Language change

This chapter considers how and why language changes. It also lists the main methods used to reconstruct past stages of languages for which written records are sparse or unavailable.

All languages are continually changing – their sounds, their syntax and their meaning. This gradual alteration is mostly unnoticed by the speakers of a language, since the syntax in particular gives a superficial impression of being static. Yet one glance at the works of Chaucer or Shakespeare shows how much English has changed in a relatively short time.

A closer look at the English of today reveals several sounds and constructions in the process of changing. [j], the

-like sound which occurs before [u

] in words such as

tune

muse

duty

, seems to be dropping out. It has already disappeared in words such as

rule

lute

. Soon, it may have dropped out entirely, as it has in the East Anglian region of England.

Meanwhile, a change in syntax is occurring in the use of the pronouns I and me. It used to be considered correct to say It’s I. Nowadays, the majority of people say It’s me. Me tends to be used after the verb, and I before it. And there are signs that this rule is being extended so that I occurs only in a position directly preceding the verb. The line in the popular song, ‘Me and the elephant, we still remember you’, is only partially a joke. Such sentences possibly encapsulate this changing rule of grammar.

People working on this branch []()of linguistics are interested above all in how and why language changes. They are also interested in reconstructing an earlier state of affairs in cases where we have no written records of the previous stages of the language.

[]() How language changes

Until relatively recently, language change was considered to be a mysterious, unobservable phenomenon which crept up on one unawares. Like the movement of the planets, it was regarded as undetectable by the unaided human senses. However, advances in sociolinguistics have led to a growing understanding of the mechanisms behind both the spread of change from person to person, and its dissemination through a language.

The American sociolinguist William Labov was one of the first people to examine in detail how a change spreads through a population. He found a new pronunciation creeping in among the permanent inhabitants on the island of Martha’s Vineyard, a popular holiday resort off the coast of Massachusetts. Judging by previous accounts of the islanders’ speech, the vowel sounds in words such as I, my and out, about were altering their character, being produced with the mouth considerably less wide open than the standard American pronunciation. Labov did a survey of these vowel sounds, interviewing the islanders, and asking them to read passages containing the crucial words.

He found that the change seemed to be radiating out from a group of fishermen who were regarded as typifying the old true values of the island, in contrast to the despised summer visitors. The fishermen’s speech had always been somewhat different from the standard American pronunciation, though in recent years it appeared to have become more extreme. The non-standard vowels were being picked up and imitated in particular by people aged 30–45 who had made a firm decision to stay permanently on the island.

The fishermen’s strange vowels, then, were not a totally new []()invention, they were simply an exaggeration of existing vowels. Other inhabitants who came into contact with these respected old fishermen perhaps subconsciously imitated aspects of their speech in an effort to sound like ‘true’ islanders. At first, the adopted fisherman-type vowels fluctuated with their existing more standard vowels, then gradually the new ones took over. At this point, the change started to spread to others who came into contact with this second group, and so on.

There is some truth, therefore, in the notion that changes are infectious. Parents sometimes complain that their children ‘pick up’ dreadful accents at school. But children are not infected against their will. Subconsciously at least, humans imitate those they admire, or desire to be associated with. It is as impossible to stop children acquiring the accents of their friends as it is to stop them wanting to wear the same clothes, or admiring the same pop stars.

[]()Insight

Changes spread from one human being to another often because humans tend to want to talk like others they admire and respect.

Some changes occur ‘from above’, meaning ‘from above the level of consciousness’, when people consciously imitate the accent of others. For example, in British English, someone who comes from an area where \[h] is omitted at the beginning of words such as hot, high, might make a decision to gradually incorporate it, to fit in with the more usual pronunciation. Other changes are ‘from below’, meaning ‘from below the level of consciousness’, as with the Martha’s Vineyard changes, where those involved might have been unaware of which parts of their speech were changing.

But whether the changes are ‘from above’ or ‘from below’, the mechanism of spread from person to person appears to be the same: alternatives creep in, usually copied from those around, then gradually replace the existing pronunciation.

[]()Insight

Change inevitably involves variant []()forms while it is in progress, so speech variation is often a sign that a change is taking place.

[]() Spread of change within a language

The spread of a change through a language is a topic which at one time seemed even more mysterious than its spread through a population. One puzzling phenomenon was the so-called

‘regularity’

of sound change. If one sound changes, the alteration does not only occur in an isolated word. It affects all similar words in which the same sound occurs. So, in English

wyf

became

wife

, just as

lyf

became

life

, and

bryd

became

bride

, all showing a change from [i

] to [a

]. In the nineteenth century, linguists claimed that sound changes were ‘laws’ which worked with ‘blind necessity’, sweeping all before them like snowploughs. But one problem remained. How did odd words get left behind? For example, British English [æ] normally changed into [a

] before [s], as in

pass

fast

disaster

. So why do we still get

gas

mass

aster

tassel

? It seems very puzzling that these sweeping ‘laws’ should rush through a language, yet somehow accidentally miss some words.

