CHAPTERS6 3 страница

 

 

180 Sentence-meaning and prepositional content

have taken over many of the functions of the Old English sub­junctive as part of a process which has been going on for centu­ries and has made Modern English, in this respect as in others, morphologically more analytic (or periphrastic) and less syn­thetic for inflecting). A similar long-term process has been tak­ing place in other Germanic languages and in the Romance languages, though most of these still have a somewhat richer sys­tem of verbal inflections than Modern English. One of the conse­quences of this, as we shall see, is that it is much easier to objectify and propositionalize the inherently expressive and sub­jective, non-propositional, components of the meaning of utter­ances in English than it is in many other languages.

The point that I have been making here about the need to dis­tinguish sentence-type from mood is of more than purely termi­nological interest. As we shall see in Part 4, this distinction can be seen as supporting a tripartite analysis of the logical structure of both sentences and utterances in preference to the bipartite, or even unitary, analysis favoured by many logicians and formal semanticists. Terminology is, in any case, especially important in this area of semantics, since it helps us to keep apart, not only sentence-type and mood (which are frequently confused even by linguists), but also form and function.

As was mentioned earlier, some terms, such as 'conditional', are traditionally used to label one of the moods in certain languages (e.g., in French or Italian), as well as being used more generally to label sentences (typically complex, but in some cases compound or paratactie) which are used, character­istically, to express composite propositions, or implications (see 6.4). Whenever one employs such terms, one must be careful not to confuse either the formal with the functional or the nar­rower with the broader formal category. To take the French or Italian so-called conditional mood, for example: on the one hand, it does not occur in all conditional sentences, but only in that subclass of conditional sentences which are used character­istically to express counterfactual conditional propositions (and it occurs in the main clause, rather than the subordinate, condi­tional, clause); and, on the other, it has other functions in addi­tion to its use in conditional sentences. One of these uses, which

6.6 Sentence-type, clause-type and mood 181

is of particular interest in the context of a discussion of the need to distinguish form from function and sentence-type (or clause-type) from mood, is in declarative sentences to express a particu­lar kind of subjective epistemic modality, comparable with that expressed by what is called the evidential mood in the many languages throughout the world that have such a mood (e.g., in addition to the Siouan family referred to above, Turkish and Bulgarian).

Much the same point that I have made about the term 'condi­tional' can also be made about 'subjunctive' and 'optative', which are sometimes used in philosophical and logical seman­tics, in contrast with 'indicative', with reference to function rather than form or to sentence-type rather than mood. Having made this point, in the remaining sections of this chapter I will let the term 'declarative sentence' (abbreviated as 'declarative') stand for 'indicative declarative sentence'. This is how it is usually interpreted in recent work in linguistic semantics. The important thing to remember is that in many languages there are also various kinds of non-indicative declarative sentences.

We must now return briefly to the question of prosodic struc­ture. In this section (and throughout this book), we have opted for the view that the classification of sentences (and clauses) by type is wholly a matter of their grammatical structure, in both the written and the spoken language. It has already been noted, however, that, in normal conversation, spoken utterances, in all languages, are punctuated and modulated - i.e., invested with various kinds of subjective, non-propositional, meaning -by superimposing upon the string of forms of which they are composed a particular prosodic contour (see 1.3). In speech, the grammatical structure and the prosodic structure of utter­ances are generally complementary and mutually supportive, but, as we shall see presently, they may also be in apparent con­flict. For example, a declarative sentence may be uttered ironi­cally to express a proposition that contradicts the proposition which, taken at face-value, it purports to express (e.g., That's a clever thing to do!); an interrogative sentence may be uttered to make, indirectly, a statement of the kind that is traditionally (and somewhat misleadingly) referred to as a rhetorical question

182 Sentence-meaning and proportional content

(e.g.. Who could possibly think that such negotiations would bring lasting peace to the region!}.

We shall look at some of these apparent conflicts between sen­tence-meaning and utterance-meaning in Part 4. The point being made here is the more general one, that in speech the pro-sodic (and paralinguistic) structure of the utterance would nor­mally resolve the apparent conflict or contradiction. The fact that we have excluded prosodic structure from sentence-structure (and that we have therefore drawn a distinction between sentence-meaning and utterance-meaning, for >both the written and the spoken language, where it has been drawn) is well motivated from a methodological point of view. It does not follow that in drawing the distinction in this way and at this point, we are providing a realistic analysis of the production and interpretation of utterances.

