by Daniel von Wachter*
Preprint version of: 1999, "What has Necessity to do with Analyticity?", Metaphysics in the Post-Metaphysical Age, ed. U. Meixner and P. Simons, Wien: Verlag Hölder-Pichler-Tempsky, pp. 326-330.
Author:
Dr. phil. Daniel von Wachter
Oriel College, Oxford, OX1
4EW, U.K.
Email: Daniel.Wachter@oriel.ox.ac.uk
The claim that analytic statements are necessary is as old as the concept of analyticity and seldom taken to be contentious. The claim that necessary statements are analytic is a bit younger and usually taken to be a bit more contentious, but it gained a remarkably wide following. Both claims are sometimes not really claims but rather parts of stipulative definitions of ‘necessary’ or ‘analytic’. There is no totally non-arbitrary philosophical use of the word ‘necessary’, and how philosophers use it often depends on how they want to relate necessity to analyticity. What we should try to do is to think about what phenomena there are in that area of objects of philosophical investigation where philosophers speak about ‘necessity’ and ‘analyticity’. That is, we have to investigate what the matter is with these statements which philosophers call ‘necessary’ or ‘analytic’. Then we have to look for a fitting classification whereby we should be anxious that we do not put too different phenomena in the same class.
In what follows I want to make suggestions about how the words ‘necessary’ and ‘analytic’ are fittingly used in philosophy. In the end, I shall reconsider the question whether analytic statements are necessary, and whether necessary statements are analytic.
1. Analyticity
Let us start with the worn-out example ‘All bachelors are unmarried’, which is the most often quoted example of an analytic statement. There is, of course, something peculiar about this statement. We should investigate what it is, and if it is not something totally different from what philosophers usually meant when they spoke about ‘analyticity’, then we will be well advised to design our concept of analyticity such that it captures the particular peculiarity of the statement ‘All Bachelors are unmarried’.
To make it short, I suggest that the peculiarity of this statement should be accounted for as follows. The word ‘bachelor’ in English is used to say about something that it is a man who has never married or to refer to objects which are men who have never married. So it is part of the stipulative (nominal) definition of the word ‘bachelor’ that bachelors are unmarried. The only thing which is stated by the statement that bachelors are unmarried is that the word ‘bachelor’ has such and such a meaning. Several things which Kant said about analytic statements are true here: The concept of being unmarried is already contained in the concept of a bachelor. To quote Kant ‘Nothing is said in the predicate which is not already thought in the subject-concept’. That is the reason why there is a sort of ‘redundant predication’ (Katz 1998, 557). A.J. Ayer’s way of describing the peculiarity of this statement is to say that its truth depends solely on the definitions of the symbols it contains’ (Ayer 1936, 73). Note that the word ‘solely’ here is crucial. Of course all statements depend in their truth on the definitions of the words, but what is peculiar about the statement ‘All Bachelors are married’ is that it is made true solely by the fact that the words have the meaning they have. I think that is the right analysis of the peculiarity of the statement ‘All bachelors are unmarried’, and I think all this is close enough to what is usually said about analyticity so that we can coin a corresponding definition of analyticity. I propose that:
A statement is to be called ‘analytic’ if and only if it is made true solely by the linkage of a certain word to its meaning, and if it does not explicitly have the form ‘A means B’.
This yields us a criterion for analyticity. To deny a true analytic statement is, in effect, to deny falsely that a certain word has a certain meaning. That is, to deny a true analytic statement is to commit a mistake of language.
2. Necessity
Our next task is to approach the concept of necessity and the connected modal concepts, i.e. impossibility, possibility, and contingency. What is necessity? What should we call ‘necessary’? There is no one non-controversial paradigm example of necessity. We can start from some very basic intuitions. What is necessary is, for some reason, such that it could not possibly be otherwise. What is necessarily not the case is such that it is not only not actually the case, it is even somehow excluded that the course of history could have gone such or will go such that it would be the case. Some things would obviously be otherwise had history gone otherwise. There could be unicorns, had history gone otherwise, and NATO would not have intervened in Kosovo in 1999 had Slobodan Milosevic acted differently. To say that something is necessarily so-and-so is, at least, to rule out that the course of history at any time goes such that it comes otherwise.
It is helpful to look at statements with which we make claims about something being necessarily so-and-so. Let us consider the statement ‘Nobody can be guilty for something he did not do freely’. By ‘guilt’ we mean that thing about which our conscience occasionally tells us that we have it. By somebody ‘being free’ we mean that the person has it in his power to choose between alternative actions and to perform the one he chooses. Or, more precisely, somebody does something freely if and only if his undertaking the action is not fully caused by earlier events. With the statement ‘Nobody can be guilty for something which he did not do freely’ one claims that there is a certain thing which cannot occur without a certain other thing. We understand what somebody is speaking about when he says about somebody that he is guilty for something. We also understand what somebody is speaking about when he says about somebody that he did something freely. And we also understand what somebody claims if he claims that somebody is guilty for something he did not do freely. But some of us believe that any such a claim is false because, we believe, one cannot be guilty for something one did not do freely.