One answer was to deny that there really were any exceptions and suggest that such words were borrowed from neighbouring dialects. For example, British \[æ] in words such as gas might be due to American influence. But this feeble attempt to evade the problem was not really satisfactory and detailed questionnaires and surveys have revealed a better answer.

Linguists have now shown that sound changes do not occur simultaneously in all words at once. They move across the language going from one word to another, like apples on a tree, which ripen at round about the same time, but not simultaneously. When charted on a diagram, the progress of a change typically shows a characteristic S-curve: the timescale goes along the bottom, the number of words affected up the []()side (). First, the change touches relatively few words, and affects them variably, in that the new pronunciation is likely to exist alongside the old. In the Martha’s Vineyard vowels, for example, the word I was affected early, but the ‘new’ pronunciation of I did not happen each time, only sometimes. At this point, the change is merely a mild tendency which could be reversed, or even fade out altogether. The words affected early are sometimes the commonest, but phonetic factors are also important: words beginning with vowels tended to be the most affected in Martha’s Vineyard. So I and out were changed early partly because they are common words, partly because of their linguistic shape.

[]()

Figure 13.1.

The early stage of a change, with just a few words intermittently affected, may last a long time. But, at some point, the change is likely to ‘catch on’. It will in all probability then spread fairly fast to a considerable amount of vocabulary, as shown by the steep part of the S in . Towards the end, a change tends to lose its impetus and peter out, so there may be a few words it never reaches. This scenario, then, of a change creeping from word to word accounts for why changes are for the most part regular, but also why some words can get left out. The process is known as lexical diffusion. The words affected by a change fluctuate at first, with the new and old pronunciation coexisting. But eventually, the newer pronunciation wins out.

[]()Insight

Change occurs via a process of lexical []()diffusion, in that it creeps from word to word, slowly at first, then faster and faster, then again slowly towards the end.

[]() Causes of language change

‘There is no more reason for language to change than for jackets to have three buttons one year and two the next,’ asserted one well-known linguist, arguing that all change is due to accidental, social factors. This viewpoint cannot be correct, for two reasons. First, similar changes recur the world over. There are certain tendencies inherent in language, which possibly get triggered by social factors, but which are there waiting in the wings, as it were, for something to set them off, as with an avalanche: a lone skier who disturbed the snow was perhaps the immediate trigger, but deeper underlying causes already existed, before that skier arrived.

Furthermore, language patterning never breaks down. This is the second reason why changes cannot be simply accidental. The patterns within language enable the mind to handle large amounts of linguistic information without strain. If change was random, the organization would collapse. The mind would be overloaded with a junk-heap of disorganized material, and communication would be impossible.

[]()Insight

Changes cannot be random happenings, because similar changes happen all over the world, and because language never loses its underlying patterns.

These factors provide two major causes of change: on the one hand, there are underlying natural tendencies in language, which can get triggered by social factors. On the other hand, there is a therapeutic tendency, a predisposition to make readjustments in order to restore broken patterns. Let us briefly consider these.

[]() Natural tendencies

There are numerous natural tendencies, and []()some of them are stronger than others. They can be triggered by social factors, or may be held at bay for centuries, perhaps held in check by other, opposing tendencies.

A widespread tendency is for the ends of words to disappear. In cases where this has largely occurred already, as in the Polynesian languages, Italian, and French, many English speakers claim the language ‘sounds beautiful’, ‘has flowing sounds’. But when it begins to happen to our own language, and people leave \[t] off the end of words such as hot, what, and replace it with a ‘glottal stop’ – a closure at the back of the vocal tract with no actual sound emitted – then many people get upset, and talk about ‘sloppiness’, and ‘disgraceful swallowing of sounds’. However, such an incident is certainly not ‘sloppiness’, since producing a glottal stop requires as much muscular tension as the sound it replaces. Furthermore, the change is creeping on inexorably: even those who criticize it usually fail to notice that they themselves are likely to replace \[t] with a glottal stop in football, hot milk, a bit more. In some areas, the change has affected \[k] as well, and also, to a lesser extent, \[p]. At the rate at which it seems to be spreading, \[t] and \[k] and \[p] may have disappeared from the end of British English words by the end of the twenty-first century.

