IELTS Academic Reading Practice Test 449

From the flat ground a mile or two to the north, the figure on the hillside looks unmistakably like a horse. It has a long body, an arched neck and a small head with what appears to be a beak, and its tail is a single sweeping stroke. The horse lies on a steep slope of the Berkshire Downs in Oxfordshire, just south of the village that gives it its name, and it measures roughly 110 metres from nose to tail. Seen from close up, it dissolves into trenches of dazzling white stone.

The white is chalk. The figure was made by cutting shallow trenches through the turf and the dark soil until the chalk bedrock was reached. The hollows were then packed with crushed chalk, which reflects the light far more strongly than the surrounding grass. The chalk ridge here belongs to the same band of rock that runs across southern England and ends in the cliffs at Dover. Left alone, the trenches would quickly be colonised by grass and scrub, and within a few decades the horse would vanish; the figure survives only because people have repeatedly cleared it.

The hill above the horse is crowned by Uffington Castle, an Iron Age hillfort whose earth ramparts still enclose a large grassy area. Below the horse, a natural chalk knoll called Dragon Hill rises from the valley floor, and tradition holds that this is where Saint George killed the dragon; the bare patch on its summit is said to be ground on which no grass will grow, because the dragon's blood soaked into it. A short walk along the Ridgeway, an ancient trackway that passes close by, brings the visitor to Wayland's Smithy, the remains of a Neolithic tomb.

For most of the twentieth century the horse was assumed to be the work of Iron Age people, partly because of the hillfort above it and partly because stylised horses of very similar design appear on British coins of that period. That assumption was challenged in the 1990s, when soil buried beneath the figure was dated using a technique called optically stimulated luminescence, which measures how long mineral grains have been hidden from sunlight. The results pointed to a period between roughly 1400 and 600 BC, in the later Bronze Age, making the horse several centuries older than the fort.

Some archaeologists have questioned the new date, noting that samples were taken from only a few points and that the figure may have been repaired or remade many times. The name of the hill, at least, is old: it appears in documents from the twelfth century onwards, and medieval records show that the figure was already a familiar landmark.

Medieval writers offered a different explanation of the horse's origin. According to a legend repeated for centuries, it commemorated a victory by King Alfred over the Danes at the Battle of Ashdown in 871. The story is appealing, but it cannot be reconciled with a Bronze Age date, and historians now treat it as an invention of later centuries. Other suggestions — that the figure marked a territorial boundary, that it represented a goddess worshipped by local people, or that it guided travellers along the Ridgeway — remain speculation.

Its meaning is not the only puzzle. Scholars have also asked why a horse should be drawn in this peculiar way, with a body reduced to a few bold curves and legs that are barely indicated. One answer may be that the designers were working for viewers at a distance: details that would be legible at close range would blur into nothing when seen from the valley below. The same economy of line appears on Iron Age coinage, where horses are reduced to a handful of strokes.

Keeping the horse white has always required effort. In the eighteenth and nineteenth centuries the scouring of the figure was an occasion for a fair, with food, drink and wrestling matches, and parishes were expected to send labourers to clear the chalk. Records of these events survive in local newspapers. Scouring continues today, though on a smaller scale, and volunteers work with hand tools, removing turf and weeds and adding fresh chalk to worn patches.

During the Second World War the figure was covered with turf and enclosed by barbed wire, so that it could not serve as a landmark for enemy aircraft. It was uncovered again once the war had ended. Since then the surrounding land has been protected: the horse is scheduled as an ancient monument, and visitors are asked to admire it from a distance rather than walk on it, because the chalk compacts underfoot and the edges of the trenches crumble.

Britain has many other white horses, most of them cut into hillsides in Wiltshire during the eighteenth and nineteenth centuries, when such figures became fashionable. The Uffington horse is by far the oldest, and the only one whose origins lie in prehistory. It remains, as it has been for perhaps three thousand years, a shape that makes sense only when you step back; and it is still, for the thousands of people who walk the Ridgeway each year, one of the most surprising sights in southern England.