Having established the distinction 'between sentence-types (and clause-types) and mood and having noted that not all declarative sentences (or clauses) are in the indicative mood (in all languages that have such a mood), we shall now move on to consider the relation between interrogatives and declaratives.

6.7 THE MEANING OF INTERROGATIVE AND DECLARATIVE SENTENCES

It is generally recognized that sentences other than declaratives present problems for truth-conditional theories of sentence-meaning. In this section, we shall be concerned with'one class of non-declaratives, namely interrogatives, and shall be comparing them semantically with declaratives. In the following section we shall look at two other classes of non-declarative sentences, drawing upon the points made here and introducing others. The general conclusion towards which we are proceeding is that not even declarative sentences are fully analysable semanti­cally in terms of a standard truth-conditional theory of meaning. In English, as in many other languages, there are two gram­matically distinct subclasses of interrogative sentences, which can be put into correspondence (by means of the notion of characteristic use and face-value meaning) with two subclasses

6.7 The meaning of interrogative and declarative sentences 183

of questions: yes-no questions and what I will call x-questions.

We shall restrict our attention initially to what may be referred to, derivatively, as yes-no interrogatives, such as

(44) 'Is the door open?'.

This is systematically related, in terms of its grammatical and lexical structure, to the declarative sentence

(45) 'The door is open'.

And the systematic grammatical and lexical relation between the two would seem to reflect a no less systematic semantic relation. But what is the nature of this semantic relation? Intuitively, it would seem that they share much, if not all, of their prepositional content, but differ with respect to the totality of their sentence-meaning.

There are several ways of assigning truth-conditions to (44), such that both the similarity and difference between its meaning and that of (45) are systematically accounted for. One is to say that it has the same meaning as

(46) 'I ask whether the door is open'.

But this is readily shown to be unsatisfactory. First of all, it seems clear that the meaning of (44) is independent of its being used to ask a question. For example, there is nothing illogical or con­tradictory about the utterance

(47) Is the door open? - that is a question which I refuse to ask.

And yet there should be if (44) and (46) have the same meaning. Secondly, if we adopt this approach, we are presumably com­mitted to the view that the meaning of the grammatically com­plex sentence (46) is simpler than the meaning of the grammatically simple sentence (44). This is in ifself counter­intuitive; and it is in conflict with the principle of compositional-ity (which was mentioned in Chapter 4 and will be discussed with reference to sentence-meaning in section 7.2). But, to make matters worse, we also have to reckon with the fact that the subordinate clause which operates as the complement, or direct object, of the verb 'ask' in (46), is generally regarded as

184 Sentence - meaning and propositional content

being grammatically comparable with the that -clause which operates as the complement of the verb 'say' in

(48) 'I say that the door is open',

the former, whether the door is open, being related to, and perhaps derived from, (44) in exactly the same way as the latter, that the door is open, is related to (45). But it is generally agreed that the truth-conditions of (48) are clearly different from the truth con­ditions of (45). And there is no good reason to challenge this con­sensus, especially as (i) English is, in this respect, by no means untypical of languages which grammaticalize the distinction between so-called direct-discourse and indirect-discourse constructions and (ii) there are many languages which do not have indirect-discourse constructions but few, if any, that do not have direct-discourse constructions. It is clearly unsatisfac­tory to treat indirect-discourse constructions as more basic and grammatically simpler than direct-discourse constructions.

A third, and conclusive, reason for rejecting the view that (44) and (46) - and (45) and (48) - are truth-conditionally equivalent is that acceptance of this view presupposes that we have a satisfactory and independently motivated truth-conditional analysis of (46) - and of (48). But, as we shall see in Chapter 8, it is only when (46) and (48) are given a special performative interpretation (and have a particular aspectual meaning) that they can be said to be semantically equivalent to (44) and (45), respectively. The performative analysis of sentences (in contrast with the performative analysis of some or all kinds of utterances), though favoured by several of the so-called generative semanticists in the early 1970s, has now been universally rejected on both grammatical and semantic grounds. Another way of accounting for the meaning of interrogative sentences such as (44) within the framework of truth-conditional semantics is by identifying it, semantically, with the set of declaratives, including 'The door is open', that may be used correctly or acceptably to answer it when it is uttered to ask a question. This approach to the semantic analy­sis of interrogatives has been adopted, and developed with great subtlety, in much recent work in formal semantics. All

6.7 The meaning of interrogative and declarative sentences 185

that needs to be said about it here is that, whatever its advan­tages from a purely logical point of view, it is hardly the approach that would be chosen by someone working in linguis­tic semantics who was not determined, for metatheoretical reasons, to force the whole of sentence-meaning into a truth-conditional straitjacket.

