Language and information
Lecture 4: The Nature of Language
[Audio recording—Opens a separate, blank window. Return to this window to read the transcript.]
Now, there is one other thing that is universal to languages, and that is that languages change. For certain purposes, it has been customary in modern linguistics to draw a line between the present and the past—the present of a single language, one at a time. And this was important because very many grammatical relations in a language hold only among the forms in one language at the same time; in other words, they have to relate to each other. But nevertheless the line between the present and the past cannot be drawn precisely. In a detailed grammar, once a grammar is sufficiently detailed, there are always some forms that are in a process of change, naturally if the language is changing more or less all of the time. And these cannot be really regular in a time-slice. One can attribute them, regularize them by assigning them to the nearest sort of thing to which they can be attached by some reduction or something of that sort, but the language is not really completely regular in a true sense.
In many forms also—this is also a reason why one cannot entirely separate the time-slice from the history—in many forms one can see the earlier construction which has changed into this form, and it is more regular, very often than the form into which it has changed, and has certain importances. I'll give you one or two very simple examples.
There are certain verbs, like give and take, that are used as pre-verbs attached to other verbs. Really, they are operators on the following verbs, which are their argument. And they have also much of their meaning, like to give a look at something, and to take a look at something is not so terribly different. Nevertheless, the grammar which belongs to those words by themselves has not entirely disappeared. Give, where you can say to give a book to somebody and to give somebody a book: you can still say give it a look as well as give a look at it. You cannot say take it a look. In other words give and take still differ from each other even in this situation. These are things which were not so to speak explained in the time-slice grammar; they are explained historically.
So in describing a particular grammar, one has to take into account certain results of change. In a theory of grammar, one has to take into account the fact of change. It has to show up in an explanation of language structure as a whole. And of course also, by a well-known principle, one has to assume that change has been going on as long as language has been going on, not that it started at a certain point.
It might be of interest to say one other thing: that different parts of the grammar change at different rates, and not the same rates in all languages. What is most likely to change is the word use, the likelihoods of word combinations; less, the word stock; still less, the grammatical constructions and the grammatical subclasses of words; still less, the phonemic distinctions. What doesn't change at all, as far as we know about language, is the partial ordering. The thing that makes sentencehood, as far as we know, is universal, and does not change.