Translated's Research Center

Question of the year – 2021

“Is language a technology or a culture?”

We have asked this question to the major experts in the field of language, digital media and communication, strongly believing that their ideas are important to our multilingual and multicultural community.

In 2020, the lockdown we have been forced to has increased the digital dimension and has evolved new multicultural communications, while experimenting the growing role of artificial intelligence in enabling multilingual communication. The ability to communicate in a multicultural environment has never been more important. We should be able to evaluate this importance as a new wealth generator in terms of technology, culture, economy, society. We think that a set of answers like the one they provide us could gather facts and opinions so diverse and rich as to create a new knowledge of the importance of languages and multicultural relations.

Andy Way

Deputy Director at ADAPT Centre

Full Professor in Computing at Dublin City University. His research interests include all areas of machine translation: statistical MT, example-based MT, neural MT, rule-based MT, hybrid models of MT, MT evaluation, teaching MT, etc.

We can of course apply technology to language to increasingly good effect. In the machine translation (MT) field in which I operate, translation quality has improved significantly over the past 5 years or so. It is no coincidence that this improvement came about as we transitioned to neural architectures which surpassed the statistical models that had been the state-of-the-art for 30 years or so. But unless we system designers learn how to incorporate more and more linguistic knowledge into the models, we will just hit another glass ceiling which will largely curtail the rather dramatic improvements in quality that we have seen in recent times. And, of course, in order to do that, we must either know about language and linguistics ourselves, or we must team up with linguists who can help your system improve even further. My own view is that once some of the more engineering-oriented newcomers to MT realise how hard it is to process a language – and in MT we have two to deal with!! – they will move on to an easier application area, while those of us who have been operating in the discipline for a much longer time will remain!

It is still the case that while NMT definitely produces translations of  ‘better’ quality than SMT, some of the outputs may appear deceptively fluent albeit producing terrible translations of the source. Human translators make mistakes, of course, but the mistakes they make are quite different from those made by today’s systems. We have seen many claims of human parity, which many researchers (including people in my team) have shown to be less clear than originally expounded. But surely, if these systems were as good as human translators, wouldn’t they also make the same sorts of mistakes that humans do?


Imminent Annual Report 2021

Imminent Annual Report 2021

Fit your business in global shape. Get your copy of Imminent Annual Research Report 2021. And let us know what you think.

Get your copy now

So, to return to the question, language embodies much of our culture, and it is a fact that different languages carve the world up in different ways: I know perfectly well what the difference is between an ‘inside corner’ and an ‘outside corner’, but it’s just in Spanish that they have different words for them. Translating either ‘rincon’ or ‘esquina’ as ‘corner’ misses something; adding the words ‘inside’ or ‘outside’ gives rise to a clumsy English phrase. And while German has different words for ‘inside wall’ and ‘outside wall’, I know perfectly well not to hang my collection of Van Goghs on the ‘Mauer’, but instead on the ‘Wand’, even though English doesn’t have this distinction at lexical level! How today’s improving MT systems address the issue of ‘lexical gaps’ is an interesting one, as it is really only through world knowledge – which these systems will never have – that human parity has a chance of being attained. Human translators need not fear ‘the rise of the machines’, far from it …

Joseph Mariani

Directeur de Recherche chez CNRS-IMMI & LIMSI

Director of the French-German Institute for Multilingual and Multimedia Information (immi).  He was Director of Limsi-CNRS and Head of its Human-Machine Communication department, then Director of the Information and Communication Technology Department at the French Ministry of Research.

Language technologies are mandatory to ensure language sustainability through multilingualism, aiming both at language preservation and communication. They are now widely spreading in everyday life through many applications, but they still need more research for more advanced applications, for better quality and especially for more languages.


“Presently, less than 150 languages benefit from language technologies, with very variable levels of quality. This represents only 2% of the more than 7,000 languages that are used in the world.”

Joseph Mariani

We should now find a way to address the remaining 98%.

Those statements follow the conclusions of the LT4All (“Language Technologies for All”) conference that we organised for Unesco last December.

