Statement of the problem

Natural languages contain many expressions which are grammatically well-formed but meaningless; they are assembled from meaningful words or morphemes in accordance with the syntactic rules of the language but no meaning is conferred upon them by the semantic rules of the language. When we call expressions or utterances ‘meaningless’ here without further qualification, that will just be for the sake of brevity. We want to indicate by that term that the expressions or utterances are semantically anomalous in such a way that they will generally evoke responses like ‘What do you mean?’ or ‘What are you talking about?’ There is no implication that they are on a par with totally meaningless expressions as Krz is thwing.

Let us take six typical examples of such well-formed but meaningless expressions:

(1) Incompletely defined functors: Many predicates are not defined for all syntactically permissible arguments. Thus the verb to run is defined for animals with locomotive appendages, for humans, machines, fluids and for noses, not however for plants, minerals or numbers. And the German verb lachen is defined only for humans and the sun. The sentence Der Mond lacht, though constructed grammatically just as Die Sonne lacht, has, in distinction to the latter, no meaning.

(2) Non-existing objects: Sentences about objects which do not exist or no longer exist form a significant sub-category of example (1). The sentences Odysseus is (now) shaving himself and Eisenhower is (now) sick are meaningless but not the sentences Professor Snell is dreaming of Odysseus or Nixon remembers Eisenhower.

Purpose

In this paper I want to present the general framework of a semantic theory which can describe partial meaning relations between sentences of natural languages; relations such as partial synonymy, partial consequence, partial incompatibility. I will call such a theory henceforth semantics and sometimes also grammar. Three features are important for the kind of semantics proposed here: first, the theory only makes use of transformational rules (henceforth: T-rules) which are stated within a formalism that is a modified and generalized version of the formulation of Ginsburg and Partee (1968) and Brockhaus (1972). Second, logical systems developed within the framework of predicate logic can be embedded into this theory without difficulties. The system can be connected with a model theory. And third, the theory lends itself to later justification from speech act theory.

I mention these points explicitly because I think that a semantics of natural language should be conceived in such a way as to make use of the results of three important contemporary linguistic schools: transformational grammar as worked out by Harris, Chomsky and others; formal logic in the widest sense including model theory, and speech act theory (Austin, Searle and others).

Nevertheless, a semantic theory cannot simply be a chaotic mixture of these or other different approaches, but the goals of such a theory should be determined independently in a clear way.

In this paper I discuss the goals of a semantic theory, and then propose a criterion of adequacy.

I would like to discuss two aspects of pragmatics that in recent years have been treated very differently: indexicals and conversational implicatures. Montague and Scott proposed to handle indexicals by adding to points of reference (sometimes called ‘indices’) extra coordinates for speaker, hearer, time and place of utterance. This proposal places indexicals among those phenomena to be dealt with by formal logic, and such systems have in recent years been articulated by Lewis and Kamp, among others. Implicatures on the other hand, were taken by Grice to be by nature informal inferences of a fundamentally different kind than logical inferences, and hence not to be dealt with by the apparatus of formal logic. In other papers I have dropped hints to the effect that indexicals and implicatures should be treated somewhat differently than they are in the Montague-Scott and Grice proposals. I would like to elaborate a bit on those hints.

The basic suggestion is this:

1. (I) If the goals of what I have called natural logic are adopted, then it should in time be possible to handle indexicals without any extra coordinates for speaker, hearer, and time and place of utterance, and it should also be possible to handle implicatures without any kinds of extralogical inference.

The basic ingredients of the suggestion are as follows:

1. (A) The so-called performative analysis for imperatives, questions, statements, promises, etc.

2. (B) The limitation of points of reference to assignment coordinates for variables and atomic predicates. […]

One of the oldest of metaphysical distinctions is that between fact and value. Underlying the belief in this distinction is the perception that values somehow derive from persons and cannot lie in the world, at least not in the world of stones, rivers, trees, and brute facts. For if they did, they would cease to be values and would become simply another part of that world. One trouble with the distinction in the history of philosophy is that there have been many different ways of characterizing it, and they are not all equivalent. Hume is commonly supposed to have been alluding to it in a famous passage in the Treatise where he speaks of the vicissitudes of moving from “is” to “ought”. Moore saw the distinction in terms of the difference between “natural” properties like yellow, and what he called “non-natural” properties, like goodness. Ironically, Moore's successors, reversing the usual order of metaphysical progression, have read this metaphysical distinction back into language as a thesis about entailment relations in language. So construed it is a thesis that no set of descriptive statements can entail an evaluative statement. I say “ironically” because language, of all places, is riddled with counter-instances to the view that no evaluations can follow from descriptions. As we saw in chapter 6, to call an argument valid is already to evaluate it and yet the statement that it is valid follows from certain ‘descriptive’ statements about it.

In this chapter we shall attempt to complete our characterization of the illocutionary act by giving an analysis of the propositional act of predication. Predication, like reference, is an ancient (and difficult) topic in philosophy, and before attempting to give a speech act analysis of predication I shall consider certain well known theories of predication and the problems of “ontological commitment” with which they are related. I begin with Frege's account.