[]()Insight

Loss of \[t] from the ends of words in English is a natural tendency which has already happened in several other languages. It is quite unnecessary for people to get upset over it, though speakers should always be careful to make their speech comprehensible to others, and should slow down and pronounce \[t] if the person they are talking to is finding a conversation hard to understand.

Not all tendencies are major, noticeable ones. Others can be minor, affecting only one sound in a particular position: the sound [e] tends to become

[]()

] before [

], so

England

is now pronounced as if it were spelled ‘Ingland’. A [b] tends to be inserted between [m] and [l], so the word

bramble

is from an earlier

bremel

. And so on.

Some tendencies can have repercussions throughout the language, as with the loss of the ends of words. This means that in French, for example, an alternative means of expressing ‘plural’ has had to be developed: it is essential to put les \[le] at the beginning of plural words, as in les chats \[le ∫a] ‘the cats’, since by itself the word chat \[∫a] ‘cat’ is pronounced the same in both the singular and plural, so this distinction is now marked by the determiner placed in front of words.

[]() Therapeutic changes

Therapeutic changes restore patterns which have been damaged by previous changes. A number of examples of this are provided by the use of analogy, reasoning from parallel cases, which is a fundamental feature of human language. It is most obvious in the case of child language, when children create past tenses such as taked, drinked, after hearing forms such as baked, blinked.

In language change, analogy tends to restore similar forms to items which have become separated by sound changes. For example, changes in the vowel system resulted in the separation of the adjective old from its comparative form elder. So a new comparative form older has been formed by analogy with forms such as young, younger, where the first part of both words is the same. The form elder has now been relegated to a few specialized uses and phrases, such as elders of the church, his elder brother.

[]()Insight

Analogy, defined as reasoning from parallel cases, can sometimes restore broken patterns.

However, it is a mistake to regard analogy []()purely as a restorer of broken patterns. This is an oversimplification, because analogy is not only found in a therapeutic role: it can itself disrupt sound changes. For example, we would expect \[d] in the middle of the word father (as in Gothic fadar). But father was influenced by the word brother, and the expected \[d] appears as \[ð]. In addition, there are a number of different types of analogy. In some ways, it is a ‘rag-bag’ category used to explain a variety of changes. So it is misleading to make vague general claims about analogy, unless the statement can be narrowed down.

[]() Changes that trigger one another

Chain shifts

– that is, changes which seem to occur in linked sequences – are a particularly interesting therapeutic phenomenon. For example, in Chaucer’s time the word

lyf

‘life’ was pronounced [li

f] (like today’s

leaf

). The vowel [i

] changed to [e

] (and later to [a

]). But this was not an isolated change. At around this time, all the other long vowels shifted: [e

] became [i

], [

] became [e

], and [a

] became [

]. It looks as if each vowel rushed to fill the empty space left by the one ahead of it (

[]()

Figure 13.2.

Yet the situation is not necessarily as straightforward as the example given here – which is itself disputed. Quite often, several sounds move simultaneously, making it difficult to say whether one sound is being dragged into an empty space, or whether it is being pushed out of its rightful place. In

, is [e

] being dragged into the space left by [i

]? Or is it being pushed by [

Controversy surrounds the actual mechanism []()by which sounds affect one another. The only certain fact is that changes seem to occur in linked sequences, and in so doing preserve the basic patterning of human language.

[]()

Figure 13.3.

[]()Insight

Changes sometimes occur in linked sequences – though it is difficult to tell which sound moved first, and consequently whether we are dealing with pattern maintenance via a drag chain, or a push chain.

Some linguists have suggested that chain shifts occur not only in the sounds of a language, but in the syntax also. For example, some languages have relative clauses (clauses introduced by which, that) placed before the nouns attached to them. They say, as it were:

[]()Which is sour wine

where English says ‘Wine which is sour’. Such languages also tend to have objects which precede their verbs. A literal translation of English ‘Harry dislikes wine which is sour’ might be:

[]()Harry \[which is sour wine] dislikes.

If, however, a change occurs so that relative clauses move to a position after their nouns, then the verb and object are likely to change places also. Once again, most people would []()agree that the changes are linked in some way, though the precise mechanism is disputed.

[]() Interacting changes

So far, the changes examined have been fairly straightforward. But some types of change are more complex. In order to give some idea of the numerous factors sometimes involved in a language change, let us finally look at some interacting changes. In the following example, loss of word endings has combined with changes in word order, to bring about both the disappearance of a construction and a change of meaning.