Specifically, the fact that UNESCO organised the International Year of Indigenous Languages this year (IYIL 2019: https://en.iyil2019.org/) was important in itself. The event included the LT4All conference and results in the decision of the United Nations to launch the International Decade of Indigenous Languages (2022-2032), which will support language diversity and intercultural communication.

More generally, I believe that the availability of Language Technologies, including Machine Translation and cross-lingual Information Retrieval, and the expansion of social networks will also contribute to preserving languages and developing communication.

Daniel Pimienta

Head of the Observatory of Linguistic & Cultural Diversity in the Internet

Board member and former Exec. Sec. of the World Network for Linguistic Diversity (MAAYA. He manages the Observatory of languages and cultures on the Internet since 1989 being a pioneer in cybermetrics and the measure of the space of languages on the Internet.  In 1993, he founded the Networks & Development Foundation (Funredes). 

“Is language a technology or a culture?” According to the correct definition of technology (the study of tools and techniques, not the tools nor the techniques themselves) languages or cultures are certainly not technologies… but the very question which is asked is indeed itself a good example of technology. This would be too short an answer for this Imminent proposal … so let’s do what everybody (including me) has always done and keep doing: ignore the true definition of technology and abuse it as the set of tools and techniques which humanity has created to facilitate its endeavors within its living environment. Then the question translates into “Is language a tool, a technique or a culture?”. Note that we could ask the same question about culture: “Is culture a technology?” and if both answers would be yes then the final statement would turn into “both cultures and languages are technologies”.

In reality, we can take the initial question simply as a pretext to ask language-minded persons to communicate on a subject that is so open that they expect some creative thinking. I will not continue teasing and ask if pretexts are technologies!  Communication is clearly the main role of languages to facilitate the endeavors of humans within their living environment, enabling them to create the condition for effective collaboration between humans, a requirement for the Human Species to survive, facing, as a group, or better said, as a community, the challenges and dangers of their living environment. Within this operational perspective of communication, culture could be perceived as the implicit part of the group communication, which does not need to be expressed because it is granted (given to children by family implicit and explicit education) and the group can now focus on the explicit part of communication and therefore achieve more efficient communication. Note that languages are not only technologies to communicate with others; they are also useful for self-communication, enabling each human being to create the condition for reflection.

For effective reflection, culture needs to be put aside as much as possible, in order to free reflection from the implicit realm embedded into one’s culture and open the mind to new ideas, to other cultures and to creative thinking. Part of keeping a child mind (a mind able to ask the naïve questions which lead to creative thinking) is having the capacity to extract one’s mind from the too predictive and closed answers provided by one’s culture. In that sense, languages are more powerful than cultures as they are not only a technology but also a meta-technology, a tool required for practically all other technologies. This recursive property of languages (capable of describing themselves) is the essence of languages, the most powerful technology that humanity has ever created and probably would ever create, except maybe Mediterranean gastronomy (but, after reflection, this could be a cultural bias).

Then there is the question of diversity and the Babel Tower myth. Many people  mistakenly believe that the lack of a unique language shared by all people is a malediction for humanity and the cause of misunderstandings leading to conflicts. First of all, careful observation would tell us that the source of misunderstandings lies much more in cultural differences than in language differences (in Latin America, the subtle Spanish differences between countries and the misunderstandings which they provoke are much more a cause of pleasant and frequent conversations and good humour than a cause of conflict). Technologies are not neutral, all of them carry specific biases. Most of the biases of languages are probably related to the culture in which they are embedded and have evolved.

Cultures carry the vision, values, rules, moral concepts of the community who share them and cultural differences are not always easy to overcome, although communication (language) does help, especially the one tailored for reflection which manages as much as possible to escape from the cultural biases. Complex system theories, which are the best tools to observe phenomena characterised by the interaction/correlation of a large number of factors and where the causal effect approach is unable to capture the essence, states the intuitively obvious (at least for who want to use reflection) in the Ashby Law of requisite variety. This law can be translated intuitively as: if a system does not hold sufficient diversity, it is prone to be unable to react to external thread and collapse.


“Diversity is a condition for humanity to survive, including and starting by language’s diversity.”

