Translators, the future and the role of training
Interview with Deborah Fry on 15 July 2024
Deborah Fry is a specialist German-to-English translator and language consultant based in Wiesbaden, Germany.
“To be in control of our own destinies, we’d better be – and stay – good.” (Deborah Fry)
Edith Vanghelof: Do you sometimes wonder whether what we do as translation providers still has a future? The widespread use of machine translation by corporate customers is creating a situation in which some translators feel that they must choose between providing post editing for machine translation or being pushed out of the market. This is driving translators who have built a reputation by delivering quality to switch completely to editing and revision work. Are you seeing this, too? And do you think this is a viable outlook for translators for future work?
Deborah Fry: Can I zoom out a bit here to start with? I think we shouldn't forget just how many different translation submarkets there are. This makes generalizations difficult, but it also gives providers a lot of niches, and we shouldn't forget that, or develop tunnel vision. I'd also say that what we, as translators, do or can do going forward depends on understanding who our customers are, what we can offer them, and the point in the project life cycle at which we get involved.
So if you're working with direct clients when the source texts are still being produced, and are very possibly going through multiple iterations, you can offer translation plus terminology, editing, proofreading and consulting, for example. If on the other hand you’re in the mass market – if you work for big agencies or corporate customers producing raw MT output of finished texts – often the only service they actually need will be post-editing. This can indeed be an option, though I’d be a bit cautious as this work may change and decline in future as the machines get better. Also, we must remember that only some of the texts produced in this way will actually need post editing anyway.
Of course, editing and revising text produced by humans is another option for translators, and as we all know, the ability to understand the source text where one actually exists or to understand what a non-native author is actually trying to say can be a real advantage. However, editing and revising well requires stylistic skills and in-depth subject area knowledge that goes well beyond what is taught at university. Translators therefore often need more training to be able to deliver in these areas.
Edith Vanghelof: Let me pick up on one of the points you mentioned: Some corporate customers are also using internal systems to prepare annual reports, for example. And in some cases they make machine translation part of their internal processes without considering the know-how and skills necessary to actually produce good quality translations. What role can translation sellers, regardless of their business model, play in this process? Do they need to rethink their business models and switch to more of an advisory role?
Deborah Fry: Personally, I don't have much experience of such a negative trend but then, as I said, the market is highly diverse and I’m sure there are cases of poor implementation. But let me be a bit provocative at this point. We must get used to the fact that companies are using neural machine translation and other AI tools internally. This is the new normal and it’s not going to go away, and it means that the role and use of external suppliers is changing as a result. Even allowing for the hype – which definitely does exist – machine translation offers real advantages in some areas. So the first thing that’s needed is for everyone concerned to know when and under what conditions it makes sense to use it, when it doesn't, and when it's downright dangerous to do so. After this, the question is how we as external service providers can add value in all these cases. If we can't demonstrate this, we have a problem: Why should anyone use us? We need to sell something that people need to buy.
I think you’re right that translators can play an important advisory role here. But again, this means knowing our stuff. It also means talking to everyone involved at the client in the language that they're used to hearing, and this includes management and IT departments, who often make the decisions. This isn't something we're always good at, even though building bridges at one level is our core business. So we need to work out how to transpose our skills here.
Another key point is that some translators like myself don't have a background in computational linguistics, which means that some more technical consulting roles are too much of a stretch without further training. But we still have real value to add when it comes to understanding and communicating the practical aspects of the new technology, and its advantages and limitations. At the same time, we need to understand our clients' business and the transformations that they are going through. It’s not business as normal for them at the moment, either.
Edith Vanghelof: That opens up some new perspectives, especially about consulting and advisory roles in the context of translation and technology, but let's go on. Specialized knowledge of specific domains is still relevant to post-editors. What would you recommend they focus on when looking to improve their skills?
Deborah Fry: This is very important. Domain knowledge has always been vital to quality translation, but the point is now that the bar is getting higher. Not only are the amount of knowledge and the pace of change growing rapidly across all subject areas, but AI will change the equation further in future. More information will be in, and produced by, the machine and we will have to know with confidence if we can accept it or whether it needs changing, and how. This is the new baseline for us when deciding if we can add value.
How can we do this, and how do we decide where to specialize? We need to analyze our individual skill sets, understand what exactly it is that we love doing – because this is important – and what the market needs, and then make the connections between these. And then we also need to plug any gaps and invest in training. Everyone will have their own individual answers here, and that's a good thing. There are some sectors that stand out in today's new world as being in strong demand: technology, environmental protection, legal and financial, or healthcare for example. And often in practice a mix of areas is needed. But equally there are some niche specialists who can also do very well indeed if the market is there, and these people are often very passionate and knowledgeable about their areas of interest. That can be a very happy mix. And in all these areas very strong editing and stylistic skills will be needed.
At the same time, to repeat myself, we need to look outside the box. I have been banging on for years about what I call the spider's web of different skills that a good translator needs, and source and target language expertise are only one part of this. As we’ve said, you need domain knowledge in both languages – because these may be different – plus general management and project management skills and tech smarts. And that's just the start. This basic thesis of mine is still valid, and it’s good risk management. None of us knows what precisely the future will bring, so having both breadth and depth, is a real advantage. Finally, the other thing we need to bear in mind every day is that we are service providers, and that we need to focus on what our customers want and need.
All this means that translation isn't actually a low-maintenance career option. But then there are very few jobs that really are, especially not today. We have to keep moving all the time. That's actually one of the great things about this profession. You can't avoid learning something new any every day, anyway.
Edith Vanghelof: So leading on from this, what is the point of attending specialist training and seminars for translators, editors, and others?
