Kites— or Language Cluster Finland — supports the internationalization of companies in Finland through development in multilingual communication and language technology.
Acolad has been active member of Kites from the very beginning – in fact, we are one of its founders. The main theme for 2017 was the influence of technology on language services.
What kind of collision between language services and technology will we get? Car meets train, or bowling ball meets marble?
We are now at a stage in the development of both language and technology where a collision is imminent – some might even say it has already happened, and we’re just waiting for the dust to settle. Digitalization, the Internet of Things, artificial intelligence, and – going further back – the smart phone, voice recognition, even the advent of the internet and social media. It is hard to imagine any other time when not just what we can say, but how we can say it has changed so much, and so quickly.
The ability for companies to engage with customers globally via electronic devices has led to a huge growth in the amount of consumer content. Technological innovation is needed to help us create, manage, and curate all of this content.
Let’s ask a fundamental question: What happens when two objects, in this case language services and technology, collide?
As you can imagine, the results of the collision can be very dramatic – think car meets train – and one of the objects is altered dramatically. Or, the result can be that one object is flung off in an entirely different and unexpected direction – think bowling ball meets marble.
I suspect that technology will spin language services off into new territory, rather than wrecking it, but that may be a matter of perspective! To get that perspective, I spoke to the people here at Acolad who teach, train, design, translate, and produce for a multilingual client base. I asked them one rather simple question, “How will technology impact language services in the next three to five years?”
Translation services and automation collide
Machine translation, or MT to give it its shorter name, is probably the most well-known of language technologies, given its near-seventy years of research history. We are all familiar with Google Translate and its development from a source of ridicule for its sometimes clumsy, humorous, or downright bizarre translations to a serious driving force in automated text-to-text and text-to-speech translation, as well as simultaneous live translation of speech.
Some translators in the future will specialize in “remanufacturing” machine translations
Taina Saarela-Rissanen knows more than most here about the end product in the translation industry. She is our team leader for quality assurance, and ensures daily that the translated text keeps not just the meaning, but also the tone, feeling, or intent of the source text. How does she see MT developing?
As with most of the others here, I think that the abilities of machine translation will increase dramatically over the next five years. After all, technology has been helping translators in the form of translation memories for a long time. However, machine translation will develop at a steady pace, and at the very least, all technical texts will be translated by machines for closely related languages. The number of translators needed remains to be seen, however, depending on the language pair and the nature of the text.
It seems likely that there will always be translations where machine translation cannot get beyond being just a rough draft, and the actual translation needs to be completed by a human translator or editor. Therefore, I imagine that the need for translators, editors, and proofreaders may not necessarily diminish, but that the nature of the work will change, and require, for example, additional training.
Machine translation systems are being developed at two levels: at the technical level: improving the learning processes and developing hybrid systems that incorporate more than one basic MT method; and at the level of individual languages: removing problems specific to them.
In addition, I assume that some translators in the future will specialize in “remanufacturing” machine translations, but that most will continue to do the other jobs, such as marketing texts – and, of course, also fiction – in which machine translation will not be of great help. Idioms, homonyms, and colloquial expressions – just a few of the unorthodox language structures that are typically found in creative writing – will continue to prove difficult for translation engines to deal with accurately.
Machine translation as an emergent technology was examined from the perspective of smaller languages such as Finnish for one of our eBooks earlier this year. The main author of the book is Annamari Korhonen, who has been working as a translator at Acolad for a long time.
Her areas of expertise are technical and marketing translations. In our eBook she highlighted that there was a very valid reason for examining MT vis-à-vis smaller languages: the development of machine translations for any language requires a lot of resources, meaning that not all languages are created equal.
Machine translation systems are being developed at two levels: the technical level: improving the learning processes and developing hybrid systems that incorporate more than one basic MT method; and the level of individual languages: removing problems specific to them. Most of the resources are directed at major Western and Asian languages, not at Finnish, which means that viable machine translation for Finnish lags behind.
The even bigger problem for Finnish? As it is practically without close relatives, Finnish is not structurally similar to any language with a large number of speakers.
Finnish is also a morphologically complex language: endings are added to words – often several at the same time. Languages like this are notoriously difficult for translation engines to handle. It is a noun-centric language, while English, for example, relies on verbs, making the text sound more dynamic. Machine translation from Finnish to English, for example, will often result in noun-heavy language. To a speaker of English, it will be dull to read, and will seem stylistically inadequate. It is also filled with the passive voice, which is considered acceptable for nearly all styles in Finnish but should rather be avoided in most style registers of other languages.
A well-trained example-based translation engine might be able to avoid these problems to some extent, but it is unlikely that these pitfalls could ever be completely avoided. A huge amount of work is required to teach machines to translate Finnish. Even as machine translation technology is progressing for many large language pairs, it will take time to include Finnish on these lists.
Important work is being done in Finnish translation agencies to investigate the potential that the different translation engines and systems have for translating to and from the Finnish language.
- Chapter 1: Computer-assisted translation (CAT) and the translation process. What role is technology already playing in the translation industry?
- Chapter 2: Machine translation (MT) – is it a game-changer? What computer translation technology can and cannot do.
- Chapter 3: How do computers actually translate? Rule-based, statistical, and example-based machine translation.
- Chapter 4: Why are languages so difficult to translate? What are the main challenges posed by natural language?
- Chapter 5: Why smaller languages, such as Finnish are so challenging for machine translation?
- Chapter 6: Man vs. machine – what is the future for machine translation?
Learning meets automation
Artificial Intelligence, or AI for short, has been around as a sci-fi concept for many decades, but the self-learning computer of Hollywood movies is still a technological leap too far.
However, smart as a prefix has been added to everything from lightbulbs to cars in recent years. In the context of learning, smart means personalized and tailored, and coupled with automation it may well give the impression of intelligence at work in a learning service.