Daniel Pimienta

The main thread of today’s age is the systemic trend towards diversity in all its fundamental aspects: the variety of biological species, diversity among surviving languages, diversity of cultures and, if we ever reach the point where we threaten diversity in thinking, then we will come to a ‘meta threat’, the one which could recursively spiral humanity to extinction: the reduction of epistemic diversity, the meta-reduction of diversity the no-return threshold.

Je ne peux pas terminer sur une note aussi pessimiste, alors je vais tenter de terminer ma réflexion de manière positive et, pour la cohérence, dans d’autres langues.

La salida de la crisis de la diversidad que plantea el mundo de hoy solo se encuentra en proteger, potenciar y empoderar la diversidad, eso debe comenzar con las lenguas y por supuesto traductores e intérpretes son actores claves de esta lucha.

Shalom, Salam, Salud, Cheer

Esther Tan

Senior researcher at Leiden-Delft-Erasmus Center for Education and Learning (LDE-CEL) at Delft University of Technology

Education technology officer with the Educational Technology Division (ETD), Ministry of Education (MOE), Singapore; head of IT at Shanghai Singapore International School (SSIS), Shanghai and a research associate with the Singapore Future School Project to advance research in the 21st century digital skills and competencies at the Learning Sciences Lab (LSL), National Institute of Education (NIE), Singapore.

Translation, is a very controlled industry. 
It is controlled by big players and everyone is actually serving them. Smaller vertical operations have little leverage 
and since they think they are big or probably they truly are big, they dictate everything. 
That’s the reality and as a business or as an industry, that is bad because once an industry is controlled by a monopoly, 
that means it is time to get out of that industry. 

A monopoly controls everything via the price, and also how they want to treat other people. 
There is no room for questioning. We need to “worship” them, and nothing they say can be challenged. 

Even if they make fatal mistakes, the minor parties carry the guilt. 
That’s the reality in our industry. And when I heard about the integration of RWS and SDL … 

OK – that’s the end of this industry.

Michaël Oustinoff

Associate Professor in Translation Studies at the Institute of the Anglophone World, University of Paris 3 Sorbonne

Nouvelle and currently on sabbatical leave at the ISCC, the Institute for Communication Sciences of the CNRS. His third book, Traduire et communiquer à l’heure de la mondialisation (Translating and Communicating in a Globalised World) was published by CNRS Éditions in 2011.

You are absolutely right: the Covid-19 pandemic has transformed the world of intercultural communication and changed our digital habits, increasing the amount of time people spend online. But while everyone is aware that this has increased their time spent online, I think few people are still aware of its unprecedented impact on the “world of intercultural communication”. Most people have access to local news, so to speak, even though viruses know no borders, of course. 


Imminent Annual Report 2021

Imminent Annual Report 2021

Fit your business in global shape. Get your copy of Imminent Annual Research Report 2021. And let us know what you think.

Get your copy now

I am fortunate to be able to keep up to date with news in seven languages: French and English, but also Portuguese (my mother tongue), Spanish, Italian, German and Russian, not to mention some others, to a lesser extent, through intercomprehension. 

I am also fortunate to have been invited, especially since 2009, to many places abroad. “Travel broadens the mind”, as they say, and I have had several occasions to verify that this is true, but travelling is not enough. I wouldn’t say, like Levi-Strauss in the “Sad Tropics” that “I hate travelling”, but I suppose he meant that travelling without real intercultural communication is useless or, at best, deceptive. 

Today we can travel without having to go abroad, even if this is more precisely “virtual travel”. The Internet and new technologies have enabled us to go almost everywhere while staying at home. On the other hand, the progress of “machine translation” has been spectacular in the last ten years or so. 

Although young people tend to take advantage of all these opportunities – including low-cost travel – which were still unimaginable for their parents’ generation (the Internet was born in the late 1990s), they do not use them to the fullest and, it seems to me, they tend to underuse them, because they have not yet understood the full potential of that acceleration of history. Do not blame them, we are all inclined to look to the future through lenses of the past. 

 But that is precisely why we have to look at the current situation with new eyes and this is the meaning I see in your question :

“Is language a technology or a culture?”