Anyone who listens to recordings of English speech made a century ago notices how much has altered. The voices on early gramophone records sound, to modern ears, clipped and strange, with vowels in unexpected places and consonants that are now usually dropped carefully pronounced. Such change is not a matter of a few individuals: whole communities shift, so that a way of speaking that once marked a particular town or class gradually becomes something else. Pronunciation is the part of language that changes most readily, and it does so continuously, though usually too slowly for any single generation to notice. Recordings from the middle of the last century, played beside those made today, make the process audible.

The most dramatic English example occurred between roughly 1400 and 1700, when the long vowels of Middle English were rearranged in what is known as the Great Vowel Shift. The vowel of 'time' moved upwards, the vowel of 'see' moved too, and sounds that had been pronounced at the front of the mouth migrated towards the back. Because printing preserved the old spellings, modern English is full of words whose written form records a pronunciation that no longer exists. Change of this kind is not a simple drift in one direction, however; it tends to proceed as a series of linked movements. Similar rearrangements have occurred in other languages, though rarely so well documented.

One reason for those linked movements is that vowels within a language must remain distinct from one another. If one vowel begins to move, it may encroach on the space occupied by its neighbour, and the neighbour in turn shifts to keep out of the way. Linguists call the result a chain shift, and a striking example is under way in the cities of the American Great Lakes region. From the middle of the twentieth century the vowel of 'cat' began to be pronounced higher in the mouth, and other vowels have since followed it in a stately sequence of displacements. The change has spread across cities such as Chicago, Detroit and Buffalo.

The southern hemisphere offers parallel cases. Recordings made in the 1940s, compared with those made today, show the movement clearly. In New Zealand the short vowels of 'ship', 'sheep', 'bed' and 'bad' have all moved, so that visitors from Australia sometimes hear 'fish and chips' as something close to 'fush and chups'. Australian English has its own constellation of changes, and although the two varieties began from similar foundations, they are drifting apart rather than together. Comparable shifts have been described among young speakers in Canada and in parts of South Africa.

In England the changes most often noticed are consonantal. In many urban accents the 'th' sounds at the start of words such as 'think' have become 'f', giving 'fink'; the 'l' at the end of 'milk' has turned into a vowel; and the 't' in the middle of 'butter' is increasingly replaced by a catch in the throat. These features were once confined to London and a few other cities, but they have spread outwards along the roads and railways into the surrounding towns. The term 'Estuary English', coined in the 1980s, was an early attempt to name the result. It has since been applied to a very wide range of speech, and linguists disagree about how useful the label is.

Why do speakers adopt such features? Part of the answer is that pronunciation carries social meaning. In the 1960s the American linguist William Labov studied the island of Martha's Vineyard, off the coast of Massachusetts, and found that younger islanders were exaggerating a vowel sound in words such as 'house' and 'right'. The exaggeration was strongest among those who intended to stay on the island and resented the summer visitors, and it appears to have served as a badge of local identity. Speech, in other words, is a way of saying who you are, and change may be driven as much by social allegiance as by any internal pressure in the sound system.

Children and adolescents, rather than parents, appear to be the engine of such change. Studies of families in which the parents speak one way while their children speak another show that young people acquire much of their pronunciation from their friends. Contact between speakers matters too: when large numbers of people move into a city, or meet at a university, accents tend to mix, and some features spread while others disappear. It is often assumed that television and the internet must make everyone sound the same, but researchers who have looked for evidence that heavy viewing reshapes local speech have found little of it.

What remains genuinely uncertain is why a particular change begins when it does. Linguists can describe a shift in detail and track its progress across generations, but they cannot predict which vowel will move next, and attempts to forecast the direction of change have a poor record. What can be said with confidence is that accents have never been stable. Nor is there any sign that the process is slowing down. The belief that speech is in decline is very old and has been expressed in every century for which records survive, yet no generation has produced evidence that the language is the worse for it.

