From the very origins of philosophy, philosophers have been especially conscious of the inadequacies and limitations of language. But they have disagreed on whether this reflects a corresponding deficiency in thought, on what its explanation and implications are, and on how we should respond—in short, on what this itself means. ‘Do you know what I mean?’ explains that in actual cases of analysis, three different dimensions—interpretive, decompositional, and regressive—are typically involved. It also introduces G. E. Moore (1873–1958) who, alongside Russell, inaugurated analytic philosophy in Britain. Moore’s main interests lay with epistemology (the theory of knowledge) and ethics. The paradox of analysis is also discussed.
If an anthropologist from outer space were to come to Earth and listen in on our conversations, then they might be struck by the frequency with which we ask if the person we are talking to knows what we mean. We use many phrases in both spoken and written language that express uncertainty as to whether we have expressed ourselves properly, or that show recognition that we have not, or that seek for reassurance along the way as to whether we are being understood. Know what I mean? As well as ‘know what I mean’, such phrases include ‘as it were’, ‘so to say’, ‘roughly speaking’, ‘as a first approximation’, ‘somehow’, ‘sort of’, ‘kind of’, ‘et cetera’, ‘and so on’, ‘you know’, and ‘like’. (There are analogous phrases in many other languages, so the phenomenon is not restricted to the English-speaking world.) At a more sophisticated level, we also use a variety of analogies, diagrams, examples, illustrations, metaphors, and similes in an attempt to explain ourselves. We also have a range of expressions that attempt to describe this process and its intended result—‘getting the message across’, ‘understanding the gist’, ‘grasping the point’, ‘catching the drift’, and ‘cottoning on’, to mention just some.
From the very origins of philosophy (both Western and non-Western), philosophers have been especially conscious of the inadequacies and limitations of language. But they have disagreed on whether this reflects a corresponding deficiency p. 44↵in thought, on what its explanation and implications are, and on how we should respond—in short, on what this itself means. In the conversations of everyday life, we may talk loosely and inaccurately, but given time, or if we think hard enough about something, are we not bound, sooner or later, to come up with the right way to express ourselves? How many times have you thought of the perfect response to someone’s witticism or criticism, but only after the opportunity to give it has passed? And, in any case, do we not regularly re-express what someone has told us, whether to clarify what they have said or in telling someone else later? So are these not all cases in which there are clear thoughts to be grasped, the task just being to find the appropriate articulation?
On the other hand, how many times have you found someone else putting the point you were struggling to make so much more elegantly or succinctly than you could have done? And we have a variety of idioms, such as ‘That’s spot on’ and ‘You’ve hit the nail on the head’, to express this common experience. Are these not cases where someone else not merely knows what I mean but knows better what I mean? But what does this mean? Does it mean, for example, that I did not really know what I meant in such cases? Yet, if I can indeed recognize that someone else has neatly articulated what I was trying to say, then must I not have known what I meant? If I had not already grasped the thought in some form, then how could I say that someone else has captured just that thought? I do not say, for example, ‘That’s a good thought you’ve just had’, but ‘That’s precisely my thought—well put!’. Alexander Pope made the point, with poetic elegance and succinctness, in An Essay on Criticism over 300 years ago:
True wit is nature to advantage dressed,What oft was thought, but ne’er so well expressed;Something whose truth convinced at sight we find,That gives us back the image of our mind.
p. 45The same questions can be raised concerning philosophical debate; indeed, they seem to arise in an even starker form, as our discussion in Chapters 1 and 2 illustrates. When I say that there are ‘infinite’ square numbers, for example, do I really know what I mean? If I become convinced that there are ℵ0 such numbers, is that what I really meant? What do I mean when I say that unicorns do not exist? That the first-level concept of being a unicorn falls within the second-level concept of being not instantiated? Here there seems a stronger case for saying that this is what I meant than in the example of the number of square numbers. But in both cases I am being given new conceptual resources to express more accurately what it was I meant—or if not this, then maybe what it was I should have meant. And this is exactly what philosophical analysis is intended to achieve.
What does ‘analysis’ mean? If asked today, many people would say that it involves breaking something down to reveal its constituent parts and structure. This is indeed an important sense of ‘analysis’, which we can call its ‘decompositional’ sense, but it is not the only sense. The word has its origins in the ancient Greek term ‘analusis’, which first acquired a technical sense in Euclidean geometry, where it meant the process of working back to first principles or more basic figures by means of which something could then be proved or constructed—through a corresponding process of ‘synthesis’. This can be called its ‘regressive’ sense, which we also find illustrated throughout the history of mathematics, philosophy, and science. When Frege and Russell sought to ‘reduce’ arithmetic to logic, for example, they were seeking the supposedly more fundamental principles of logic (axioms, definitions, and rules of inference) by means of which to prove the laws and theorems of arithmetic. This is regressive analysis.