The shortest way to answer is to adapt what Umberto Eco said about the language of Europe: “The language of globalisation is translation” and to combine it with Karl Jasper’s quotation in Vom Ursprung und Ziel der Geschichte (1983, 161):

“« Das jedenfalls ist offenbar: Technik ist nur Mittel, an sich weder gut noch böse. […] Die Technik ist unabhängig von dem, was mit ihr zu machen ist, als selbstständiges Wesen eine leere Macht, ein schließlich lähmender Triumph des Mittels über den Zweck.” (This at least is obvious: technology is only a means, in itself it is neither good nor bad. […] Regardless of what one has to do with it, technology, as an independent actor, is an empty power, a paralysing final triumph of means over the end).

 As for the “imminent future”, the greatest risk seems to me to be the current blind faith in the power of technology, which tends to consider language as a mere instrument of thought. In this context, all languages are interchangeable and mere tools, which incidentally is the way languages are considered in the Greco-Roman tradition. But languages are not mere tools, if we follow the school of thought developed by Wilhelm von Humboldt. 

For this reason, a language cannot be considered as a technology, because a technology, according to Karl Jasper, is only a means to an end, in other words, an instrument, however sophisticated it may be. A striking example of the fascination of technology today is this announcement for Microsoft’s “universal translator”: https://mspoweruser.com/microsoft-translator-announces-the-worlds-first-personal-universal-translator/

 The “universal translator” is an obvious allusion to the “universal translator” used by the crew of the USS Enterprise in the science fiction film Star Trek which enables them to communicate instantly in all the languages of the Universe, terrestrial or extraterrestrial. 

These new technologies are undoubtedly an extraordinary innovation that should not be underestimated: now we can use machine translation to read online texts in languages that until now were completely opaque for those who did not know them.

 They allow us to communicate, but at a “statistical” and “predictable” level: it is essentially a functional mode of communication. But language is much more than a technology. 

In 2003, the French theorist Dominique Wolton wrote a book entitled “L’autre mondialisation” (the other globalisation) in which he explained that in reality there were three globalisations and not just one. The first was the political one, with the rise of the UN in the wake of the Second World War; the second was the one generally thought of when we talk about ” globalisation”: economic globalisation that goes hand in hand with the advent of new technologies; but it is the third that is the most important, the other globalisation”, i.e. cultural globalisation. 

 As George Bernard Shaw half-joked, “England and America are two countries divided by the same language”. Knowing a language, even to perfection, is therefore not enough.


“Language is inseparable from culture and, more precisely, from cultural diversity. “Global English” can therefore be nothing more than a language of communication that is certainly useful, but artificial for its so-called “planetary” universality.”

Michaël Oustinoff

The same goes for “automatic translation”, which some would like to elevate to the rank of a truly global language, all the more practical and democratic because it dispenses with the need to learn a foreign language. 

Consequently, language is a culture, not a technology. But we are “after Babel”, as George Steiner would say: there is not just one language, but many languages, from 6,000 to 7,000, not counting dialects. Even if we only look at the most widely spoken languages in the world, there are at least 200. Hence the central place occupied by translation, because we cannot learn all languages. 

What the Covid-19 pandemic has demonstrated once again is the importance, not only of translation, but also of its indispensable role in intercultural communication, without which there can be no in-depth understanding of the contemporary world in all its complexity. It is also understandable that such an understanding makes sense only in an interdisciplinary framework: who alone could take into account all the linguistic, technological, geo-historical, political and communicative dimensions involved? 

And since we tend to see the future through the spectacles of the past, this implies that new fields of research open up for us to explore. 

This exploration is only just beginning. 


Imminent Annual Report 2021

Imminent Annual Report 2021

Fit your business in global shape. Get your copy of Imminent Annual Research Report 2021. And let us know what you think.

Get your copy now

This is how I could answer the question in English, which is actually a very good language to demonstrate the need to promote linguistic diversity and not just global English, which fortunately many native English speakers do. 

 As for “The Wealth of Languages”, it reminds me of Adam Smith’s “The Wealth of Nations”, which is actually the shortened version of the original title “An Inquiry into the Nature and Causes of the Wealth of Nations”. It really is time to undertake an “Inquiry into the Nature and Causes of the Wealth of Languages”!

