"It seems to me that the notion of 'a language' should not be regarded as scientifically reconstructable at all. We can say in very broad terms that a human language is a characteristic way of structuring expressions shared by a speech community; but that is extremely vague, and has to remain so. The vagueness is ineliminable, and unproblematic. Human languages are no more scientifically definable than human cultures, ethnic groups, or cities. The most we can say about what it means to say of a person that they speak Japanese is that the person knows, at least to some approximation, how to structure linguistic expressions in the Japanese way (with object before verb, and postpositions, and so on). But in scientific terms there is no such object as 'Japanese'."