It is dif?cult to see adequately the functions of language, because it is so deeply rooted in the whole of human behaviour that it may be suspected that there is little in the functional side of our conscious behaviour in which language does not play its part. Sapir (1933) Language is a complicated business. In everyday talk, we use the word ‘language’ in many different ways. It isn’t clear how ‘language’ should be de?ned or what the person on the street thinks it actually is! We talk about how miraculously a child’s ‘language’ is developing but how they make charming ‘grammar mistakes’, like me maden that instead of ‘I made that’.
Here, language is an ability that is blossoming in the child. But the word is used in a myriad of different ways. For example, people have strong views about how beautiful or how hideous the ‘language’ is of some region or country or age group; how it sounds to the ear. People say ‘I just adore Italian or an Irish accent. ’ They grimace or smile at teenager talk on television. Here ‘language’ is being judged aesthetically. By contrast, we remark that you can’t really appreciate a culture without knowing the ‘language’, as when we learn French or Japanese for that reason.
Then pupils struggle with rules for tenses like the passe compose and imparfait or have to memorize genders and irregular verb conjugations, matters of grammar which seem a million miles from cuisine, ?lm, high tech or Zen Buddhism. ‘Language’ here equates with grammar. Then, people relate the word ‘language’ to the expression of thoughts. They often say that they ‘can’t ?nd the words’ for their thoughts or express feelings. Or they are ‘hunting for the right words’. Alternatively, we say that language is a means of communication.
Politicians often use as an excuse the fact that their message ‘just isn’t getting across’ because the media distorts what they say. In negotiations or relationships, when communication fails, we say, ‘they just don’t speak the same language’. In another sense, ‘language’ refers to a school subject. It makes sense to say that ‘little Mary is behind in her English’, although you’d never know it when you hear her chatting with her friends. ‘Language’ is being viewed as a set of skills acquired in school. We are taught to write Standard English and spell correctly.
At the same time, we use the term ‘language’ analogically, as a metaphor. We talk of such things as ‘body language’, or the ‘languages’ of music, painting or dance. It is fairly clear that these various ordinary uses of the word refer to different aspects of language, and take different perspectives on the sort of thing language is. Or, alternatively, we have simply grouped together under the heading of ‘language’ a range of diverse phenomena which are only partially related to each other. In order to clarify our thoughts about language, let’s look at some of the ways language is viewed by linguists.
We can then give a precise statement of the speci?cally sociolinguistic view of language, and contrast it to other views of language assumed in linguistics proper. The primary aim of all linguistic scholarship is to determine the properties of natural language, the features it has which distinguish it from any possible arti?cial language. This means that linguistics will be universalistic in its basic aims. It will examine individual natural languages in the course of constructing a theory of universal grammar that explains why the whole set of natural languages are the way they are.
Natural languages, English, French and so on, are in fact the data for this theory of natural language. Arti?cial languages are of interest too since they can exhibit certain properties any language has, but they also have features that can sharply distinguish them from any naturally evolved language. We will look at some arti?cial languages to illustrate this. The linguist Noam Chomsky, in his in?uential book Syntactic Structures (1957), employed the following languages in the course of his arguments: (i) (ii) (iii) ab, aabb, aaabbb, . . . and all sentences of the same type. a, bb, abba, baab, aaaa, bbbb, aabbaa, abbbba, . . . and all sentences of the same type. aa, bb, abab, baba, aaaa, bbbb, aabaab, abbabb, . . . and all sentences of the same type. Why would we want to call (i), (ii) or (iii) languages? The answer is that they have certain properties of any language. They have a vocabulary of symbols, in this case two letters of the alphabet ‘a’ and ‘b’. Also, they have a syntax. That is, each of the languages has speci?c rules for joining together their symbols to produce the sentences or strings of that language.
If the rule of syntax is not followed, then the string or sentence produced is not a sentence of that language. Consider the syntactic rules of the three languages. In language (i) the rule seems to be that for each sentence, whatever the number of occurrences of the ?rst symbol, a, it is immediately followed by exactly the same number of occurrences of the second symbol, b. In language (ii), the rule is that, for each sentence, whatever the arrangement of a and b in the ?rst half of that sentence, then that arrangement is repeated in reverse in the second half of the same sentence.
I’ll leave the reader to work out the equally simple syntax of language (iii). Note that the output of the application of their respective syntactic rules to the symbols of these languages is an in?nite set of strings which are members of the language sharply distinguishable from another in?nite set of strings which are not members of the language. In brief, then, these arti?cial languages have vocabularies and syntactic rules for joining their symbols together. And, by following the rules of their syntax, an in?nite set of strings can be produced.