Jaap van der Meer

Founder of TAUS

Jaap van der Meer founded TAUS in 2004. He is a language industry pioneer and visionary, who started his first translation company, INK, in The Netherlands in 1980. Jaap is a regular speaker at conferences and author of many articles about technologies, translation and globalisation trends.

It all depends, of course, on the story and the context. In our world (at TAUS), language is primarily data. Think about how a child learns to speak a language by endlessly listening to and echoing words and sequences of words. It’s not so different from how we now train computers to translate: we feed them with all the translations we already have.


“Language is data and this data is unreasonably effective in teaching humans and machines to improve their communications skills.”

Jaap van der Meer

TAUS was founded in 2005, a year before Google Translate was launched, with a clear vision that the dream of automated translation would finally become reality. In 2008, we launched the TAUS Data Cloud, an industry-shared repository of translation memory data, which helped many of the early adopters of SMT to improve their engines. Soon thereafter, we invited governments and industry to join us in the Human Language Project: a very large-scale collaboration with 1 billion euros in funding to build a giant database that would store all the utterances of all languages on our planet and make this universally accessible. Our vision was inspired by the Human Genome Project which succeeded in unravelling the secrets of the human body through an inventory of the 3.1 billion base pairs that constitute the human genome. The difference, of course, is that, unlike the physical world, language can generate an infinite amount of data. It is changing as we speak and throughout our lives. This highlights the cultural and literary sides of language. We will probably never completely unravel the secrets of human language but with a good investment in language data, we will be able to make global communications work a lot better.

The closest we can get to our grand vision of a universally accessible language data network, for now, is the CEF Data Marketplace, an EU funded collaboration between Translated, FBK Trento and TAUS, to be launched in November.

Sandeep Nulkar

Founder & CEO @ BITS & Vernac Language Technologies

Sandeep Nulkar is the Founder and CEO of one of India’s oldest translation companies, BITS Private Limited and the Co-founder of Vernac Language Technologies that developed India’s first crowdsourced translation platform. He was recently re-elected as the President of India’s language industry body, CITLoB. Sandeep is also a columnist, an author, a competitive sprinter and an aspiring drummer.

But for technology, language would have always been primarily a culture. However, with the advent of the Internet, our conversations moved online and thus began the process of change, a change that would hit at the base of both the language as well as culture.

Language, that was always a medium that was used to communicate with people who we could see, touch and feel, would now be used increasingly to communicate with profile pictures and avatars. Significant portions of our verbal communication thus came to be replaced by typed chats, and those too, of the online variety. Our ability to read emotions using both facial cues as well as our interlocutor’s body language was severely compromised since we rarely spoke to a face. These, I would say, were the first signs of technology entering the realm of language.


“Eventually, businesses also moved online and that is when, in my judgement, language ceased to primarily be a culture, becoming instead, more of a technology of its own kind.”

Sandeep Nulkar

Look at the impact this has had on a country like India, a land of 22 official, 122 major and 1566 other languages, where over a billion of the nearly 1.4 billion people do not speak English as their first, second or even third language. The Internet represents neither their language nor their culture. To them, the language of the The Internet is simply intuitive technology. This problem is only compounded further by applications and systems that are not UA-ready. A seamless experience for the average Indian in his/her language is thus a distant dream.

The question now is whether language will ever be more of culture again in an increasingly technological world. I think it can because the only reason it became more of a technology, in the first place, was because the language of the Internet failed to represent a large majority of cultures. The Internet was primarily English and mostly Western. Stakeholders of the translation, localization and intercultural communication world would therefore do well to recognize this as a unique opportunity to build products and technologies of the future that are truly inclusive and designed for culturally diverse peoples.

Tess Whitty

ATA and ITI Certified English-Swedish translator

English-Swedish freelance translator specializing in digital marketing, software and website localization. She has an M.Sc. in Economics and an M.A. in marketing and previously worked as a marketing manager. She shares her knowledge and experience in marketing and business as an award-winning speaker, trainer, consultant, author, and podcaster

Is language technology or culture?

I think you cannot separate these two. A language has elements of both.