There is a third sense of ‘analysis’, however, which is just as important as the first two, and which is especially characteristic p. 46↵of analytic philosophy. This is ‘interpretive’ analysis, and we have already seen it illustrated in Chapters 1 and 2. Frege’s analysis of existential and number statements, for example, consists in interpreting them as assertions about concepts, and likewise, Russell’s theory of descriptions interprets sentences involving definite descriptions as expressing claims about concepts. Based on these interpretive analyses, decompositional analysis can then be used to identify the constituent elements in each case, such as the relevant concepts, whether empirical (e.g., the first-level concept of a unicorn, which might in turn be ‘decomposed’ into the concepts of a horse and of a horn) or logical (e.g., the second-level concept of being instantiated). We can also see such analyses as having a regressive dimension, in that the aim is to work back to what are seen as the more basic elements (in these cases, the relevant concepts). So it should be stressed that in actual cases of analysis, all three dimensions—interpretive, decompositional and regressive—are typically involved.
The relationship between these three dimensions of analysis will be an important theme in the rest of the book, and we will draw together some of the threads in Chapter 6. But let us focus here on interpretive analysis and consider how this raises the questions we have just asked concerning meaning. Let us take Russell’s interpretive analysis of ‘The present King of France is a philosopher’ as ‘The concept “King of France” is uniquely instantiated and whatever instantiates this concept also instantiates the concept “philosopher”’. Is this what someone means when they understand the original sentence? If asked what that sentence means, most people are highly unlikely to give you the Russellian analysis: one needs acquaintance with the theory of descriptions (and the account given in Chapter 2) to rustle that up! But that is precisely the point. What interpretive analysis typically offers are richer conceptual resources to elucidate that meaning. It would be tempting to say that interpretive analysis tells us what something really means or what we should mean. But how we are to understand, explain, or even just describe all this? As we will see, p. 47↵the issues raised by interpretive analysis came to be addressed and explored by analytic philosophers themselves.
Introducing G. E. Moore
G. E. Moore (1873–1958) is generally regarded, alongside Russell, as having inaugurated analytic philosophy in Britain through their joint rebellion against British idealism at the turn of the 20th century. Indeed, Russell credits Moore with having made the first move, with him following close behind. Whereas Russell’s main interests at the time lay in the philosophy of mathematics and logic, however, Moore’s concern was with epistemology (the theory of knowledge) and ethics. Like Russell he was educated at Cambridge and was elected to a Fellowship, in Moore’s case in 1898, two years after Russell. He left Cambridge when his Fellowship ended in 1904 but returned in 1911 to take up a lectureship in moral science. He was Professor from 1925 until he retired in 1939 (when Wittgenstein succeeded him), and was Editor of Mind, one of the top journals of philosophy, from 1921 to 1944. He spent most of the Second World War in the States, lecturing in New York and California, among other places, thereby encouraging the development of analytic philosophy in North America.
His most famous work was Principia Ethica, published in 1903, but he also published a number of influential papers, including ‘A Defence of Common Sense’ (1925) and ‘Proof of an External World’ (1939), aimed at refuting both idealism and scepticism. Moore made it his task to distinguish and clarify as carefully as he could the precise questions asked by philosophers and the various answers that might be given to these questions, without committing himself, in many cases, to any definitive answers. As a result, he can often come across as unduly pedantic in his writing, but he had an enormous influence on those he taught and with whom he engaged in philosophical discussion. What he bequeathed was more his analytic approach than any characteristic set of doctrines or ideas.
The central question of Principia Ethica is ‘What is good?’, and Moore’s main claim is that ‘good’ is indefinable, or, as it might also be put, that goodness is unanalysable and hence has to be regarded as a simple quality. His main argument for this is the ‘open question argument’, as it has come to be called. Consider a possible definition of ‘good’, say, as ‘that which we desire to desire’, which Moore himself takes as one of the more plausible. It seems, though, that we can quite genuinely ask ‘Is that which we desire to desire good?’ This question seems open in a way that the question ‘Is what is good good?’ or ‘Is that which we desire to desire that which we desire to desire?’ is not. The answer to these last two questions is obviously ‘yes’, as we are simply being asked to agree to a tautology—to what is self-evidently true. Yet if ‘good’ (or ‘what is good’) and ‘that which we desire to desire’ had exactly the same meaning, then all of these questions should be equally closed. Since a similar argument could be run for any purported definition of ‘good’, then it would seem that ‘good’ must be taken as indefinable.
