Object language and metalanguage
Like any other branch of linguistics, semantics deals with the words, phrases and sentences with which we communicate. But for semantics the immediate objects of study are not these words, phrases and sentences themselves, in the sense of the sounds, sequences of letters or handsigns which we utter or perform and can then write down or record. As the study of meaning, semantics is interested in something which cannot be perceived directly through our senses, but which, in one way or another, we experience in using and thinking about language. We cannot see, hear or touch a word’s meaning: meanings are things we understand. It is not meanings that go between speaker and hearer: the only things that are transferred from one speaker to the other are sound waves in the air. This means that in order to get started in semantics, we need a way of identifying meanings and bringing them to light in an unambiguous way so that we can begin to study them.
The main way in which we normally reveal the meanings of linguistic expressions is, quite simply, by describing them in language. But since it is language that we’re interested in in the first place, we need to distinguish between the language whose meanings we want to describe and the language in which we couch the descriptions. The language whose meanings we are describing is called the object language. The language in which we describe these meanings is called the metalanguage.
When we propose a metalanguage description of the meaning of an object language expression, we are using one type of meaning (the meaning of the metalanguage expression) to explain another kind (the meaning of the object language expression). Let us take the example of English as the metalanguage for Dutch, as would be the case if we were explaining the meaning of the Dutch word groot to an English speaker. One possible metalanguage explanation, or definition, that we could give of groot in English would be the word ‘tall’, as in (35):
The object language expression groot is here defined by the metalanguage expression ‘tall’. But note that not all aspects of the word ‘tall’ are relevant to this definition: it is completely irrelevant to the definition of groot that the metalanguage definition we have chosen, ‘tall’, has four letters, or is a monosyllable, or starts with the consonant /t/. All these phonetic and orthographic details are irrelevant to semantics, since the only thing that matters for the purpose of defining groot is what ‘tall’ means. To tell someone that groot means ‘tall’ is to make a statement about two meanings, and to say that these two meanings are the same. The particular phonetic and other characteristics of the metalanguage term are therefore irrelevant: as long as the person for whom the definition is intended understands the meaning of ‘tall’ in English, the definition is successful.
This confronts us with an interesting problem. For couldn’t it be objected that, in one way, we haven’t actually explained anything when we define groot as ‘tall’? We have certainly given a definition of the word which will help an English speaker to understand the meaning of the Dutch sentence. But if we want to go beyond the problem of allowing people to translate from one language to another, hasn’t our analysis left something crucial out? On hearing our explanation that the meaning of groot is ‘tall’, someone might easily object by pointing out that this explanation only shows an equivalence between two words in English and Dutch, and does nothing to explain what this meaning, which both groot and tall express, actually is. ‘I know what groot means in English’, they might say, ‘but you haven’t told me what it actually is for something to mean something.’ And even though we could go on to give a more detailed explanation of ‘tall’, perhaps using terms like ‘elevated in stature’, ‘not short’, etc., for as long as we continue to explain the meaning of a word by using the meanings of other words, we will not have satisfied our objecter’s curiosity.
For many linguists, this objection is rather forceful. As long as we go on defining meanings by other meanings, we leave out the essential task of explaining what meaning actually is. We can see this very clearly by considering the case of circular definitions. Consider someone who wants to find out the meaning of the English word ‘humorous’. One possible definition of ‘humorous’ would be ‘droll’. But this definition would only be effective if the meaning of ‘droll’ was already known. If it was not, it too would need to be explained: ‘droll’, perhaps, could be plausibly explained through the metalanguage definition ‘amusing’. ‘Amusing’, in turn, could be defined as ‘funny’, as in (36).
Depending on the person for whom the definition was intended, this chain of definitions would sooner or later achieve its purpose: if the per son knew the meaning of ‘funny’, we could stop the explanation at this point, so that ‘humorous’ would have been defined through ‘droll’, ‘droll’ through ‘amusing’, and ‘amusing’ through ‘funny’. It is obvious, however, that this chain could not go on forever. Sooner or later we would run out of new words: if the language learner did not know even what ‘funny’ meant, we can imagine giving up in frustration, and saying, simply, ‘funny just means “humorous”’. In this case, it’s clear that our unfortunate language learner would be none the wiser, since ‘humorous’ was the word whose meaning was originally in question. Since ‘humorous’ has been used both as an object language term and a metalanguage term, the definition is circular and does not succeed in telling us anything new:
Clearly, then, for as long as we remain within the circle of definitions by substituting one word or phrase as the definition of another, we remain confined within language. The lexical resources of any language are limited: at some point, the metalanguage definitions will have to include object language terms, and thereby introduce circularity. We can continue to refi ne our definitions and search out the most precise and explanatory ways of couching them, but in contenting ourselves with this task we will not have provided any account of what the meanings we are defining actually are, nor of how they relate to any of the three points of the semiotic triangle. In particular, we will have left it completely obscure what it is for a speaker to understand the meaning of a word. If I understand the meaning of ‘droll’, then the definitional chain can be stopped. But what does it mean to say that I understand the meaning of ‘droll’? What is it that I actually understand? For many linguists, the fact that we cannot answer these questions about meaning by remaining inside the definitional circle means that we have to look outside language for answers. If linguistics is to play a part in explaining the way language can be actually used by real speakers, we need to find a point at which the circle can be broken in order to link meaning in with something non-linguistic. We will consider a few proposals about how this could be done in the next section.
