Language and information
Lecture 1. A formal theory of syntax
1.8. Properties of the reduced sentences
[Audio recording—Opens a separate, blank window. Return to this window to read the transcript.]
The sentences outside the base have their own structure. As was seen, they result only from a fixed set of reduction types, and only when specified high-likelihood conditions are satisfied. Now, once you get them, they are optional, that is to say, you can say the long form and you can say the short form. There are only a very few that are required. Those that are required, that are not optional, there is a way, which is justified, actually, historically, of filling out a source sentence which would be suppletive to—to take the place of the missing source sentence required for the reduction. I won't go into it because that will be touched upon later. If all sentences—this is given a correction in the case of required reductions, which are not many—then all reduced sentences have corresponding to them sentences in the base. Any reduced sentence has a base sentence that corresponds to it. If also reduced sentences do not change the meaning of a sentence, as we have seen, they do not even change the presence of a word, they only make it invisible, and some of them don't make it invisible, they only make it shorter, like going to to gonna, or something like that. This means that the meanings of all the reduced sentences are found in simple form in the base sentences. This means that the simple structures of the base carry all the information expressed in language, so that the well-known complexity of grammar, most of which is created by the reductions, is not due to complexity in the information, and is not needed for information. Many people think that language is complex because information is complex and requires it, but that is perhaps not really the case.
The set of reduced sentences has structural features lacking in the base, as will be seen next week.There are marginal sentences due to fuzzy domains of various reductions. There are ambiguous sentences—in the reduced set, not in the base set—made be degenerate—well, there are different kinds of ambiguity in the base. There are ambiguous sentences made by degeneracies in reducing, that is, when different reductions on different source sentences may yield the same word sequence, so that you can't tell which it is. But that word sequence preserves the two meanings of the different sources. That's what makes it ambiguous.
The reductions also create many word subsets when the domain of a reduction is not the whole of a base word-class, but only a high-likelihood section of it. In many languages, the reductions have created morphology, as a reduced phonetic distance between certain words and their argument, merging them into one composite word. Reductions also create many special constructions. In the new phonemic shapes which reductions make, there are places with similar grammatical relations having similar forms, some of which lead to structural patterns, and even to grammatical paradigms, such as conjugations of personal endings on verbs. Such grammatical paradigms give prominence to particular grammatical meanings, such as tense, plurality, and person, but all this is done in the reductions, and all this has equivalents from which it can be derived in the base with help from the reductional apparatus.
These are the major specific properties of the base and derived sentences as they result from the constraints. They give language its form and its semantic capabilities. The more general properties of language, including its general mathematical structure, will be noted in the last session.