In Chapter 10 we focused on pragmatic theories which emphasised ideational or conceptual meanings and regarded communication as the efficient transfer or implication of information-bearing propositions. But just as we balanced the ideational/conceptual meaning of semantics in Chapters 3 and 4 with other kinds of meaning in Chapter 5, including the interpersonal, so the last section of Chapter 9 balanced the co-operative principle with the politeness principle. To begin this chapter, I revisit the ideational–interpersonal distinction in the context of information theory and the concepts of entropy and redundancy. The chapter proceeds to discuss collocational predictability as a way of introducing Hoey's theory of lexical priming. Taking a hint from his book, I show that ambiguity (entropy) is much overestimated, and that, indeed, many of the jokes and humorous texts in this book depend upon the overriding of the most obviously primed meanings.

Predictability and information theory

As noted earlier (1.5), grammatical/function words carry less information than lexical words because they are more predictable. Technically, information is measured in bits, short for binary digits. One bit of information is carried by a symbol which has 50% predictability, as in binary computer code, notated 0 or 1. A symbol or unit of meaning which is 100% predictable carries no information, and is completely redundant. A symbol which is highly unpredictable carries a great deal of information and can be called entropic.