This argument is far too quick to be true; indeed, it threatens to rule out any definition of anything, not just of ‘good’, other than obvious tautologies. Yet there seem to be many examples of true definitions that are far from obvious tautologies. The standard example is ‘Water is ’. Can we not easily imagine situations in which the question ‘Is water ?’ is open in a way in which ‘Is water water?’ is clearly not? For example, someone could know what water is, in the sense of knowing that it is something we drink, what is found in rivers and oceans, what comes out of taps, and so on, and they could also know what is, in the sense of knowing that it is a molecule formed of two hydrogen atoms and one oxygen atom. But they might not have put the two together. For them, the question as whether water is would be open: it would come as a discovery to learn what the scientific definition of ‘water’ is.
p. 49Why should the situation not be the same in ethics? As a ‘moral scientist’, should Moore not have been trying to discover the nature of goodness just as a chemist seeks to discover the nature of a particular substance? It is here we touch on one of the deepest problems of philosophy, and not just of analytic philosophy. Are ethical questions ones that can be answered by using the methods and results of natural science? Those who say ‘yes’ are called naturalists, and the debate between naturalism and non-naturalism is not just confined to ethics, but rages fiercely in many other areas of philosophy. Frege, for example, attacked all forms of naturalism in the philosophy of mathematics; one major motivation behind his attempt to show that arithmetic is basically logical was to repudiate empirical and psychological accounts of number.
Moore was convinced that naturalistic accounts of ethics were mistaken: they all commit what he called the ‘naturalistic fallacy’, as his open question argument was intended to demonstrate. Any attempt to define ‘good’ naturalistically, whether in terms of pleasure, happiness, desires, or whatever, falls foul of the open question argument. So how did he see the situation as different from natural science? Much has been written about this; but let me try to identify one of his underlying thoughts. In the case of an ethical question, such as in asking whether something is ‘good’ or a course of action ‘right’, we do not need to do any scientific investigation; we have the resources to answer the question ourselves. We will need to know all the relevant facts, of course, but these alone will not give us the answer; we must also apply our ethical concepts and beliefs, which we already have. In principle these are sufficient: we do not need to wait for natural science to supply the answer.
Moore spoke here of ‘intuition’, and this has also generated a great deal of controversy, but again the basic idea is simple. All we can do is put ourselves in a position (by considering all the facts) to see the simple quality that is called ‘good’ (if it is there) in the same way as we need to put ourselves in an appropriate position p. 50↵(in the right lighting conditions, etc.) to see what colour something is. It is in elaborating and defending some such account of how we apprehend supposedly non-natural qualities that is perhaps the biggest challenge to non-naturalism, while the biggest challenge to naturalism is in providing convincing definitions of ‘good’ and other ethical terms. The question of whether ‘good’ can be defined remains very much open today.
How can analyses be both correct and informative?
Moore’s open question argument makes a key assumption: that if you understand the meanings of two phrases, then you can immediately tell whether those meanings are the same or not. You can understand the meaning of ‘good’ and ‘that which we desire to desire’, for example, and immediately tell, according to Moore, that they are different. But as noted above, this threatens any attempt to define or analyse something. Moore’s assumption, in fact, generates what has since been called the paradox of analysis (Box 3). If analyses are what analytic philosophy essentially seeks to provide, then it would seem to be fatally flawed at its very core.
Box 3. The paradox of analysis
Consider an analysis of the form ‘A is B’, where ‘A’ represents what is analysed and ‘B’ what is offered as the analysis. Then either ‘A’ and ‘B’ have the same meaning, in which case the analysis expresses a trivial identity; or else they do not, in which case the analysis is incorrect. So no analysis can be both correct and informative.
One might accept that some definitions are indeed uninformative. Consider the definition of ‘2’ as ‘’ or that of ‘puppy’ as ‘young dog’. Can anyone understand the various terms here and not immediately know that ‘2’ and ‘’, and ‘puppy’ and ‘young dog’, p. 51↵have the same meaning and hence that the definitions are trivially true? But surely the whole point of a definition that encapsulates a successful analysis is that it is informative? When we considered the definition of water as , I suggested that someone could indeed understand the meanings of ‘water’ and ‘’ without knowing (at a certain time) that the definition was true: for them it could be informative. And are not the analyses discussed in Chapters 1 and 2 intended to be informative? Indeed, did you not find at least some of them informative? How could this be if the relevant meanings were not different—in the way that the meanings of ‘water’ and ‘’ are different? But if that is the case, then how can the definition be true?
Sense and reference
Many paradoxes arise because of an ambiguity in a key term. The paradox of analysis might seem to be no exception. Surely it cries out for a disambiguation of ‘meaning’—for an analysis of ‘meaning’! In a definition (recording an analysis) of the form ‘A is B’, ‘A’ and ‘B’ must have the same meaning, in one sense of ‘meaning’, for the definition to be correct; but must have different meanings, in a different sense of ‘meaning’, for the definition to be informative. What are these two different senses of ‘meaning’?
