propositions are defined to be the bearers of a determinate and unchanging truth-value, it is quite clear that they must be distinguished from sentences. For the same sentence can be used on one occasion to say what is true

142 Meaningful and meaningless sentences

and on another to say what is false. And it is worth noting in this connexion, that even sentences such as

(11) 'Napoleon was defeated at Waterloo in 1815'

can be used to assert a variety of true and false propositions. There are certain natural languages in which personal names and place-names are in one-to-one correspondence with their bearers. But neither English nor French is among them. In Eng­lish the relation between a proper name and the set of entities or places which each bear that name is completely arbitrary. (The situation in French is slightly different, in that in France there are certain legal restrictions relating to the choice and assignment of personal names.) If 'Napoleon' happens to be the name of my dog and I am referring to my dog when I utter the above sentence, the proposition that I have asserted is presum­ably false.

Nor should it be thought that I have gratuitously or fa­cetiously introduced the qualification 'presumably' here. I have done so in order to remind readers of the very important point that here, as always, whenever one says that something is or is not true, one is making certain background assumptions that others may not share. For example, I have tacitly ruled out the possibility that Napoleon Bonaparte may have been reincar­nated as my dog. And there are indefinitely many such ontologi-cal assumptions - often loosely and inaccurately referred to as world-knowledge - which have a bearing upon the interpreta­tion of sentences such as (11) on particular occasions of utter­ance. There is nothing in the structure of English which commits us to the denial of unfashionable or eccentric ontologi-cal assumptions.