Factory
Will 2018 change the way we are working? Our colleagues had a think about workflow-related issues and they are certain some things will take off and some will fade.
Will post-editing start its decline in the year to come? Is the role of translators safe in the future or will it transform in the changing environment? Is there a real danger that Machine Translation and our ages-old reflexes, alongside with our far-to-comfortable workflows turn against us – just to transform translation work into a sweatshop job?
Is project management about to be reformed entirely? Can we finally sort out the huge question of quality standards? Will the agile revolution that we can see across multiple industries alter the landscape of the translation business too?
These are the questions we are looking at in this section.
Lights-out project management
Solution Engineer at memoQ
As the language market is under a lot of pressure to produce faster results and to keep prices in check, Jure Dernovšek, Solution Engineer, believes 2018 is probably the time for a new buzz word to gain some ground.
Lights-out project management (LOPM) is an automated project management workflow where human intervention is not necessary.
In the translation industry, a typical life-cycle would be that a requester requests a translation through a web platform, portal or a similar shared access area. There the content would be automatically picked up by the processing system, imported into the translation environment, prepared for translation, automatically assigned to linguists and eventually delivered to the requester. LOPM also means that the linguists would be invoiced automatically.
Unfortunately due to the rapidly changing text-production systems and environments, it’s very difficult to create a universal translation environment that would allow a fully operational LOPM. Nevertheless, it can be a trend in the upcoming year, since the demand for translation is increasing and the pressure on prices, especially for projects that are highly automated, is getting greater and greater.
Post-editing RIP
Kilgray Owner and Board Member
Post-editing has been a familiar activity for decades in the industry. Peter Reynolds, Kilgray Owner and Board Member, believes the time may have come for the post-editing model to retire.
When Machine Translation (MT) started becoming more widely used, many predicted that it would use a model which saw the entire text being translated and then this would be post-edited to correct and improve the translated text. Since then many translators have used MT systems within CAT tools where they see the MT translation of one segment at a time.
We have also seen new MT tools such as Lilt which expects a translator to be using MT to translate one segment at a time. 2017 has seen the widespread use of neural machine translation systems. One issue with these systems, which poses a difficulty for the post-editing model, is the fact the translated text may look correct but could be wrong. Unless the translator sees the source text they will not know this.
This would suggest that the post-editing model will be used less and less.
Quantifiable end-user values in QA
Kilgray founder and Head of Innovation at memoQ
Quality has been a somewhat arbitrary term for the industry. Not because market actors are disinterested in upholding sound procedures but mainly because of a lack of common understanding. Gábor Ugray, Kilgray founder and Head of Innovation, believes this era is now coming to an end.
The days when everyone differentiates on “quality” without a consensus on the meaning of quality are coming to an end. End buyers will begin to measure the efficiency of localized messages the same way source-language content is measured in digital marketing. We will see the first reports of localization A/B tests before 2018 is out.
Guest commentary
It all comes down to attitude
Vice President at Asling - Geneva & Independent Translation tools consultant
It was in 1995, after working as a freelance medical/technical translator and reviewer for 20 years from Italian to English, that Juliet Macan, Vice President at Asling in Geneva, and an independent translation tools consultant encountered a computer assisted translation tool for the first time. This initial experience must have gone very well, as she now has in-depth knowledge of a long and impressive range of various CAT tools. She focussed on testing, trouble-shooting and using translation environment tools on networks and helping customers, project managers and translators. She naturally jumped at the QA tools when they appeared, and she fell for memoQ when it was launched. Juliet now looks up-stream a lot, trying to help technical writers understand the importance of clarity and consistency. “There are always plenty of problems to be solved, and you never stop learning, and finding new and better ways of doing things is very stimulating,” says Juliet.
Despite the progress and all the special software that have been developed over the last fifteen years in the field, quality assurance is still a problem. In the last couple of years, I have been working on a lot of quality issues (for translation agencies and translators but also customers) – and it is amazing how many people are still not taking quality assurance seriously. Unfortunately, this may also be due to the quality of original source materials – people do not always realize just how important the quality of the source text is – except for the translators who are trying to understand it.
Despite the investments in technology, standardizing QA is difficult for a host of reasons that are often not just linguistic, or technological. Then there is the human factor: authors and translators should get “feedback”, and post-editors and reviewers should be recognized for their expertise and, even remunerated better – if only to make them feel it is worthwhile.
Maybe other industries, where clearer processes exist for A/B testing, may drive the change. All in all, I think we need more control of source material to ensure accuracy and consistency; technical writers should be more aware of the importance of terminology. We need to show clients the benefits of quality assurance of their communications with the world.
Although some very good experts, including people at Taus, have been working on this for some time now, currently I see no industry-wide consensus over the issues of quality. Communication, and translation, are hard to pin down, there are so many details that are very difficult to standardize. I think it eventually all comes down to attitude: first we must work on changing it for the better.
The well-trained translator
Content Manager at memoQ
What will the future bring for the individual translator when the present is already full of riddles? Szabolcs Kincse, Content Manager at memoQ, believes recent developments in the industry will have their winners as well as their losers.