The forerunner of the English word like was lician, a verb which meant to ‘give pleasure to’. This verb was in common use in English at a time when objects normally preceded the verb, and the subject of the sentence did not necessarily come at the beginning of a sentence. A typical sentence might be:


tham	cynge	licodon	peran
to the	king	gave pleasure	pears

[]()‘Pears pleased the king.’

Then, over the course of several centuries, two things happened. First, the noun and verb endings were dropped. The king became the standard form for the noun, and the plural ending dropped from the verb. This eventually led to the form:

[]()The king liked pears.

Meanwhile, it gradually became normal to put the subject before the verb, and the object after it. So the sentence ‘The king liked pears,’ which originally meant ‘To the king were a pleasure pears,’ was reinterpreted as ‘The king took pleasure in pears.’ Therefore, loss of word endings and a word order change have triggered []()off two further changes: the loss of the construction-type ‘to someone, something is a pleasure’, and a change of meaning in the word like from ‘to give pleasure to’ to ‘to take pleasure in’. This example illustrates the fact that, in many types of language change, a multiplicity of factors are involved.

[]() Reconstruction

An enormous amount has been found out about language change by examining changes in progress. However, language change is a relatively slow process, so we need in addition to consider how languages have altered over the centuries. Yet our written records are inevitably incomplete. In order to supplement these, historical linguists extend their knowledge by reconstructing stages of language for which there are no written documents.

There are a number of different types of reconstruction. The best known of these is external reconstruction, also known as comparative historical linguistics. In this, a linguist compares the forms of words in genetically related languages, that is, languages which have developed from some common source, and then draws conclusions about their common ancestor. This type of reconstruction will be discussed further in the next chapter. An older name for it is ‘comparative philology’, which sometimes causes confusion, since in the USA, France and Germany, ‘philology’ normally refers to the study of literary texts.

A second type of reconstruction is known as

internal reconstruction

. In this, linguists look at the state of one language at a single point in time. By comparing elements which are likely to have had a common origin, they are often able to draw conclusions about their earlier form. To take a simple example, consider the words

long

and

longer

. /l

ð/ and /l

ðg/ (with /g/ on the end) are both allomorphs of the morpheme

[]()

long

. This suggests that, originally, they were identical, and that the word

long

was once pronounced with [g] at the end, as it still is in some parts of England, such as Liverpool.

A third type of reconstruction is typological reconstruction. This is newer than the other two. Linguists try to divide languages into different types, and to specify the basic characteristics attached to each type, a branch of linguistics known as language typology. For example, languages such as Hindi, which have verbs after their objects, also tend to have auxiliary verbs after the main verb. On the other hand, languages such as English which have verbs before their objects tend to have auxiliary verbs before the main verb. If, therefore, we were to find the remnants of a language which had its auxiliaries attached after the main verb, we would be able to predict that it might also have its object before the verb, even if we had no direct evidence for this. Typology will be discussed further in the next chapter.

Let us now see how we might use these three types of reconstruction. Suppose we have three related languages, Twiddle, Twuddle and Twoddle. Let us also assume that we have no past records, merely a record of their present-day speech. First, we would use internal reconstruction (IR) to reconstruct an earlier state of each of these languages – Early Twiddle, Early Twuddle and Early Twoddle. Then we would use external reconstruction (ER) to reconstruct Proto-T, the common ancestor of these three. Then, once again, we would employ internal reconstruction, this time combined with typological reconstruction (TR) to reconstruct an earlier form of Proto-T: Pre-Proto-T ().

[]()

Figure 13.4.

In this way, we might be able to reconstruct []()a probable history of these languages stretching over hundreds, and perhaps even thousands of years.

In this chapter, we have considered how and why language changes. We have also briefly looked at how linguists reconstruct past stages for which they have no written evidence. This process will be discussed further in the next chapter.

[]()THINGS TO REMEMBER

[]()Once, language changes were considered []()to be mysterious and unpredictable phenomena.
[]()The American linguist William Labov was the pioneer who first studied language changes in progress.
[]()Changes spread from one person to another, primarily because humans want to talk like those they admire and respect.
[]()Sometimes people intentionally try to change their speech, at other times they are unaware that they are doing so.
[]()Within a language, changes often spread via a process of lexical diffusion, in which the change creeps from word to word.
[]()Changes cannot be random, because similar changes happen all over the world, and because a language never loses its underlying patterns.
[]()Social factors can trigger a change, which may be caused by underlying natural tendencies in a language, or by a therapeutic tendency, the need to restore broken patterns.
[]()Changes may be complex, because there may be several interacting causes.
[]()We can find out more about changes by including those which happened long ago.
[]()In the case of languages for which we have no written records, there are various ways of reconstructing past stages.