Veja agora mesmo a nova edição #71 da Revista New Routes na íntegra!

Editorial NRInglês

The automated english teacher?

WHY MACHINES MIGHT NOT BE MAKING US OBSOLETE SOON

BORIS AND THE TURK

On a cold day in December of 2018 the Russian youth technology forum called Proyectoria was held in city of Yaroslavl, around 150 miles north‑east of Moscow. It was a televised event, with a stage and two presenters sharing various technology related projects with the young people in attendance. At one point they brought out the piece‑de‑resistance of the forum: a human‑sized robot called Boris. The presenters said that Boris could walk, talk, do maths and even dance.

However, many Russian bloggers were not convinced. Why had nobody heard about Boris online? Where were the sensors? And why did it look as though Boris was made exactly big enough to have a human being inside?

The answer was, of course, that there was a human being inside. After the forum, a picture emerged of Boris backstage, with his helmet removed and the head of a young man visible inside the robot. It turns out that the robot was made by a costume company. In the end, the show and the news channel both said they had never really implied it was a real robot and denied trying to mislead viewers.

Now this could just be chalked up to another one of those odd news pieces you find on the internet. But it is not the first time that a robot has been hailed as the next unbelievable technology, only to not appear what it seemed.

The Mechanical Turk, also known simply as The Turk, or the Automaton Chess player was a chess‑playing machine built in the late 1700s. It was constructed by a Hungarian named Wolfgang von Kempelen, originally to impress the Empress Maria Theresa of Austria. The Mechanical Turk was toured around several royal courts of Europe and was able to beat many humans it played against, including such illustrious opponents as Napoleon Bonaparte and Benjamin Franklin. It was hailed as an incredible piece of technological wizardry.

The “robot” Boris.
Photograph: MBKh Media

Until, of course, the secret eventually got out. There was a very small, uncomfortable cubby hole where a human could hide inside the Turk. At least five chess masters had hidden inside it during its operation over nearly a century.

Boris the Robot and the Mechanical Turk are both, I think, useful stories. They fit into a larger picture of stories we are told, and that we tell, about magical technology ‑ and specifically robots.

AUTOMATION AND THE COMING OF THE ROBOTS

Today, most of the talk about robots in the news or in popular culture is about automation. The first wave of this was the automation of elements of factory or agricultural work, by robot arms and machines. For example, from 1900 to 2000 the US agricultural workforce went down from 41% of the entire workforce to a mere 2%. More recently, advances in artificial intelligence, robots and machine learning are predicted to replace workers in many other sectors. Transport, accounting, clerical and administration work and even many services are said to be at risk. At least one presidential candidate for the Democratic Party of the United States, Andrew Yang, has made the anxiety about the coming of automation the centerpiece of his campaign. And while the numbers quoted above are from the United States, similar patterns are repeating around the world. A 2017 report, compiled by the McKinsey Global Institute, claimed that advances in AI and robots could cause the loss of as many as 800 million jobs worldwide.

Are there signs of this in education, and specifically in English language teaching? There have been some experiments. EngKey was a white, egg‑shaped robot developed in the early 2010s in South Korea. It wasn’t entirely a machine, as the ‘head’ was in fact a screen, which connected to a real teacher in another country such as Australia or the Philippines. EngKey was deployed to teach children in South Korean classrooms. A European project from 2018, called L2TOR (pronounced ‘el tutor’) introduced a small humanoid robot that was designed to teach limited English (colours, prepositions) to five‑year olds. There have also been experiments with teacher robots in Singapore, New Zealand the UK and the US to name but a few.

These experiments have had limited success. For example, one newspaper article about a 2009 Japanese robot teacher named Saya dryly claimed that the robot could not do “much more than take the register and shout orders like ‘Be quiet’”. Designers of the L2TOR robot say that they are training the robot to recognize human emotions in order to react better to what is happening in the classroom, but this still seems to be some way off.

It seems therefore that these robots have not taken off in popularity in the same way as other new technologies. For example, it is now common to see projectors and computers in many language classrooms around the world. Teachers use the web to find materials for their classes. Online elements of language courses are also becoming increasingly common. But robots? Speaking personally, over the past fifteen years of travelling around the world and visiting schools with all manner of technology (from high‑tech to low‑tech to no tech) I have yet to see one robot in a classroom. And although at least one expert has boldly predicted that robots will replace teachers by 2027, the world of classrooms being led by robot teachers seem to be based more in faith than in fact.

[…] The language teacher is not to be replaced by a humanoid robot in the classroom. Rather, the whole concept of the language classroom […] will be completely obsolete as everyone will be communicating to each other via translators built into their phones, and eventually into little earbuds.

ARTIFICIAL INTELLIGENCE AND LANGUAGE LEARNING

If the mechanical robot teacher has yet to explode in popularity, it does not mean that other forms of technology or automation aren’t being touted as replacements for teachers. The popularity of home‑based artificial intelligence systems such as Amazon’s Alexa, Apple’s Siri or the Google Assistant and the ubiquity of mobile phones have presented what many consider to be another existential threat to the profession of language teaching: AI and machine‑based translation.