Almost every debate about online speech arrives eventually at the same question: when a post is defamatory, threatening or simply false, who should answer for it? For two decades the dominant answer in many countries has been that the platform hosting the material is not its author. That settlement is now being revised on several continents at once, and the arguments for and against it are rarely as simple as the headlines suggest. Platforms remove millions of items each year, yet the rules they apply are written by themselves rather than by parliaments, and they vary from one service to another.

The best-known legal foundation is a short provision in American law, section 230 of the Communications Decency Act of 1996. It states that an online service should not be treated as the publisher or speaker of information supplied by its users, and it adds that a company which removes objectionable material in good faith should not be penalised for doing so. The provision was written when the internet was young, and it is widely credited with allowing platforms to grow without facing a lawsuit over every post. It also means that a company hosting a defamatory statement is generally not liable for it, even after being told that the statement is false.

The companies defend this arrangement with an argument about volume. Several billion items are posted every day, and no workforce could examine more than a fraction of them; automated systems deal with the rest and make mistakes. If a platform were legally answerable for what users publish, its lawyers would advise it to remove anything that might be questionable, since the cost of leaving a post up would always exceed the cost of taking it down. Smaller services, which cannot afford large compliance teams, would be driven out of business, and the largest firms would be strengthened. The result, they say, would be a service that is safe but dull, stripped of the argument and humour that make it worth using.

Critics reply that the picture of the platform as a passive conduit has been out of date for more than a decade. The material a user sees is chosen by software that ranks, recommends and sometimes promotes posts in order to hold attention, and advertising is sold against that attention. When a company's own systems decide which of a million items reaches a particular teenager, the claim that the company merely hosts content begins to look strained. The distinction between hosting and publishing, once clear, has been blurred by personalisation. Advertising, meanwhile, rewards whatever holds attention longest, whether or not it is accurate.

Civil liberties organisations make a different objection. If private firms are made responsible for deciding what is lawful, they will do so quickly, cheaply and without the safeguards of a court. The people most likely to lose their voices are those whose speech is already marginal: minority groups, journalists in small countries and individuals without the money to sue. Appeals against removal are slow, and a small publisher rarely has the resources to pursue one. The organisations add that automated filters cannot recognise irony, context or quotation, and that the power to police content is public power transferred to unaccountable bodies.

Campaigners whose families have been damaged by online material take the opposite view. For them, the past twenty years have shown that voluntary codes and promises of self-regulation achieve little, because a company's commercial interest in engagement points the other way. Their testimony before legislators has been influential in several countries, and they argue that a business which profits from a system designed to spread material should be answerable for the consequences when that system spreads harm. Several governments have cited such cases when introducing new laws, and the families themselves have usually asked for regulation rather than for the deletion of particular posts.

Legislators have responded with two broad models. Germany's Network Enforcement Act, passed in 2017, required large platforms to delete clearly illegal material within a short fixed period or face heavy fines. Researchers note that the law produced swift removal but also a rise in over-removal, as companies erred on the side of caution. The European Union's Digital Services Act, agreed in 2022 and applied from 2024, takes a different route. It requires the biggest platforms to evaluate the dangers that their services pose, to submit to independent audits, and to explain their decisions to users. The Act also requires platforms to publish reports on the action they have taken, a requirement that supporters regard as a first step towards genuine accountability.

In my view, the question has been framed in the wrong way. A company should not be treated as the author of everything its users say; that would turn it into a censor with an impossible task. But the systems that decide what millions of people see are the company's own creation, and a firm can reasonably be required to take care in operating them. A duty of care, backed by public transparency and a genuine right of appeal, is the right shape for the law. What matters is not the size of a company's moderation budget but whether its design choices make harm likely, and whether it can show that it considered the alternatives. Responsibility, yes; liability for every individual post, no.