Last revised 7 February 2005.
So far we have looked at some aspects of our thinking about education. I offered you a very rough "sociological" account of what is going on when people use the concept (this was NOT meant to be a definition of 'education') and we also looked briefly at some of the things people often claim to be aiming at when education is going on. I indicated that none of this fits what traditionally people have wanted to do in offering a conceptual analysis or a definition of a term like 'education'. Now I want to look a litle longer at the general issue of definition and the nature of concepts.
It is remarkably easy to fall into talking about concepts, but what are they? or what is it to have a concept or use it correctly? Psychologists will tell you that "concepts are the most fundamental constructs in theories of the mind" (Margolis and Laurence (eds.), Concepts: a useful collection of papers with a long introductory essay [I will adopt their convention of using capital letters as names of concepts]), but then you will find any number of conflicting theories, so you are not much better off. Concepts are whatever play certain roles in explaining our cognitive activities; we should allow the possibility that our best account of our cognitive activities splits up our common-sense idea of concepts (I'm not saying it will, but just that it might, so we shouldn't assume we have a clear grasp of what concepts are now, before the best theory is discovered).
Psychologists are interested in a variety of animals; but educationalists tend to be interested only in humans. That gives us an easy way out, because in the human case we can say there is a rough correspondence between concepts and words. How do I tell whether you have the concept of 'TYRANNY' or 'MAGENTA' or 'IF ... THEN ---'? One straightforward way is to see how you use the corresponding words.
This is very rough. You can reveal a grasp of a concept by picking things out in a particular way without using language at all. You may be a very young child with little or no language, and yet you may have certain concepts. We may all have concepts that do not correspond to words in any simple way (consider the shades of colour you can distinguish). But, for educational purposes at any rate, most of these qualifications are of little importance. Education is largely a linguistic transaction; its concerns for concepts are mediated by language.
Ever since Socrates, philosophers and others have thought that progress can be made, and perhaps can only be made, by seeking for accounts of key concepts we use. Typically these accounts are then taken to be definitions of those concepts (or of the words that correspond to them). (An account doesn't have to be specifiable as a definition. One could take most of Plato's Republic as an account of justice, an account that provides a lot more than any short quasi-definition he might offer in the course of it. But typically, investigators have wanted to classify things using taxonomies modelled after Aristole's, where one does offer a neat story of genus and differentia.)
I think it fair to say that traditionally language has seemed "transparent", to quote a phrase Russell used to characterise his early views. So one would not pay much attention to the difference between giving an account of what a bit of language means (of what a concept is) and giving an account of the things in the world the language picks out. But once one sees that there is a difference, one can mark it by saying that "real" definitions seek to characterise the things, "nominal" definitions seek to describe the language.
A pretty immediate consequence of noticing that distinction is to wonder how far, and by what methods, we can do either of these things. It might seem that competent speakers of a language automatically know what their words mean/what concepts they are expressing. But the variety of beliefs among different groups, or one's own occasional errors, might encourage a certain modesty about our ability to characterise things in the world just like that.
Actually, it seems to me that in neither respect can we assume that the answers are directly available to us, just by reflection. That is pretty obvious as far as the nature of the world around us is concerned. Whether to classify whales as a fish is not something you can decide without actual investigation. But it is also true of our own language in many important respects. We may know what we mean in some sense, but that does not mean that we can accurately characterise it in other terms, analyse it, explicitly state its relations to other notions, etc., etc. All that is theoretically loaded activity that isn't an added extra when one is a competent speaker.
The traditional ideal for a definition of 'T' was to provide a set of individually necessary and jointly sufficient conditions for being T. Let us take what may be a decent example: 'brother' =def male sibling. To say that these conditions are individually necessary is to say that all brothers are male and all brothers are siblings (if X is necessary for Y then roughly all Y are X - work out why it is that way round). To say that they are jointly sufficient is to say that all male siblings are brothers (roughly, if X is sufficient for Y then all X are Y).
You can see two obvious ways in which a proffered definition can go wrong. What it offers may not be necessary (if you included 'adult' among the features for a brother) or it may not be sufficient (if you simply said 'sibling'). Definitions can go wrong in both ways at once.
Despite what some people say, it is obvious that it would be impossible to demand that one always define ones terms in this way. If you did make this absurd demand, in offering the account of brother above, you would immediately have to define 'male' and 'sibling', and if you tried to do either of those, you would immediately have to define whatever further words you used in those definitions, and so on. You could never get started, in fact, since whatever word you tried to use would stand in need of a definition.
One traditional response to this problem is to think that there are some basic concepts that we cannot define but in terms of which we can then define all the others.
What does seem to be true is that humans find some categories more immediate or natural than others: mango and fruit rather than Julie mango or multi-celled creature. For practical purposes we have basic concepts versus more specialised or more general ones; but these structures do not match those that philosophers have usually assumed when talking about their basic and derived concepts.
Given that it is impossible to define all your terms, you need to decide when to define any. I suspect that those people who insist on ritual definitions of terms have not considered this question.
What is now pretty clear is that there is much that is wrong with the traditional approach. I indicate a number of problems. Since quite a lot of what is here is due to Lakoff's work, you ought also to consult a critical response to Lakoff by John Vervaeke and Christopher D. Green: 'Women, Fire, and Dangerous Theories: A Critique of Lakoff's Theory of Categorization'.
