Language and information
Lecture 1. A formal theory of syntax
[Audio recording—Opens a separate, blank window. Return to this window to read the transcript.]
Now, the fourth constraint. This gives a linear order to the words of the sentence. Since the relation that makes sentences out of words is a partial order, while speech is linear, a linear projection is involved from the start. Every language has one or more linear projections. In English, the operator is said after the first argument, so that wear, which has two arguments, appears in the middle, as in Men wear coats, or something like that. Now, the fact that the lineal order a separate thing, a separate step in sentence-making, just a projection of something else, leaves room for alternative linear orders to exist without prejudicing the grammatical relations, which are related to the partial order.
This gets us the grammatical source of modifiers, which are a very, very important thing in grammar, because there are altogether two main alternative linearizations in many languages, including English. One of them is bringing a word toward the front of the sentence, especially one that is so to speak a topic of the sentence, as one can say John I have long distrusted, from I have long distrusted John. So this is bringing a word in a sentence to its front. The other one—and these are the two main ones anyhow—the other other alternative linearization is to interrupt a sentence with a whole subsidiary sentence. For instance, one can say—in the sentence The opposition was unprepared one can interrupt it and say The opposition, so John said, was unprepared. And you can interrupt it with many—at various points.
Now, one case is especially important. It's a case that uses both of these together. It's a case where the interrupting sentence begins with the same word as that before the interruption. For instance, if a person says I believe that John is responsible. Then you insert John I have long distrusted. Then one says I believe that John—John I have long distrusted—is responsible. In this situation, the adjacency of the repeated words—the two occurrences of John—enables the repetition to be reduced to a pronoun. This is not an easy matter, because language does not have a counting system, it doesn't have an addressing system for words, and therefore can only do things under favorable conditions. When the two words are next to each other, it can say these two words are the same. When the two words are at a distance, it is very hard in language to say what is the distance between the two words, and therefor to identify what the antecedent is. But this is something I will touch upon in a later lecture.
So in this situation, where you have I believe that John—John I have long distrusted—is responsible, in that situation one can reduce the second occurrence of John to a pronoun: which, who, and so forth. When you do that, you get I believe that John, whom I have long distrusted, is responsible. This is the crux of the relative clause, from which all modifiers in English are derivable. It's a very important construction, and it comes simply from the fact that linearization is a separate step from partial ordering, and gets there there with a certain play—there's a certain play in linearization. Obviously you can do all kinds of linearizations.