Natural languages can also be considered in this way. Thus, English can be viewed as a set of strings. And this in?nite set is produced by the vocabulary and syntactic rules of English. If linguists can construct a device, a grammar, which can specify the grammatical strings of English and separate them from the combinations of symbols which are not English, they have gone a considerable distance towards making explicit the syntactic properties of the language. And if the types of rule in that rammar are also necessary for the grammar of any natural language, then they might have discovered some of those universal properties of language which it is the aim of linguistics to discover. Chomsky, in fact, used languages (i), (ii) and (iii) to rule out a certain class of grammars as candidates for grammars of natural language. Of course, these arti?cial languages are also extremely unlike natural languages. One very noticeable difference is that the symbols and strings don’t bear any relation to the world. They have no senses or meanings, but are purely syntactic.
The study of meaning and how it relates symbols to the world is called semantics. There are other arti?cial languages which have strings of symbols which are meaningful. An example is arithmetic. Consider ‘2 + 2 = 4’ or ‘3 ? 3 = 9’. These formulae have a syntax and a semantics. And they are true, while ‘2 + 2 = 5’ is false. These are language-like properties. But there is also something very unlike natural language, the language spontaneously acquired by children, about these formulae. Nothing in the world (we feel) could ever make ‘2 + 2 = 4’ false, as long as the symbols themselves don’t change their meanings.
The formulae appear to be analytic or ‘always true by de?nition’. Contrast this with some sentences from natural language: 1. 2. 3. 4. Arthur is taller than Brenda. Brenda is taller than Tom. Doreen is taller than Brenda. Tom is shorter than X? We can use these sentences to make statements which are true or false, express our beliefs that each sentence designates a state of affairs in the actual world. These sentences are synthetic, true or false according to the facts. (Strictly speaking, it isn’t the sentences which are true or false, but the propositions which they express.
A ‘sentence’ may express many different ‘propositions’. However, I will ignore the distinction in this book. ) We can capture a sentence’s relation with the world by giving its truth conditions. These are precisely the possible worlds – possible states of affairs – in which it is true. For example, 1 is true in worlds where the individual designated by ‘Arthur’ is a member of the class of individuals who are ‘taller than the individual designated by “Brenda” ’; otherwise it is false. Similarly, if ‘Doreen’ is also a member of that class, then 3 would be true, otherwise false.
Only if we know these truth conditions, can we use the sentences to state what we ourselves believe. Or understand what somebody else using the sentence is claiming to be the case. Intuitively, to know truth conditions is part of the ‘meaning’ of the sentences. But sentences also relate to each other. For example, if 1 is true, then Arthur is ‘bigger’ or ‘greater’ than Brenda with respect to her ‘height’ or her ‘tallness’. Synonymy is one example of sense or semantic relations. Such semantic properties constitute inferential relationships between the sentences.
Another example. We know that, if both Doreen and Arthur ‘are taller than’ Brenda, and Brenda ‘is taller than’ Tom, then Doreen and Arthur ‘are taller than’ Tom. We don’t have to look at the world to know this fact. It is a result of a semantic property of the language; the ‘transitivity’ of the predicate ‘taller than’. Similarly, ‘is shorter than’ in 4 bears a systematic semantic relation to its converse ‘is taller than’. Example 1 entails ‘Brenda is shorter than Arthur’. Entailments are inferences that depend on semantic relations.
If one thinks about it, this web or network of sense relations seems to describe features of the very same possible worlds in which the sentences are true. Of course it would, wouldn’t it? This is because inferential relations between sentences are just those relations where the two sentences are both true! Hence, to specify sense relations is a way to partially describe the ‘worlds’ of the truth conditions – the ones in which the sentences are true. Hence it is a way of giving the ‘meaning’ of the sentences. So far, no social factors have been mentioned. How do social factors ?gure in the explanation of language?
They don’t seem directly related to either syntax or semantics. We can begin a treatment of this question by mentioning a few social aspects of semantics. A fundamental factor in making both the arithmetic and natural language examples work is convention. In the ?rst case, of the arithmetic symbols ‘2’, ‘4’ etc. , we have con?dence that when we use them, our addressee will understand that we intend to refer to sets of two and four, 2 and 4, etc. , respectively. This is an example of co-operative social co-ordination. It connects the sound [tu:] or the mark ‘2’ with any set of two things.
It allows an English speaker to use the term with con?dence that their intention will be understood. The ‘sign’ and its ‘object’ have a coded relationship. Similarly with the predicates ‘is taller than’ and ‘is shorter than’. They have a coded relation with the states of affairs they represent. It is important to note that any intrinsic properties that the signs ‘2’ or ‘4’ or ‘tall’ or ‘short’ might have do not explain the link with their objects. Any noise or mark could just as easily be chosen. This is the property of the arbitrariness of the linguistic sign.
Signs and objects are arbitrarily linked, by convention. And this is a social phenomenon. From a different perspective, the connection of world and words isn’t arbitrary, though it is equally social. Consider the web of inferential relations sketched above. The semantic structure of language describes the possible worlds in which sentences are true. Now to even establish this structure it is necessary for us to use the signs to express belief, what we take as actually true, to coordinate ‘taller than’ and ‘shorter than’ with the world as we take it to be.
