IELTS Academic Reading Practice Test 48

Of all the dyes obtained from plants, none has had a longer or more eventful history than indigo. Blue is a difficult colour to extract from nature, and for thousands of years indigo was one of the very few reliable sources of it. The dye can be obtained from several plants, the most important being Indigofera tinctoria, a shrub of tropical Asia, and woad, a hardier plant that was grown widely in Europe. Today the colour is so familiar, above all in the denim of blue jeans, that it is easy to forget how much trade, conflict and scientific effort lie behind it.

Curiously, the leaves of these plants are not blue at all. What they contain is a colourless compound called indican. When freshly cut leaves are soaked in water and allowed to ferment, the indican breaks down, and when the resulting liquid is stirred so that it comes into contact with air, a blue substance forms and sinks to the bottom as a sediment, which can be dried into cakes for storage and transport. This blue substance, however, does not dissolve in water, so it cannot simply be mixed into a bath like most dyes.

Instead, dyers prepare an alkaline vat in which, starved of oxygen, the indigo is converted into a soluble form that is yellowish-green rather than blue. Cloth is dipped into the vat and then lifted out, and as it meets the air it changes before the dyer's eyes from green to blue. Each additional dip produces a deeper shade. Because the dye remains on the surface of the fibres rather than penetrating them, indigo-dyed cloth gradually fades with wear, which explains the characteristic appearance of old jeans.

The technique is extraordinarily ancient. In 2016 researchers reported that fragments of cotton fabric excavated at Huaca Prieta, on the northern coast of Peru, contained indigo and were around 6,000 years old. This was some 1,500 years earlier than the oldest indigo-dyed textiles previously known, which came from Egypt. The English name of the dye, however, points to a different region: it derives from a Greek word meaning from India, and the Romans imported indigo from the subcontinent as a costly substance used mainly as a paint pigment rather than for dyeing cloth.

In medieval Europe, blue cloth was dyed with woad, whose cultivation and processing supported prosperous industries in regions such as Thuringia in Germany and the area around Toulouse in France. After Portuguese navigators opened a sea route to India at the end of the fifteenth century, however, increasing quantities of Asian indigo reached European ports.

Indigo produced a stronger colour than woad, and the threat to their livelihoods led woad producers to campaign for protection. In 1577 the German imperial assembly banned what it called the devil's dye, a prohibition repeated in later decades, and similar restrictions were introduced in France and England. The bans could not hold back the superior product indefinitely, and by the eighteenth century indigo had largely displaced woad.

European demand encouraged the growing of indigo in colonies across the Americas, almost always using the forced labour of enslaved Africans. In South Carolina in the 1740s, Eliza Lucas, a young woman managing her absent father's plantations, conducted experiments with indigo seed that helped to establish the dye as one of the colony's most valuable exports. After the American colonies won their independence, British merchants turned increasingly to India, and in particular to Bengal, as a source of supply.

In Bengal, European planters persuaded or compelled local farmers to grow indigo by making advance payments that trapped them in debt, often forcing them to plant the crop on land that could have produced food. Resentment eventually erupted in 1859, when farmers in several districts refused to sow indigo and attacked planters' property. The colonial government responded by appointing a commission of inquiry, which in 1860 criticised the planters' methods and confirmed that farmers could not be forced to grow the crop. The same year, the playwright Dinabandhu Mitra dramatised the farmers' suffering in Nil Darpan, a play that became one of the most celebrated works of protest literature in nineteenth-century India.

The final blow to the natural indigo trade came from the chemistry laboratory. The German chemist Adolf von Baeyer succeeded in synthesising indigo around 1880, but his method was far too expensive for industrial use. The chemical company BASF invested heavily in the search for a cheaper process and, in 1897, began selling synthetic indigo made from naphthalene, a substance obtained from coal tar. The synthetic dye was chemically identical to the natural product, purer and more consistent, and within two decades the area of land devoted to indigo in India had fallen dramatically. Baeyer was awarded the Nobel Prize in Chemistry in 1905, partly in recognition of his work on dyes.

