IELTS Academic Reading Practice Test 113

In the earliest days of the telephone, a subscriber who wanted to speak to someone else needed a direct wire running all the way to that person's instrument. A network linking every pair of subscribers in a town would therefore have required an impossible number of wires: ten subscribers need forty-five connections, and a hundred need nearly five thousand. The solution, hit upon within two years of the telephone's invention, was to bring every line into a single building and connect them there. This central switching point became known as the telephone exchange, and it made the telephone a practical public service rather than a toy for the wealthy.

The first exchange opened in New Haven, Connecticut, in January 1878. It was the work of George W. Coy, a local contractor who had seen a demonstration of Bell's telephone and obtained a franchise to operate a system in the city. Coy's exchange was little more than a wooden board fitted with sockets and cables. When a caller turned a crank on the telephone, a shutter fell on the board to attract an operator's attention. The operator asked which number was wanted, inserted a plug into the corresponding socket and rang the other subscriber.

At first the operators were teenage boys, but they proved impatient and rude, and within months the company replaced them with young women. Women were thought to be more courteous and to have better manual dexterity, and by the 1880s operating had become one of the few skilled occupations open to them. The work was demanding: an operator might handle hundreds of calls an hour, and the job required intense concentration.

Manual exchanges had obvious weaknesses. Calls took time to set up, operators could overhear conversations, and a mistake could cut a caller off. As networks grew, switchboards became enormous, with thousands of sockets and teams of operators working in shifts. Labour costs rose, and engineers began to seek a machine that could make connections automatically. The first successful system was devised by Almon B. Strowger, an undertaker from Kansas City, who was granted a patent for an automatic telephone exchange in 1891.

Strowger's exchange used electromagnetic switches known as selectors. Early subscribers pressed buttons to signal, but in the form that spread, a caller operated a rotary dial, and the dial interrupted the current a fixed number of times, sending a train of electrical pulses down the line. Each pulse drove a ratchet in the selector forward by one step, so that dialling a digit moved a wiper to the corresponding row of contacts. A second selector then hunted for the required line. The first automatic exchange was installed at La Porte, Indiana, in 1892, and although early systems were unreliable, they removed the need for an operator.

Automatic working spread slowly. Many companies retained manual exchanges well into the twentieth century, because the equipment was expensive and customers were used to asking for a number by name. The next major advance was the crossbar switch, first tried in Sweden and widely adopted from the late 1930s. Instead of a wiper moving across a bank of contacts, the crossbar used a grid of horizontal and vertical bars; electromagnets selected one bar in each direction, and a contact was made where they crossed. Crossbar exchanges were faster and more reliable than stepping switches.

The arrival of electronics transformed switching again. In May 1965 a new exchange opened at Succasunna, New Jersey, using stored-program control. Rather than relying on electromechanical relays, the system used a computer to decide which path a call should take and to record billing information. The computer could be reprogrammed to provide new services, such as call waiting and abbreviated dialling, without rewiring the exchange. Electronic exchanges gradually replaced crossbar equipment in the 1970s and 1980s, and the software that controlled them became as important as the hardware.

Transmission changed at the same time. Telephone companies began to convert speech into digital signals using pulse code modulation, a technique invented in 1937 by Alec Reeves. The continuous sound wave is sampled thousands of times a second, and each sample is recorded as a number. Digital signals can be regenerated without accumulating noise, and they allow many conversations to be carried together by time-division multiplexing, in which each call is allotted a tiny slice of time in a repeating cycle. Digital exchanges switched these streams directly.

Fibre-optic cables, which carry information as pulses of light rather than electricity, replaced copper trunk lines from the 1980s onwards. A single fibre can carry tens of thousands of calls, and the cost of sending a call over long distances fell sharply. Meanwhile, the internet introduced packet switching, in which speech is broken into small packets that travel independently and are reassembled at the destination. Internet telephony now carries a large share of the world's calls, and the old distinction between telephone and data networks has largely disappeared.