Deborah Fry: Let's pick up on what we already talked about: Delivering a superior service means keeping up with developments and investing in learning to stay there. All businesses need to do this, and the laws of nature aren't suspended for translators. And greater innovation everywhere means we have to learn about more and more, and more complicated stuff, and to do this faster and faster. The result is a need for high level, life-long learning. What is more, the fact that initial translator training will be focusing more and more on project management and technology in future actually offers specialist translators an added opportunity here, since we can potentially add value at the content and stylistic level. Conversely, though, it's also a threat if we fail to do so. Finally, we should remember that expanding our core business into, say, editing actually means that we're moving into somebody else's core business and there are already trained editors out there. In other words, we have to be qualified to their level.
What role does formalized training offer here? You can find vast amounts of useful information on the Internet, and we all do every day. However, you have to go out and look for it, assess it correctly, and adapt it to your context. Specialist seminars shorten the learning curve here. Somebody has already gone to the trouble of putting together precisely what we need, and at a high level. Being unable to travel is often no excuse today either, since hybrid seminars make courses accessible wherever you are. Such seminars also improve outcomes in other ways: They're interactive rather than passive, so you can ask questions and get more out of them. You also learn from the entire group and not just the trainer. And you understand that you're not alone. So they're an objective necessity, but they're also often a subjective pleasure as well.
Edith Vanghelof: Yes, I agree. Let’s get back to MT again for a moment. Would you agree that quality is losing ground to speed and cost due to the widespread use of machine translation? For example, within some organizations non-translators are now being tasked with editing MT output.
Deborah Fry: Remember the old joke about how many translators it takes to change light bulb? The answer is that it depends on the context. That's true here, in the sense that it depends on the submarket you're in. The nearer you are to the mass market, which is driven by volume and cost and time and standardized quality, the more likely this is to apply. But it's up to you to decide what market you're in.
And let me also add a word of caution about looking down on non-translators. Levels of competence in my native language, English, have risen immensely over my lifetime – that’s globalization for you. And the more specialized a text is, the more important domain knowledge becomes, so the more finely balanced the value proposition for using domain experts versus translators is. If we can't understand a text or its implications, we can't translate or edit it properly, and the consequences of introducing mistakes may be severe. So once again, it depends on the context and who the non-translator is. We can't afford to assume that we offer a superior service simply because we have a translation qualification. This may or may not be true in practice, and we have to make sure we can walk the talk as well.
Edith Vanghelof: You mentioned consequences. Do you think that we will see an increased awareness among customers of the reputational risk of poor communication in foreign languages? In my opinion, this is one of the risks of using a lot of machine translation.
Deborah Fry: Well, I predicted that this would happen some years ago, and I think we're already seeing it in some areas. Take key press releases, for example. The rise of social media, the politicization of many areas of business activity and the global reach of publication mean that source texts that are often meticulously crafted and checked increasingly need the same treatment in the target language, too. This is another example of market differentiation. There is more precise of analysis of the purposes and consequences of, and the target groups for, specific communications, and hence of their translation needs.
I expect this differentiation process to continue. The good news is that this will offer opportunities for translators who can meet these quality requirements, but those who cannot may have problems. We need to understand that human-generated and human-assisted translation are now part of a larger translation market and that the boundaries between the different segments are particularly fluid at the moment. Translators need to add value compared to the machine, and I think they can add value in areas like this. And at the same time, clients must be aware of this value and must be prepared to pay for it where they need it.
Edith Vanghelof: Yes, I agree with you in all of these points. One final question: There are studies being conducted on the way that post editing machine translation output is changing the language used by translators. One finding is a standardization of language and loss of creativity in language. What do you think?
Deborah Fry: I'm not an expert in this area, though I have seen some research on it. And again, I think it depends on the context. I also think we can be laid back about this one. To start with, don’t forget that in a lot of business contexts ‒ I'm not talking about literary translation here ‒ creativity can be downright dangerous. Technical and legal translators are required to stick to defined terms for a very good reason, for example. And with rough for-your-information translations or texts that are only needed as raw input for something else the benefit of speed often outweighs any stylistic or other restrictions. In contrast, other sorts of texts live from their use of synonyms and stylistic excellence. In other words, here, too, as always, it all depends.
More generally, too, we shouldn't forget that translation memories already aid standardization. And I'm old enough to remember the howls of pain at the time from “creatives” who thought this technology would cramp their style. In fact, though, standardized quality turned out to be part of the added value of these tools: they not only increased productivity but also helped reduce unwanted variations in the target textures against the source. So, if you do it right, you end up with a repository of high-quality, reusable texts. At present, neural machine translation often reintroduces inconsistencies compared to TMs, although over time we shall very probably see improvements here, with TMs playing a role. So no, I don't think technology necessarily kills creativity. It depends on how you use it.
I sometimes also wonder whether this fear isn't actually really a worry about declining agency, by which I mean not translation agencies, but the freedom that we have as translators to do our own thing, Problems with technology may sometimes be inherent in the technology itself, but they are also often about the processes used to implement it. This leads on to an important point that I can't emphasize too much: innovations change our own and our customers’ workflows as well as the technology we use. And we must be aware of this.
We also need to take care that creativity isn't used as an argument by translators who simply don't want to change. Change is inevitable, and my job is certainly radically different to what it was when I started out. Despite its bad rap at the moment, progress still does have its advantages and going backwards isn’t an option in practice. So for me, the bottom line is that, whatever tools they use, “content-heavy” translators should aim to be in control of their own destinies. And that means they had better be, and stay, good. And that also means training, and attending seminars.
Edith Vanghelof: Thank you for the interview.
About Deborah Fry, see Deborah Fry and www.frytranslations.com
About Edith Vanghelof: Camels - Team