Swimming and Walking

The contrast between the intentional and purposive (chess and language) and the non-intentional and non-purposive (swimming), which is the pivotal point of Dummett's argument, rests on a vacillation between two senses of the term `swim.' Sometimes he uses it to mean `keeping one's head above water' or `staying afloat'; at other times, he uses it to mean doing a recognizable stroke, for example the crawl.

When he says that we can reasonably say: `I don't know; I've never tried,' in response to the question `Can you swim?', `swim' must be taken in the first sense. By contrast, when swimming counts as a genuine skill, the argument turns on the second sense. I shall show that in the first sense swimming is not a skill, and hence any features of how we know how to swim shed no light on the question of our knowledge of language--indeed, they shed no light on knowledge of swimming understood in the second sense.

There is a sense in which one may turn out to be able to swim without ever having learned to do so, without, that is, ever acquiring a skill. This is the sense in which swimming is nothing more than keeping one's head above water. But, this is not the sense in which one knows how to swim in the manner required to compete in swimming meets, pass swimming classes, and so forth; moreover, it is not even a step on the way to the acquisition of those skills. The contrast between staying afloat and doing the crawl lies in the fact that the former involves no criteria, it does not accord with rules, and most certainly is not guided by rules. There are no rules governing staying afloat, and this entails that, strictly speaking, it cannot be learned or taught.

Under the heading `keeping one's head above water,' we count everything from frantic doggy-paddling to graceful breaststroke-like movements. It is not grace that makes for success here, all that matters is that one avoid drowning or getting too much water up one's nose. Apart from this, there is nothing that counts as right or wrong. It would make no sense to say to the `swimmer': `No, no, you move your legs like this, not like you're doing.' If he doesn't getting too much water up his nose and spends very little time below the surface, he's doing just fine. If we choose, we may say that he's swimming, but we need to be clear that in such a case, `swimming' does not count as a skill in any ordinary sense. However, it is only in this sense--this `non-skill' sense--that it is reasonable, when asked whether I can swim, to say, `I don't know; let's see what happens when I jump into the deep end of the pool.''

Compare this with the second sense--call it `proper-swimming.' Here corrections are in order; it is possible to get it wrong, as well as right. Unless I am willing to count some movements as correct and others as incorrect, I am simply not engaged in proper-swimming. I may be keeping my head above water, I may be going back and forth across the pool faster than anyone else, but I'm not proper-swimming. Proper-swimming, unlike staying afloat, is a skill because the actions involved in it are subject to correction.

Skills have criteria associated with them. In virtue of these criteria (or rules), skills may be taught and learned, may be performed more or less skillfully. Bouncing a ball off a wall isn't a skill. However, if we keep score--1 point for a clean catch; 0 points for a miss; -1 point if the ball goes into the street--then we may have a skill. Dummett's argument requires the contrast between a skill which can be accounted for entirely in terms of knowledge-how, and an intentional, purposive activity which requires knowledge-that. Swimming (as contrasted with proper-swimming) does not provide this contrast. In Dummett's arguments, swimming is an activity which is not properly classified as a skill, with the result that the comparison tells us nothing about knowledge of language.

If we try to find something which is a skill without criteria, we might consider walking. Here we have an activity which is not subject to assessment or correction (i.e., which has no rules), but which may nevertheless be said to be learned. In the normal course of events, a child is said to have `learned to walk,' and in saying this we do not imply that the child has mastered a set of rules or even been corrected when first trying to move about on two feet. There is no getting it right or wrong--unless the child proudly announces `Look at me, I'm walking' as she crawls across the floor. So long as the child is navigating on her own two feet without hanging onto the couch too often, she's walking. Perhaps we can substitute walking for swimming in Dummett's argument to make the case that skills which rest entirely upon knowledge-how are not like language with respect to how we know them.

But even this will not do. When we say that a child has `learned' to walk, we only mean that the child has reached a certain point in her natural development; we do not have in mind anything like what we mean when we say that she has learned her scales or times tables. To classify walking as a skill is not an accurate a description of it. Like staying afloat, walking is less a skill, and more an instinctive behavior, a quasi-instinctive behavior if you will.¹⁵ As such, it doesn't provide the contrast with language that Dummett needs for his argument.

In fact, there are three categories at play in Dummett's argument, and one of them is irrelevant to the discussion of a speaker's knowledge of her language.

C1

Random or non-purposive activities:¹⁶ these have no criteria or rules; as a consequence they are not properly classified as skills at all. Staying afloat, walking and bouncing a ball off a wall fall into this category. Some are quasi-instinctive (walking and doggy-paddling to stay afloat), others are just unguided sorts of things we do to fill the time between lunch and tea (bouncing a ball off a wall).

C2

Skills which we know-how to perform: these have associated rules and criteria, they properly count as skills because there are correct and incorrect ways of doing them. Behavior in this category is neither random nor non-purposive, but, at the same time, agents are not required to have any knowledge-that of the rules. Proper-swimming is an example of this.

C3

Skills which, for want of a better term, I call `Reasoned skills': these have rules and criteria associated with them, and one might well consider them to be skills, though most likely we would call them `intellectual' activities to set them apart from the skills of C2. Rules play a special role in the performance of these activities; they may be explicitly appealed, not only to correct oneself and others, but to plan one's next move. Hence to acquire and practice these skills, agents must possess knowledge-that of the rules. Chess fits in this category.

Since it doesn't concern skills at all, C1 is simply irrelevant here. What Dummett needs is the contrast between the different sorts of skills represented by the activities in C2 and C3. Between these, the crucial difference lies in the role played by rules vis-á-vis the agent's behavior, both in acquiring the skill and in its subsequent performance. The issue of central concern is what counts as mastery of these rules.