Modern exchanges are no longer rooms full of clicking relays. They are data centres, where software running on servers routes calls, connects mobile phones to the network and manages the signalling that sets up a conversation. Mobile networks have their own core switches, and cloud services increasingly replace physical equipment. The technology has changed beyond recognition. Yet the principle established in New Haven in 1878 remains unchanged: a call is possible because a central point connects one line to another, if only for a few minutes.

Agriculture has always depended on information: which part of a field is short of water, where weeds are spreading, whether a crop is ready to harvest. Traditionally this information came from walking the land or from satellites, but both have limits. A farmer on foot can inspect only a small area each day, and satellite images are often too coarse or too cloudy to reveal problems in individual rows. In the past decade, unmanned aerial vehicles, commonly called drones, have given growers a third option. A drone can fly slowly and low over a field, carry a range of sensors and return within minutes with detailed images of the crop below.

Most agricultural drones are either multirotors or fixed-wing aircraft. A multirotor, with four or more rotors, can take off vertically, hover and fly slowly along a precise path; it is ideal for inspecting a small area or for spraying a few rows. Its weakness is endurance, since battery power limits most models to twenty or thirty minutes of flight. A fixed-wing drone can cover hundreds of hectares in a single flight and is more efficient for mapping large fields, but it needs a runway, a catapult or a hand launch, and it cannot hover. A third design, the hybrid vertical-take-off aircraft, combines the two.

The sensors carried by a drone determine what it can detect. An ordinary colour camera records what the human eye would see and is useful for counting plants, measuring canopy size and spotting lodged or damaged crops. A multispectral camera records light in narrow bands, including near-infrared, which is invisible to the eye. Healthy leaves reflect strongly in the near-infrared and absorb red light for photosynthesis. By comparing the two, software calculates a vegetation index, commonly NDVI, which gives a numerical measure of how much living, active plant material is present. Thermal cameras detect differences in temperature and can reveal water stress, because a crop that lacks water closes the pores in its leaves and warms up.

To turn images into a map, the drone flies a programmed route, taking overlapping photographs at regular intervals. A GPS receiver records the position of each image, and software matches features shared by neighbouring photographs, a process known as photogrammetry. The result is an orthomosaic, a single image of the whole field in which every pixel is linked to a coordinate. Vegetation-index maps are then laid over this image, so that a farmer can see exactly where growth is poor. The same data can produce a three-dimensional model of the crop, from which plant height and volume are estimated.

The first use of these maps is diagnosis. A patch of yellowing crop may be caused by a nutrient shortage, a disease, waterlogging or pests, and a drone image cannot always distinguish between them. But it can direct the farmer to the affected area, where a closer inspection or a soil test can establish the cause. Repeated flights show whether a problem is spreading and whether an intervention has worked. In this role the drone is less a decision-maker than a guide, narrowing the area that needs attention.

The second use is action. Spraying drones carry a tank of pesticide or liquid fertiliser and a set of nozzles, and they can apply a treatment only where the map shows that it is needed. This targeted approach can reduce the volume of chemicals used, although the savings depend on the crop and the pest. Some drones are designed to scatter seeds or beneficial insects, and others drop small capsules of nutrients. The technology is not fully autonomous: a human operator must supervise the flight and refill the tank, and in many countries spraying from the air is tightly regulated.

Drones are also used with livestock. A thermal camera can pick out sheep or cattle against a cold field, making it easier to count animals and to find one that has separated from the herd. Regular flights can show whether animals are grazing evenly or crowding around a single water trough. Some farmers use drones to move livestock gently from one pasture to another, although the animals may be disturbed by the noise.

Several limitations remain. Battery life restricts flight time, and the payload of a small drone is only a few kilograms, so spraying must be repeated often. Wind and rain can make flying unsafe or blur the images. Regulations in most countries require the drone to remain within the operator's sight, which limits the area that can be covered in one flight. The volume of data is another challenge: a single flight may produce hundreds of images, and analysing them requires software and a trained agronomist. For these reasons, many growers use a specialist service rather than owning an aircraft.

Research is addressing these problems. Engineers are developing lighter batteries and more efficient rotors, while agronomists are training machine-learning systems to recognise specific diseases from images rather than relying on a general vegetation index. Others are experimenting with swarms of small drones that cooperate to cover a field, or with drones that carry soil-moisture sensors and land to take readings. The most promising direction may be integration: drones, satellites and sensors fixed in the ground each provide information at a different scale, and software that combines them can give a more complete picture than any one source alone.