Culture plays an important part on how a language develops and continues to exist, and the pronunciations, words, sentence structure, intonation etc. are all very much dependent on cultural aspects and developments. However, to structure a language, to teach it and maintain it, technology is necessary.

Cultural influences, such as invasions, migration and global trade affects the evolution of languages. The world’s languages have transformed throughout history due to cultural shifts.

However, technological developments have also influenced languages, from the printing press, to the internet and social media.


“Therefore my answer is that language is neither technology, nor culture, but both.”

Tess Whitty

Kind regards/Vänliga hälsningar

Renato Beninatto

CEO of Nimdzi Insights LLC

Co-author of The General Theory of the Translation Company and leads Nimdzi Insights, a think-tank and consulting company that focuses on growth strategies for localisation leaders. A former owner of an LSP, an executive in some of the leading companies in the industry and a linguist in his own right, this Brazilian-Italian-American citizen can’t shut up in Portuguese, English, Italian, French or Spanish.

Language is a vehicle. It is one of the ways that people have found to communicate their emotions, experiences, and learning with each other. Language is a code. It is an agreement between people to attach meaning to certain sounds, images, signs, or other forms of representation. Language is history. It carries with it the legacy of generations and geographies from distant times.

I like to think of language as water. It is essential to survival and it can change according to the environmental conditions. It can contract and expand, and even mix with other liquids to create new flavours and varieties. But language can also be a barrier and the source of controversy and misunderstanding. 

Language is an intrinsically human experience with all the positive and negative consequences that it entails.


“In a sense, technology and culture would not exist without language, because they need language to express themselves.”

Renato Beninatto

Technology and culture need language to be shared, learned, and improved. 

Language develops from the primordial need of people to work together towards a common goal, whether that is hunting an animal, avoiding poisonous berries, keeping a fire burning, building computers or curing diseases.

Language is what allows you and me to be here.

Kirti Vashee

Language technology evangelist at SDL

Formerly an independent consultant focusing on machine translation (MT) and translation technology. He was also with Asia Online and was previously responsible for worldwide business development and marketing strategy at statistical MT pioneer Language Weaver, prior to its acquisition by SDL. Kirti has extensive sales and marketing experience in the enterprise software industry, working both for large global companies (EMC, Legato, Dow Jones, Lotus, Chase) and several successful startups. 

It is neither. Language is a means of communication and an information-sharing protocol that employs sounds, symbols, and gestures. Language can sometimes use technology to enable amplification, extend the reach of messages, and accelerate information and knowledge sharing. Language can create a culture when shared with(in) a group and used with well-understood protocols and norms. Intercultural communication can also cross species, e.g., when communicating with dogs and horses.

Jan Hinrichs

Founder and CEO of Beluga

Founder and CEO of Beluga, a translation company helping fast-moving companies run efficient localization programs and positively engage with their global audiences. In the past 15 years, Jan has been deeply involved in the emerging market of web-service and mobile app localization and the supportive software infrastructure of online TMS to guarantee smooth and continuous translation workflows. In early 2019, he started the global LocLunch movement.


“For me, language is the basis of culture.”

Jan Hinrichs

It has emerged as a way to express and share what we see, smell, hear, taste, and feel in the world around us. This shared understanding through language defines who we are and how we are. Technology is about the tools we create and use to accomplish certain things, which influences our language and culture. Learning another language is hard work, as we all know, but it helps us travel, do business, and engage with other cultures. Technologies like machine translation, natural language processing, or natural language generation, increase our capacity to communicate with less effort and on a global scale. Language technology plays a crucial role in connecting the next 4 bn people to the internet and helping businesses engage with their audiences worldwide and in a personalised way. With annual double-digit growth, Africa, India, and Southeast Asia will join the global e-commerce and service market with 60+ major languages. Those who embrace this technological opportunity will benefit the most.

Alexander Drechsel & Josh Goldsmith

Techforword

Alexander works as a conference interpreter at the European Commission. He loves languages and communicating with people and is enthusiastic about technology. Alexander has been an iPad user since day one and also knows a thing or two about Android tablets. As Co-Founder and Chief Technology Officer of techforword, Alexander shares his knowledge in courses, on Twitter, and on his website. An avid podcaster, Alex writes and produces two podcasts for language professionals, Troublesome Terps and LangFM.