Speaking at the InnovateELT conference in Barcelona in 2019, the writer Scott Thornbury called machine‑based translation ‘the elephant in the room’ of language teaching. Similar claims were also made at the closing plenary of the 2019 IATEFL conference in Liverpool in a panel on the future of ELT. In this version of the future, the language teacher is not to be replaced by a humanoid robot in the classroom. Rather, the whole concept of the language classroom, and indeed language learning, will be completely obsolete as everyone will be communicating to each other via translators built into their phones, and eventually into little earbuds. The technology for simultaneous translation via machine now exists. Google and Skype both have software programmes that can understand spoken words and speak the translations. And while historically people have criticized the accuracy of Google translate (in its earlier written form) this software is getting better and better all the time.

LANGUAGE ROBOTS AND INTERPRETERS IN FICTION

The idea of machines or artificial intelligence acting as language interpreters has long been a feature of science fiction. In the Star Trek universe of TV shows, members of federation crew starships communicate with alien races via the Universal Translator, a piece of software that can translate simultaneously any language. In the comedy book series the Hitchhiker’s Guide to the Galaxy, people use a Babel fish for interpretation. This is literally a small fish that you stick in your ear that enables you to understand anything in another language. Finally, the famous robot C3PO in the Star Wars franchise of films is, originally, an interpreter robot. In the films C3PO refers to himself as a protocol droid, fluent in ‘over six million forms of communication’.

These translation/interpretation devices or beings all served the same purpose in these stories: to remove the problem of translating between alien languages. This means that the plot to progress without getting bogged down in a realistic examination of linguistic issues. But the popularity of these robots and devices is large. It is this popularity which probably what accounts for, in some way, the desire of technology companies to be the first to invent a widely used machine translator.

I’m sure if you asked many students if they’d like a device that could speak English for them, they would love it. But even in a world where such devices existed and worked well, I believe that people would still value actually knowing a foreign language.

ASSUMPTIONS ABOUT ROBOT LANGUAGE TEACHERS OR TRANSLATORS

This all still begs the question: will language teachers be made obsolete by robots or machines? Like with so many other features of everyday life supposedly ‘disrupted’ by technology (music, travel accommodation, taxis etc) the temptation for many teachers is to say ‘yes’, even if we secretly hope to live out the rest of our careers before it happens. However, I believe that to say that teachers will be replaced by machines often belies several assumptions about the nature of language learning, language teaching and translation. Here are three such assumptions:

  1. People learn other languages mostly for transactional and utilitarian reasons.
  2. People would be fine not being able to speak another language if they could get something to do it for them.
  3. Translation and interpretation are processes that can be stripped of human influence.

Are these assumptions true? I’d argue that they are not.

Looking at modern General English courses of the past twenty years, one could be forgiven for believing that English is primarily a utilitarian tool. An important part of any syllabus is to teach learners how to function in English.

The Babel fish from Hitchhiker’s Guide to the Galaxy

This ‘functional language’ strand of a syllabus in books is often called things like ‘Real World English’, ‘Function Globally’, ‘English for Life’ and so on. And at lower levels especially, it does contain a lot of transactional language such as ‘Can you tell me the way to the city centre’ and ‘Does this dish contain cheese?’. Indeed, many learners and teachers will say that this is among the most useful language they learn as it will help them travel. And it is precisely this kind of language that is most easily used by artificial intelligence and machines. So if a learner merely wants to learn a language to be able to move around the world buying things, then that learner would probably be best served by a translation device. But language lessons serve two other purposes. First of all, fortunately or unfortunately, English is seen as a gatekeeper to education and jobs. A knowledge of B2 level English is becoming necessary for university studies in many countries now. Access to job markets is often dependent on a certain level of English, even if it isn’t actually used that much in the job. These requirements go hand in hand with the huge industry of language testing and examinations. Robot teachers or machine translators may exist, but will have a hard time disrupting this model of gatekeeping. Secondly, some people like learning languages because they just really like the language. It sounds old‑fashioned to say, but there are almost always some people in a language classroom who are there because they love the way the language sounds, or want to know it to access its literature or culture. For these people, a translation device might help in a pinch, but they would want to learn it themselves.

I’m sure if you asked many students if they’d like a device that could speak English for them, they would love it. But even in a world where such devices existed and worked well, I believe that people would still value actually knowing a foreign language. Where I live in Spain, a former president did not speak or understand any English. This was someone who was accompanied at all times by an interpreter (a human version of Google Translate, if you will). And he was ruthlessly mocked for this fact. Knowing several languages is seen as a sign of intelligence and respected by many people. It’s also seen as something important for children, to learn a second language at school. This has been the case almost since schools began and students were forced to learn ancient Latin or Greek as it was seen to be ‘good for their mind and soul’.

