Logic and language
The preceding account is only the most modest sketch of the bases of a logical approach to meaning. Before ending the chapter, we should register some of the advantages and controversies surrounding the use of logic as a tool for the analysis of natural language. As we have shown at a number of points, there seem to be areas of clear incompatibility between logical constructs and the natural language terms which partly translate them. One such area of incompatibility has just been noted in Strawson’s critique of Russell’s theory of definite descriptions. A more serious incompatibility between logic and natural language, however, is the one discussed in 6.3: it would seem that natural language connectives frequently do not behave in anything like the same way as their logical counterparts. This immediately problematizes any attempt to advance logical constructs as somehow underlying or as basic to natural language meanings. Another problem for the suggestion that logical constructs are relevant to the understanding of natural language relates to the role of truth in the two systems. As discussed in Chapter 3, there are many reasons to doubt the centrality of truth to everyday language. As Halliday and Matthiessen (2004: 117) put it, ‘[s]emantics has nothing to do with truth; it is concerned with consensus about validity, and consensus is negotiated in dialogue’. A factor which deserves special emphasis in the context of logic is that truth can only be a relevant consideration to factual sentences, i.e. to declarative statements. Questions, requests, commands and apologies, to name only a few, are neither true nor false, and as a result need a different logical formal ism from the one introduced here.
Even in the case of declarative sentences, the question of truth is far from straightforward. As we have seen, logically oriented semanticists would claim that knowing the truth conditions for a sentence is at least necessary for knowing that sentence’s meaning. For example, the claim would be that one cannot know the meaning of the sentence in (111) unless one knows the kind of situation in which the sentence would be true.
However, this claim seems questionable. One can perfectly well know what (111) means without knowing whether it’s true in a certain case: the notions of truth and falsity are immensely obscure and complex. For example, is it true that the door is open if it is slightly ajar? What if the door has been taken off its hinges and leant against the wall, in exactly the same position it would be in if it had been opened normally? It would seem, in other words, that there are more options than simply true or false. We return to this point in Chapter 7. The incorporation of these considerations into logical accounts of language is a lively area of ongoing research, and a very necessary one for anyone committed to maintaining the relevance of logic to language.
On the positive side, the study of linguistic meaning from a logical point of view brings a number of important advantages. First, in its attention to declarative sentences it promises a formalization of a very important subset of natural language sentences. Declarative sentences are certainly not the only sentence-type in language: far from it. But they are, by any account, an important one (see Givón’s remarks in 4.1). In particular, they are a crucial format for the presentation of many culturally important types of knowledge in our society, such as scientific statements, and fictional, journalistic and historical narrative. If a logical approach can help to illuminate the underlying structure of this particular sentence type, then it will have advanced our understanding of an important part of language. Second, the logical approach permits a degree of rigour and formalization which entirely outstrips that of the more descriptive approaches to meaning discussed in most of the rest of this book. As a result, it is eminently amenable to manipulation by computer, and logical principles form the basis of computational approaches to language. As a result, the development of programs that mimic human language behaviour have a vital reliance on logical ways of modelling language and meaning (see 8.2 for discussion). Lastly, the focus of logical analysis on propositions has been seen by some researchers as supported by psycho logical evidence. As noted in 6.2, experimental evidence suggests that what people remember are not the actual words of utterances, but their content or gist. Propositions can be taken as one way of representing this remembered content (Barsalou et al. 1993). In other words, logical symbol ism may not always very accurately mirror the apparent use of words in natural language, but it may well serve as a valuable way of capturing the underlying structure of certain aspects of their meaning.