Better cultural skills will also be more important in the future ...(a) smart or AI translation system will probably not recognize or translate body language. In the future it will still be necessary for humans to read between the lines to figure out what someone is really saying.
So, what will the smart language service of the future look like? I spoke to Mike Aaja, a senior communication trainer here who also works with instructional design.
Technology will make technical language learning less important than other language skills. By that I mean there will be less focus on learning grammar and vocabulary. Smart email programs will correct your writing, for example, or in-ear headphones will translate everything that is said during a negotiation for you and the other people in the meeting room, regardless of the language. With the caveat, of course, that Finnish is a special case!
The focus of what language and communication trainers like me offer to clients will move to the area of soft skills. If the AI system handles what we say in a negotiation, we can concentrate on how and why we say it. The same applies to presentation skills, or customer service, for example.
Better cultural skills will also be more important in the future. If there is no language barrier, then global companies can assemble teams staffed by people from any local office. But they will need to learn how to overcome their cultural barriers in order to take advantage of the lack of a language barrier. A smart or AI translation system will probably not recognize or translate body language! In the future it will still be necessary for humans to read between the lines to figure out what someone is really saying.
Update your presentation skills for the digital age with our free eBook, Presentation Skills 2.0, with special focus on giving digital presentations – what to consider when presenting via digital channels, such as virtual meetings, webinars or video blogs.
As we start to see technology applied to training, there will be a transformative effect on how we learn. We asked Johanna Jalava from our sales team for her insights. She is a keen follower of technological influences and trends in training and competence development.
The jury is still out on nascent technologies such as virtual reality or other similar digital ways to create new learning possibilities. It is hard to imagine those becoming mainstays in education if they do not achieve the commercial success to justify the very high development costs at present.
There are so many positive developments right now. Automation is bringing a deeper understanding and usage of analytics, and data-driven cognitive systems are starting to support personalized learning and learning experiences. I’m sure that there will be much more choice and personalization in the future. There will be the ability to deliver learning content on an individualized basis, shifting the balance of power in instructional design to the end user. Users will choose where, when, and how to learn.
Learning as a life-long process, including micro-learning and skill-based learning, will become a mainstay of training and education. Aside from learning hard skills, we will see more skill-based learning: communication, teamwork, problem-solving, empathy, people management, critical thinking, etc.
Innovations that we have seen emerge recently will have become standard elements of learning, such as gamification to support learning and learning from each other. Data filtering and packaging (what data is relevant, reliable, and useful) will also be a less glamorous, but critical, component of effective learning.
Automation, at some level, will impact on digital learning. Content will be packaged and delivered according to the individual needs of the learner. It will be the engine behind personalized learning in the workplace.
In my opinion, the jury is still out on nascent technologies such as virtual reality or other similar digital ways to create new learning possibilities. It is hard to imagine those becoming mainstays in education if they do not achieve the commercial success to justify the very high development costs at present.
The digitalization of the learning environment
Next, I spoke with Kimmo Oksanen, a project manager for digital learning projects here at Acolad, and asked him how he saw technology impacting on customer needs.
As technology develops, content production becomes easier and more flexible. More of the production work will be done in-house by the customer as the tools become easier to use and more intuitive. Look at website development, for example. Ten or fifteen years ago, only someone with coding skills could build a website; now, anyone can build a website with a drag and drop.
However, companies now more than ever need to focus on ensuring that content meets the needs of customers, is useful, is SEO optimized so it can be found by search engines, and is available in all of the languages that their customers speak.
The use of video content will also increase dramatically in the future. There are some purely logistical reasons for this, such as having greater availability of higher bandwidth that can handle video and rich, interactive content. It is also “easier” to localize visual content by adding voiceovers or closed-caption subtitling.
So, customers will have new tools at their disposal that will allow them to share their subject-area expertise. Pedagogy, however, will remain a critical part of digital learning. Great content needs a pedagogical focus if it is to be great learning content.
Automation, at some level, will impact on digital learning. Content will be packaged and delivered according to the individual needs of the learner. It will be the engine behind personalized learning in the workplace.
VR and AR technologies will bring amazing benefits to certain learning situations. Imagine a fighter pilot taking part in true simulations without leaving the ground, or a first-aid responder role-playing a disaster relief operation, and it is easy to see that the considerable financial investment would be worth it. However, VR will not be a standard component of learning in all areas, in my opinion. We must always ask the question, “From the learning perspective, what is the added value in using VR?”
It is a similar question that we urge clients to ask today regarding another popular innovation in learning: gamification. While its benefits are well-known, it is not always the magic bullet that customers think it is going to be for the learning challenge.
Creating e-learning content is no longer the preserve of coders; technological advances have made it more accessible and cheaper to produce. Discover how e-learning can be the solution to your organizational learning needs with our 48-page eBook.
Anu Rummukainen regularly talks to customers about their future training needs. She echoes Kimmo’s assessment of the growing use of video content.
Video content is being increasingly used to deliver three main types of content: marketing, learning, and instructional. It is accessible and shareable, and importantly, production costs are coming down significantly.
One of the growth areas that I see emerging is in localization. Traditionally, we have thought of localization as multilingual subtitling, or audio localization, i.e. voiceovers. With digital video content, we can more and more easily localize the visuals, too. It will be easy to localize instructional videos in the future, for example. Or, a marketing content video will feature the right local product for the right market simply by using digital substitution done with a laptop, instead of an expensive re-shoot.
How will technology and languages develop in the coming years?
Few of us can predict how things will go, though technologies are constantly developing in every sector. Technology may reduce the need for language training, though at the same time the tools needed in remote interpreting must be made as user-friendly and effective as possible. When a need is catered for, another need will emerge.
Updated