Much more attractive is the view taken by Gottlob Frege, the German scholar whose seminal work on the philosophy of language in the late nineteenth century has been of central importance in the formalization of semantics. According to Frege, and his present-day followers, the meaning of 'Is the door open?' is composed of both a propositional and a non-propositional component. The propositional component, "The door is open", it shares with 'The door is open'; the non-propositional component is that part of its meaning by virtue of which it is used, characteristically, for questions rather than statements. But 'The door is open' also has a non-propositional component, namely that part of its meaning which makes it appropriate for uttering statements. Frege's formulation was slightly different from the one that I have just given, partly because he did not distinguish between sen­tences and utterances - or indeed, at times, between sentences, clauses and propositions ('Satz' in German covers all three). But my formulation preserves the substance of Frege's and adjusts it, terminologically and conceptually, to the broader notion of meaning adopted in this book.

Frege's view, which does not require us to assign truth-conditions to non-declaratives, saves the appearances. For the appearances, across a large sample of the world's languages, certainly suggest that the meaning of corresponding open declaratives and interrogatives of the kind exemplified by 'The door is open' and Ts the door open?' respectively can be factorized into two parts. Generally speaking, in languages in which there is a clearly identifiable distinction between declaratives and interrogatives, the latter differ from the former in one of three ways: by a difference of word-order, by the occurrence of a special interrogative particle, or by morphological variation in the verb. It is sometimes said

186 Sentence-meaning and prepositional content

that there is another way of distinguishing declaratives and interrogatives: by means of intonation.

On the view taken here and made explicit above, however, this kind of intonational difference, which in many languages distin­guishes questions from statements, should be attributed, not to the structure of sentences, but to the process and products of utterance. This means that there are languages (e.g., Italian, Spanish, Modern Greek - to name but a few of the more familiar European languages) in which there is no difference, at the sen­tence-level, between declaratives and yes -no interrogatives. The difference between statements and yes -no questions is normally marked prosodically in speech and by punctuation in writing.

Sentences that are grammatically neutral with respect to the distinction between declaratives and interrogatives (but can be used appropriately in the utterance of either statements or ques­tions) are the only sentences whose meaning may be exhausted by their prepositional content. (Whether even, such sentences, in Italian. Spanish, Modern Greek, etc., can be said to be wholly devoid of non-propositional meaning depends on the way the grammatical categories of tense and mood are handled semanti-cally: see 10.3, 10.5.) Sentences whose grammatical structure marks them as either, declarative or interrogative have as the non-propositional component of their meaning an indication of their potential for use, characteristically, with one communica­tive function rather than another: that of making statements, on the one hand, or of asking questions, bn the other. And in many languages the grammatical structure of such sentences is often readily analysable into a prepositional and a non-proposi­tional part. As we shall see in Chapter 7, several versions of transformational grammar, including the earliest version developed by Chomsky (1957) and subsequently adopted (with mod­ifications; by Katz and Postal (1964), have exploited this fact.

So far we have discussed only neutral, or unmarked, yes-no interrogatives: i.e., interrogatives which do not encode, grammatically or lexically, the speaker's presuppositions or expecta­tions with respect to the addressee's response. Non-neutral, or marked, interrogatives differ from neutral interrogatives in that they do encode such information. For example, so-called

6.7 The meaning of interrogative and declarative sentences 187

tag-interrogatives — more precisely, reversed-polarity tag-interrogatives - in English, such as

'The door is open, isn't it?'
and

'The door isn't open, is it?'

encode the speaker's expectation that the question will be answered in the affirmative or the negative, respectively: i.e.. that, when these sentences are used with their characteristic function of presenting a proposition to an addressee and asking him or her to assign a truth-value to the proposition presented, by using these marked, non-neutral, constructions speakers (a) indicate (whether sincerely or not) what they themselves con­sider the truth-value to be and (b) in the tag explicitly seek the addressee's agreement or confirmation. Thus (49) would be used, characteristically, to present the proposition "The door is open" as one to which the speaker is disposed to assign the value true and (50) would be used, characteristically, to present the same proposition ("The door is open") as one to which the speaker is disposed to assign the value false or, alternatively, to present the corresponding negative proposition ("The door is not open") as one to which he or she is disposed to assign the value true. Many languages (including Latin) have distinct marked, or non-neutral, yes - no interrogatives, which are semantically, if not grammatically, comparable with (49) and (50).