The last assumption, that translation and interpretation can be stripped of human influence and left to machines, is also dubious. As Thornbury himself has written, modern machine learning translation apps ‘need to be able to prove their worth in a much wider variety of registers, both informal and formal’, as they still seem to be stuck in routine exchanges of information. Nuance, humour, poetry, history and symbolism are all important in interpretation or translation of texts. Any translator of literature will tell you that there is a difference between rendering a text readable and understandable, and that there are differences between good and great translations. And until a robot can effectively and consistently pass the Turing test, they will not be able to render language as well that whoever writes the codes and algorithms for the robots might also have their own biases that can lead to a whole minefield of issues. For example, big technology companies such as Google and Apple have been quite prudish when it comes to using profanity (as anyone who has had their autocorrect write ‘duck you’ will attest). There is no guarantee that translation of certain words or terms (profane, or political) might not be ‘massaged’ by the software that does the translating. If this is the case, or suspected even to be the case, people will always trust more what they understand exactly someone else to be saying than have it mediated by a machine.

[…] Until a robot can effectively and consistently pass the Turing test, they will not be able to render language as well as another human. Additionally, it’s worth remembering that whoever writes the codes and algorithms for the robots might also have their own biases that can lead to a whole minefield of issues.

“Language teaching is more than about grammar structures, or words and more words. The best classes create a space where people share, explore and learn from each other. It’s an intimate act that a computer would find impossible to do.”

THE HUMAN BEHIND THE ROBOT

At the beginning of this article I mentioned two robots that weren’t what they seemed, that in fact they had humans inside them. It’s one of technology’s dirty little secrets that many algorithms and AI programmes still have, to this day, people in the background operating them. In 2019, several news reports came out about the real‑life people being paid to listen in to Siri and Alexa to help fine tune the responses. I myself have worked on vocabulary learning software, whose algorithm supposedly gave intelligent feedback on learner errors. But myself and the other writers had to anticipate and respond to all the errors ourselves. On another occasion, an EL T startup experiment creating a translation chatbot. It would immediately translate words and phrases across the platform Whatsapp. The bot was in fact operated by humans, who desperately fielded all the questions that came in in real time. Amazon has an entire service set up that offers piecemeal jobs such as clicking on photos, finding prices, correcting sentences or checking translations for companies. Many of these ‘gigs’ are in fact people doing the work that supposedly an algorithm or machine is doing. Ironically, this Amazon service is called Amazon’s Mechanical Turk.

THE ROBOT APOCALYPSE – COMING TO ELT?

This article is not meant to say that robots or machine learning aren’t real. Nor that they should be shunned. There are many times when a machine translation will be helpful, or that a piece of automated technology could help in the classroom (creating and marking assignments, providing language support). The reasons I suggest above are all ones why robots might not replace language learning. Now whether robots will replace language teaching? Well, I don’t believe that either.

As any language teacher knows, there is a personal side of teaching – the affective side – that robots or algorithms just cannot capture. Language teaching is more than about grammar structures, or words and more words. The best classes create a space where people share, explore and learn from each other. It’s an intimate act that a computer would find impossible to do. Those of us who believe that language teaching is like this, that strive for classrooms to become these spaces, should not be afraid to say so. And we shouldn’t be afraid of robots coming for our jobs.

Suggestions

Studio Pre-intermediate Sb + E-zone

Studio is the exciting new six-level course for adult learners from A1 to C1 (Beginner to Advanced), with an innovative lesson-based approach to topics. The Studio ‘space’ allows students to be creative and provides teachers with clear, easy-to-teach lessons. Three main lessons (two at Beginner) cover vocabulary, grammar and skills work. Video-based functional lessons (101 Things to do in English) and comprehensive Review lessons complete a course that gives students scope to be creative with la.

Studio Intermediate Sb + E-zone

Studio is the exciting new six-level course for adult learners from A1 to C1 (Beginner to Advanced), with an innovative lesson-based approach to topics. The Studio ‘space’ allows students to be creative and provides teachers with clear, easy-to-teach lessons. Three main lessons (two at Beginner) cover vocabulary, grammar and skills work. Video-based functional lessons (101 Things to do in English) and comprehensive Review lessons complete a course that gives students scope to be creative with la

Studio Elementary Sb + E-zone

Studio is the exciting new six-level course for adult learners from A1 to C1 (Beginner to Advanced), with an innovative lesson-based approach to topics. The Studio ‘space’ allows students to be creative and provides teachers with clear, easy-to-teach lessons. Three main lessons (two at Beginner) cover vocabulary, grammar and skills work. Video-based functional lessons (101 Things to do in English) and comprehensive Review lessons complete a course that gives students scope to be creative with la.

THE AUTHOR

Lindsay Clandfield

Lindsay Clandfield is an English teacher, trainer and writer based in Spain. You can find out more about him at his website www.lindsayclandfield.com

Related posts
Editorial NRInglês

Bonnie Scotland

BlogInglês

Are we really teaching speaking?

Editorial NRInglês

Disal Indica - Look (American English) 1 Reading Anthology

Editorial NRInglês

How do you say... in English?

Assine nossa Newsletter e
fique informado
E-mail

Deixe uma resposta

O seu endereço de e-mail não será publicado. Campos obrigatórios são marcados com *

Espere um pouquinho!
Queremos mantê-lo informado sobre as principais novidades do mercado acadêmico, editorial e de idiomas!
Suas informações nunca serão compartilhadas com terceiros.