Frege on concept and object

In a statement made using the sentence “Sam is drunk” what if anything stands to “—is drunk” as Sam stands to “Sam”? Or is this an improper question? Frege, who assumed it was a proper question, gave the following answer. Just as “Sam” has a sense and in virtue of the sense has a referent namely Sam, so “—is drunk” has a sense and in virtue of that sense has a referent. But what is the referent of “—is drunk”? To this Frege's answer is: “a concept”. To which one's natural response would be: “which concept?” And to this the tempting answer is, “the concept drunkenness”. But clearly, as Frege sees, this answer will not do, for on that account “Sam is drunk” must be translatable or at any rate must have the same truth value as “Sam the concept drunkenness”, in accordance with a version of the axiom of identity which Frege accepts, that whenever two expressions refer to the same object one can be substituted for the other in a sentence without changing the truth value of the corresponding statement.

In this chapter and the next we shall delve inside the proposition to consider the propositional acts of reference and predication. Our discussion of reference will be confined to singular definite reference and will be to that extent an incomplete theory of reference. As we shall see that alone will provide us with plenty of problems, but until we get clear about them we are hardly likely to get clear about other kinds of reference.

The notion of singular definite reference is a very unsatisfactory one, but one we can hardly do without. The most obvious cases of referring expressions are proper names, but as soon as we consider other kinds of expressions such as singular definite descriptions we find that some of them are referring expressions, some obviously not, and some seem to fall in between. Furthermore, some occurrences of proper names are not referential, as in, e.g., “Cerberus does not exist”. Philosophers who discuss definite descriptions almost invariably fasten onto examples like “the king of France”, or “the man”, and seldom onto examples like “the weather”, “the way we live now”, or “the reason why I like beans”. This ought to arouse our suspicions. Consider for example the difficulty of applying Russell's theory of descriptions, without any paraphrases of the original, to a sentence like “The weather is good”: “(∃x) (x is a weather (y) (y is a weather → y = x) · x is good)” hardly makes any sense. Yet one is inclined to say the expression “the weather” performs a similar role in “The weather is good” to that of the expression “the man” in “The man is good”.

The hypothesis then of this work is that speaking a language is engaging in a rule-governed form of behavior. To put it more briskly, talking is performing acts according to rules. In order to substantiate that hypothesis and explicate speech, I shall state some of the rules according to which we talk. The procedure which I shall follow is to state a set of necessary and sufficient conditions for the performance of particular kinds of speech acts and then extract from those conditions sets of semantic rules for the use of the linguistic devices which mark the utterances as speech acts of those kinds. That is a rather bigger task than perhaps it sounds, and this chapter will be devoted to preparing the ground for it by introducing distinctions between different kinds of speech acts, and discussing the notions of propositions, rules, meaning, and facts.

Expressions and kinds of speech acts

Let us begin this phase of our inquiry by making some distinctions which naturally suggest themselves to us as soon as we begin to reflect on simple speech situations. (The simplicity of the sentences in our examples will not detract from the generality of the distinctions we are trying to make.) Imagine a speaker and a hearer and suppose that in appropriate circumstances the speaker utters one of the following sentences:

1. Sam smokes habitually.

2. Does Sam smoke habitually?

3. Sam, smoke habitually!

4. Would that Sam smoked habitually.

The ground has now been prepared for a full dress analysis of the illocutionary act. I shall take promising as my initial quarry, because as illocutionary acts go, it is fairly formal and well articulated; like a mountainous terrain, it exhibits its geographical features starkly. But we shall see that it has more than local interest, and many of the lessons to be learned from it are of general application.

In order to give an analysis of the illocutionary act of promising I shall ask what conditions are necessary and sufficient for the act of promising to have been successfully and non-defectively performed in the utterance of a given sentence. I shall attempt to answer this question by stating these conditions as a set of propositions such that the conjunction of the members of the set entails the proposition that a speaker made a successful and non-defective promise, and the proposition that the speaker made such a promise entails this conjunction. Thus each condition will be a necessary condition for the successful and non-defective performance of the act of promising, and taken collectively the set of conditions will be a sufficient condition for such a performance. There are various kinds of possible defects of illocutionary acts but not all of these defects are sufficient to vitiate the act in its entirety. In some cases, a condition may indeed be intrinsic to the notion of the act in question and not satisfied in a given case, and yet the act will have been performed nonetheless.

The philosophy of language

How do words relate to the world? How is it possible that when a speaker stands before a hearer and emits an acoustic blast such remarkable things occur as: the speaker means something; the sounds he emits mean something; the hearer understands what is meant; the speaker makes a statement, asks a question, or gives an order? How is it possible, for example, that when I say “Jones went home”, which after all is in one way just a string of noises, what I mean is: Jones went home. What is the difference between saying something and meaning it and saying it without meaning it? And what is involved in meaning just one particular thing and not some other thing? For example, how does it happen that when people say, “Jones went home” they almost always mean Jones went home and not, say, Brown went to the party or Green got drunk. And what is the relation between what I mean when I say something and what it means whether anybody says it or not? How do words stand for things? What is the difference between a meaningful string of words and a meaningless one? What is it for something to be true? or false?

Such questions form the subject matter of the philosophy of language. We must not assume that in the versions I have stated they even make sense.