Let us now turn to x -interrogatives. In English these contain one of a set of interrogative forms, adjectives, pronouns or adverbs, including who/whom, what, which, when, where and how. (Since all of these, except how, in their written form begin with wh-, the sentences that contain them are often referred to as wh-sentences. And the terms ' wh -sentence' and ' wh -question' are often extended to the description of languages other than English.)

The reason for calling such sentences x -interrogatives is almost self-evident. Looked at from the point of view of their logical structure, they can be thought of as sentences which con­tain a restricted variable (x) in their prepositional component,

 

 

188 Sentence-meaning and propositional content

for which, when such sentences are used to ask a question, the addressee is invited to supply a value falling within the range of the variable. For example, 'who' in the form who or whom restricts the value of x to persons (of which the prototypical exemplars are human beings). Thus

(51) 'Who has been eating my porridge?',

when used to ask a question, solicits from the addressee an answer which will identify the person who has been eating the speaker's porridge, by supplying as the value of x an appropriate referring expression, such as 'Goldilocks', or 'the little bear from next door', or 'the person who left these footprints on the path', or 'whoever it was who saw us going out this morning'. As always, reference is context-dependent: it is determined, first of all, by the speaker's general ontological beliefs and assump­tions and, then, by his or her more specific background beliefs and assumptions relevant to the particular context of utterance and often acquired in the course of the particular conversation to which the utterance contributes and of which it constitutes a part. So too. and for the same reasons, is the range of the restricted variable in the propositional content of x -questions.

But what is the propositional content of (51)? It is intuitively clear that the x -interrogative (51) is closely related semantically to

(52) 'Someone has been eating my porridge',

which differs from (51) formally in that it has the indefinite pro­noun 'someone', rather than the interrogative pronoun 'who' in subject position. Looked at from a logical point of view, 'some­one' can be thought of as a free (or unbound) restricted variable whose range is the same as that of the interrogative pronoun 'who'. To say that it is a restricted variable, as we have noted above, is to say that it does not range over all the entities in the universe of discourse, but over a (proper) subset of these: in the present case, entities that are (more precisely, are assumed or presumed to be) persons - entities that belong to the class {x: x is a person}. To say that a variable is free is to say that it is not bound - its reference is not fixed within its range - either by a

6.7 The meaning of interrogative and declarative sentences 189

logical operator (such as the universal or existential quantifier) or otherwise. In standard systems of logic, formulae which con­tain free variables are not regarded as propositions, but as pro-positional functions: they are converted into propositions either by binding the variables they contain or by substituting for them constants, whose reference is fixed (within any given uni­verse of discourse).

The logical distinction between bound and free variables and its correlates in natural languages have been of immense impor­tance recently, not only in logical and linguistic semantics, but also in grammatical theory. This is why it has been explained here, where its applicability is especially easy to appreciate. We shall be exploiting it later, as we shall also be exploiting the dif­ference between propositions and propositional functions in our discussion of reference (10.1).

But we have still not established the nature of the semantic relation between (51) and (52). It is obviously not the same as that which holds between (44) and (45), since (52) has its own yes-no interrogative. In fact, it has two:

(53a) 'Has someone been eating my porridge?'

(53b) 'Has anyone been eating my porridge?'.

What difference there is, semantically, between (53a) and (53b) is difficult to determine: the some/any distinction which exists in English is notoriously controversial and will not be dealt with in this book. In any case, it is not directly relevant to the point at issue. For present purposes, let us simply agree that (53b) is the normal yes-no interrogative which corresponds with (52) -when (52) is also being used normally - in the same way that (44) corresponds with (45). It follows that (53b) has the same, propositional content as (52). But so too, apparently, has (51).

The difference between (53b) and (51) - more generally, between yes—no interrogatives and x -interrogatives - has to do with the scope of the interrogativity that is encoded in them and with what are commonly referred to as the presupposi­tions of the questions that the two subclasses of interrogatives are (characteristically) used for. In (53b), as in (44), the whole of the propositional content is within the scope of the

 

 

190 Sentence-meaning and prepositional content,

interrogativity; and, if either of these sentences is used to ask a question (unless there is some contextual, or in speech prosodic, limitation of scope), it will be the proposition expressed by the corresponding declarative (uttered as a straightforward, un­qualified, statement) that is queried. And in uttering (53b) or (44), in these circumstances, the speaker gives no indication of his or her presuppositions as to the truth or falsity of the proposition expressed. In (51), in contrast, it is only part of the prepositional content that is within the scope of the interrogativity. In uttering (51) to ask a question, in normal circumstances, the speaker takes for granted, or presupposes, the truth of the proposition that would be expressed by the utterance of (52) in the same context and, by using the pro­noun 'who' in what might be referred to as the x -position, focuses upon the identity of the person referred to by 'someone'.