A lot has been happening in the past half-decade in the language market and technology. The appearance of a host of good-quality Machine Translation (MT) applications, the possible decline of post-editing, and the expected rise of time-based calculation all point towards the changing role of the translator.
Translating simple strings will be devalued and the “open this” and “click-here”-types of simple and monotonous string translation tasks will be surely taken over by MT. Those translators will be able to ride the waves in the future who can do more than that.
Translators on the other hand, who cannot be trusted with proofreading and editing, will be out of business – just think about all the challenges MT hits represent: perfectly formulated sentences that seem ok but just do not convey the original meaning, or are not in touch with the expected style. If a translator is not able to catch these nuances, it is difficult to imagine how they will be able to cope in the future.
In 2018, therefore, I expect to see the rise of the translator who is always keen on learning new skills.
Agile translation
Q&A Consultant at memoQ
The entire world is going agile. What is this madness? It is something the translation industry will pick up on – or so believes Gusztáv Jánvári, Q&A Consultant at memoQ.
You turn on your radio (or your smart device) and the term agile pops up from everywhere. While you enable your device to update your radio app the second time this month, you may ask yourself what agile really means. Exactly this. In a nutshell, agile, from the consumer point of view, means a product or a service receives new and improved features in small, yet rather frequent iterations.
Agile development is distributed into 2-3 week long sprints, and sprints deliver new versions, complete with the new translated or localized content. To manage this speed and ensure resource efficiency, players need highly automated tools and advanced practices (like kanban). In exchange, work is divided into small installments – usually lasting 1 or 2 days.
The key concepts and components for LSPs to go agile:
- Automation. Using project templates, a new fully configured project can be launched by a couple of key strokes. Automated actions can replace sleepy PMs and move the tasks forward by, for example, assigning them to free resources automatically. PMs (even the sleepy ones) can focus on the bigger picture of progress visualized in dashboards.
- Integration. Getting stuff out from a source system (manually), converting it to another format (manually) and pushing it into a CAT tool (manually), and, at the end, performing the entire process the other way around, again, manually – this is history now. It takes much more time than performing the TEP procedure altogether. Integrated systems can do all this for you. Couple with automation, and you have a super weapon!
- Quick access to information. To meet tight deadlines and to keep your good human resources, ensure access to relevant information, such as expected terminology and context information without the need to switch over to different applications, or you will lose money instead of earning.
- Standardized QA to ensure quality all along the workflow. Quality is something not to be messed with.
Guest commentary
Agility and consolidation
Founder and CEO of onTheGoSystems and WPML
Amir Helzer, founder and CEO of onTheGoSystems and WPML, also believes that translation business is also going agile. This may have some curious implications to the landscape of the translation industry as well.
Technically, we can say that the market has been moving towards agile translation in the past few years. About five to seven years ago, technical translation would have required a lot of manual work: manual exporting, sending, processing and bookkeeping would have been commonplace for delivering various translation projects. I think, however, most of this is already the past. What it means in practice is that people who manage multilingual sites do not need to think about internal processes; they just translate things on demand, as they need. There is a very little set-up: just click a button, send and receive – it is a lot less scary process than it used to be. Client input is dramatically shortening as customers do not have to think about what processes are taking place in the background. This certainly all points towards a more agile operation in the industry.
I am sure that most LSPs are taking pride in their internal differentiation. LSPs we talk to do not want to see themselves becoming a commodity – so they are probably not interested in providing fine-grained and interchangeable services. While they are using the same freelancers and the same technological solutions all over the world, the main differentiator among the various service providers is their internal work and their internal processes. Some translation services are trying to differentiate themselves in terms of speed, some in terms of pricing, some by fields of expertise and some in terms of scale (as scale will make them look more reliable with large clients). However, technically they are still very similar.
We can already see the signs of the four areas highlighted in Gusztáv’s suggestion gaining ground rapidly. For example, clients who receive translations from any vendor in WordPress via WPML can access a feedback form which appears on the site’s front end. This allows visitors to mark pages where they encounter problematic translations. This feedback goes to the site admin who can approve it, and then it will go all the way back to the translation provider, packed together with the information the visitor provided – the translation service can then alter and modify the objected translation. It is very useful for the site administrator to know that there is an express way to the service provider if something goes wrong. From the outside it is a very simple action, but in fact it comprises the technical integration, the human interface, and the entire workflow to make things happen. It is a complex multistage process but implemented as a simple text input on the website.
Essentially, just like in any other industry, I expect there will be some kind of consolidation. Just a short while ago, translation services with physical offices at many locations had a competitive advantage: having a branch in a city would have provided an advantage for an LSP over other companies who were not present at that specific location. Now that everything is becoming web-based, this advantage is diminishing, similarly to many other industries. Since the scope is now on web-based operation, I expect some companies will disappear, and some will go on. It is not a very large market and margins are relatively low. Not everyone can survive this kind of business atmosphere I think, but this is what happens to literally every market: as soon as the processes become web-based, the advantage will be with those who can enjoy the perks of scalability. It is difficult for the small players to stay above the water in this situation and MT is getting better and better, which makes it even more difficult for the small ones to survive.