Today almost all indigo is synthetic, and it is produced in enormous quantities for the denim industry. Its manufacture and use are not free of problems, since the chemicals used to dissolve the dye in modern factories can pollute wastewater if they are not properly treated. These concerns have encouraged a revival of interest in natural indigo among craft dyers and fashion designers, and some scientists are exploring ways of using microorganisms to produce the dye. The long story of indigo, it seems, is not yet finished.

The cane toad, Rhinella marina, is a large, heavy-bodied amphibian native to Central and South America. Like many toads, it defends itself chemically: glands on its shoulders produce a powerful toxin that can kill animals that attempt to eat it. Adults commonly measure between ten and fifteen centimetres in length, and exceptionally large individuals can weigh more than a kilogram. In its native range, where predators have evolved alongside it, the toad is an unremarkable part of the ecosystem. In Australia, where it was deliberately introduced less than a century ago, it has become one of the most notorious invasive species in the world.

The introduction was intended to solve an agricultural problem. In the 1930s the sugar industry of northern Queensland was suffering heavy losses to cane beetles, whose larvae feed on the roots of sugar cane. Reports from Hawaii and Puerto Rico suggested that cane toads had helped to control beetles there, and in June 1935 about a hundred toads were brought from Hawaii to Gordonvale, near Cairns. They bred rapidly in captivity, and within a few months more than 2,400 young toads had been released into the surrounding cane fields. Some naturalists warned that the toads might harm native wildlife, but their objections had little effect, and further releases followed.

The toads were a failure at the task they had been given. There is no evidence that they reduced the number of cane beetles, partly because the adult beetles spend much of their time high on the cane stalks, well out of reach of a ground-dwelling toad. The toads simply ate whatever other small animals they could find. Meanwhile, they thrived in their new home. With abundant food, few diseases and almost no predators able to tolerate their poison, they multiplied rapidly, helped by the fact that a single female can lay many thousands of eggs at a time.

From their release sites, the toads gradually spread across northern Australia. In the early decades their range expanded by roughly ten kilometres a year, but the pace later increased considerably, and in some recent periods the invasion front in the tropical north has advanced by more than fifty kilometres a year. The toads crossed into the Northern Territory in the 1980s and reached Western Australia by the late 2000s, while other populations moved south along the coast into New South Wales. Human activity has helped: toads have been carried accidentally in vehicles and cargo, and they use roads and cleared tracks as convenient routes through dense vegetation. They now occupy more than a million square kilometres of the continent.

Why the invasion accelerated became clear through research led by the ecologist Rick Shine and his colleagues. In a paper published in the journal Nature in 2006, they reported that toads at the invasion front had longer legs than those in long-established populations, and that longer-legged toads travelled further and were the first to arrive in new areas. The explanation is a process sometimes called spatial sorting. The fastest-moving individuals inevitably end up together at the edge of the range, where they breed with one another, so that each generation at the front tends to be faster than the last.

For Australian wildlife, the arrival of the toads has often been disastrous. Many native predators have never encountered a toxic toad and attack it readily, and a single bite can be fatal. Northern quolls, small carnivorous marsupials, have suffered severe local declines as the toads have advanced, and large lizards such as the yellow-spotted monitor and freshwater crocodiles have also died in substantial numbers. The impacts are not uniform, however. Some predators appear to learn quickly to avoid the toads, and populations of certain species have recovered after an initial fall.

Efforts to halt the invasion have so far been unsuccessful. Volunteers have collected and killed millions of toads, and traps using lights and recorded calls have been developed, but the animals reproduce far faster than they can be removed. Biological control using diseases or genetic techniques has also been investigated, but no method has yet been shown to be both effective against toads and safe for native frogs. More promising are methods that target the toads' weaknesses. In the arid parts of northern Australia, toads must reach water to survive the dry season, and researchers have shown that fencing off artificial water sources such as farm dams can cause large numbers to die of dehydration without harming native animals.

Since eliminating the toads appears impossible, scientists have also tried to help native predators coexist with them. In one study, captive northern quolls were fed a small dead toad treated with a chemical that caused nausea. The quolls were then fitted with radio collars and released, and those that had been trained in this way survived considerably longer than untrained quolls. Researchers have since developed sausages made from minced toad meat and a nausea-inducing chemical, which can be laid out ahead of the invasion front so that wild predators learn to avoid toads before the real animals arrive. There are encouraging signs that the lesson lasts, since some trained quolls have gone on to survive and breed in the wild. The cane toad's history remains a powerful reminder of the risks of introducing one species to control another.