Many different kinds of presupposition have been recognized by logicians and linguists; and it is not clear how they relate to one another and to different kinds of implication. We shall return to this question in Part 4. What has been said here about presupposition (and scope) is relatively informal and theory-neutral. It also applies to the full range of x -interrogatives that is found in English (and in other languages), not only pronom­inal, but also adjectival and adverbial.

At this point, it is important to note that formally and to some extent functionally there are overlaps and parallels in many languages, not only between x -interrogatives and declaratives containing indefinite pronouns, adjectives and adverbs, but also between x -interrogatives and declaratives containing demonstrative and relative pronouns, adjectives and adverbs. It must also be added that in many, if not most, of the lan­guages of the world, it is impossible to identify all of these as grammatically and semantically distinct constructions. We must be careful, therefore, not to assume that every natural language grammaticalizes differences and equivalences of sentence-meaning in exactly the same way.

In this section we have concentrated upon the meaning of interrogative sentences in relation to that of declarative sen­tences. We have seen that, not only interrogatives (as one sub-

6.7 The meaning of interrogative and declarative sentences 191

class of non-declaratives), but also declaratives, grammaticalize a non-propositional component of meaning, which expresses their characteristic use (as does that of interrogatives and other non-declaratives) and combines this with their propositional content and, in certain languages more obviously than in Eng­lish, with yet another component of sentence-meaning expressed ^; by mood. We have also noted that, although it is presumably possible to make statements and to ask questions in all languages (though not necessarily statements and questions that are purely neutral, or unmarked, in terms of modality), there are languages Which do not grammaticalize the distinction between declara­tives and interrogatives.

Interrogativity has been dealt with here as a property of sen­tences which is distinct from, but may: combine, with, mood (indicative, subjunctive, etc.) in those languages that have such a grammatical category. This is certainly the way it should be dealt with in the grammatical and semantic analysis of the Indo-European languages and many other languages through­out the world. In other languages, however, interrogativity may well be grammaticalized in one of the moods. Whether, and to what degree, this is the case is difficult to establish.

One reason for this difficulty is that it is hard to draw a functional distinction (unless the language itself clearly grammatica­lizes or lexicalizes the distinction) between asking a question and expressing doubt. There are several American-Indian languages (including Menomini, Serrano and Hidatsa) which have what is traditionally called a dubitative mood; and the use of the term 'dubitative' implies that grammarians describing these languages have decided that the characteristic, if not the sole, function of the mood so labelled is that of expressing the speaker's doubt. But if speakers express doubt as to the truth of a particular proposition, in conversation rather than in solilo­quy, they may well be understood in context (and expect to be understood) to be inviting the addressee to resolve their doubt for them: i.e., to be asking (and not merely posing) a question.

Conversely, of course, a sentence whose characteristic function is deemed to be that of asking questions - and which is for that reason said to be interrogative (either in sentence-type or

194 Sentence-meaning and prepositional content

traditionally called exclamations. Let us begin with exclamative sentences.

In English, and many other-languages, there is a structural similarity between exclarnative sentences and dependent inter­rogative clauses. For example,

(58) 'How tall he is'

has the same structure, at least superficially, as the subordinate

clause in

(59) 'I wonder how tall he is'.

Functionally, however, there is a clear difference between excla-matives of the kind exemplified by 'How tall he is' and interrog-atives. In fact, exclamatives of this kind are best seen, semanti-cally, as a subclass of expressive declaratives, in which the non-propositional part of what distinguishes the meaning of 'How tall he is' from the meaning of

(60) 'He is very tall'

is grammaticalized, rather than being expressed, in utterance, by a particular prosodic contour. It is because it is grammatica­lized and is correlated with systematic restrictions on polarity, the use of modal verbs, etc., that 'How tall he is' is rightly regarded by grammarians as an exemplar of a distinct sen­tence-type. It is, of course, important not to confuse exclama­tives with exclamations. Sentences of all types may be uttered with that particular expressive modulation which is conveyed in the spoken language by stress and intonation, and in the writ­ten languages by means of the exclamation-mark. Exclamation is something very different from making statements, issuing commands and requests, and asking (or posing) questions. Let us now turn to imperatives.