Few products of the smartphone era have been adopted as enthusiastically as language-learning apps. At the last count, the most popular of them claimed hundreds of millions of registered users, and its green owl mascot has become a familiar figure on screens around the world. The appeal is easy to understand. Lessons are short, free at the basic level and available at any hour; progress is measured in streaks and virtual rewards; and the app promises that a few minutes a day will be enough to master a language. For a generation that has grown up with games and social media, the format feels natural.

The claims are not entirely empty. A large body of research in cognitive psychology supports spaced repetition, the technique by which an app brings back a word or phrase just before the learner is likely to forget. Testing, rather than mere rereading, is another well-established aid to memory, and apps that ask users to translate or to choose the correct form are in effect giving them frequent tests. The best apps also adapt to the learner, presenting more difficult items more often. These are genuine advantages, and they help to explain why even a few weeks of regular use can produce measurable gains in vocabulary.

Yet the evidence for broader fluency is much weaker. Reviews of research on mobile-assisted language learning have found that reported gains are modest and often short-lived, and that few studies measure whether learners can hold a conversation. The reason is not hard to find. An app can teach a word, but it cannot replicate the uncertainty of a real exchange, in which the other person speaks quickly, uses unfamiliar idioms and expects an immediate response. Speaking a language is a skill, and like any skill it improves with practice under realistic conditions. Multiple-choice tapping is not that practice.

This is not to say that apps are useless for speaking. Some now include speech recognition, which compares the learner's pronunciation with a model and offers a score; others pair users for text or voice chat. But speech recognition is better at judging whether a word has been pronounced intelligibly than at teaching the rhythm and intonation of a language. And conversation partners found through an app may be patient and encouraging, valuable for confidence but unlike the demands of a workplace or a shop. The learner who wishes to speak well still needs hours of unscripted interaction with fluent speakers.

Gamification is another area where the design of apps has been criticised. Streaks, points and leaderboards motivate users to return, and for a beginner the habit of daily practice is worth a great deal. But engagement is not the same as learning. A learner who repeats an easy exercise to preserve a streak may be spending time on the app without improving. Some apps have been accused of making the free version deliberately frustrating so that users pay for more; others offer so many rewards that the language becomes a secondary concern. Design that maximises time in the app may not maximise progress.

Cost and access are often presented as the great democratic advantage of apps. A subscription is far cheaper than a course at a language school, and the free versions cost nothing at all. For people who live far from a teacher, or who cannot afford one, an app may be the only structured introduction to a language. That is a real benefit. But the advantage is less complete than it appears. Effective use of an app requires a phone with a good screen, a reliable internet connection and a quiet place to practise; and the learners who progress furthest are often those who also have access to books, films and native speakers.

Teachers, meanwhile, have developed a more nuanced view. Many encourage students to use apps for vocabulary revision, a task at which they are genuinely efficient, while reserving class time for conversation, feedback and explanation. This blended approach treats the app as a tool rather than a syllabus. It also addresses the danger of false confidence. A learner who has completed a hundred lessons may believe that they are ready to negotiate a contract or understand a news bulletin, when in fact they have met only a narrow range of language. The app's cheerful progress bar can conceal how much remains.

The commercial interests of app developers add another complication. A company that charges a subscription benefits from retaining users, and the features that keep people coming back are not always those that teach them most. Independent evaluation is difficult, because much of the published evidence comes from studies funded by the companies themselves. This does not make the findings false, but it should make us cautious. We need more research by teams with no financial stake, comparing app use with other forms of study and measuring real communicative ability rather than scores inside the app.

My own view is that language-learning apps are neither the revolution their advertisements suggest nor the waste of time their critics describe. They are excellent for the early stages: for building a vocabulary, for learning the sounds of a new script and for establishing a daily habit. They are poor at taking a learner to fluency on their own. The most sensible policy is to use them as part of a larger programme, alongside conversation, reading and, above all, contact with people who speak the language. An app can open the door; it cannot walk through it.