One answer is to distinguish between ‘sense’ and ‘reference’, which is a distinction that Frege introduced to account for the informativeness of identity statements, of which definitions are one kind. Frege’s most famous example is the following:
(HP) Hesperus (the Evening Star) is Phosphorus (the Morning Star).
‘Hesperus’ was the name given to the planet Venus as it appears in the evening, and ‘Phosphorus’ the name given to Venus as it appears in the morning. It came as a discovery to astronomers that Hesperus and Phosphorus were indeed one and the same heavenly body, namely, Venus. (HP) is thus both true and informative. p. 52↵What Frege said is that ‘Hesperus’ and ‘Phosphorus’ have the same reference (they both refer to Venus) but different senses. A sense, according to Frege, expresses a ‘mode of determination’ or ‘mode of presentation’ of a reference. Venus can be presented to us in the morning, as reflected in our using ‘Phosphorus’ (‘the Morning Star’) to refer to it, or in the evening, as reflected in our using p. 53↵‘Hesperus’ (‘the Evening Star’). The identity statement can be informative insofar as we learn that the object referred to in one way is in fact the same object referred to in another way.
(HP) is not a definition, but Frege’s distinction can clearly be applied in suggesting how successful analyses can be stated in true and informative definitions. Expressed in the form ‘A is B’, ‘A’ and ‘B’ must have the same reference for the definition to be correct but different senses for it to be informative. ‘Water is ’, for example, could be seen as representing a correct chemical analysis in virtue of ‘water’ and ‘’ referring to the same substance, and as being informative to the extent that ‘water’ and ‘’ have different senses (along the lines suggested above).
I think that some such distinction between sense and reference is the first step in resolving the paradox of analysis, although the notion of ‘sense’ has generated enormous debate ever since Frege introduced the distinction and calls for clarification and elaboration. But something more is needed anyway, in my view, to resolve the paradox. Let us go back to Frege’s analysis of ‘Unicorns do not exist’ as ‘The concept “unicorn” is not instantiated’. It would be misleading to describe this as just two ways of conceiving the same thing, on the model of ‘Hesperus is Phosphorus’. The point of the analysis is to correct misunderstandings that we might have in thinking ‘Unicorns do not exist’. The senses are not, as it were, on a par; and it might be better to think of the analysis as sharpening or refining the sense of ‘Unicorns do not exist’, bringing the senses into line by enriching our original understanding with the new conceptual resources upon which our analysis draws. If talk of ‘sense’ is intended to capture what we understand about something, then this can change over time, even when we use the same expressions, and analysis itself effects such change. In a nutshell, I think that analysis is informative by being transformative, and we need to give the idea of transformation a central role in our account of analysis. We have already illustrated the idea in Chapters 1 and 2, and we will be returning to it in what follows.
Moore’s open question argument, we noted, makes a key assumption: that if you understand the meanings of two phrases, then you can immediately tell whether those meanings are the same or not. In the light of the disambiguation of ‘meaning’ just suggested, this assumption is only plausible, at best, for senses. If this is right, then there is one way in which I can be said to know what I mean: when I understand the senses of the expressions I use, taken as implying the ability to tell whether those senses are the same as those of any other expressions that I understand. Correspondingly, there is one way in which you can be said to know what I mean: when you understand the senses of the expressions I use. And you can be said not to know what I mean when you do not understand the senses of the expressions I use.
With the disambiguation of ‘meaning’ into ‘sense’ and ‘reference’ in mind, however, there is another way in which you can be said not to know what I mean, even when you understand the senses of the expressions I use: when you do not know what the reference is of at least one of those expressions. If I say ‘My friend is coming’, for example, you may well ask ‘Who do you mean?’ if you do not know to whom I am referring. To take a more extreme case, imagine waking up in the middle of a pitch black room after having been drugged and abducted, so that you have no idea where you are, what time of the day or night it is, or even who you are. You may well be able to utter a truth in saying ‘I am here now’, but do you really know what you mean if you do not know what ‘I’, ‘here’ and ‘now’ refer to? If I am with you in the same state, then I do not know what you mean, either.
For present purposes, though, the interesting cases concern what we say when someone cannot provide relevant analyses and definitions. In his later work, Moore came to stress that one can understand the meaning of an expression without knowing its p. 55↵analysis. Interpreting ‘meaning’ here as sense rather than reference, this might seem right: we can understand the sense of ‘water’, for example, without knowing that it is . In the light of what we have said about the transformative role of analysis, however, there is room for argument here. There would seem to be some grounds for claiming that someone does not ‘really’ know what they mean by ‘water’ if they do not know that it is . And correspondingly, if you do know what the relevant analyses and definitions are, and I do not, then there would be grounds for saying that you know better than me what I mean. The question ‘Do you know what I mean?’, then, itself has various meanings, and we need to know which meaning is meant before we can answer. It’s all kind of like, ‘well, it all depends’. Know what I mean?