Semantics

“If meaning is as important as all that”, a harsh reader might say, “then it is about time that he got to semantics”. It might indeed seem odd that linguistics should take over fifty years to get its priorities right. And to be fair, linguists and earlier scholars of language often had very clear ideas about the importance of meaning and the need for study of it. It was simply not the kind of subject which it proved easy to study. There were, to begin with, numerous preconceptions and false ideas about the nature of meaning which hindered clear thinking, but which it was difficult to get rid of because of their respectable ancestry. One was the tendency [...] to identify words and things, to think that meanings were somehow concrete entities – words would be called dirty, dangerous, beautiful, and so on, instead of the objects or events being referred to. This conception we have seen goes back to Plato and before, and is still with us. Another was the attempt to see the meaning wholly in terms of behavioural stimulus and response, [...] such clearly semantic considerations as personal connotations or abstract ideas show that it is not possible to determine the meaning of a linguistic event solely by observation of the environment and behaviour.

Simplifications of the idea of meaning, for example by defining it as the relationship between words and things, or as the use of language, were common in the early part of this century. And when more sophisticated accounts of meaning developed, there was a tendency to talk about the philosophical or rhetorical implications of the questions, and to state over the theoretical questions presupposed by any systematic analysis. This point can be illustrated from a number of sources — for instance, from the educational doctrines of general semantics, associated with the name of Alfred Korzybski in particular, which aimed at the general improvement of human beings by better training in the use of words and other symbols; or for the semantic analysis practised by the logical positivists; or by the rhetorical and literary discussion of the concept by C. K. Ogden and I. A. Richards in the Meaning of Meaning, published in 1923. Another example of this very general theorizing was in the debates about the relationship between language and reality, carried on by psychologists, philosophers, anthropologist and linguists alike. In particular, there was the question of whether there are universal concepts existing independently of language, or whether language imposes a conceptual framework on our thinking, without our noticing it. “We dissect nature along lines laid down by our native languages”, said Benjamin Lee Whorf; and his arguments, along with those of Edward Sapir, led to the development of a position known as the Sapir-Whorf hypothesis, which distracted attention from more central and testable aspects of semantics for a number of years. The history of semantics is a peculiarly complex one, because so many fields of study are involved: it is well surveyed by Stephen Ullman, in his book Semantics, published in 1962. It is clear from this that the subject of meaning is by no means a discovery of modern linguistics; but at the same time, little of the early work has proved to be of permanent value.

There was, then, much theorizing, but little strict theory. There was little attempt to actually analyse the structure of a language to determine in detail the way in which it parcelled out reality, or to see how words (and other units) defined each other’s meaning in various ways. The use of semantic field techniques was a European development of the thirties which proved to be of value, but it was very limited in scope. This was a method whereby one studied the different sets of words used by different languages to identify the features of a particular area of experience — such as establishing how the English vocabulary of colour cuts up the colour spectrum, and seeing how this differs from the way French, or Japaneese, or Welsh might do it. [...]

More recent than this is the attempt to study all the various relationships of meaning which exist between words in a particular language, an approach, known as structural semantics. Some of these relationships were of course traditionally taught under such headings as antonyms and synonyms; but research has shown that there are far more kinds of relationship binding the meanings of words together than this old pair of labels suggests. However, the problem of how to analyse words which have more than one meaning; the problem of defining idioms accurately; the problem how much detail to allow into the definition of a word; the problem of styles of usage affecting the meanings of words: these are all questions which have received no satisfactory explanation, and an adequate semantic theory, which would have to account for such things, is accordingly a long way off. The existence of large dictionaries listing the meanings of words in a language should not blind us to this fact. There is nothing infallible about a dictionary (or a grammar, for that matter). All dictionaries are artefacts; they have principles about what words they will include and what they will leave out, about how to define meanings, about the order in which meanings should be defined and how they should be grouped, about ways of presenting pronunciation and spelling, about ways of dealing with etymologies, about how much information to introduce concerning stylistic or idiomatic information and so on. It is extremely instructive to take two dictionaries and compare their treatment of any set of words. The differences are remarkable; and for intelligent use, they should be evaluated.

But much of this is fringe linguistics. Lexicography (dictionary-writing) is one of the applications of language study, putting semantic information over in a certain way. Obtaining that information in the first place is independent of any particular dictionary format. And semantics, for the linguist, must be primarily concerned with the problems of how the semantic system hypothesized for a language is organized, and what kind of model might most usefully be constructed in order to facilitate analysis. At the moment, there have been few positive suggestions. There has been some discussion about procedures for obtaining semantic information; a few suggested techniques for analysing certain areas of meaning have been made (such as componential analysis, which is used a great deal in the analysis of kinship systems in anthropology); and there have been one or two theoretical excursions of a highly programmatic kind. [...]