Problems with classical categories
The classical view of categorization is open to a number of criticisms. First, there are remarkably few examples of adequate definitions in the classical mould. In fact, as discussed in Chapter 2, some researchers doubt that there are any. We noted in 2.6 that many definitions do not seem successful in specifying necessary and sufficient conditions for membership of a given category. This is certainly true of dictionary definitions, but the same problem applies to more technical and detailed definitions like those given in semantics. To pick an example almost at random, the Concise Oxford’s definition of food, ‘substance(s) (to be) taken into the body to maintain life and growth’ applies just as much to medicine as it does to food like bread or apples, a circumstance which invalidates that particular definition. Similarly, the same dictionary’s definition of game as ‘contest played according to rules and decided by skill, strength or luck’ does not apply to card games like patience (solitaire), which involve a single participant and are thus not contests, nor to a game in which a child throws a ball against a wall. Further, it also applies to wars and exams, which are decidedly not examples of games. As discussed in Chapter 2, the history of semantics is full of examples of a proposal for the correct definition of a term being shown to be inaccurate. A famous example is the previously standard definition of kill as ‘cause to die’. Imagine that someone has tampered with the sheriff’s gun in such a way as to cause it not to fire in a shoot-out with an outlaw. As a result, the outlaw is able to shoot the sheriff to death. In a case like this, we would say that the tamperer has caused the sheriff to die, but has not actually killed the sheriff (for further problems with this case, see Fodor 1970). Furthermore, even longer and more detailed definitions like those advanced by Wierzbicka and her colleagues apparently do not resolve these problems. Cases like this occur time and time again in the history of definitional semantics. The problems of definition are discussed at length in Chapter 2 (see especially 2.6).
Rosch and Mervis outline a more influential criticism of the classical view of categorization (1975: 573–574):
As speakers of our language and members of our culture, we know that a chair is a more reasonable exemplar of the category furniture than a radio, and that some chairs fit our idea or image of a chair better than others. However, when describing categories analytically, most traditions of thought have treated category membership as a digital, all-or-none phenomenon. That is, much work in philosophy, psychology, linguistics, and anthropology assumes that categories are logical bounded entities, membership in which is defined by an item’s possession of a simple set of criterial features, in which all instances possessing the criterial attributes have a full and equal degree of membership.
In other words, the classical interpretation of categories (and hence meanings) as sets of necessary and sufficient conditions fails to do justice to the fact that there seem to be different statuses of category membership: some members of a category seem to be better examples of that category than others.
We can illustrate this with an example which has played an important role in critiques of classical categorization. Consider a colour category like RED. We can think of many shades of red, including the red of a fire-engine, the deep reds found on fruit like plums, which might also be described as purple, and very pale reds which might also be described as pink. It seems impossible to identify any single point along the scale of redness that constitutes the boundary between red and other colours, and as a result it seems clear that the category RED is not defined by any necessary and sufficient conditions, or anything else that might provide a clear category boundary for it. Yet there is a clear sense in which the red of a fire engine seems a better example of red than the colour of a ripe plum. In order to give an idea of the type of colour referred to by red, we would obviously do much better pointing to a fire-engine or a standard red rose, than to a ripe plum or the orangey-pink of a sunset, even though both of these might also be described as ‘red’. RED, then, seems to be a category of which some members are better examples than others.
QUESTION What are some other categories in which some members are better examples of the category than others?
Colours are by no means the only example of categories with different statuses of category membership. Consider Figure 7.1 below, a series of representations of various cup- and mug-like objects, taken from an influential study by Labov (1973).
It seems obvious that some of these objects, like (1), are very good examples of cups, and that others, like (11), are very good examples of mugs. There also seem to be several intermediate cases, like (7), in which it is not clear whether cup or mug is the better description, as well as others, like (17) and perhaps (4), where we might hesitate to apply either label. (If some of the objects were represented with accompanying saucers this might reduce the ambiguity, of course.) This is, in fact, exactly what Labov found when he asked subjects to decide which was the appropriate label in each case.
We could make similar observations about many other categories in natural language. The category CHAIR is a case in point (Figure 7.2). The chair in the centre of the diagram seems a particularly good example of the category, unlike the high chair on the middle left or the deck chair in the bottom row. The arm chair and the rocking chair also seem clear examples of the category, but somehow less obvious than the original ordinary four-legged chair. That, indeed, is the only one of the pictured chairs which is precisely that: an ordinary chair of the sort we might refer to through expressions like a normal chair, an ordinary chair, a standard chair, and so on. There are two important points to draw from these examples:
• There are categories in which some members are better exemplars of the category than others.
• There are categories in which the boundaries of membership are not clear-cut: it is not always possible to say whether or not something is a member of the category.
If categories are constituted by nothing other than sets of necessary and sufficient conditions, neither of these points is expected. The second one in particular is very unexpected: if there is a finite set of necessary and sufficient conditions for a category, we should be able to state unambiguously what a given category’s members are.
What conclusions can we draw about the nature of the categories? One possible answer is that these categories are not structured in terms of necessary and sufficient conditions, but that membership in them is graded: a matter of degree.