Let us consider one more example, say the statement ‘Nothing can cause something which took place earlier’. What do we mean by ‘cause’? We take cases like somebody throwing a stone into a window which then breaks, or one billiard ball hitting another, as paradigm cases of causation. We believe that there is something special about such ordered pairs of events or things. We say in such cases one thing brings about the other, or one thing ‘causes’ the other. Ordered pairs of things are such that one of the things is rightly called the cause of the other if and only if these pairs resemble our standard cases in the relevant respect. If somebody tells us that a window in his house broke because somebody had thrown a stone into it, we understand that this is a claim about a case of causation and we believe that what we are told might have happened. If somebody tells us that somebody on Wednesday performed a rain dance to bring about rain on the Monday before at some other place about which nobody knows whether it was raining there on Monday, then we also understand that this is a claim about a case of causation. But some of us believe that any such claim is false, not only because we do not believe that dances are the sort of thing which brings about rain, but because we believe that backward causation is impossible. To claim that backward causation is impossible is to claim that there cannot be ordered pairs of events or things where one is cause of the other but is later than the other.
Examples like the impossibility of guilt without freedom or the impossibility of backward causation, seem to me to be examples of the sort of thing for which it is fitting to use the term ‘necessity’ in philosophical discourse. They are claims about something being necessarily so-and-so, they are not just claims about some hypothetical or conditional necessity like ‘If you want to have healthy teeth it is necessary that you brush your teeth regularly’, they makes sense, they are easy to understand, they are interesting, and they are even about philosophically relevant issues. So I think we are well advised if we design our philosophical concept of necessity such that it captures what we mean in such examples.
As we understand such claims about impossibility or necessity quite easily and we have a good grasp of the meaning of talk about ‘necessity’ here, there is perhaps no need for a further going definition of the term ‘necessary’. But it would be nice if we could say something more general about necessity, or if we could say something about the source of necessity. Here is an attempt to say more about the source of necessity: We human beings have the ability to have views about things. One essential part of this is our ability to recognise and to refer to causal aspects, or ‘properties’, of a thing. We form predicates with which we can refer to properties and therefore we are able to describe things and situations. We can not only describe them, we can conceive of them, we can consider them. In that sense we can say that we can ‘construe things in our mind’. For any set of predicates, as long as these predicates – like in our examples – are not explicitly defined in terms of each other like ‘bachelor’ is defined in terms of ‘being unmarried’, we can consider the existence of a thing or a situation to which they all apply. If we have the predicates ‘A’, ‘B’, and ‘C’, we can consider that there is a thing which is A and B and C. Predicates can be combined arbitrarily. But, I suggest, we have no reason to assume that the objects of these predicates are to be combined equally arbitrarily. Why should all properties be combinable with each other. (Cf. Armstrong (1989), who holds that all properties are combinable.)
We can express the belief that the objects of the predicates ‘A’ and ‘B’ are not combinable by saying that it is impossible that there is something which is A and B. We can construe all sorts of things or situations in our mind, and the world may or may not allow for the existence of such a thing or situation. We can consider and claim that there is somebody who is guilty for something which he did not do freely, but presumably it is impossible that there is such a thing.
We may say then that every non self-contradictory description of a thing or a situation is such that either necessarily there is such a thing, or necessarily there is no such thing, or contingently there is such a thing, or contingently there is no such thing.
3. Necessity and Analyticity
Now we are in a position to reconsider the claim that analytic statements are necessary and the claim that necessary statements are analytic. An analytic statement is one which is made true solely by the linkage of a certain word to its meaning. I linked the concept of necessity to the phenomenon that we can construe all sorts of things in our mind, some of which the world might not allow for. A necessary statement is one where the existence of a thing or situation is described and it is claimed that the world does not allow for the existence of such a thing. A necessary statement is made true by the fact that the world does not allow for the existence of a thing or situation of a certain type. It is made true by the fact the the objects of the predicates involved are not compatible in the way in question. But that means that it is never made true by the linkage of a certain word to its meaning, because any word could have any meaning. So analytic statements are not necessary.
Are necessary statements analytic? No, they are not, because they are not made true by the linkage of a certain word to its meaning. They are made true by the objects of the predicates involved. Further, to contradict an analytic statement is to commit a linguistic mistake. To contradict a true necessary statement does not involve a linguistic mistake – it is a mistake about what is possible.
One might draw the conclusion from this result that the concepts of analyticity and necessity from which I derived this result must be inadequate. I do not think they are. These concepts of analyticity and necessity capture what is essential about cases which are very good candidates for being paradigm cases for analyticity or necessity respectively.
Of course, one may widen the concepts such that one can say at least that analytic statements are necessary. As I said, there is no totally non-arbitrary usage of the word ‘necessary’. Or one may use ‘necessary’ in the sense of ‘analytic’. But then the term ‘necessary’ would be quite useless, because we have already the term ‘analytic’, and we would need another term for what I called ‘necessary’. Besides that would this usage of ‘necessary’ be far away from our normal talk about ‘necessity’. That does not seem advisable. And I think if one widens the concept of necessity such that it covers also analytic statements, then the concept of necessity becomes too much a mixed bag. It is true, both, analytic and necessary statements are in some way immune against being false. But the reason for that lies in the one case in linguistic conventions and in the other case in the objects of the predicates. Necessary statements are claims that such-and-such a thing cannot exist. They are claims that something which can be described consistently in fact cannot exist. Analytic statements, on the other hand, are true by stipulative definition. They are about linguistic conventions. These conventions could be different. There is nothing necessary about analytic statements.
Armstrong, D.M. (1989), A Combinatorial Theory of Possibility. Cambridge University Press.
Ayers, A.J. (1936), Language, Truth and Logic. London: Penguin, 1971.
Katz, J. (1998), "The Problem in Twentieth-Century Philosophy". Journal of Philosophy 95: 547-575.