The Language of Uncertainty
Uncertainty spreads through our lives so thoroughly that it dominates our language. Our everyday speech is made up in large part of words like probably, many, soon, great, little. What do these words mean? "Atomic war," declared a recent editorial in the London Times, "is likely to destroy forever the nation that wages it." How exactly are we to understand the word likely? Lacking any standard for estimating the probability, we are left with the judgment of the editorial writer.
Such verbal imprecision is not necessarily to be criticised. Indeed, it has a value just because it allows us to express judgments when a precise quantitative statement is out of the question.
The language of uncertainty has three main categories: (1) words such as probably, possibly, surely, which denote a single subjective probability and are potentially quantifiable; (2) words like many, often, soon, which are also quantifiable but denote not so much a condition of uncertainty as a quantity imprecisely known; (3) words like fat, rich, drunk, which can not be reduced to any accepted number because they are given different values by different people.
We have been trying to pin down by experiments what people mean by these expressions in specific contexts, and how the meanings change with age. For instance, a subject is told "There are many trees in the park" and is asked to say what number the word many mean to him. Or a child is invited to take "some" sweets from a bowl and we then count how many he has taken. We compare the number he takes when he is alone with the number when one or more other children are present and are to take some sweets after him, or with the number he takes when told to give "some" sweets to another child.
First, we find that the number depends, of course, on the items involved. To most people some friends means about five, while some trees means about twenty. However, unrelated areas sometimes show parallel values. For instance, the language of probability seems to mean about the same thing in predictions about the weather and about politics: the expression is certain to (rain, or be elected) signifies to the average person about a 70 per cent chance; is likely to, about a 60 per cent chance; probably will, about 55 per cent.
Secondly, the size of the population of items influences the value assigned to an expression. Thus, if we tell a subject to take "a few" or "a lot of" glass balls from a box, he will take more if the box contains a large number of glass balls than if it has a small number. But not proportionately more: if we increase the number of glass balls eight times, the subject takes only half as large a percentage of the total.
Thirdly, there is a marked change with age. Among children between six and fourteen years old, the older the child, the fewer glass balls he will take. But the difference between a lot and a few widens with age. This age effect is so consistent that it might be used as a test of intelligence.