In January 1954, researchers from Georgetown University and IBM demonstrated a computer that translated more than sixty Russian sentences into English. The system had a vocabulary of around 250 words and only six grammatical rules, and the sentences had been carefully selected to suit it, but the press reported the event with great excitement, and some of those involved predicted that fully automatic translation would arrive within a few years. Seventy years later, machine translation is used by hundreds of millions of people, and its quality would have astonished the pioneers. Yet I believe that we are in danger of repeating their central mistake, which was to confuse an impressive demonstration with a solved problem.

The early optimism did not last. In 1966 a committee advising the United States government, known as ALPAC, concluded that machine translation was slower, less accurate and more expensive than human translation, and saw little prospect of improvement in the near future. American funding for the field fell sharply. The systems of that era were rule-based: linguists wrote dictionaries and grammatical rules by hand, and the computer applied them. Such systems could work well in narrow fields with predictable language. The METEO system, for instance, translated Canadian weather forecasts between English and French for some twenty years from the early 1980s. But ordinary language, full of ambiguity, idiom and exception, required more rules than anyone could write.

The next breakthrough abandoned rules altogether. From the late 1980s, researchers at IBM showed that a computer could learn to translate by analysing large collections of texts that had already been translated by humans, notably the proceedings of the Canadian parliament, which are published in both English and French. Instead of following grammatical instructions, these statistical systems calculated which translation of a phrase was most probable. The method had an important advantage, since it could be applied to any pair of languages for which enough translated text existed, without linguists having to describe each language in detail. Its weakness was that translations were assembled from fragments and often read awkwardly. When Google launched its free translation service in 2006, it was based on this approach.

A further leap came with neural networks. In 2016 Google switched to a neural system that processed whole sentences rather than individual phrases, and reported that, in evaluations by human judges, it reduced translation errors substantially for several major language pairs. Neural systems, and the large language models that have followed them, produce text that is remarkably fluent. Very often it is also accurate. I use these tools myself, and it would be foolish to pretend that they are not enormously useful.

My concern is that fluency is precisely what makes their errors dangerous. A clumsy translation announces its own unreliability; readers are alerted by the awkward phrasing and treat the content with caution. A smooth translation that is subtly wrong offers no such warning. The risk is greatest where mistakes have serious consequences.

A study published in 2021 in the Journal of General Internal Medicine asked native speakers to assess how well Google Translate rendered common hospital discharge instructions into seven languages. The meaning was preserved in 94 per cent of cases for Spanish, but in only 55 per cent for Armenian. I am not arguing that such tools should be banned from hospitals, but they cannot yet be relied on where a misunderstanding could harm a patient.

The study also illustrates a deeper inequality. Machine translation learns from data, and the quality of its output depends heavily on how much translated text exists for a given pair of languages. For English, French or Spanish the supply is vast. For thousands of other languages, digital text of any kind is scarce. Initiatives such as Meta's No Language Left Behind project, which in 2022 released a model covering 200 languages, many of them with little available data, deserve credit. But a model that covers a language is not necessarily one that serves its speakers well, and researchers have warned that low-quality machine-translated text published online may itself become training data, so that errors are recycled rather than corrected.

Then there is the question of what translation actually involves. A literary translator is, in a sense, a second author. Translating a technical manual is largely a matter of conveying information accurately. Translating a novel, a poem or a political speech is something else: it requires decisions about tone, rhythm, cultural reference and what the writer leaves unsaid, decisions on which skilled human translators often disagree. Machines can produce plausible versions of such texts, but they make these choices without understanding the purposes they serve. Professional translators increasingly work by correcting machine output, a practice known as post-editing, and many report that it is efficient for routine material but poorly suited to creative work.

None of this is an argument against machine translation. It is an argument for clarity about what it is. The most sensible course is to use machines for what they do well while keeping people responsible for what matters most. The pioneers of 1954 were wrong to think that translation was a problem that a machine could solve once and for all, and those who now claim that human translators will soon be unnecessary are making the same error. Translation is an act of judgement, and judgement still requires a human being who can be held responsible for it.