Josh is a UN and EU accredited translator and interpreter working from Spanish, French, Italian and Catalan into English. He splits his time between interpreting, translating, and working as a trainer and researcher focusing on the intersection between interpreting, technology and education. As Co-Founder and Chief Educational Officer of techforword, Josh shares tips about technology, translation and interpreting in conferences and workshops, academic articles, and on Twitter.


Imminent Annual Report 2021

Imminent Annual Report 2021

Fit your business in global shape. Get your copy of Imminent Annual Research Report 2021. And let us know what you think.

Get your copy now

Will technology replace translators and interpreters?

We’ve all done it – even professional translators and interpreters. 

What do you do when you come across an article in a language you don’t speak? Run it through Google Translate. When you need to have a quick chat over video with someone and don’t have a common language? Skype Translator can help. Or that exotic menu written in a foreign language while on holiday? Just open the handy translator app on your smartphone, point it at the text, and let the magic begin.

For many basic tasks, these automatic tools work well enough. They’re certainly not perfect, but they do help us grasp the gist of the message and make ourselves understood.

All of this leads us to the million dollar question: will translators and interpreters be out of a job soon?

We don’t think so. But as machine translation improves, it is becoming useful in many situations that go beyond the basic examples we’ve just mentioned.

Why is this the case? Basically, the answer boils down to two key developments: big data and a new, better approach to machine translation.

First, big data. Strip away the jargon, and you’re left with a key idea: that software can analyse huge amounts of data and detect useful patterns far faster and better than a human ever could. This applies to plenty of things other than language, of course – from real-time traffic reports to targeted advertising or Netflix’s recommendations for what you should watch next.

In language technology, companies like Google and DeepL draw on huge bilingual and multilingual corpora. Their databases include millions of sentences from original documents along with their translations produced by humans. These treasure troves of well-translated content form the backbone of modern machine translation.

Second, new machine translation models. Until a few years ago, the statistical approach to machine translation was widely used. In this model, computers identify sequences of words in the original text, look up possible translations, and then use statistical models to decide which translation is most likely to be right.

The new kid on the block, neural machine translation, uses a different approach: instead of relying on probability for the right translation, artificial intelligence recognises patterns in the source and target languages and matches the two. These patterns go beyond single words, taking in entire phrases and sentences. Recent studies have shown that this method yields better translations than before.

Yet machine translation still has plenty of hurdles to overcome. Whereas human translators understand a sentence in the context of an entire text (and more!), machine translation often operates on a sentence-by-sentence basis. Plus, it needs huge amounts of data in the form of human-translated text, which is not available in the same volumes for every language pair. Machines still struggle with concepts that were not in their data set and cannot grasp jokes, irony, cultural references, puns, rhymes, and the other fun stuff that makes communication so rich.

Nevertheless, technological developments are changing the face of the language industry. Many companies already save time and money by having their texts translated by a machine and polished by a professional linguist; the jury’s still out on quality and job satisfaction. Others use controlled language with a smaller subset of words and strict rules. After all, a simple text is easier for a machine to translate. 

As for machine interpretation, it’s even harder than machine translation to get right. That’s because machine interpretation has traditionally included three stages: transcribing speech, using machine translation to convert that text into another language, and then using speech synthesis for the spoken output. An error in the first step can be compounded in later phases. And new machine interpreting models that leave out the machine translation “middleman” are still embryonic at best.

Most importantly, machine translation and interpretation are based on the fallacy that language professionals just “translate words”. But we do far more than that. As research shows, interpreters draw on body language, information on screens, word lists and more to form meaning; we add information, explain cultural references, advise speakers, change language registers when needed, and detect and solve all sorts of problems.

In summary, although machine translation and interpreting can facilitate basic understanding, they still fall far short of what humans can do. So while we encourage you to keep an eye on technological developments – which may help to streamline the work of language professionals – you should still talk to a professional translator or interpreter for your next big project.


“You should still talk to a professional translator or interpreter for your next big project.”

Alexander Drechsel & Josh Goldsmith

Photo credit: Damon Lam, Unsplash