In essence, semantics de?nes possible states of the world based on our beliefs. Truth has to do with ‘senses i. e. the inferential net’, the relation of ‘words and world’, and ‘our beliefs’. Without the ‘possibly true’ world set given through meanings, we couldn’t inquire, because we couldn’t think hypothetically. Without the inferential relations, we couldn’t reliably think out the consequences of our hypotheses to test them and thus be right or wrong in our beliefs, assent or dissent in the light of experience.
Thus, crucially, the semantic structure of a language is the very resource necessary for humanity to form any empirical theory of the world and use language to inquire – to ?x belief and hence deal with everyday experience, be able to live. That the set of sentences can form a coherent theory can be seen by the fact that, if you believe that 1–3 are true, then you can give a true answer to 4, without further looking at the world. Tom must be the ‘shortest one of all’ in this particular universe of discourse. There is no doubt that the process of inquiry is social.
We have to co-ordinate our beliefs and inferences for language to work. Are there other properties of natural language which require social explanation? The answer is, ‘Yes, there are many such properties. ’ Next we will look at one of the most de?nitive social properties of language. This property is called variability. Consider the English word ‘butter’. On the levels of syntax, vocabulary and semantics, it is a single English item; a mass noun which means something like an edible, yellow, dairy product used in cooking and as a spread.
Yet although it is one item, if I asked you to describe its pronunciation in English, you would not be able to give a single answer: there are various phonetic realizations of ‘butter’. In British English Received Pronunciation the t is made by putting the tongue tip on the ridge behind the teeth, and releasing the air in a small explosion without vibration of the vocal chords. The r, however, is not pronounced, although it is present in the written form. Instead, a vowel sound, schwa (phonetically transcribed as e) follows the t. The schwa is the same sound that is normally ?nal in the word sofa.
Thus, the RP speaker and many other British English speakers say [bvte]. In Canadian and American accents there is a rule that when explosive sounds like t are made between two vowels, the vibration of the vocal chords, called voice, continues through the whole sequence. This has the effect of turning the [t], which is voiceless, into [d], which is its voiced counterpart. Thus, a Canadian saying ‘butter’ in fact pronounces it as if it were ‘budder’. However, Canadians and many of their American neighbours also have r-full accents (as do the Scots and Irish).
This means that, unlike the RP British English speaker, they pronounce the written r in butter, giving us the ?nal form [bvter ]. In many British English accents there is yet another variation in the pronunciation of t in this environment. The vocal chords themselves are closed tightly and then released abruptly, giving the impression that t is missing. In fact, the gap is ?lled by a socalled glottal stop, symbolized by ?. So ‘butter’ is pronounced [bv? e]. Such a pronunciation would typify London working-class speech, familiar to North Americans as a Cockney accent from ?lms like My Fair Lady.
This ?lm, from George Bernard Shaw’s Pygmalion, introduces another feature of the variability we have been describing. For Professor Higgins (modelled by Shaw on the famous phonetician, Henry Sweet) to take such pains to train Eliza Doolittle to pronounce words like ‘butter’ as [bvte], as opposed to [bv? e], indicates that the variation must mean something. There is no conceptual difference in the word-meaning itself. The meaning difference of the variation is socially signi?cant and relates to those groups in a social structure who typically use one form rather than another.
Such social meanings of variants can be further illustrated by looking at two other versions of ‘butter’. In the West Country of England there are some local accents which, like Canadian and some American accents, are r-full. Speakers would typically pronounce the r in ‘butter’. And this can be combined with the use of the glottal stop to give the form [bv? er ]. On British television an advertisement promoting butter used this regional form, presumably because it had a social meaning to British audiences suggestive of honest West Country farmers genuinely in touch with real, non-synthetic cows.
In New York City a working-class accent will, in casual speech, be largely r-less like the British RP. But this would be combined with the voicing of the written ‘t’ between vowels giving the form [bvde]. Followers of the 1970s Kojak detective series on television will recognize this form. Imagine, however, the different social meaning that would be conveyed if Lieutenant Kojak pronounced the word [bvder ] as might an upper-middle-class New Yorker, or [bvte] as might an upper-middle-class Englishman. It would not be the impression of the ‘tough New York cop’.
The diagram opposite gives a summary of the various ways ‘butter’ can be pronounced which we have looked at. The actual situation is far more complex and interesting than I have indicated, but we will be studying this in more detail later in the book. The purpose here is to merely illustrate the property of variability which natural languages possess. It is clear that this property requires social explanation. This is in contrast with the arbitrary property of language mentioned earlier. In characterizing the variant forms of ‘butter’, I needed to make reference to the geographical location in which the form