Philosophy now usually acknowledges Wittgenstein as the source for an appreciation of the fact that many terms apply to a set of things that do not share a unique set of necessary and sufficient defining characteristics. The point had been seen before -- J.S. Mill quotes Bain on the use of the word 'stone':
Take the familar term Stone. It is applied to mineral and rocky materials, to the kernels of fruit, to the accumulations in the gall-bladder and in the kidney; while it is refused to polished minerals (called gems), to rocks that have the cleavage suited to roofing (slates), and to baked clay (bricks). It occurs in the designation of the magnetic oxide of iron (loadstone), and not in speaking of other metallic ores. Such a term is wholly unfit for accurate reasoning, unless hedged round on every occasion by other phrases; as building stone, precious stone, gall stone, &c. Moreoever, the methods of definition are baffled for want of sufficient community to ground upon.but its implications for business as usual had not been appreciated.
Wittgenstein's claim was that what is going on is that we apply a word to things that share criss-crossing "family resemblances", not a genuinely common set of features. His famous example is "game". Whatever one wants to say about that word, his thesis applies to family resemblance terms themselves: e.g., Socratic. And arguably, to many other words in our languages.
One might attempt to recoup something for tradition by invoking the more complex notion of inus conditions: iinsufficient but necessary parts of an unnecessary but sufficient condition. But even then, one would usually have to add something like "and so on" to cover tomorrow's intelligible extension of a word to a new sort of case - motivated but not rule-governed in Lakoff's terms.
If our concepts worked in the way tradition assumed, every example of T would equally be a T. But the way we actually think involves various complexities. We make "default" assumptions: birds fly, except for odd ones like penguins. A sparrow is taken to be more typically a bird than a penguin or ostrich. We operate with various special cases: typical/normal examples; ideal examples (notice how different your idea of a typical husband is from your idea of an ideal husband), and these special cases may be extended to further levels of detail. If it is fair to say, with Margolis and Laurence, that "most concepts are complex representations whose structure encodes a statistical analysis of the properties their members tend to have" then we can see how it becomes impossible to make a principled distinction between those matters that belong to the concept as a matter of its meaning and those that we take to be true of its instances (this is one way to see the import of Quine's celebrated attack on the notion of analytic truth).
We have already suggested that "education" is a term that different people will use with very different associations and aims. Other similar terms are "democracy", "terrorism", or "equality". One could say that these terms have very different conceptions associated with them. However one characterises it, the point is that we will not make much progress trying to find a minimal skeleton shared by all divergent uses. What matters much more are the differences, the different principles or aims espoused by the different groups. If progress is possible, it will have to proceeed by addressing those differences rather than seeking a generally acceptable definition.
This is another kind of case where defining one's terms is of no practical value. I can define "democracy" so that what goes in Burma or China is ruled out, but so what? I have not in that way done anything to convince a Burmese general of the error and evil of his ways.
I have just suggested that the use of some words can insinuate a host of principles or assumptions. These may form part of what we intend to convey about the world when we use them. But there is also the question of what the world is actually like when we so use the words. Attending to this last question could be described as offering a "factual analysis" of the language in question. While we can always make ther distinction, it is of particular philosophical interest when what we think most plausible as factual analysis does not measure up to what we intend to convey (e.g., the traditional use of the word "witch"). Most examples will be contentious here, but consider our thoughts about free will: we may intend to suggest something about autonomous origination of action, but what goes on in the world may only amount to the absence of various sorts of obstacle to acting.
When we use language it is plausible to think that our behaviour is responsive to certain features, both of the world and of how we think. In neither case do we necessarily know explicitly what guides our usage. Some theorists of concepts have accepted something like this in distinguishing between superficial perceptual criteria for using a term correctly and the actual meaning we intend to convey by it (so we pick out women by how they look but what we mean is something that is not impugned by transvestites).
In the case of what I am calling "logical cues" our explicit grasp is, I would guess, less than in the perceptual case. We know how to make some inferences from what we think, but we are often unable to state the relevant rules. Indeed, even explicit theorists have yet to agree on such apparently simple matters as what the word "large" means. We can see fairly easily that a large X is something like an X larger than the average, but exactly how to fill out this story is still obscure.
What that example suggests and illustrates is what I take to be very common: the words we actually use do not exhaust the content of what we mean; we leave a lot unsaid that is vital to there being anything definite for us to say (anything definitely true or false, anything we can obey or disobey, etc.). An elementary example is provided by our talk of left and right. It all presupposes a frame of reference or an orientation, but we hardly ever bother to specify this explicitly. But it is vital - without implicitly putting it in we simply wouldn't know what is being said. You can see more about this phenomenon, ellipsis, in my elementary notes on critical thinking, or in a more advanced discussion.
The last point I want to make now takes us back to the nominal/real distinction we met earlier. The tradition has often assumed that the basic concepts to which others will be reduced are matters of observation. So people have thought that they could define "gold", say, as a heavy yellow metal that .... Here their attempted definition would be close to the perceptual cues I mentioned above. But what we noted then was that while these cues may be what we use to make quick judgments, they do not correspond to our intended meaning. What we mean can rather be revealed in what we would say about various imaginary situations. If something was a heavy yellow metal that did ... but it was not made up of the same sort of molecules as gold, then we would think it wasn't gold (not just another variety of gold). So what we mean in this sort of case is something to do with the actual constitution of gold, its "real essence", rather than its observable characteristics. But once again, if that is what we mean, it is not something we will know about simply because we are competent speakers of the language. People have used the word "gold" and its forbears competently for centuries without the least notion of its chemical constitution.
Last revised 7 February 2005.
© E.P. Brandon, 2005.